Git Inbox Mirror of the ffmpeg-devel mailing list - see https://ffmpeg.org/mailman/listinfo/ffmpeg-devel
 help / color / mirror / Atom feed
* [FFmpeg-devel] [PATCH v9 01/13] avcodec/vaapi_encode: introduce a base layer for vaapi encode
@ 2024-05-20 14:52 tong1.wu-at-intel.com
  2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 02/13] avcodec/vaapi_encode: add async_depth to common options tong1.wu-at-intel.com
                   ` (12 more replies)
  0 siblings, 13 replies; 18+ messages in thread
From: tong1.wu-at-intel.com @ 2024-05-20 14:52 UTC (permalink / raw)
  To: ffmpeg-devel; +Cc: Tong Wu

From: Tong Wu <tong1.wu@intel.com>

Since VAAPI and future D3D12VA implementation may share some common parameters,
a base layer encode context is introduced as vaapi context's base.

Signed-off-by: Tong Wu <tong1.wu@intel.com>
---
 libavcodec/hw_base_encode.h | 56 +++++++++++++++++++++++++++++++++++++
 libavcodec/vaapi_encode.h   | 39 +++++---------------------
 2 files changed, 63 insertions(+), 32 deletions(-)
 create mode 100644 libavcodec/hw_base_encode.h

diff --git a/libavcodec/hw_base_encode.h b/libavcodec/hw_base_encode.h
new file mode 100644
index 0000000000..1996179456
--- /dev/null
+++ b/libavcodec/hw_base_encode.h
@@ -0,0 +1,56 @@
+/*
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#ifndef AVCODEC_HW_BASE_ENCODE_H
+#define AVCODEC_HW_BASE_ENCODE_H
+
+#define MAX_DPB_SIZE 16
+#define MAX_PICTURE_REFERENCES 2
+#define MAX_REORDER_DELAY 16
+#define MAX_ASYNC_DEPTH 64
+#define MAX_REFERENCE_LIST_NUM 2
+
+enum {
+    PICTURE_TYPE_IDR = 0,
+    PICTURE_TYPE_I   = 1,
+    PICTURE_TYPE_P   = 2,
+    PICTURE_TYPE_B   = 3,
+};
+
+enum {
+    // Codec supports controlling the subdivision of pictures into slices.
+    FLAG_SLICE_CONTROL         = 1 << 0,
+    // Codec only supports constant quality (no rate control).
+    FLAG_CONSTANT_QUALITY_ONLY = 1 << 1,
+    // Codec is intra-only.
+    FLAG_INTRA_ONLY            = 1 << 2,
+    // Codec supports B-pictures.
+    FLAG_B_PICTURES            = 1 << 3,
+    // Codec supports referencing B-pictures.
+    FLAG_B_PICTURE_REFERENCES  = 1 << 4,
+    // Codec supports non-IDR key pictures (that is, key pictures do
+    // not necessarily empty the DPB).
+    FLAG_NON_IDR_KEY_PICTURES  = 1 << 5,
+};
+
+typedef struct HWBaseEncodeContext {
+    const AVClass *class;
+} HWBaseEncodeContext;
+
+#endif /* AVCODEC_HW_BASE_ENCODE_H */
+
diff --git a/libavcodec/vaapi_encode.h b/libavcodec/vaapi_encode.h
index 0eed9691ca..f5c9be8973 100644
--- a/libavcodec/vaapi_encode.h
+++ b/libavcodec/vaapi_encode.h
@@ -33,34 +33,27 @@
 
 #include "avcodec.h"
 #include "hwconfig.h"
+#include "hw_base_encode.h"
 
 struct VAAPIEncodeType;
 struct VAAPIEncodePicture;
 
+// Codec output packet without timestamp delay, which means the
+// output packet has same PTS and DTS.
+#define FLAG_TIMESTAMP_NO_DELAY 1 << 6
+
 enum {
     MAX_CONFIG_ATTRIBUTES  = 4,
     MAX_GLOBAL_PARAMS      = 4,
-    MAX_DPB_SIZE           = 16,
-    MAX_PICTURE_REFERENCES = 2,
-    MAX_REORDER_DELAY      = 16,
     MAX_PARAM_BUFFER_SIZE  = 1024,
     // A.4.1: table A.6 allows at most 22 tile rows for any level.
     MAX_TILE_ROWS          = 22,
     // A.4.1: table A.6 allows at most 20 tile columns for any level.
     MAX_TILE_COLS          = 20,
-    MAX_ASYNC_DEPTH        = 64,
-    MAX_REFERENCE_LIST_NUM = 2,
 };
 
 extern const AVCodecHWConfigInternal *const ff_vaapi_encode_hw_configs[];
 
-enum {
-    PICTURE_TYPE_IDR = 0,
-    PICTURE_TYPE_I   = 1,
-    PICTURE_TYPE_P   = 2,
-    PICTURE_TYPE_B   = 3,
-};
-
 typedef struct VAAPIEncodeSlice {
     int             index;
     int             row_start;
@@ -193,7 +186,8 @@ typedef struct VAAPIEncodeRCMode {
 } VAAPIEncodeRCMode;
 
 typedef struct VAAPIEncodeContext {
-    const AVClass *class;
+    // Base context.
+    HWBaseEncodeContext base;
 
     // Codec-specific hooks.
     const struct VAAPIEncodeType *codec;
@@ -397,25 +391,6 @@ typedef struct VAAPIEncodeContext {
     AVPacket        *tail_pkt;
 } VAAPIEncodeContext;
 
-enum {
-    // Codec supports controlling the subdivision of pictures into slices.
-    FLAG_SLICE_CONTROL         = 1 << 0,
-    // Codec only supports constant quality (no rate control).
-    FLAG_CONSTANT_QUALITY_ONLY = 1 << 1,
-    // Codec is intra-only.
-    FLAG_INTRA_ONLY            = 1 << 2,
-    // Codec supports B-pictures.
-    FLAG_B_PICTURES            = 1 << 3,
-    // Codec supports referencing B-pictures.
-    FLAG_B_PICTURE_REFERENCES  = 1 << 4,
-    // Codec supports non-IDR key pictures (that is, key pictures do
-    // not necessarily empty the DPB).
-    FLAG_NON_IDR_KEY_PICTURES  = 1 << 5,
-    // Codec output packet without timestamp delay, which means the
-    // output packet has same PTS and DTS.
-    FLAG_TIMESTAMP_NO_DELAY    = 1 << 6,
-};
-
 typedef struct VAAPIEncodeType {
     // List of supported profiles and corresponding VAAPI profiles.
     // (Must end with AV_PROFILE_UNKNOWN.)
-- 
2.41.0.windows.1

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

^ permalink raw reply	[flat|nested] 18+ messages in thread

* [FFmpeg-devel] [PATCH v9 02/13] avcodec/vaapi_encode: add async_depth to common options
  2024-05-20 14:52 [FFmpeg-devel] [PATCH v9 01/13] avcodec/vaapi_encode: introduce a base layer for vaapi encode tong1.wu-at-intel.com
@ 2024-05-20 14:52 ` tong1.wu-at-intel.com
  2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 03/13] avcodec/vaapi_encode: add picture type name to base tong1.wu-at-intel.com
                   ` (11 subsequent siblings)
  12 siblings, 0 replies; 18+ messages in thread
From: tong1.wu-at-intel.com @ 2024-05-20 14:52 UTC (permalink / raw)
  To: ffmpeg-devel; +Cc: Tong Wu

From: Tong Wu <tong1.wu@intel.com>

Signed-off-by: Tong Wu <tong1.wu@intel.com>
---
 libavcodec/hw_base_encode.h     | 10 +++++++++-
 libavcodec/vaapi_encode.c       | 13 ++++++++-----
 libavcodec/vaapi_encode.h       |  7 -------
 libavcodec/vaapi_encode_av1.c   |  1 +
 libavcodec/vaapi_encode_h264.c  |  1 +
 libavcodec/vaapi_encode_h265.c  |  1 +
 libavcodec/vaapi_encode_mjpeg.c |  1 +
 libavcodec/vaapi_encode_mpeg2.c |  1 +
 libavcodec/vaapi_encode_vp8.c   |  1 +
 libavcodec/vaapi_encode_vp9.c   |  1 +
 10 files changed, 24 insertions(+), 13 deletions(-)

diff --git a/libavcodec/hw_base_encode.h b/libavcodec/hw_base_encode.h
index 1996179456..5272f2836d 100644
--- a/libavcodec/hw_base_encode.h
+++ b/libavcodec/hw_base_encode.h
@@ -50,7 +50,15 @@ enum {
 
 typedef struct HWBaseEncodeContext {
     const AVClass *class;
+
+    // Max number of frame buffered in encoder.
+    int             async_depth;
 } HWBaseEncodeContext;
 
-#endif /* AVCODEC_HW_BASE_ENCODE_H */
+#define HW_BASE_ENCODE_COMMON_OPTIONS \
+    { "async_depth", "Maximum processing parallelism. " \
+      "Increase this to improve single channel performance.", \
+      OFFSET(common.base.async_depth), AV_OPT_TYPE_INT, \
+      { .i64 = 2 }, 1, MAX_ASYNC_DEPTH, FLAGS }
 
+#endif /* AVCODEC_HW_BASE_ENCODE_H */
diff --git a/libavcodec/vaapi_encode.c b/libavcodec/vaapi_encode.c
index f54b2579ec..9373512417 100644
--- a/libavcodec/vaapi_encode.c
+++ b/libavcodec/vaapi_encode.c
@@ -669,7 +669,8 @@ static int vaapi_encode_set_output_property(AVCodecContext *avctx,
                                             VAAPIEncodePicture *pic,
                                             AVPacket *pkt)
 {
-    VAAPIEncodeContext *ctx = avctx->priv_data;
+    HWBaseEncodeContext *base_ctx = avctx->priv_data;
+    VAAPIEncodeContext       *ctx = avctx->priv_data;
 
     if (pic->type == PICTURE_TYPE_IDR)
         pkt->flags |= AV_PKT_FLAG_KEY;
@@ -699,7 +700,7 @@ static int vaapi_encode_set_output_property(AVCodecContext *avctx,
             pkt->dts = ctx->ts_ring[pic->encode_order] - ctx->dts_pts_diff;
     } else {
         pkt->dts = ctx->ts_ring[(pic->encode_order - ctx->decode_delay) %
-                                (3 * ctx->output_delay + ctx->async_depth)];
+                                (3 * ctx->output_delay + base_ctx->async_depth)];
     }
 
     return 0;
@@ -1320,6 +1321,7 @@ static int vaapi_encode_check_frame(AVCodecContext *avctx,
 
 static int vaapi_encode_send_frame(AVCodecContext *avctx, AVFrame *frame)
 {
+    HWBaseEncodeContext *base_ctx = avctx->priv_data;
     VAAPIEncodeContext *ctx = avctx->priv_data;
     VAAPIEncodePicture *pic;
     int err;
@@ -1365,7 +1367,7 @@ static int vaapi_encode_send_frame(AVCodecContext *avctx, AVFrame *frame)
             ctx->dts_pts_diff = pic->pts - ctx->first_pts;
         if (ctx->output_delay > 0)
             ctx->ts_ring[ctx->input_order %
-                        (3 * ctx->output_delay + ctx->async_depth)] = pic->pts;
+                        (3 * ctx->output_delay + base_ctx->async_depth)] = pic->pts;
 
         pic->display_order = ctx->input_order;
         ++ctx->input_order;
@@ -2773,7 +2775,8 @@ static av_cold int vaapi_encode_create_recon_frames(AVCodecContext *avctx)
 
 av_cold int ff_vaapi_encode_init(AVCodecContext *avctx)
 {
-    VAAPIEncodeContext *ctx = avctx->priv_data;
+    HWBaseEncodeContext *base_ctx = avctx->priv_data;
+    VAAPIEncodeContext       *ctx = avctx->priv_data;
     AVVAAPIFramesContext *recon_hwctx = NULL;
     VAStatus vas;
     int err;
@@ -2966,7 +2969,7 @@ av_cold int ff_vaapi_encode_init(AVCodecContext *avctx)
     vas = vaSyncBuffer(ctx->hwctx->display, VA_INVALID_ID, 0);
     if (vas != VA_STATUS_ERROR_UNIMPLEMENTED) {
         ctx->has_sync_buffer_func = 1;
-        ctx->encode_fifo = av_fifo_alloc2(ctx->async_depth,
+        ctx->encode_fifo = av_fifo_alloc2(base_ctx->async_depth,
                                           sizeof(VAAPIEncodePicture *),
                                           0);
         if (!ctx->encode_fifo)
diff --git a/libavcodec/vaapi_encode.h b/libavcodec/vaapi_encode.h
index f5c9be8973..02410c72ec 100644
--- a/libavcodec/vaapi_encode.h
+++ b/libavcodec/vaapi_encode.h
@@ -374,8 +374,6 @@ typedef struct VAAPIEncodeContext {
     int             has_sync_buffer_func;
     // Store buffered pic
     AVFifo          *encode_fifo;
-    // Max number of frame buffered in encoder.
-    int             async_depth;
 
     /** Head data for current output pkt, used only for AV1. */
     //void  *header_data;
@@ -491,11 +489,6 @@ int ff_vaapi_encode_close(AVCodecContext *avctx);
       "Maximum B-frame reference depth", \
       OFFSET(common.desired_b_depth), AV_OPT_TYPE_INT, \
       { .i64 = 1 }, 1, INT_MAX, FLAGS }, \
-    { "async_depth", "Maximum processing parallelism. " \
-      "Increase this to improve single channel performance. This option " \
-      "doesn't work if driver doesn't implement vaSyncBuffer function.", \
-      OFFSET(common.async_depth), AV_OPT_TYPE_INT, \
-      { .i64 = 2 }, 1, MAX_ASYNC_DEPTH, FLAGS }, \
     { "max_frame_size", \
       "Maximum frame size (in bytes)",\
       OFFSET(common.max_frame_size), AV_OPT_TYPE_INT, \
diff --git a/libavcodec/vaapi_encode_av1.c b/libavcodec/vaapi_encode_av1.c
index b868f5b66a..a5a475dca6 100644
--- a/libavcodec/vaapi_encode_av1.c
+++ b/libavcodec/vaapi_encode_av1.c
@@ -965,6 +965,7 @@ static av_cold int vaapi_encode_av1_close(AVCodecContext *avctx)
 #define FLAGS (AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM)
 
 static const AVOption vaapi_encode_av1_options[] = {
+    HW_BASE_ENCODE_COMMON_OPTIONS,
     VAAPI_ENCODE_COMMON_OPTIONS,
     VAAPI_ENCODE_RC_OPTIONS,
     { "profile", "Set profile (seq_profile)",
diff --git a/libavcodec/vaapi_encode_h264.c b/libavcodec/vaapi_encode_h264.c
index d656b1020f..ebb1760cd3 100644
--- a/libavcodec/vaapi_encode_h264.c
+++ b/libavcodec/vaapi_encode_h264.c
@@ -1276,6 +1276,7 @@ static av_cold int vaapi_encode_h264_close(AVCodecContext *avctx)
 #define OFFSET(x) offsetof(VAAPIEncodeH264Context, x)
 #define FLAGS (AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM)
 static const AVOption vaapi_encode_h264_options[] = {
+    HW_BASE_ENCODE_COMMON_OPTIONS,
     VAAPI_ENCODE_COMMON_OPTIONS,
     VAAPI_ENCODE_RC_OPTIONS,
 
diff --git a/libavcodec/vaapi_encode_h265.c b/libavcodec/vaapi_encode_h265.c
index 2f59161346..77bd5e31af 100644
--- a/libavcodec/vaapi_encode_h265.c
+++ b/libavcodec/vaapi_encode_h265.c
@@ -1394,6 +1394,7 @@ static av_cold int vaapi_encode_h265_close(AVCodecContext *avctx)
 #define OFFSET(x) offsetof(VAAPIEncodeH265Context, x)
 #define FLAGS (AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM)
 static const AVOption vaapi_encode_h265_options[] = {
+    HW_BASE_ENCODE_COMMON_OPTIONS,
     VAAPI_ENCODE_COMMON_OPTIONS,
     VAAPI_ENCODE_RC_OPTIONS,
 
diff --git a/libavcodec/vaapi_encode_mjpeg.c b/libavcodec/vaapi_encode_mjpeg.c
index c17747e3a9..fb5c0d34c6 100644
--- a/libavcodec/vaapi_encode_mjpeg.c
+++ b/libavcodec/vaapi_encode_mjpeg.c
@@ -540,6 +540,7 @@ static av_cold int vaapi_encode_mjpeg_close(AVCodecContext *avctx)
 #define OFFSET(x) offsetof(VAAPIEncodeMJPEGContext, x)
 #define FLAGS (AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM)
 static const AVOption vaapi_encode_mjpeg_options[] = {
+    HW_BASE_ENCODE_COMMON_OPTIONS,
     VAAPI_ENCODE_COMMON_OPTIONS,
 
     { "jfif", "Include JFIF header",
diff --git a/libavcodec/vaapi_encode_mpeg2.c b/libavcodec/vaapi_encode_mpeg2.c
index c9b16fbcfc..d0980c52b0 100644
--- a/libavcodec/vaapi_encode_mpeg2.c
+++ b/libavcodec/vaapi_encode_mpeg2.c
@@ -639,6 +639,7 @@ static av_cold int vaapi_encode_mpeg2_close(AVCodecContext *avctx)
 #define OFFSET(x) offsetof(VAAPIEncodeMPEG2Context, x)
 #define FLAGS (AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM)
 static const AVOption vaapi_encode_mpeg2_options[] = {
+    HW_BASE_ENCODE_COMMON_OPTIONS,
     VAAPI_ENCODE_COMMON_OPTIONS,
     VAAPI_ENCODE_RC_OPTIONS,
 
diff --git a/libavcodec/vaapi_encode_vp8.c b/libavcodec/vaapi_encode_vp8.c
index 8a557b967e..4e284f86e2 100644
--- a/libavcodec/vaapi_encode_vp8.c
+++ b/libavcodec/vaapi_encode_vp8.c
@@ -216,6 +216,7 @@ static av_cold int vaapi_encode_vp8_init(AVCodecContext *avctx)
 #define OFFSET(x) offsetof(VAAPIEncodeVP8Context, x)
 #define FLAGS (AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM)
 static const AVOption vaapi_encode_vp8_options[] = {
+    HW_BASE_ENCODE_COMMON_OPTIONS,
     VAAPI_ENCODE_COMMON_OPTIONS,
     VAAPI_ENCODE_RC_OPTIONS,
 
diff --git a/libavcodec/vaapi_encode_vp9.c b/libavcodec/vaapi_encode_vp9.c
index c2a8dec71b..88f951652c 100644
--- a/libavcodec/vaapi_encode_vp9.c
+++ b/libavcodec/vaapi_encode_vp9.c
@@ -273,6 +273,7 @@ static av_cold int vaapi_encode_vp9_init(AVCodecContext *avctx)
 #define OFFSET(x) offsetof(VAAPIEncodeVP9Context, x)
 #define FLAGS (AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM)
 static const AVOption vaapi_encode_vp9_options[] = {
+    HW_BASE_ENCODE_COMMON_OPTIONS,
     VAAPI_ENCODE_COMMON_OPTIONS,
     VAAPI_ENCODE_RC_OPTIONS,
 
-- 
2.41.0.windows.1

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

^ permalink raw reply	[flat|nested] 18+ messages in thread

* [FFmpeg-devel] [PATCH v9 03/13] avcodec/vaapi_encode: add picture type name to base
  2024-05-20 14:52 [FFmpeg-devel] [PATCH v9 01/13] avcodec/vaapi_encode: introduce a base layer for vaapi encode tong1.wu-at-intel.com
  2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 02/13] avcodec/vaapi_encode: add async_depth to common options tong1.wu-at-intel.com
@ 2024-05-20 14:52 ` tong1.wu-at-intel.com
  2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 04/13] avcodec/vaapi_encode: move pic->input_surface initialization to encode_alloc tong1.wu-at-intel.com
                   ` (10 subsequent siblings)
  12 siblings, 0 replies; 18+ messages in thread
From: tong1.wu-at-intel.com @ 2024-05-20 14:52 UTC (permalink / raw)
  To: ffmpeg-devel; +Cc: Tong Wu

From: Tong Wu <tong1.wu@intel.com>

Signed-off-by: Tong Wu <tong1.wu@intel.com>
---
 libavcodec/hw_base_encode.h | 5 +++++
 libavcodec/vaapi_encode.c   | 4 +---
 2 files changed, 6 insertions(+), 3 deletions(-)

diff --git a/libavcodec/hw_base_encode.h b/libavcodec/hw_base_encode.h
index 5272f2836d..a578db8c06 100644
--- a/libavcodec/hw_base_encode.h
+++ b/libavcodec/hw_base_encode.h
@@ -25,6 +25,11 @@
 #define MAX_ASYNC_DEPTH 64
 #define MAX_REFERENCE_LIST_NUM 2
 
+static inline const char *ff_hw_base_encode_get_pictype_name(const int type) {
+    const char * const picture_type_name[] = { "IDR", "I", "P", "B" };
+    return picture_type_name[type];
+}
+
 enum {
     PICTURE_TYPE_IDR = 0,
     PICTURE_TYPE_I   = 1,
diff --git a/libavcodec/vaapi_encode.c b/libavcodec/vaapi_encode.c
index 9373512417..2d22e4bd85 100644
--- a/libavcodec/vaapi_encode.c
+++ b/libavcodec/vaapi_encode.c
@@ -38,8 +38,6 @@ const AVCodecHWConfigInternal *const ff_vaapi_encode_hw_configs[] = {
     NULL,
 };
 
-static const char * const picture_type_name[] = { "IDR", "I", "P", "B" };
-
 static int vaapi_encode_make_packed_header(AVCodecContext *avctx,
                                            VAAPIEncodePicture *pic,
                                            int type, char *data, size_t bit_len)
@@ -277,7 +275,7 @@ static int vaapi_encode_issue(AVCodecContext *avctx,
 
     av_log(avctx, AV_LOG_DEBUG, "Issuing encode for pic %"PRId64"/%"PRId64" "
            "as type %s.\n", pic->display_order, pic->encode_order,
-           picture_type_name[pic->type]);
+           ff_hw_base_encode_get_pictype_name(pic->type));
     if (pic->nb_refs[0] == 0 && pic->nb_refs[1] == 0) {
         av_log(avctx, AV_LOG_DEBUG, "No reference pictures.\n");
     } else {
-- 
2.41.0.windows.1

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

^ permalink raw reply	[flat|nested] 18+ messages in thread

* [FFmpeg-devel] [PATCH v9 04/13] avcodec/vaapi_encode: move pic->input_surface initialization to encode_alloc
  2024-05-20 14:52 [FFmpeg-devel] [PATCH v9 01/13] avcodec/vaapi_encode: introduce a base layer for vaapi encode tong1.wu-at-intel.com
  2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 02/13] avcodec/vaapi_encode: add async_depth to common options tong1.wu-at-intel.com
  2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 03/13] avcodec/vaapi_encode: add picture type name to base tong1.wu-at-intel.com
@ 2024-05-20 14:52 ` tong1.wu-at-intel.com
  2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 05/13] avcodec/vaapi_encode: move the dpb logic from VAAPI to base layer tong1.wu-at-intel.com
                   ` (9 subsequent siblings)
  12 siblings, 0 replies; 18+ messages in thread
From: tong1.wu-at-intel.com @ 2024-05-20 14:52 UTC (permalink / raw)
  To: ffmpeg-devel; +Cc: Tong Wu

From: Tong Wu <tong1.wu@intel.com>

When allocating the VAAPIEncodePicture, pic->input_surface can be
initialized right in the place. This movement simplifies the send_frame
logic and is the preparation for moving vaapi_encode_send_frame to the base layer.

Signed-off-by: Tong Wu <tong1.wu@intel.com>
---
 libavcodec/vaapi_encode.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/libavcodec/vaapi_encode.c b/libavcodec/vaapi_encode.c
index 2d22e4bd85..227cccae64 100644
--- a/libavcodec/vaapi_encode.c
+++ b/libavcodec/vaapi_encode.c
@@ -878,7 +878,8 @@ static int vaapi_encode_discard(AVCodecContext *avctx,
     return 0;
 }
 
-static VAAPIEncodePicture *vaapi_encode_alloc(AVCodecContext *avctx)
+static VAAPIEncodePicture *vaapi_encode_alloc(AVCodecContext *avctx,
+                                              const AVFrame *frame)
 {
     VAAPIEncodeContext *ctx = avctx->priv_data;
     VAAPIEncodePicture *pic;
@@ -895,7 +896,7 @@ static VAAPIEncodePicture *vaapi_encode_alloc(AVCodecContext *avctx)
         }
     }
 
-    pic->input_surface = VA_INVALID_ID;
+    pic->input_surface = (VASurfaceID)(uintptr_t)frame->data[3];
     pic->recon_surface = VA_INVALID_ID;
     pic->output_buffer = VA_INVALID_ID;
 
@@ -1332,7 +1333,7 @@ static int vaapi_encode_send_frame(AVCodecContext *avctx, AVFrame *frame)
         if (err < 0)
             return err;
 
-        pic = vaapi_encode_alloc(avctx);
+        pic = vaapi_encode_alloc(avctx, frame);
         if (!pic)
             return AVERROR(ENOMEM);
 
@@ -1345,7 +1346,6 @@ static int vaapi_encode_send_frame(AVCodecContext *avctx, AVFrame *frame)
         if (ctx->input_order == 0 || frame->pict_type == AV_PICTURE_TYPE_I)
             pic->force_idr = 1;
 
-        pic->input_surface = (VASurfaceID)(uintptr_t)frame->data[3];
         pic->pts = frame->pts;
         pic->duration = frame->duration;
 
-- 
2.41.0.windows.1

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

^ permalink raw reply	[flat|nested] 18+ messages in thread

* [FFmpeg-devel] [PATCH v9 05/13] avcodec/vaapi_encode: move the dpb logic from VAAPI to base layer
  2024-05-20 14:52 [FFmpeg-devel] [PATCH v9 01/13] avcodec/vaapi_encode: introduce a base layer for vaapi encode tong1.wu-at-intel.com
                   ` (2 preceding siblings ...)
  2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 04/13] avcodec/vaapi_encode: move pic->input_surface initialization to encode_alloc tong1.wu-at-intel.com
@ 2024-05-20 14:52 ` tong1.wu-at-intel.com
  2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 06/13] avcodec/vaapi_encode: extract the init and close function " tong1.wu-at-intel.com
                   ` (8 subsequent siblings)
  12 siblings, 0 replies; 18+ messages in thread
From: tong1.wu-at-intel.com @ 2024-05-20 14:52 UTC (permalink / raw)
  To: ffmpeg-devel; +Cc: Tong Wu

From: Tong Wu <tong1.wu@intel.com>

Move receive_packet function to base. This requires adding *alloc,
*issue, *output, *free as hardware callbacks. HWBaseEncodePicture is
introduced as the base layer structure. The related parameters in
VAAPIEncodeContext are also extracted to HWBaseEncodeContext. Then DPB
management logic can be fully extracted to base layer as-is.

Signed-off-by: Tong Wu <tong1.wu@intel.com>
---
 libavcodec/Makefile             |   2 +-
 libavcodec/hw_base_encode.c     | 597 ++++++++++++++++++++++++
 libavcodec/hw_base_encode.h     | 124 +++++
 libavcodec/vaapi_encode.c       | 793 +++++---------------------------
 libavcodec/vaapi_encode.h       | 102 +---
 libavcodec/vaapi_encode_av1.c   |  35 +-
 libavcodec/vaapi_encode_h264.c  |  84 ++--
 libavcodec/vaapi_encode_h265.c  |  53 ++-
 libavcodec/vaapi_encode_mjpeg.c |  13 +-
 libavcodec/vaapi_encode_mpeg2.c |  33 +-
 libavcodec/vaapi_encode_vp8.c   |  18 +-
 libavcodec/vaapi_encode_vp9.c   |  24 +-
 12 files changed, 988 insertions(+), 890 deletions(-)
 create mode 100644 libavcodec/hw_base_encode.c

diff --git a/libavcodec/Makefile b/libavcodec/Makefile
index 2443d2c6fd..998f6b7e12 100644
--- a/libavcodec/Makefile
+++ b/libavcodec/Makefile
@@ -165,7 +165,7 @@ OBJS-$(CONFIG_STARTCODE)               += startcode.o
 OBJS-$(CONFIG_TEXTUREDSP)              += texturedsp.o
 OBJS-$(CONFIG_TEXTUREDSPENC)           += texturedspenc.o
 OBJS-$(CONFIG_TPELDSP)                 += tpeldsp.o
-OBJS-$(CONFIG_VAAPI_ENCODE)            += vaapi_encode.o
+OBJS-$(CONFIG_VAAPI_ENCODE)            += vaapi_encode.o hw_base_encode.o
 OBJS-$(CONFIG_AV1_AMF_ENCODER)         += amfenc_av1.o
 OBJS-$(CONFIG_VC1DSP)                  += vc1dsp.o
 OBJS-$(CONFIG_VIDEODSP)                += videodsp.o
diff --git a/libavcodec/hw_base_encode.c b/libavcodec/hw_base_encode.c
new file mode 100644
index 0000000000..ec7178d2dc
--- /dev/null
+++ b/libavcodec/hw_base_encode.c
@@ -0,0 +1,597 @@
+/*
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include "libavutil/avassert.h"
+#include "libavutil/common.h"
+#include "libavutil/internal.h"
+#include "libavutil/log.h"
+#include "libavutil/mem.h"
+#include "libavutil/pixdesc.h"
+
+#include "encode.h"
+#include "avcodec.h"
+#include "hw_base_encode.h"
+
+static void hw_base_encode_add_ref(AVCodecContext *avctx,
+                                   HWBaseEncodePicture *pic,
+                                   HWBaseEncodePicture *target,
+                                   int is_ref, int in_dpb, int prev)
+{
+    int refs = 0;
+
+    if (is_ref) {
+        av_assert0(pic != target);
+        av_assert0(pic->nb_refs[0] < MAX_PICTURE_REFERENCES &&
+                   pic->nb_refs[1] < MAX_PICTURE_REFERENCES);
+        if (target->display_order < pic->display_order)
+            pic->refs[0][pic->nb_refs[0]++] = target;
+        else
+            pic->refs[1][pic->nb_refs[1]++] = target;
+        ++refs;
+    }
+
+    if (in_dpb) {
+        av_assert0(pic->nb_dpb_pics < MAX_DPB_SIZE);
+        pic->dpb[pic->nb_dpb_pics++] = target;
+        ++refs;
+    }
+
+    if (prev) {
+        av_assert0(!pic->prev);
+        pic->prev = target;
+        ++refs;
+    }
+
+    target->ref_count[0] += refs;
+    target->ref_count[1] += refs;
+}
+
+static void hw_base_encode_remove_refs(AVCodecContext *avctx,
+                                       HWBaseEncodePicture *pic,
+                                       int level)
+{
+    int i;
+
+    if (pic->ref_removed[level])
+        return;
+
+    for (i = 0; i < pic->nb_refs[0]; i++) {
+        av_assert0(pic->refs[0][i]);
+        --pic->refs[0][i]->ref_count[level];
+        av_assert0(pic->refs[0][i]->ref_count[level] >= 0);
+    }
+
+    for (i = 0; i < pic->nb_refs[1]; i++) {
+        av_assert0(pic->refs[1][i]);
+        --pic->refs[1][i]->ref_count[level];
+        av_assert0(pic->refs[1][i]->ref_count[level] >= 0);
+    }
+
+    for (i = 0; i < pic->nb_dpb_pics; i++) {
+        av_assert0(pic->dpb[i]);
+        --pic->dpb[i]->ref_count[level];
+        av_assert0(pic->dpb[i]->ref_count[level] >= 0);
+    }
+
+    av_assert0(pic->prev || pic->type == PICTURE_TYPE_IDR);
+    if (pic->prev) {
+        --pic->prev->ref_count[level];
+        av_assert0(pic->prev->ref_count[level] >= 0);
+    }
+
+    pic->ref_removed[level] = 1;
+}
+
+static void hw_base_encode_set_b_pictures(AVCodecContext *avctx,
+                                          HWBaseEncodePicture *start,
+                                          HWBaseEncodePicture *end,
+                                          HWBaseEncodePicture *prev,
+                                          int current_depth,
+                                          HWBaseEncodePicture **last)
+{
+    HWBaseEncodeContext *ctx = avctx->priv_data;
+    HWBaseEncodePicture *pic, *next, *ref;
+    int i, len;
+
+    av_assert0(start && end && start != end && start->next != end);
+
+    // If we are at the maximum depth then encode all pictures as
+    // non-referenced B-pictures.  Also do this if there is exactly one
+    // picture left, since there will be nothing to reference it.
+    if (current_depth == ctx->max_b_depth || start->next->next == end) {
+        for (pic = start->next; pic; pic = pic->next) {
+            if (pic == end)
+                break;
+            pic->type    = PICTURE_TYPE_B;
+            pic->b_depth = current_depth;
+
+            hw_base_encode_add_ref(avctx, pic, start, 1, 1, 0);
+            hw_base_encode_add_ref(avctx, pic, end,   1, 1, 0);
+            hw_base_encode_add_ref(avctx, pic, prev,  0, 0, 1);
+
+            for (ref = end->refs[1][0]; ref; ref = ref->refs[1][0])
+                hw_base_encode_add_ref(avctx, pic, ref, 0, 1, 0);
+        }
+        *last = prev;
+
+    } else {
+        // Split the current list at the midpoint with a referenced
+        // B-picture, then descend into each side separately.
+        len = 0;
+        for (pic = start->next; pic != end; pic = pic->next)
+            ++len;
+        for (pic = start->next, i = 1; 2 * i < len; pic = pic->next, i++);
+
+        pic->type    = PICTURE_TYPE_B;
+        pic->b_depth = current_depth;
+
+        pic->is_reference = 1;
+
+        hw_base_encode_add_ref(avctx, pic, pic,   0, 1, 0);
+        hw_base_encode_add_ref(avctx, pic, start, 1, 1, 0);
+        hw_base_encode_add_ref(avctx, pic, end,   1, 1, 0);
+        hw_base_encode_add_ref(avctx, pic, prev,  0, 0, 1);
+
+        for (ref = end->refs[1][0]; ref; ref = ref->refs[1][0])
+            hw_base_encode_add_ref(avctx, pic, ref, 0, 1, 0);
+
+        if (i > 1)
+            hw_base_encode_set_b_pictures(avctx, start, pic, pic,
+                                          current_depth + 1, &next);
+        else
+            next = pic;
+
+        hw_base_encode_set_b_pictures(avctx, pic, end, next,
+                                      current_depth + 1, last);
+    }
+}
+
+static void hw_base_encode_add_next_prev(AVCodecContext *avctx,
+                                         HWBaseEncodePicture *pic)
+{
+    HWBaseEncodeContext *ctx = avctx->priv_data;
+    int i;
+
+    if (!pic)
+        return;
+
+    if (pic->type == PICTURE_TYPE_IDR) {
+        for (i = 0; i < ctx->nb_next_prev; i++) {
+            --ctx->next_prev[i]->ref_count[0];
+            ctx->next_prev[i] = NULL;
+        }
+        ctx->next_prev[0] = pic;
+        ++pic->ref_count[0];
+        ctx->nb_next_prev = 1;
+
+        return;
+    }
+
+    if (ctx->nb_next_prev < MAX_PICTURE_REFERENCES) {
+        ctx->next_prev[ctx->nb_next_prev++] = pic;
+        ++pic->ref_count[0];
+    } else {
+        --ctx->next_prev[0]->ref_count[0];
+        for (i = 0; i < MAX_PICTURE_REFERENCES - 1; i++)
+            ctx->next_prev[i] = ctx->next_prev[i + 1];
+        ctx->next_prev[i] = pic;
+        ++pic->ref_count[0];
+    }
+}
+
+static int hw_base_encode_pick_next(AVCodecContext *avctx,
+                                    HWBaseEncodePicture **pic_out)
+{
+    HWBaseEncodeContext *ctx = avctx->priv_data;
+    HWBaseEncodePicture *pic = NULL, *prev = NULL, *next, *start;
+    int i, b_counter, closed_gop_end;
+
+    // If there are any B-frames already queued, the next one to encode
+    // is the earliest not-yet-issued frame for which all references are
+    // available.
+    for (pic = ctx->pic_start; pic; pic = pic->next) {
+        if (pic->encode_issued)
+            continue;
+        if (pic->type != PICTURE_TYPE_B)
+            continue;
+        for (i = 0; i < pic->nb_refs[0]; i++) {
+            if (!pic->refs[0][i]->encode_issued)
+                break;
+        }
+        if (i != pic->nb_refs[0])
+            continue;
+
+        for (i = 0; i < pic->nb_refs[1]; i++) {
+            if (!pic->refs[1][i]->encode_issued)
+                break;
+        }
+        if (i == pic->nb_refs[1])
+            break;
+    }
+
+    if (pic) {
+        av_log(avctx, AV_LOG_DEBUG, "Pick B-picture at depth %d to "
+               "encode next.\n", pic->b_depth);
+        *pic_out = pic;
+        return 0;
+    }
+
+    // Find the B-per-Pth available picture to become the next picture
+    // on the top layer.
+    start = NULL;
+    b_counter = 0;
+    closed_gop_end = ctx->closed_gop ||
+                     ctx->idr_counter == ctx->gop_per_idr;
+    for (pic = ctx->pic_start; pic; pic = next) {
+        next = pic->next;
+        if (pic->encode_issued) {
+            start = pic;
+            continue;
+        }
+        // If the next available picture is force-IDR, encode it to start
+        // a new GOP immediately.
+        if (pic->force_idr)
+            break;
+        if (b_counter == ctx->b_per_p)
+            break;
+        // If this picture ends a closed GOP or starts a new GOP then it
+        // needs to be in the top layer.
+        if (ctx->gop_counter + b_counter + closed_gop_end >= ctx->gop_size)
+            break;
+        // If the picture after this one is force-IDR, we need to encode
+        // this one in the top layer.
+        if (next && next->force_idr)
+            break;
+        ++b_counter;
+    }
+
+    // At the end of the stream the last picture must be in the top layer.
+    if (!pic && ctx->end_of_stream) {
+        --b_counter;
+        pic = ctx->pic_end;
+        if (pic->encode_complete)
+            return AVERROR_EOF;
+        else if (pic->encode_issued)
+            return AVERROR(EAGAIN);
+    }
+
+    if (!pic) {
+        av_log(avctx, AV_LOG_DEBUG, "Pick nothing to encode next - "
+               "need more input for reference pictures.\n");
+        return AVERROR(EAGAIN);
+    }
+    if (ctx->input_order <= ctx->decode_delay && !ctx->end_of_stream) {
+        av_log(avctx, AV_LOG_DEBUG, "Pick nothing to encode next - "
+               "need more input for timestamps.\n");
+        return AVERROR(EAGAIN);
+    }
+
+    if (pic->force_idr) {
+        av_log(avctx, AV_LOG_DEBUG, "Pick forced IDR-picture to "
+               "encode next.\n");
+        pic->type = PICTURE_TYPE_IDR;
+        ctx->idr_counter = 1;
+        ctx->gop_counter = 1;
+
+    } else if (ctx->gop_counter + b_counter >= ctx->gop_size) {
+        if (ctx->idr_counter == ctx->gop_per_idr) {
+            av_log(avctx, AV_LOG_DEBUG, "Pick new-GOP IDR-picture to "
+                   "encode next.\n");
+            pic->type = PICTURE_TYPE_IDR;
+            ctx->idr_counter = 1;
+        } else {
+            av_log(avctx, AV_LOG_DEBUG, "Pick new-GOP I-picture to "
+                   "encode next.\n");
+            pic->type = PICTURE_TYPE_I;
+            ++ctx->idr_counter;
+        }
+        ctx->gop_counter = 1;
+
+    } else {
+        if (ctx->gop_counter + b_counter + closed_gop_end == ctx->gop_size) {
+            av_log(avctx, AV_LOG_DEBUG, "Pick group-end P-picture to "
+                   "encode next.\n");
+        } else {
+            av_log(avctx, AV_LOG_DEBUG, "Pick normal P-picture to "
+                   "encode next.\n");
+        }
+        pic->type = PICTURE_TYPE_P;
+        av_assert0(start);
+        ctx->gop_counter += 1 + b_counter;
+    }
+    pic->is_reference = 1;
+    *pic_out = pic;
+
+    hw_base_encode_add_ref(avctx, pic, pic, 0, 1, 0);
+    if (pic->type != PICTURE_TYPE_IDR) {
+        // TODO: apply both previous and forward multi reference for all vaapi encoders.
+        // And L0/L1 reference frame number can be set dynamically through query
+        // VAConfigAttribEncMaxRefFrames attribute.
+        if (avctx->codec_id == AV_CODEC_ID_AV1) {
+            for (i = 0; i < ctx->nb_next_prev; i++)
+                hw_base_encode_add_ref(avctx, pic, ctx->next_prev[i],
+                                       pic->type == PICTURE_TYPE_P,
+                                       b_counter > 0, 0);
+        } else
+            hw_base_encode_add_ref(avctx, pic, start,
+                                   pic->type == PICTURE_TYPE_P,
+                                   b_counter > 0, 0);
+
+        hw_base_encode_add_ref(avctx, pic, ctx->next_prev[ctx->nb_next_prev - 1], 0, 0, 1);
+    }
+
+    if (b_counter > 0) {
+        hw_base_encode_set_b_pictures(avctx, start, pic, pic, 1,
+                                      &prev);
+    } else {
+        prev = pic;
+    }
+    hw_base_encode_add_next_prev(avctx, prev);
+
+    return 0;
+}
+
+static int hw_base_encode_clear_old(AVCodecContext *avctx)
+{
+    HWBaseEncodeContext *ctx = avctx->priv_data;
+    HWBaseEncodePicture *pic, *prev, *next;
+
+    av_assert0(ctx->pic_start);
+
+    // Remove direct references once each picture is complete.
+    for (pic = ctx->pic_start; pic; pic = pic->next) {
+        if (pic->encode_complete && pic->next)
+            hw_base_encode_remove_refs(avctx, pic, 0);
+    }
+
+    // Remove indirect references once a picture has no direct references.
+    for (pic = ctx->pic_start; pic; pic = pic->next) {
+        if (pic->encode_complete && pic->ref_count[0] == 0)
+            hw_base_encode_remove_refs(avctx, pic, 1);
+    }
+
+    // Clear out all complete pictures with no remaining references.
+    prev = NULL;
+    for (pic = ctx->pic_start; pic; pic = next) {
+        next = pic->next;
+        if (pic->encode_complete && pic->ref_count[1] == 0) {
+            av_assert0(pic->ref_removed[0] && pic->ref_removed[1]);
+            if (prev)
+                prev->next = next;
+            else
+                ctx->pic_start = next;
+            ctx->op->free(avctx, pic);
+        } else {
+            prev = pic;
+        }
+    }
+
+    return 0;
+}
+
+static int hw_base_encode_check_frame(AVCodecContext *avctx,
+                                      const AVFrame *frame)
+{
+    HWBaseEncodeContext *ctx = avctx->priv_data;
+
+    if ((frame->crop_top  || frame->crop_bottom ||
+         frame->crop_left || frame->crop_right) && !ctx->crop_warned) {
+        av_log(avctx, AV_LOG_WARNING, "Cropping information on input "
+               "frames ignored due to lack of API support.\n");
+        ctx->crop_warned = 1;
+    }
+
+    if (!ctx->roi_allowed) {
+        AVFrameSideData *sd =
+            av_frame_get_side_data(frame, AV_FRAME_DATA_REGIONS_OF_INTEREST);
+
+        if (sd && !ctx->roi_warned) {
+            av_log(avctx, AV_LOG_WARNING, "ROI side data on input "
+                   "frames ignored due to lack of driver support.\n");
+            ctx->roi_warned = 1;
+        }
+    }
+
+    return 0;
+}
+
+static int hw_base_encode_send_frame(AVCodecContext *avctx, AVFrame *frame)
+{
+    HWBaseEncodeContext *ctx = avctx->priv_data;
+    HWBaseEncodePicture *pic;
+    int err;
+
+    if (frame) {
+        av_log(avctx, AV_LOG_DEBUG, "Input frame: %ux%u (%"PRId64").\n",
+               frame->width, frame->height, frame->pts);
+
+        err = hw_base_encode_check_frame(avctx, frame);
+        if (err < 0)
+            return err;
+
+        pic = ctx->op->alloc(avctx, frame);
+        if (!pic)
+            return AVERROR(ENOMEM);
+
+        pic->input_image = av_frame_alloc();
+        if (!pic->input_image) {
+            err = AVERROR(ENOMEM);
+            goto fail;
+        }
+
+        pic->recon_image = av_frame_alloc();
+        if (!pic->recon_image) {
+            err = AVERROR(ENOMEM);
+            goto fail;
+        }
+
+        if (ctx->input_order == 0 || frame->pict_type == AV_PICTURE_TYPE_I)
+            pic->force_idr = 1;
+
+        pic->pts = frame->pts;
+        pic->duration = frame->duration;
+
+        if (avctx->flags & AV_CODEC_FLAG_COPY_OPAQUE) {
+            err = av_buffer_replace(&pic->opaque_ref, frame->opaque_ref);
+            if (err < 0)
+                goto fail;
+
+            pic->opaque = frame->opaque;
+        }
+
+        av_frame_move_ref(pic->input_image, frame);
+
+        if (ctx->input_order == 0)
+            ctx->first_pts = pic->pts;
+        if (ctx->input_order == ctx->decode_delay)
+            ctx->dts_pts_diff = pic->pts - ctx->first_pts;
+        if (ctx->output_delay > 0)
+            ctx->ts_ring[ctx->input_order %
+                        (3 * ctx->output_delay + ctx->async_depth)] = pic->pts;
+
+        pic->display_order = ctx->input_order;
+        ++ctx->input_order;
+
+        if (ctx->pic_start) {
+            ctx->pic_end->next = pic;
+            ctx->pic_end       = pic;
+        } else {
+            ctx->pic_start     = pic;
+            ctx->pic_end       = pic;
+        }
+
+    } else {
+        ctx->end_of_stream = 1;
+
+        // Fix timestamps if we hit end-of-stream before the initial decode
+        // delay has elapsed.
+        if (ctx->input_order < ctx->decode_delay)
+            ctx->dts_pts_diff = ctx->pic_end->pts - ctx->first_pts;
+    }
+
+    return 0;
+
+fail:
+    ctx->op->free(avctx, pic);
+    return err;
+}
+
+int ff_hw_base_encode_receive_packet(AVCodecContext *avctx, AVPacket *pkt)
+{
+    HWBaseEncodeContext *ctx = avctx->priv_data;
+    HWBaseEncodePicture *pic = NULL;
+    AVFrame *frame = ctx->frame;
+    int err;
+
+    av_assert0(ctx->op && ctx->op->alloc && ctx->op->issue &&
+               ctx->op->output && ctx->op->free);
+
+start:
+    /** if no B frame before repeat P frame, sent repeat P frame out. */
+    if (ctx->tail_pkt->size) {
+        for (HWBaseEncodePicture *tmp = ctx->pic_start; tmp; tmp = tmp->next) {
+            if (tmp->type == PICTURE_TYPE_B && tmp->pts < ctx->tail_pkt->pts)
+                break;
+            else if (!tmp->next) {
+                av_packet_move_ref(pkt, ctx->tail_pkt);
+                goto end;
+            }
+        }
+    }
+
+    err = ff_encode_get_frame(avctx, frame);
+    if (err < 0 && err != AVERROR_EOF)
+        return err;
+
+    if (err == AVERROR_EOF)
+        frame = NULL;
+
+    err = hw_base_encode_send_frame(avctx, frame);
+    if (err < 0)
+        return err;
+
+    if (!ctx->pic_start) {
+        if (ctx->end_of_stream)
+            return AVERROR_EOF;
+        else
+            return AVERROR(EAGAIN);
+    }
+
+    if (ctx->async_encode) {
+        if (av_fifo_can_write(ctx->encode_fifo)) {
+            err = hw_base_encode_pick_next(avctx, &pic);
+            if (!err) {
+                av_assert0(pic);
+                pic->encode_order = ctx->encode_order +
+                    av_fifo_can_read(ctx->encode_fifo);
+                err = ctx->op->issue(avctx, pic);
+                if (err < 0) {
+                    av_log(avctx, AV_LOG_ERROR, "Encode failed: %d.\n", err);
+                    return err;
+                }
+                pic->encode_issued = 1;
+                av_fifo_write(ctx->encode_fifo, &pic, 1);
+            }
+        }
+
+        if (!av_fifo_can_read(ctx->encode_fifo))
+            return err;
+
+        // More frames can be buffered
+        if (av_fifo_can_write(ctx->encode_fifo) && !ctx->end_of_stream)
+            return AVERROR(EAGAIN);
+
+        av_fifo_read(ctx->encode_fifo, &pic, 1);
+        ctx->encode_order = pic->encode_order + 1;
+    } else {
+        err = hw_base_encode_pick_next(avctx, &pic);
+        if (err < 0)
+            return err;
+        av_assert0(pic);
+
+        pic->encode_order = ctx->encode_order++;
+
+        err = ctx->op->issue(avctx, pic);
+        if (err < 0) {
+            av_log(avctx, AV_LOG_ERROR, "Encode failed: %d.\n", err);
+            return err;
+        }
+
+        pic->encode_issued = 1;
+    }
+
+    err = ctx->op->output(avctx, pic, pkt);
+    if (err < 0) {
+        av_log(avctx, AV_LOG_ERROR, "Output failed: %d.\n", err);
+        return err;
+    }
+
+    ctx->output_order = pic->encode_order;
+    hw_base_encode_clear_old(avctx);
+
+    /** loop to get an available pkt in encoder flushing. */
+    if (ctx->end_of_stream && !pkt->size)
+        goto start;
+
+end:
+    if (pkt->size)
+        av_log(avctx, AV_LOG_DEBUG, "Output packet: pts %"PRId64", dts %"PRId64", "
+               "size %d bytes.\n", pkt->pts, pkt->dts, pkt->size);
+
+    return 0;
+}
diff --git a/libavcodec/hw_base_encode.h b/libavcodec/hw_base_encode.h
index a578db8c06..2667ae61cd 100644
--- a/libavcodec/hw_base_encode.h
+++ b/libavcodec/hw_base_encode.h
@@ -19,6 +19,8 @@
 #ifndef AVCODEC_HW_BASE_ENCODE_H
 #define AVCODEC_HW_BASE_ENCODE_H
 
+#include "libavutil/fifo.h"
+
 #define MAX_DPB_SIZE 16
 #define MAX_PICTURE_REFERENCES 2
 #define MAX_REORDER_DELAY 16
@@ -53,13 +55,135 @@ enum {
     FLAG_NON_IDR_KEY_PICTURES  = 1 << 5,
 };
 
+typedef struct HWBaseEncodePicture {
+    struct HWBaseEncodePicture *next;
+
+    int64_t         display_order;
+    int64_t         encode_order;
+    int64_t         pts;
+    int64_t         duration;
+    int             force_idr;
+
+    void           *opaque;
+    AVBufferRef    *opaque_ref;
+
+    int             type;
+    int             b_depth;
+    int             encode_issued;
+    int             encode_complete;
+
+    AVFrame        *input_image;
+    AVFrame        *recon_image;
+
+    void           *priv_data;
+
+    // Whether this picture is a reference picture.
+    int             is_reference;
+
+    // The contents of the DPB after this picture has been decoded.
+    // This will contain the picture itself if it is a reference picture,
+    // but not if it isn't.
+    int                     nb_dpb_pics;
+    struct HWBaseEncodePicture *dpb[MAX_DPB_SIZE];
+    // The reference pictures used in decoding this picture. If they are
+    // used by later pictures they will also appear in the DPB. ref[0][] for
+    // previous reference frames. ref[1][] for future reference frames.
+    int                     nb_refs[MAX_REFERENCE_LIST_NUM];
+    struct HWBaseEncodePicture *refs[MAX_REFERENCE_LIST_NUM][MAX_PICTURE_REFERENCES];
+    // The previous reference picture in encode order.  Must be in at least
+    // one of the reference list and DPB list.
+    struct HWBaseEncodePicture *prev;
+    // Reference count for other pictures referring to this one through
+    // the above pointers, directly from incomplete pictures and indirectly
+    // through completed pictures.
+    int             ref_count[2];
+    int             ref_removed[2];
+} HWBaseEncodePicture;
+
+typedef struct HWEncodePictureOperation {
+    // Alloc memory for the picture structure and initialize the API-specific internals
+    // based of the given frame.
+    HWBaseEncodePicture * (*alloc)(AVCodecContext *avctx, const AVFrame *frame);
+    // Issue the picture structure, which will send the frame surface to HW Encode API.
+    int (*issue)(AVCodecContext *avctx, const HWBaseEncodePicture *base_pic);
+    // Get the output AVPacket.
+    int (*output)(AVCodecContext *avctx, const HWBaseEncodePicture *base_pic, AVPacket *pkt);
+    // Free the picture structure.
+    int (*free)(AVCodecContext *avctx, HWBaseEncodePicture *base_pic);
+}  HWEncodePictureOperation;
+
 typedef struct HWBaseEncodeContext {
     const AVClass *class;
 
+    // Hardware-specific hooks.
+    const struct HWEncodePictureOperation *op;
+
+    // Current encoding window, in display (input) order.
+    HWBaseEncodePicture *pic_start, *pic_end;
+    // The next picture to use as the previous reference picture in
+    // encoding order. Order from small to large in encoding order.
+    HWBaseEncodePicture *next_prev[MAX_PICTURE_REFERENCES];
+    int                  nb_next_prev;
+
+    // Next input order index (display order).
+    int64_t         input_order;
+    // Number of frames that output is behind input.
+    int64_t         output_delay;
+    // Next encode order index.
+    int64_t         encode_order;
+    // Number of frames decode output will need to be delayed.
+    int64_t         decode_delay;
+    // Next output order index (in encode order).
+    int64_t         output_order;
+
+    // Timestamp handling.
+    int64_t         first_pts;
+    int64_t         dts_pts_diff;
+    int64_t         ts_ring[MAX_REORDER_DELAY * 3 +
+                            MAX_ASYNC_DEPTH];
+
+    // Frame type decision.
+    int gop_size;
+    int closed_gop;
+    int gop_per_idr;
+    int p_per_i;
+    int max_b_depth;
+    int b_per_p;
+    int force_idr;
+    int idr_counter;
+    int gop_counter;
+    int end_of_stream;
+    int p_to_gpb;
+
+    // Whether the driver supports ROI at all.
+    int             roi_allowed;
+
+    // The encoder does not support cropping information, so warn about
+    // it the first time we encounter any nonzero crop fields.
+    int             crop_warned;
+    // If the driver does not support ROI then warn the first time we
+    // encounter a frame with ROI side data.
+    int             roi_warned;
+
+    // The frame to be filled with data.
+    AVFrame         *frame;
+
+    // Whether the HW supports sync buffer function.
+    // If supported, encode_fifo/async_depth will be used together.
+    // Used for output buffer synchronization.
+    int             async_encode;
+
+    // Store buffered pic.
+    AVFifo          *encode_fifo;
     // Max number of frame buffered in encoder.
     int             async_depth;
+
+    /** Tail data of a pic, now only used for av1 repeat frame header. */
+    AVPacket        *tail_pkt;
 } HWBaseEncodeContext;
 
+int ff_hw_base_encode_receive_packet(AVCodecContext *avctx, AVPacket *pkt);
+
 #define HW_BASE_ENCODE_COMMON_OPTIONS \
     { "async_depth", "Maximum processing parallelism. " \
       "Increase this to improve single channel performance.", \
diff --git a/libavcodec/vaapi_encode.c b/libavcodec/vaapi_encode.c
index 227cccae64..c4bb93c520 100644
--- a/libavcodec/vaapi_encode.c
+++ b/libavcodec/vaapi_encode.c
@@ -138,22 +138,26 @@ static int vaapi_encode_make_misc_param_buffer(AVCodecContext *avctx,
 static int vaapi_encode_wait(AVCodecContext *avctx,
                              VAAPIEncodePicture *pic)
 {
+#if VA_CHECK_VERSION(1, 9, 0)
+    HWBaseEncodeContext *base_ctx = avctx->priv_data;
+#endif
     VAAPIEncodeContext *ctx = avctx->priv_data;
+    HWBaseEncodePicture *base_pic = &pic->base;
     VAStatus vas;
 
-    av_assert0(pic->encode_issued);
+    av_assert0(base_pic->encode_issued);
 
-    if (pic->encode_complete) {
+    if (base_pic->encode_complete) {
         // Already waited for this picture.
         return 0;
     }
 
     av_log(avctx, AV_LOG_DEBUG, "Sync to pic %"PRId64"/%"PRId64" "
-           "(input surface %#x).\n", pic->display_order,
-           pic->encode_order, pic->input_surface);
+           "(input surface %#x).\n", base_pic->display_order,
+           base_pic->encode_order, pic->input_surface);
 
 #if VA_CHECK_VERSION(1, 9, 0)
-    if (ctx->has_sync_buffer_func) {
+    if (base_ctx->async_encode) {
         vas = vaSyncBuffer(ctx->hwctx->display,
                            pic->output_buffer,
                            VA_TIMEOUT_INFINITE);
@@ -174,9 +178,9 @@ static int vaapi_encode_wait(AVCodecContext *avctx,
     }
 
     // Input is definitely finished with now.
-    av_frame_free(&pic->input_image);
+    av_frame_free(&base_pic->input_image);
 
-    pic->encode_complete = 1;
+    base_pic->encode_complete = 1;
     return 0;
 }
 
@@ -263,9 +267,11 @@ static int vaapi_encode_make_tile_slice(AVCodecContext *avctx,
 }
 
 static int vaapi_encode_issue(AVCodecContext *avctx,
-                              VAAPIEncodePicture *pic)
+                              const HWBaseEncodePicture *base_pic)
 {
-    VAAPIEncodeContext *ctx = avctx->priv_data;
+    HWBaseEncodeContext *base_ctx = avctx->priv_data;
+    VAAPIEncodeContext       *ctx = avctx->priv_data;
+    VAAPIEncodePicture *pic = (VAAPIEncodePicture*)base_pic;
     VAAPIEncodeSlice *slice;
     VAStatus vas;
     int err, i;
@@ -274,52 +280,46 @@ static int vaapi_encode_issue(AVCodecContext *avctx,
     av_unused AVFrameSideData *sd;
 
     av_log(avctx, AV_LOG_DEBUG, "Issuing encode for pic %"PRId64"/%"PRId64" "
-           "as type %s.\n", pic->display_order, pic->encode_order,
-           ff_hw_base_encode_get_pictype_name(pic->type));
-    if (pic->nb_refs[0] == 0 && pic->nb_refs[1] == 0) {
+           "as type %s.\n", base_pic->display_order, base_pic->encode_order,
+           ff_hw_base_encode_get_pictype_name(base_pic->type));
+    if (base_pic->nb_refs[0] == 0 && base_pic->nb_refs[1] == 0) {
         av_log(avctx, AV_LOG_DEBUG, "No reference pictures.\n");
     } else {
         av_log(avctx, AV_LOG_DEBUG, "L0 refers to");
-        for (i = 0; i < pic->nb_refs[0]; i++) {
+        for (i = 0; i < base_pic->nb_refs[0]; i++) {
             av_log(avctx, AV_LOG_DEBUG, " %"PRId64"/%"PRId64,
-                   pic->refs[0][i]->display_order, pic->refs[0][i]->encode_order);
+                   base_pic->refs[0][i]->display_order, base_pic->refs[0][i]->encode_order);
         }
         av_log(avctx, AV_LOG_DEBUG, ".\n");
 
-        if (pic->nb_refs[1]) {
+        if (base_pic->nb_refs[1]) {
             av_log(avctx, AV_LOG_DEBUG, "L1 refers to");
-            for (i = 0; i < pic->nb_refs[1]; i++) {
+            for (i = 0; i < base_pic->nb_refs[1]; i++) {
                 av_log(avctx, AV_LOG_DEBUG, " %"PRId64"/%"PRId64,
-                       pic->refs[1][i]->display_order, pic->refs[1][i]->encode_order);
+                       base_pic->refs[1][i]->display_order, base_pic->refs[1][i]->encode_order);
             }
             av_log(avctx, AV_LOG_DEBUG, ".\n");
         }
     }
 
-    av_assert0(!pic->encode_issued);
-    for (i = 0; i < pic->nb_refs[0]; i++) {
-        av_assert0(pic->refs[0][i]);
-        av_assert0(pic->refs[0][i]->encode_issued);
+    av_assert0(!base_pic->encode_issued);
+    for (i = 0; i < base_pic->nb_refs[0]; i++) {
+        av_assert0(base_pic->refs[0][i]);
+        av_assert0(base_pic->refs[0][i]->encode_issued);
     }
-    for (i = 0; i < pic->nb_refs[1]; i++) {
-        av_assert0(pic->refs[1][i]);
-        av_assert0(pic->refs[1][i]->encode_issued);
+    for (i = 0; i < base_pic->nb_refs[1]; i++) {
+        av_assert0(base_pic->refs[1][i]);
+        av_assert0(base_pic->refs[1][i]->encode_issued);
     }
 
     av_log(avctx, AV_LOG_DEBUG, "Input surface is %#x.\n", pic->input_surface);
 
-    pic->recon_image = av_frame_alloc();
-    if (!pic->recon_image) {
-        err = AVERROR(ENOMEM);
-        goto fail;
-    }
-
-    err = av_hwframe_get_buffer(ctx->recon_frames_ref, pic->recon_image, 0);
+    err = av_hwframe_get_buffer(ctx->recon_frames_ref, base_pic->recon_image, 0);
     if (err < 0) {
         err = AVERROR(ENOMEM);
         goto fail;
     }
-    pic->recon_surface = (VASurfaceID)(uintptr_t)pic->recon_image->data[3];
+    pic->recon_surface = (VASurfaceID)(uintptr_t)base_pic->recon_image->data[3];
     av_log(avctx, AV_LOG_DEBUG, "Recon surface is %#x.\n", pic->recon_surface);
 
     pic->output_buffer_ref = ff_refstruct_pool_get(ctx->output_buffer_pool);
@@ -343,7 +343,7 @@ static int vaapi_encode_issue(AVCodecContext *avctx,
 
     pic->nb_param_buffers = 0;
 
-    if (pic->type == PICTURE_TYPE_IDR && ctx->codec->init_sequence_params) {
+    if (base_pic->type == PICTURE_TYPE_IDR && ctx->codec->init_sequence_params) {
         err = vaapi_encode_make_param_buffer(avctx, pic,
                                              VAEncSequenceParameterBufferType,
                                              ctx->codec_sequence_params,
@@ -352,7 +352,7 @@ static int vaapi_encode_issue(AVCodecContext *avctx,
             goto fail;
     }
 
-    if (pic->type == PICTURE_TYPE_IDR) {
+    if (base_pic->type == PICTURE_TYPE_IDR) {
         for (i = 0; i < ctx->nb_global_params; i++) {
             err = vaapi_encode_make_misc_param_buffer(avctx, pic,
                                                       ctx->global_params_type[i],
@@ -389,7 +389,7 @@ static int vaapi_encode_issue(AVCodecContext *avctx,
     }
 #endif
 
-    if (pic->type == PICTURE_TYPE_IDR) {
+    if (base_pic->type == PICTURE_TYPE_IDR) {
         if (ctx->va_packed_headers & VA_ENC_PACKED_HEADER_SEQUENCE &&
             ctx->codec->write_sequence_header) {
             bit_len = 8 * sizeof(data);
@@ -529,9 +529,9 @@ static int vaapi_encode_issue(AVCodecContext *avctx,
     }
 
 #if VA_CHECK_VERSION(1, 0, 0)
-    sd = av_frame_get_side_data(pic->input_image,
+    sd = av_frame_get_side_data(base_pic->input_image,
                                 AV_FRAME_DATA_REGIONS_OF_INTEREST);
-    if (sd && ctx->roi_allowed) {
+    if (sd && base_ctx->roi_allowed) {
         const AVRegionOfInterest *roi;
         uint32_t roi_size;
         VAEncMiscParameterBufferROI param_roi;
@@ -542,11 +542,11 @@ static int vaapi_encode_issue(AVCodecContext *avctx,
         av_assert0(roi_size && sd->size % roi_size == 0);
         nb_roi = sd->size / roi_size;
         if (nb_roi > ctx->roi_max_regions) {
-            if (!ctx->roi_warned) {
+            if (!base_ctx->roi_warned) {
                 av_log(avctx, AV_LOG_WARNING, "More ROIs set than "
                        "supported by driver (%d > %d).\n",
                        nb_roi, ctx->roi_max_regions);
-                ctx->roi_warned = 1;
+                base_ctx->roi_warned = 1;
             }
             nb_roi = ctx->roi_max_regions;
         }
@@ -639,8 +639,6 @@ static int vaapi_encode_issue(AVCodecContext *avctx,
         }
     }
 
-    pic->encode_issued = 1;
-
     return 0;
 
 fail_with_picture:
@@ -657,14 +655,13 @@ fail_at_end:
     av_freep(&pic->param_buffers);
     av_freep(&pic->slices);
     av_freep(&pic->roi);
-    av_frame_free(&pic->recon_image);
     ff_refstruct_unref(&pic->output_buffer_ref);
     pic->output_buffer = VA_INVALID_ID;
     return err;
 }
 
 static int vaapi_encode_set_output_property(AVCodecContext *avctx,
-                                            VAAPIEncodePicture *pic,
+                                            HWBaseEncodePicture *pic,
                                             AVPacket *pkt)
 {
     HWBaseEncodeContext *base_ctx = avctx->priv_data;
@@ -689,16 +686,16 @@ static int vaapi_encode_set_output_property(AVCodecContext *avctx,
         return 0;
     }
 
-    if (ctx->output_delay == 0) {
+    if (base_ctx->output_delay == 0) {
         pkt->dts = pkt->pts;
-    } else if (pic->encode_order < ctx->decode_delay) {
-        if (ctx->ts_ring[pic->encode_order] < INT64_MIN + ctx->dts_pts_diff)
+    } else if (pic->encode_order < base_ctx->decode_delay) {
+        if (base_ctx->ts_ring[pic->encode_order] < INT64_MIN + base_ctx->dts_pts_diff)
             pkt->dts = INT64_MIN;
         else
-            pkt->dts = ctx->ts_ring[pic->encode_order] - ctx->dts_pts_diff;
+            pkt->dts = base_ctx->ts_ring[pic->encode_order] - base_ctx->dts_pts_diff;
     } else {
-        pkt->dts = ctx->ts_ring[(pic->encode_order - ctx->decode_delay) %
-                                (3 * ctx->output_delay + base_ctx->async_depth)];
+        pkt->dts = base_ctx->ts_ring[(pic->encode_order - base_ctx->decode_delay) %
+                                     (3 * base_ctx->output_delay + base_ctx->async_depth)];
     }
 
     return 0;
@@ -817,9 +814,11 @@ end:
 }
 
 static int vaapi_encode_output(AVCodecContext *avctx,
-                               VAAPIEncodePicture *pic, AVPacket *pkt)
+                               const HWBaseEncodePicture *base_pic, AVPacket *pkt)
 {
-    VAAPIEncodeContext *ctx = avctx->priv_data;
+    HWBaseEncodeContext *base_ctx = avctx->priv_data;
+    VAAPIEncodeContext       *ctx = avctx->priv_data;
+    VAAPIEncodePicture       *pic = (VAAPIEncodePicture*)base_pic;
     AVPacket *pkt_ptr = pkt;
     int err;
 
@@ -832,17 +831,17 @@ static int vaapi_encode_output(AVCodecContext *avctx,
         ctx->coded_buffer_ref = ff_refstruct_ref(pic->output_buffer_ref);
 
         if (pic->tail_size) {
-            if (ctx->tail_pkt->size) {
+            if (base_ctx->tail_pkt->size) {
                 err = AVERROR_BUG;
                 goto end;
             }
 
-            err = ff_get_encode_buffer(avctx, ctx->tail_pkt, pic->tail_size, 0);
+            err = ff_get_encode_buffer(avctx, base_ctx->tail_pkt, pic->tail_size, 0);
             if (err < 0)
                 goto end;
 
-            memcpy(ctx->tail_pkt->data, pic->tail_data, pic->tail_size);
-            pkt_ptr = ctx->tail_pkt;
+            memcpy(base_ctx->tail_pkt->data, pic->tail_data, pic->tail_size);
+            pkt_ptr = base_ctx->tail_pkt;
         }
     } else {
         err = vaapi_encode_get_coded_data(avctx, pic, pkt);
@@ -851,9 +850,9 @@ static int vaapi_encode_output(AVCodecContext *avctx,
     }
 
     av_log(avctx, AV_LOG_DEBUG, "Output read for pic %"PRId64"/%"PRId64".\n",
-           pic->display_order, pic->encode_order);
+           base_pic->display_order, base_pic->encode_order);
 
-    vaapi_encode_set_output_property(avctx, pic, pkt_ptr);
+    vaapi_encode_set_output_property(avctx, (HWBaseEncodePicture*)pic, pkt_ptr);
 
 end:
     ff_refstruct_unref(&pic->output_buffer_ref);
@@ -864,12 +863,14 @@ end:
 static int vaapi_encode_discard(AVCodecContext *avctx,
                                 VAAPIEncodePicture *pic)
 {
+    HWBaseEncodePicture *base_pic = &pic->base;
+
     vaapi_encode_wait(avctx, pic);
 
     if (pic->output_buffer_ref) {
         av_log(avctx, AV_LOG_DEBUG, "Discard output for pic "
                "%"PRId64"/%"PRId64".\n",
-               pic->display_order, pic->encode_order);
+               base_pic->display_order, base_pic->encode_order);
 
         ff_refstruct_unref(&pic->output_buffer_ref);
         pic->output_buffer = VA_INVALID_ID;
@@ -878,8 +879,8 @@ static int vaapi_encode_discard(AVCodecContext *avctx,
     return 0;
 }
 
-static VAAPIEncodePicture *vaapi_encode_alloc(AVCodecContext *avctx,
-                                              const AVFrame *frame)
+static HWBaseEncodePicture *vaapi_encode_alloc(AVCodecContext *avctx,
+                                               const AVFrame *frame)
 {
     VAAPIEncodeContext *ctx = avctx->priv_data;
     VAAPIEncodePicture *pic;
@@ -889,8 +890,8 @@ static VAAPIEncodePicture *vaapi_encode_alloc(AVCodecContext *avctx,
         return NULL;
 
     if (ctx->codec->picture_priv_data_size > 0) {
-        pic->priv_data = av_mallocz(ctx->codec->picture_priv_data_size);
-        if (!pic->priv_data) {
+        pic->base.priv_data = av_mallocz(ctx->codec->picture_priv_data_size);
+        if (!pic->base.priv_data) {
             av_freep(&pic);
             return NULL;
         }
@@ -900,15 +901,16 @@ static VAAPIEncodePicture *vaapi_encode_alloc(AVCodecContext *avctx,
     pic->recon_surface = VA_INVALID_ID;
     pic->output_buffer = VA_INVALID_ID;
 
-    return pic;
+    return &pic->base;
 }
 
 static int vaapi_encode_free(AVCodecContext *avctx,
-                             VAAPIEncodePicture *pic)
+                             HWBaseEncodePicture *base_pic)
 {
+    VAAPIEncodePicture *pic = (VAAPIEncodePicture*)base_pic;
     int i;
 
-    if (pic->encode_issued)
+    if (base_pic->encode_issued)
         vaapi_encode_discard(avctx, pic);
 
     if (pic->slices) {
@@ -916,17 +918,17 @@ static int vaapi_encode_free(AVCodecContext *avctx,
             av_freep(&pic->slices[i].codec_slice_params);
     }
 
-    av_frame_free(&pic->input_image);
-    av_frame_free(&pic->recon_image);
+    av_frame_free(&base_pic->input_image);
+    av_frame_free(&base_pic->recon_image);
 
-    av_buffer_unref(&pic->opaque_ref);
+    av_buffer_unref(&base_pic->opaque_ref);
 
     av_freep(&pic->param_buffers);
     av_freep(&pic->slices);
     // Output buffer should already be destroyed.
     av_assert0(pic->output_buffer == VA_INVALID_ID);
 
-    av_freep(&pic->priv_data);
+    av_freep(&base_pic->priv_data);
     av_freep(&pic->codec_picture_params);
     av_freep(&pic->roi);
 
@@ -935,564 +937,6 @@ static int vaapi_encode_free(AVCodecContext *avctx,
     return 0;
 }
 
-static void vaapi_encode_add_ref(AVCodecContext *avctx,
-                                 VAAPIEncodePicture *pic,
-                                 VAAPIEncodePicture *target,
-                                 int is_ref, int in_dpb, int prev)
-{
-    int refs = 0;
-
-    if (is_ref) {
-        av_assert0(pic != target);
-        av_assert0(pic->nb_refs[0] < MAX_PICTURE_REFERENCES &&
-                   pic->nb_refs[1] < MAX_PICTURE_REFERENCES);
-        if (target->display_order < pic->display_order)
-            pic->refs[0][pic->nb_refs[0]++] = target;
-        else
-            pic->refs[1][pic->nb_refs[1]++] = target;
-        ++refs;
-    }
-
-    if (in_dpb) {
-        av_assert0(pic->nb_dpb_pics < MAX_DPB_SIZE);
-        pic->dpb[pic->nb_dpb_pics++] = target;
-        ++refs;
-    }
-
-    if (prev) {
-        av_assert0(!pic->prev);
-        pic->prev = target;
-        ++refs;
-    }
-
-    target->ref_count[0] += refs;
-    target->ref_count[1] += refs;
-}
-
-static void vaapi_encode_remove_refs(AVCodecContext *avctx,
-                                     VAAPIEncodePicture *pic,
-                                     int level)
-{
-    int i;
-
-    if (pic->ref_removed[level])
-        return;
-
-    for (i = 0; i < pic->nb_refs[0]; i++) {
-        av_assert0(pic->refs[0][i]);
-        --pic->refs[0][i]->ref_count[level];
-        av_assert0(pic->refs[0][i]->ref_count[level] >= 0);
-    }
-
-    for (i = 0; i < pic->nb_refs[1]; i++) {
-        av_assert0(pic->refs[1][i]);
-        --pic->refs[1][i]->ref_count[level];
-        av_assert0(pic->refs[1][i]->ref_count[level] >= 0);
-    }
-
-    for (i = 0; i < pic->nb_dpb_pics; i++) {
-        av_assert0(pic->dpb[i]);
-        --pic->dpb[i]->ref_count[level];
-        av_assert0(pic->dpb[i]->ref_count[level] >= 0);
-    }
-
-    av_assert0(pic->prev || pic->type == PICTURE_TYPE_IDR);
-    if (pic->prev) {
-        --pic->prev->ref_count[level];
-        av_assert0(pic->prev->ref_count[level] >= 0);
-    }
-
-    pic->ref_removed[level] = 1;
-}
-
-static void vaapi_encode_set_b_pictures(AVCodecContext *avctx,
-                                        VAAPIEncodePicture *start,
-                                        VAAPIEncodePicture *end,
-                                        VAAPIEncodePicture *prev,
-                                        int current_depth,
-                                        VAAPIEncodePicture **last)
-{
-    VAAPIEncodeContext *ctx = avctx->priv_data;
-    VAAPIEncodePicture *pic, *next, *ref;
-    int i, len;
-
-    av_assert0(start && end && start != end && start->next != end);
-
-    // If we are at the maximum depth then encode all pictures as
-    // non-referenced B-pictures.  Also do this if there is exactly one
-    // picture left, since there will be nothing to reference it.
-    if (current_depth == ctx->max_b_depth || start->next->next == end) {
-        for (pic = start->next; pic; pic = pic->next) {
-            if (pic == end)
-                break;
-            pic->type    = PICTURE_TYPE_B;
-            pic->b_depth = current_depth;
-
-            vaapi_encode_add_ref(avctx, pic, start, 1, 1, 0);
-            vaapi_encode_add_ref(avctx, pic, end,   1, 1, 0);
-            vaapi_encode_add_ref(avctx, pic, prev,  0, 0, 1);
-
-            for (ref = end->refs[1][0]; ref; ref = ref->refs[1][0])
-                vaapi_encode_add_ref(avctx, pic, ref, 0, 1, 0);
-        }
-        *last = prev;
-
-    } else {
-        // Split the current list at the midpoint with a referenced
-        // B-picture, then descend into each side separately.
-        len = 0;
-        for (pic = start->next; pic != end; pic = pic->next)
-            ++len;
-        for (pic = start->next, i = 1; 2 * i < len; pic = pic->next, i++);
-
-        pic->type    = PICTURE_TYPE_B;
-        pic->b_depth = current_depth;
-
-        pic->is_reference = 1;
-
-        vaapi_encode_add_ref(avctx, pic, pic,   0, 1, 0);
-        vaapi_encode_add_ref(avctx, pic, start, 1, 1, 0);
-        vaapi_encode_add_ref(avctx, pic, end,   1, 1, 0);
-        vaapi_encode_add_ref(avctx, pic, prev,  0, 0, 1);
-
-        for (ref = end->refs[1][0]; ref; ref = ref->refs[1][0])
-            vaapi_encode_add_ref(avctx, pic, ref, 0, 1, 0);
-
-        if (i > 1)
-            vaapi_encode_set_b_pictures(avctx, start, pic, pic,
-                                        current_depth + 1, &next);
-        else
-            next = pic;
-
-        vaapi_encode_set_b_pictures(avctx, pic, end, next,
-                                    current_depth + 1, last);
-    }
-}
-
-static void vaapi_encode_add_next_prev(AVCodecContext *avctx,
-                                       VAAPIEncodePicture *pic)
-{
-    VAAPIEncodeContext *ctx = avctx->priv_data;
-    int i;
-
-    if (!pic)
-        return;
-
-    if (pic->type == PICTURE_TYPE_IDR) {
-        for (i = 0; i < ctx->nb_next_prev; i++) {
-            --ctx->next_prev[i]->ref_count[0];
-            ctx->next_prev[i] = NULL;
-        }
-        ctx->next_prev[0] = pic;
-        ++pic->ref_count[0];
-        ctx->nb_next_prev = 1;
-
-        return;
-    }
-
-    if (ctx->nb_next_prev < MAX_PICTURE_REFERENCES) {
-        ctx->next_prev[ctx->nb_next_prev++] = pic;
-        ++pic->ref_count[0];
-    } else {
-        --ctx->next_prev[0]->ref_count[0];
-        for (i = 0; i < MAX_PICTURE_REFERENCES - 1; i++)
-            ctx->next_prev[i] = ctx->next_prev[i + 1];
-        ctx->next_prev[i] = pic;
-        ++pic->ref_count[0];
-    }
-}
-
-static int vaapi_encode_pick_next(AVCodecContext *avctx,
-                                  VAAPIEncodePicture **pic_out)
-{
-    VAAPIEncodeContext *ctx = avctx->priv_data;
-    VAAPIEncodePicture *pic = NULL, *prev = NULL, *next, *start;
-    int i, b_counter, closed_gop_end;
-
-    // If there are any B-frames already queued, the next one to encode
-    // is the earliest not-yet-issued frame for which all references are
-    // available.
-    for (pic = ctx->pic_start; pic; pic = pic->next) {
-        if (pic->encode_issued)
-            continue;
-        if (pic->type != PICTURE_TYPE_B)
-            continue;
-        for (i = 0; i < pic->nb_refs[0]; i++) {
-            if (!pic->refs[0][i]->encode_issued)
-                break;
-        }
-        if (i != pic->nb_refs[0])
-            continue;
-
-        for (i = 0; i < pic->nb_refs[1]; i++) {
-            if (!pic->refs[1][i]->encode_issued)
-                break;
-        }
-        if (i == pic->nb_refs[1])
-            break;
-    }
-
-    if (pic) {
-        av_log(avctx, AV_LOG_DEBUG, "Pick B-picture at depth %d to "
-               "encode next.\n", pic->b_depth);
-        *pic_out = pic;
-        return 0;
-    }
-
-    // Find the B-per-Pth available picture to become the next picture
-    // on the top layer.
-    start = NULL;
-    b_counter = 0;
-    closed_gop_end = ctx->closed_gop ||
-                     ctx->idr_counter == ctx->gop_per_idr;
-    for (pic = ctx->pic_start; pic; pic = next) {
-        next = pic->next;
-        if (pic->encode_issued) {
-            start = pic;
-            continue;
-        }
-        // If the next available picture is force-IDR, encode it to start
-        // a new GOP immediately.
-        if (pic->force_idr)
-            break;
-        if (b_counter == ctx->b_per_p)
-            break;
-        // If this picture ends a closed GOP or starts a new GOP then it
-        // needs to be in the top layer.
-        if (ctx->gop_counter + b_counter + closed_gop_end >= ctx->gop_size)
-            break;
-        // If the picture after this one is force-IDR, we need to encode
-        // this one in the top layer.
-        if (next && next->force_idr)
-            break;
-        ++b_counter;
-    }
-
-    // At the end of the stream the last picture must be in the top layer.
-    if (!pic && ctx->end_of_stream) {
-        --b_counter;
-        pic = ctx->pic_end;
-        if (pic->encode_complete)
-            return AVERROR_EOF;
-        else if (pic->encode_issued)
-            return AVERROR(EAGAIN);
-    }
-
-    if (!pic) {
-        av_log(avctx, AV_LOG_DEBUG, "Pick nothing to encode next - "
-               "need more input for reference pictures.\n");
-        return AVERROR(EAGAIN);
-    }
-    if (ctx->input_order <= ctx->decode_delay && !ctx->end_of_stream) {
-        av_log(avctx, AV_LOG_DEBUG, "Pick nothing to encode next - "
-               "need more input for timestamps.\n");
-        return AVERROR(EAGAIN);
-    }
-
-    if (pic->force_idr) {
-        av_log(avctx, AV_LOG_DEBUG, "Pick forced IDR-picture to "
-               "encode next.\n");
-        pic->type = PICTURE_TYPE_IDR;
-        ctx->idr_counter = 1;
-        ctx->gop_counter = 1;
-
-    } else if (ctx->gop_counter + b_counter >= ctx->gop_size) {
-        if (ctx->idr_counter == ctx->gop_per_idr) {
-            av_log(avctx, AV_LOG_DEBUG, "Pick new-GOP IDR-picture to "
-                   "encode next.\n");
-            pic->type = PICTURE_TYPE_IDR;
-            ctx->idr_counter = 1;
-        } else {
-            av_log(avctx, AV_LOG_DEBUG, "Pick new-GOP I-picture to "
-                   "encode next.\n");
-            pic->type = PICTURE_TYPE_I;
-            ++ctx->idr_counter;
-        }
-        ctx->gop_counter = 1;
-
-    } else {
-        if (ctx->gop_counter + b_counter + closed_gop_end == ctx->gop_size) {
-            av_log(avctx, AV_LOG_DEBUG, "Pick group-end P-picture to "
-                   "encode next.\n");
-        } else {
-            av_log(avctx, AV_LOG_DEBUG, "Pick normal P-picture to "
-                   "encode next.\n");
-        }
-        pic->type = PICTURE_TYPE_P;
-        av_assert0(start);
-        ctx->gop_counter += 1 + b_counter;
-    }
-    pic->is_reference = 1;
-    *pic_out = pic;
-
-    vaapi_encode_add_ref(avctx, pic, pic, 0, 1, 0);
-    if (pic->type != PICTURE_TYPE_IDR) {
-        // TODO: apply both previous and forward multi reference for all vaapi encoders.
-        // And L0/L1 reference frame number can be set dynamically through query
-        // VAConfigAttribEncMaxRefFrames attribute.
-        if (avctx->codec_id == AV_CODEC_ID_AV1) {
-            for (i = 0; i < ctx->nb_next_prev; i++)
-                vaapi_encode_add_ref(avctx, pic, ctx->next_prev[i],
-                                     pic->type == PICTURE_TYPE_P,
-                                     b_counter > 0, 0);
-        } else
-            vaapi_encode_add_ref(avctx, pic, start,
-                                 pic->type == PICTURE_TYPE_P,
-                                 b_counter > 0, 0);
-
-        vaapi_encode_add_ref(avctx, pic, ctx->next_prev[ctx->nb_next_prev - 1], 0, 0, 1);
-    }
-
-    if (b_counter > 0) {
-        vaapi_encode_set_b_pictures(avctx, start, pic, pic, 1,
-                                    &prev);
-    } else {
-        prev = pic;
-    }
-    vaapi_encode_add_next_prev(avctx, prev);
-
-    return 0;
-}
-
-static int vaapi_encode_clear_old(AVCodecContext *avctx)
-{
-    VAAPIEncodeContext *ctx = avctx->priv_data;
-    VAAPIEncodePicture *pic, *prev, *next;
-
-    av_assert0(ctx->pic_start);
-
-    // Remove direct references once each picture is complete.
-    for (pic = ctx->pic_start; pic; pic = pic->next) {
-        if (pic->encode_complete && pic->next)
-            vaapi_encode_remove_refs(avctx, pic, 0);
-    }
-
-    // Remove indirect references once a picture has no direct references.
-    for (pic = ctx->pic_start; pic; pic = pic->next) {
-        if (pic->encode_complete && pic->ref_count[0] == 0)
-            vaapi_encode_remove_refs(avctx, pic, 1);
-    }
-
-    // Clear out all complete pictures with no remaining references.
-    prev = NULL;
-    for (pic = ctx->pic_start; pic; pic = next) {
-        next = pic->next;
-        if (pic->encode_complete && pic->ref_count[1] == 0) {
-            av_assert0(pic->ref_removed[0] && pic->ref_removed[1]);
-            if (prev)
-                prev->next = next;
-            else
-                ctx->pic_start = next;
-            vaapi_encode_free(avctx, pic);
-        } else {
-            prev = pic;
-        }
-    }
-
-    return 0;
-}
-
-static int vaapi_encode_check_frame(AVCodecContext *avctx,
-                                    const AVFrame *frame)
-{
-    VAAPIEncodeContext *ctx = avctx->priv_data;
-
-    if ((frame->crop_top  || frame->crop_bottom ||
-         frame->crop_left || frame->crop_right) && !ctx->crop_warned) {
-        av_log(avctx, AV_LOG_WARNING, "Cropping information on input "
-               "frames ignored due to lack of API support.\n");
-        ctx->crop_warned = 1;
-    }
-
-    if (!ctx->roi_allowed) {
-        AVFrameSideData *sd =
-            av_frame_get_side_data(frame, AV_FRAME_DATA_REGIONS_OF_INTEREST);
-
-        if (sd && !ctx->roi_warned) {
-            av_log(avctx, AV_LOG_WARNING, "ROI side data on input "
-                   "frames ignored due to lack of driver support.\n");
-            ctx->roi_warned = 1;
-        }
-    }
-
-    return 0;
-}
-
-static int vaapi_encode_send_frame(AVCodecContext *avctx, AVFrame *frame)
-{
-    HWBaseEncodeContext *base_ctx = avctx->priv_data;
-    VAAPIEncodeContext *ctx = avctx->priv_data;
-    VAAPIEncodePicture *pic;
-    int err;
-
-    if (frame) {
-        av_log(avctx, AV_LOG_DEBUG, "Input frame: %ux%u (%"PRId64").\n",
-               frame->width, frame->height, frame->pts);
-
-        err = vaapi_encode_check_frame(avctx, frame);
-        if (err < 0)
-            return err;
-
-        pic = vaapi_encode_alloc(avctx, frame);
-        if (!pic)
-            return AVERROR(ENOMEM);
-
-        pic->input_image = av_frame_alloc();
-        if (!pic->input_image) {
-            err = AVERROR(ENOMEM);
-            goto fail;
-        }
-
-        if (ctx->input_order == 0 || frame->pict_type == AV_PICTURE_TYPE_I)
-            pic->force_idr = 1;
-
-        pic->pts = frame->pts;
-        pic->duration = frame->duration;
-
-        if (avctx->flags & AV_CODEC_FLAG_COPY_OPAQUE) {
-            err = av_buffer_replace(&pic->opaque_ref, frame->opaque_ref);
-            if (err < 0)
-                goto fail;
-
-            pic->opaque = frame->opaque;
-        }
-
-        av_frame_move_ref(pic->input_image, frame);
-
-        if (ctx->input_order == 0)
-            ctx->first_pts = pic->pts;
-        if (ctx->input_order == ctx->decode_delay)
-            ctx->dts_pts_diff = pic->pts - ctx->first_pts;
-        if (ctx->output_delay > 0)
-            ctx->ts_ring[ctx->input_order %
-                        (3 * ctx->output_delay + base_ctx->async_depth)] = pic->pts;
-
-        pic->display_order = ctx->input_order;
-        ++ctx->input_order;
-
-        if (ctx->pic_start) {
-            ctx->pic_end->next = pic;
-            ctx->pic_end       = pic;
-        } else {
-            ctx->pic_start     = pic;
-            ctx->pic_end       = pic;
-        }
-
-    } else {
-        ctx->end_of_stream = 1;
-
-        // Fix timestamps if we hit end-of-stream before the initial decode
-        // delay has elapsed.
-        if (ctx->input_order < ctx->decode_delay)
-            ctx->dts_pts_diff = ctx->pic_end->pts - ctx->first_pts;
-    }
-
-    return 0;
-
-fail:
-    vaapi_encode_free(avctx, pic);
-    return err;
-}
-
-int ff_vaapi_encode_receive_packet(AVCodecContext *avctx, AVPacket *pkt)
-{
-    VAAPIEncodeContext *ctx = avctx->priv_data;
-    VAAPIEncodePicture *pic = NULL;
-    AVFrame *frame = ctx->frame;
-    int err;
-
-start:
-    /** if no B frame before repeat P frame, sent repeat P frame out. */
-    if (ctx->tail_pkt->size) {
-        for (VAAPIEncodePicture *tmp = ctx->pic_start; tmp; tmp = tmp->next) {
-            if (tmp->type == PICTURE_TYPE_B && tmp->pts < ctx->tail_pkt->pts)
-                break;
-            else if (!tmp->next) {
-                av_packet_move_ref(pkt, ctx->tail_pkt);
-                goto end;
-            }
-        }
-    }
-
-    err = ff_encode_get_frame(avctx, frame);
-    if (err < 0 && err != AVERROR_EOF)
-        return err;
-
-    if (err == AVERROR_EOF)
-        frame = NULL;
-
-    err = vaapi_encode_send_frame(avctx, frame);
-    if (err < 0)
-        return err;
-
-    if (!ctx->pic_start) {
-        if (ctx->end_of_stream)
-            return AVERROR_EOF;
-        else
-            return AVERROR(EAGAIN);
-    }
-
-    if (ctx->has_sync_buffer_func) {
-        if (av_fifo_can_write(ctx->encode_fifo)) {
-            err = vaapi_encode_pick_next(avctx, &pic);
-            if (!err) {
-                av_assert0(pic);
-                pic->encode_order = ctx->encode_order +
-                    av_fifo_can_read(ctx->encode_fifo);
-                err = vaapi_encode_issue(avctx, pic);
-                if (err < 0) {
-                    av_log(avctx, AV_LOG_ERROR, "Encode failed: %d.\n", err);
-                    return err;
-                }
-                av_fifo_write(ctx->encode_fifo, &pic, 1);
-            }
-        }
-
-        if (!av_fifo_can_read(ctx->encode_fifo))
-            return err;
-
-        // More frames can be buffered
-        if (av_fifo_can_write(ctx->encode_fifo) && !ctx->end_of_stream)
-            return AVERROR(EAGAIN);
-
-        av_fifo_read(ctx->encode_fifo, &pic, 1);
-        ctx->encode_order = pic->encode_order + 1;
-    } else {
-        err = vaapi_encode_pick_next(avctx, &pic);
-        if (err < 0)
-            return err;
-        av_assert0(pic);
-
-        pic->encode_order = ctx->encode_order++;
-
-        err = vaapi_encode_issue(avctx, pic);
-        if (err < 0) {
-            av_log(avctx, AV_LOG_ERROR, "Encode failed: %d.\n", err);
-            return err;
-        }
-    }
-
-    err = vaapi_encode_output(avctx, pic, pkt);
-    if (err < 0) {
-        av_log(avctx, AV_LOG_ERROR, "Output failed: %d.\n", err);
-        return err;
-    }
-
-    ctx->output_order = pic->encode_order;
-    vaapi_encode_clear_old(avctx);
-
-    /** loop to get an available pkt in encoder flushing. */
-    if (ctx->end_of_stream && !pkt->size)
-        goto start;
-
-end:
-    if (pkt->size)
-        av_log(avctx, AV_LOG_DEBUG, "Output packet: pts %"PRId64", dts %"PRId64", "
-               "size %d bytes.\n", pkt->pts, pkt->dts, pkt->size);
-
-    return 0;
-}
-
 static av_cold void vaapi_encode_add_global_param(AVCodecContext *avctx, int type,
                                                   void *buffer, size_t size)
 {
@@ -2188,7 +1632,8 @@ static av_cold int vaapi_encode_init_max_frame_size(AVCodecContext *avctx)
 
 static av_cold int vaapi_encode_init_gop_structure(AVCodecContext *avctx)
 {
-    VAAPIEncodeContext *ctx = avctx->priv_data;
+    HWBaseEncodeContext *base_ctx = avctx->priv_data;
+    VAAPIEncodeContext       *ctx = avctx->priv_data;
     VAStatus vas;
     VAConfigAttrib attr = { VAConfigAttribEncMaxRefFrames };
     uint32_t ref_l0, ref_l1;
@@ -2211,7 +1656,7 @@ static av_cold int vaapi_encode_init_gop_structure(AVCodecContext *avctx)
         ref_l1 = attr.value >> 16 & 0xffff;
     }
 
-    ctx->p_to_gpb = 0;
+    base_ctx->p_to_gpb = 0;
     prediction_pre_only = 0;
 
 #if VA_CHECK_VERSION(1, 9, 0)
@@ -2247,7 +1692,7 @@ static av_cold int vaapi_encode_init_gop_structure(AVCodecContext *avctx)
 
             if (attr.value & VA_PREDICTION_DIRECTION_BI_NOT_EMPTY) {
                 if (ref_l0 > 0 && ref_l1 > 0) {
-                    ctx->p_to_gpb = 1;
+                    base_ctx->p_to_gpb = 1;
                     av_log(avctx, AV_LOG_VERBOSE, "Driver does not support P-frames, "
                            "replacing them with B-frames.\n");
                 }
@@ -2259,7 +1704,7 @@ static av_cold int vaapi_encode_init_gop_structure(AVCodecContext *avctx)
     if (ctx->codec->flags & FLAG_INTRA_ONLY ||
         avctx->gop_size <= 1) {
         av_log(avctx, AV_LOG_VERBOSE, "Using intra frames only.\n");
-        ctx->gop_size = 1;
+        base_ctx->gop_size = 1;
     } else if (ref_l0 < 1) {
         av_log(avctx, AV_LOG_ERROR, "Driver does not support any "
                "reference frames.\n");
@@ -2267,41 +1712,41 @@ static av_cold int vaapi_encode_init_gop_structure(AVCodecContext *avctx)
     } else if (!(ctx->codec->flags & FLAG_B_PICTURES) ||
                ref_l1 < 1 || avctx->max_b_frames < 1 ||
                prediction_pre_only) {
-        if (ctx->p_to_gpb)
+        if (base_ctx->p_to_gpb)
            av_log(avctx, AV_LOG_VERBOSE, "Using intra and B-frames "
                   "(supported references: %d / %d).\n",
                   ref_l0, ref_l1);
         else
             av_log(avctx, AV_LOG_VERBOSE, "Using intra and P-frames "
                    "(supported references: %d / %d).\n", ref_l0, ref_l1);
-        ctx->gop_size = avctx->gop_size;
-        ctx->p_per_i  = INT_MAX;
-        ctx->b_per_p  = 0;
+        base_ctx->gop_size = avctx->gop_size;
+        base_ctx->p_per_i  = INT_MAX;
+        base_ctx->b_per_p  = 0;
     } else {
-       if (ctx->p_to_gpb)
+       if (base_ctx->p_to_gpb)
            av_log(avctx, AV_LOG_VERBOSE, "Using intra and B-frames "
                   "(supported references: %d / %d).\n",
                   ref_l0, ref_l1);
        else
            av_log(avctx, AV_LOG_VERBOSE, "Using intra, P- and B-frames "
                   "(supported references: %d / %d).\n", ref_l0, ref_l1);
-        ctx->gop_size = avctx->gop_size;
-        ctx->p_per_i  = INT_MAX;
-        ctx->b_per_p  = avctx->max_b_frames;
+        base_ctx->gop_size = avctx->gop_size;
+        base_ctx->p_per_i  = INT_MAX;
+        base_ctx->b_per_p  = avctx->max_b_frames;
         if (ctx->codec->flags & FLAG_B_PICTURE_REFERENCES) {
-            ctx->max_b_depth = FFMIN(ctx->desired_b_depth,
-                                     av_log2(ctx->b_per_p) + 1);
+            base_ctx->max_b_depth = FFMIN(ctx->desired_b_depth,
+                                          av_log2(base_ctx->b_per_p) + 1);
         } else {
-            ctx->max_b_depth = 1;
+            base_ctx->max_b_depth = 1;
         }
     }
 
     if (ctx->codec->flags & FLAG_NON_IDR_KEY_PICTURES) {
-        ctx->closed_gop  = !!(avctx->flags & AV_CODEC_FLAG_CLOSED_GOP);
-        ctx->gop_per_idr = ctx->idr_interval + 1;
+        base_ctx->closed_gop  = !!(avctx->flags & AV_CODEC_FLAG_CLOSED_GOP);
+        base_ctx->gop_per_idr = ctx->idr_interval + 1;
     } else {
-        ctx->closed_gop  = 1;
-        ctx->gop_per_idr = 1;
+        base_ctx->closed_gop  = 1;
+        base_ctx->gop_per_idr = 1;
     }
 
     return 0;
@@ -2614,7 +2059,8 @@ static av_cold int vaapi_encode_init_quality(AVCodecContext *avctx)
 static av_cold int vaapi_encode_init_roi(AVCodecContext *avctx)
 {
 #if VA_CHECK_VERSION(1, 0, 0)
-    VAAPIEncodeContext *ctx = avctx->priv_data;
+    HWBaseEncodeContext *base_ctx = avctx->priv_data;
+    VAAPIEncodeContext       *ctx = avctx->priv_data;
     VAStatus vas;
     VAConfigAttrib attr = { VAConfigAttribEncROI };
 
@@ -2629,14 +2075,14 @@ static av_cold int vaapi_encode_init_roi(AVCodecContext *avctx)
     }
 
     if (attr.value == VA_ATTRIB_NOT_SUPPORTED) {
-        ctx->roi_allowed = 0;
+        base_ctx->roi_allowed = 0;
     } else {
         VAConfigAttribValEncROI roi = {
             .value = attr.value,
         };
 
         ctx->roi_max_regions = roi.bits.num_roi_regions;
-        ctx->roi_allowed = ctx->roi_max_regions > 0 &&
+        base_ctx->roi_allowed = ctx->roi_max_regions > 0 &&
             (ctx->va_rc_mode == VA_RC_CQP ||
              roi.bits.roi_rc_qp_delta_support);
     }
@@ -2771,6 +2217,16 @@ static av_cold int vaapi_encode_create_recon_frames(AVCodecContext *avctx)
     return err;
 }
 
+static const HWEncodePictureOperation vaapi_op = {
+    .alloc  = &vaapi_encode_alloc,
+
+    .issue  = &vaapi_encode_issue,
+
+    .output = &vaapi_encode_output,
+
+    .free   = &vaapi_encode_free,
+};
+
 av_cold int ff_vaapi_encode_init(AVCodecContext *avctx)
 {
     HWBaseEncodeContext *base_ctx = avctx->priv_data;
@@ -2782,10 +2238,12 @@ av_cold int ff_vaapi_encode_init(AVCodecContext *avctx)
     ctx->va_config  = VA_INVALID_ID;
     ctx->va_context = VA_INVALID_ID;
 
+    base_ctx->op = &vaapi_op;
+
     /* If you add something that can fail above this av_frame_alloc(),
      * modify ff_vaapi_encode_close() accordingly. */
-    ctx->frame = av_frame_alloc();
-    if (!ctx->frame) {
+    base_ctx->frame = av_frame_alloc();
+    if (!base_ctx->frame) {
         return AVERROR(ENOMEM);
     }
 
@@ -2810,8 +2268,8 @@ av_cold int ff_vaapi_encode_init(AVCodecContext *avctx)
     ctx->device = (AVHWDeviceContext*)ctx->device_ref->data;
     ctx->hwctx = ctx->device->hwctx;
 
-    ctx->tail_pkt = av_packet_alloc();
-    if (!ctx->tail_pkt) {
+    base_ctx->tail_pkt = av_packet_alloc();
+    if (!base_ctx->tail_pkt) {
         err = AVERROR(ENOMEM);
         goto fail;
     }
@@ -2910,8 +2368,8 @@ av_cold int ff_vaapi_encode_init(AVCodecContext *avctx)
             goto fail;
     }
 
-    ctx->output_delay = ctx->b_per_p;
-    ctx->decode_delay = ctx->max_b_depth;
+    base_ctx->output_delay = base_ctx->b_per_p;
+    base_ctx->decode_delay = base_ctx->max_b_depth;
 
     if (ctx->codec->sequence_params_size > 0) {
         ctx->codec_sequence_params =
@@ -2966,11 +2424,11 @@ av_cold int ff_vaapi_encode_init(AVCodecContext *avctx)
     // check vaSyncBuffer function
     vas = vaSyncBuffer(ctx->hwctx->display, VA_INVALID_ID, 0);
     if (vas != VA_STATUS_ERROR_UNIMPLEMENTED) {
-        ctx->has_sync_buffer_func = 1;
-        ctx->encode_fifo = av_fifo_alloc2(base_ctx->async_depth,
-                                          sizeof(VAAPIEncodePicture *),
-                                          0);
-        if (!ctx->encode_fifo)
+        base_ctx->async_encode = 1;
+        base_ctx->encode_fifo = av_fifo_alloc2(base_ctx->async_depth,
+                                               sizeof(VAAPIEncodePicture*),
+                                               0);
+        if (!base_ctx->encode_fifo)
             return AVERROR(ENOMEM);
     }
 #endif
@@ -2983,15 +2441,16 @@ fail:
 
 av_cold int ff_vaapi_encode_close(AVCodecContext *avctx)
 {
-    VAAPIEncodeContext *ctx = avctx->priv_data;
-    VAAPIEncodePicture *pic, *next;
+    HWBaseEncodeContext *base_ctx = avctx->priv_data;
+    VAAPIEncodeContext       *ctx = avctx->priv_data;
+    HWBaseEncodePicture *pic, *next;
 
     /* We check ctx->frame to know whether ff_vaapi_encode_init()
      * has been called and va_config/va_context initialized. */
-    if (!ctx->frame)
+    if (!base_ctx->frame)
         return 0;
 
-    for (pic = ctx->pic_start; pic; pic = next) {
+    for (pic = base_ctx->pic_start; pic; pic = next) {
         next = pic->next;
         vaapi_encode_free(avctx, pic);
     }
@@ -3008,12 +2467,12 @@ av_cold int ff_vaapi_encode_close(AVCodecContext *avctx)
         ctx->va_config = VA_INVALID_ID;
     }
 
-    av_frame_free(&ctx->frame);
-    av_packet_free(&ctx->tail_pkt);
+    av_frame_free(&base_ctx->frame);
+    av_packet_free(&base_ctx->tail_pkt);
 
     av_freep(&ctx->codec_sequence_params);
     av_freep(&ctx->codec_picture_params);
-    av_fifo_freep2(&ctx->encode_fifo);
+    av_fifo_freep2(&base_ctx->encode_fifo);
 
     av_buffer_unref(&ctx->recon_frames_ref);
     av_buffer_unref(&ctx->input_frames_ref);
diff --git a/libavcodec/vaapi_encode.h b/libavcodec/vaapi_encode.h
index 02410c72ec..13ccad8e47 100644
--- a/libavcodec/vaapi_encode.h
+++ b/libavcodec/vaapi_encode.h
@@ -29,7 +29,6 @@
 
 #include "libavutil/hwcontext.h"
 #include "libavutil/hwcontext_vaapi.h"
-#include "libavutil/fifo.h"
 
 #include "avcodec.h"
 #include "hwconfig.h"
@@ -64,16 +63,7 @@ typedef struct VAAPIEncodeSlice {
 } VAAPIEncodeSlice;
 
 typedef struct VAAPIEncodePicture {
-    struct VAAPIEncodePicture *next;
-
-    int64_t         display_order;
-    int64_t         encode_order;
-    int64_t         pts;
-    int64_t         duration;
-    int             force_idr;
-
-    void           *opaque;
-    AVBufferRef    *opaque_ref;
+    HWBaseEncodePicture base;
 
 #if VA_CHECK_VERSION(1, 0, 0)
     // ROI regions.
@@ -82,15 +72,7 @@ typedef struct VAAPIEncodePicture {
     void           *roi;
 #endif
 
-    int             type;
-    int             b_depth;
-    int             encode_issued;
-    int             encode_complete;
-
-    AVFrame        *input_image;
     VASurfaceID     input_surface;
-
-    AVFrame        *recon_image;
     VASurfaceID     recon_surface;
 
     int          nb_param_buffers;
@@ -100,31 +82,8 @@ typedef struct VAAPIEncodePicture {
     VABufferID     *output_buffer_ref;
     VABufferID      output_buffer;
 
-    void           *priv_data;
     void           *codec_picture_params;
 
-    // Whether this picture is a reference picture.
-    int             is_reference;
-
-    // The contents of the DPB after this picture has been decoded.
-    // This will contain the picture itself if it is a reference picture,
-    // but not if it isn't.
-    int                     nb_dpb_pics;
-    struct VAAPIEncodePicture *dpb[MAX_DPB_SIZE];
-    // The reference pictures used in decoding this picture. If they are
-    // used by later pictures they will also appear in the DPB. ref[0][] for
-    // previous reference frames. ref[1][] for future reference frames.
-    int                     nb_refs[MAX_REFERENCE_LIST_NUM];
-    struct VAAPIEncodePicture *refs[MAX_REFERENCE_LIST_NUM][MAX_PICTURE_REFERENCES];
-    // The previous reference picture in encode order.  Must be in at least
-    // one of the reference list and DPB list.
-    struct VAAPIEncodePicture *prev;
-    // Reference count for other pictures referring to this one through
-    // the above pointers, directly from incomplete pictures and indirectly
-    // through completed pictures.
-    int             ref_count[2];
-    int             ref_removed[2];
-
     int          nb_slices;
     VAAPIEncodeSlice *slices;
 
@@ -298,30 +257,6 @@ typedef struct VAAPIEncodeContext {
     // structure (VAEncPictureParameterBuffer*).
     void           *codec_picture_params;
 
-    // Current encoding window, in display (input) order.
-    VAAPIEncodePicture *pic_start, *pic_end;
-    // The next picture to use as the previous reference picture in
-    // encoding order. Order from small to large in encoding order.
-    VAAPIEncodePicture *next_prev[MAX_PICTURE_REFERENCES];
-    int                 nb_next_prev;
-
-    // Next input order index (display order).
-    int64_t         input_order;
-    // Number of frames that output is behind input.
-    int64_t         output_delay;
-    // Next encode order index.
-    int64_t         encode_order;
-    // Number of frames decode output will need to be delayed.
-    int64_t         decode_delay;
-    // Next output order index (in encode order).
-    int64_t         output_order;
-
-    // Timestamp handling.
-    int64_t         first_pts;
-    int64_t         dts_pts_diff;
-    int64_t         ts_ring[MAX_REORDER_DELAY * 3 +
-                            MAX_ASYNC_DEPTH];
-
     // Slice structure.
     int slice_block_rows;
     int slice_block_cols;
@@ -340,41 +275,12 @@ typedef struct VAAPIEncodeContext {
     // Location of the i-th tile row boundary.
     int row_bd[MAX_TILE_ROWS + 1];
 
-    // Frame type decision.
-    int gop_size;
-    int closed_gop;
-    int gop_per_idr;
-    int p_per_i;
-    int max_b_depth;
-    int b_per_p;
-    int force_idr;
-    int idr_counter;
-    int gop_counter;
-    int end_of_stream;
-    int p_to_gpb;
-
-    // Whether the driver supports ROI at all.
-    int             roi_allowed;
     // Maximum number of regions supported by the driver.
     int             roi_max_regions;
     // Quantisation range for offset calculations.  Set by codec-specific
     // code, as it may change based on parameters.
     int             roi_quant_range;
 
-    // The encoder does not support cropping information, so warn about
-    // it the first time we encounter any nonzero crop fields.
-    int             crop_warned;
-    // If the driver does not support ROI then warn the first time we
-    // encounter a frame with ROI side data.
-    int             roi_warned;
-
-    AVFrame         *frame;
-
-    // Whether the driver support vaSyncBuffer
-    int             has_sync_buffer_func;
-    // Store buffered pic
-    AVFifo          *encode_fifo;
-
     /** Head data for current output pkt, used only for AV1. */
     //void  *header_data;
     //size_t header_data_size;
@@ -384,9 +290,6 @@ typedef struct VAAPIEncodeContext {
      * This is a RefStruct reference.
      */
     VABufferID     *coded_buffer_ref;
-
-    /** Tail data of a pic, now only used for av1 repeat frame header. */
-    AVPacket        *tail_pkt;
 } VAAPIEncodeContext;
 
 typedef struct VAAPIEncodeType {
@@ -468,9 +371,6 @@ typedef struct VAAPIEncodeType {
                                  char *data, size_t *data_len);
 } VAAPIEncodeType;
 
-
-int ff_vaapi_encode_receive_packet(AVCodecContext *avctx, AVPacket *pkt);
-
 int ff_vaapi_encode_init(AVCodecContext *avctx);
 int ff_vaapi_encode_close(AVCodecContext *avctx);
 
diff --git a/libavcodec/vaapi_encode_av1.c b/libavcodec/vaapi_encode_av1.c
index a5a475dca6..f12fe54cff 100644
--- a/libavcodec/vaapi_encode_av1.c
+++ b/libavcodec/vaapi_encode_av1.c
@@ -360,6 +360,7 @@ static int vaapi_encode_av1_write_sequence_header(AVCodecContext *avctx,
 
 static int vaapi_encode_av1_init_sequence_params(AVCodecContext *avctx)
 {
+    HWBaseEncodeContext         *base_ctx = avctx->priv_data;
     VAAPIEncodeContext               *ctx = avctx->priv_data;
     VAAPIEncodeAV1Context           *priv = avctx->priv_data;
     AV1RawOBU                     *sh_obu = &priv->sh;
@@ -441,8 +442,8 @@ static int vaapi_encode_av1_init_sequence_params(AVCodecContext *avctx)
     vseq->seq_level_idx           = sh->seq_level_idx[0];
     vseq->seq_tier                = sh->seq_tier[0];
     vseq->order_hint_bits_minus_1 = sh->order_hint_bits_minus_1;
-    vseq->intra_period            = ctx->gop_size;
-    vseq->ip_period               = ctx->b_per_p + 1;
+    vseq->intra_period            = base_ctx->gop_size;
+    vseq->ip_period               = base_ctx->b_per_p + 1;
 
     vseq->seq_fields.bits.enable_order_hint = sh->enable_order_hint;
 
@@ -465,16 +466,17 @@ end:
 }
 
 static int vaapi_encode_av1_init_picture_params(AVCodecContext *avctx,
-                                                VAAPIEncodePicture *pic)
+                                                VAAPIEncodePicture *vaapi_pic)
 {
     VAAPIEncodeContext              *ctx = avctx->priv_data;
     VAAPIEncodeAV1Context          *priv = avctx->priv_data;
+    const HWBaseEncodePicture       *pic = &vaapi_pic->base;
     VAAPIEncodeAV1Picture          *hpic = pic->priv_data;
     AV1RawOBU                    *fh_obu = &priv->fh;
     AV1RawFrameHeader                *fh = &fh_obu->obu.frame.header;
-    VAEncPictureParameterBufferAV1 *vpic = pic->codec_picture_params;
+    VAEncPictureParameterBufferAV1 *vpic = vaapi_pic->codec_picture_params;
     CodedBitstreamFragment          *obu = &priv->current_obu;
-    VAAPIEncodePicture    *ref;
+    HWBaseEncodePicture    *ref;
     VAAPIEncodeAV1Picture *href;
     int slot, i;
     int ret;
@@ -482,8 +484,8 @@ static int vaapi_encode_av1_init_picture_params(AVCodecContext *avctx,
         { 1, 0, 0, 0, -1, 0, -1, -1 };
 
     memset(fh_obu, 0, sizeof(*fh_obu));
-    pic->nb_slices = priv->tile_groups;
-    pic->non_independent_frame = pic->encode_order < pic->display_order;
+    vaapi_pic->nb_slices = priv->tile_groups;
+    vaapi_pic->non_independent_frame = pic->encode_order < pic->display_order;
     fh_obu->header.obu_type = AV1_OBU_FRAME_HEADER;
     fh_obu->header.obu_has_size_field = 1;
 
@@ -601,8 +603,8 @@ static int vaapi_encode_av1_init_picture_params(AVCodecContext *avctx,
     vpic->frame_width_minus_1  = fh->frame_width_minus_1;
     vpic->frame_height_minus_1 = fh->frame_height_minus_1;
     vpic->primary_ref_frame    = fh->primary_ref_frame;
-    vpic->reconstructed_frame  = pic->recon_surface;
-    vpic->coded_buf            = pic->output_buffer;
+    vpic->reconstructed_frame  = vaapi_pic->recon_surface;
+    vpic->coded_buf            = vaapi_pic->output_buffer;
     vpic->tile_cols            = fh->tile_cols;
     vpic->tile_rows            = fh->tile_rows;
     vpic->order_hint           = fh->order_hint;
@@ -630,12 +632,12 @@ static int vaapi_encode_av1_init_picture_params(AVCodecContext *avctx,
 
     for (i = 0; i < MAX_REFERENCE_LIST_NUM; i++) {
         for (int j = 0; j < pic->nb_refs[i]; j++) {
-            VAAPIEncodePicture *ref_pic = pic->refs[i][j];
+            HWBaseEncodePicture *ref_pic = pic->refs[i][j];
 
             slot = ((VAAPIEncodeAV1Picture*)ref_pic->priv_data)->slot;
             av_assert0(vpic->reference_frames[slot] == VA_INVALID_SURFACE);
 
-            vpic->reference_frames[slot] = ref_pic->recon_surface;
+            vpic->reference_frames[slot] = ((VAAPIEncodePicture *)ref_pic)->recon_surface;
         }
     }
 
@@ -752,7 +754,7 @@ static int vaapi_encode_av1_init_slice_params(AVCodecContext *avctx,
 }
 
 static int vaapi_encode_av1_write_picture_header(AVCodecContext *avctx,
-                                                 VAAPIEncodePicture *pic,
+                                                 VAAPIEncodePicture *vaapi_pic,
                                                  char *data, size_t *data_len)
 {
     VAAPIEncodeAV1Context     *priv = avctx->priv_data;
@@ -760,10 +762,11 @@ static int vaapi_encode_av1_write_picture_header(AVCodecContext *avctx,
     CodedBitstreamAV1Context *cbctx = priv->cbc->priv_data;
     AV1RawOBU               *fh_obu = &priv->fh;
     AV1RawFrameHeader       *rep_fh = &fh_obu->obu.frame_header;
+    const HWBaseEncodePicture  *pic = &vaapi_pic->base;
     VAAPIEncodeAV1Picture *href;
     int ret = 0;
 
-    pic->tail_size = 0;
+    vaapi_pic->tail_size = 0;
     /** Pack repeat frame header. */
     if (pic->display_order > pic->encode_order) {
         memset(fh_obu, 0, sizeof(*fh_obu));
@@ -785,11 +788,11 @@ static int vaapi_encode_av1_write_picture_header(AVCodecContext *avctx,
         if (ret < 0)
             goto end;
 
-        ret = vaapi_encode_av1_write_obu(avctx, pic->tail_data, &pic->tail_size, obu);
+        ret = vaapi_encode_av1_write_obu(avctx, vaapi_pic->tail_data, &vaapi_pic->tail_size, obu);
         if (ret < 0)
             goto end;
 
-        pic->tail_size /= 8;
+        vaapi_pic->tail_size /= 8;
     }
 
     memcpy(data, &priv->fh_data, MAX_PARAM_BUFFER_SIZE * sizeof(char));
@@ -1038,7 +1041,7 @@ const FFCodec ff_av1_vaapi_encoder = {
     .p.id           = AV_CODEC_ID_AV1,
     .priv_data_size = sizeof(VAAPIEncodeAV1Context),
     .init           = &vaapi_encode_av1_init,
-    FF_CODEC_RECEIVE_PACKET_CB(&ff_vaapi_encode_receive_packet),
+    FF_CODEC_RECEIVE_PACKET_CB(&ff_hw_base_encode_receive_packet),
     .close          = &vaapi_encode_av1_close,
     .p.priv_class   = &vaapi_encode_av1_class,
     .p.capabilities = AV_CODEC_CAP_DELAY | AV_CODEC_CAP_HARDWARE |
diff --git a/libavcodec/vaapi_encode_h264.c b/libavcodec/vaapi_encode_h264.c
index ebb1760cd3..9182483762 100644
--- a/libavcodec/vaapi_encode_h264.c
+++ b/libavcodec/vaapi_encode_h264.c
@@ -233,7 +233,7 @@ static int vaapi_encode_h264_write_extra_header(AVCodecContext *avctx,
                 goto fail;
         }
         if (priv->sei_needed & SEI_TIMING) {
-            if (pic->type == PICTURE_TYPE_IDR) {
+            if (pic->base.type == PICTURE_TYPE_IDR) {
                 err = ff_cbs_sei_add_message(priv->cbc, au, 1,
                                              SEI_TYPE_BUFFERING_PERIOD,
                                              &priv->sei_buffering_period, NULL);
@@ -295,6 +295,7 @@ fail:
 
 static int vaapi_encode_h264_init_sequence_params(AVCodecContext *avctx)
 {
+    HWBaseEncodeContext          *base_ctx = avctx->priv_data;
     VAAPIEncodeContext                *ctx = avctx->priv_data;
     VAAPIEncodeH264Context           *priv = avctx->priv_data;
     H264RawSPS                        *sps = &priv->raw_sps;
@@ -326,18 +327,18 @@ static int vaapi_encode_h264_init_sequence_params(AVCodecContext *avctx)
         sps->constraint_set1_flag = 1;
 
     if (avctx->profile == AV_PROFILE_H264_HIGH || avctx->profile == AV_PROFILE_H264_HIGH_10)
-        sps->constraint_set3_flag = ctx->gop_size == 1;
+        sps->constraint_set3_flag = base_ctx->gop_size == 1;
 
     if (avctx->profile == AV_PROFILE_H264_MAIN ||
         avctx->profile == AV_PROFILE_H264_HIGH || avctx->profile == AV_PROFILE_H264_HIGH_10) {
         sps->constraint_set4_flag = 1;
-        sps->constraint_set5_flag = ctx->b_per_p == 0;
+        sps->constraint_set5_flag = base_ctx->b_per_p == 0;
     }
 
-    if (ctx->gop_size == 1)
+    if (base_ctx->gop_size == 1)
         priv->dpb_frames = 0;
     else
-        priv->dpb_frames = 1 + ctx->max_b_depth;
+        priv->dpb_frames = 1 + base_ctx->max_b_depth;
 
     if (avctx->level != AV_LEVEL_UNKNOWN) {
         sps->level_idc = avctx->level;
@@ -374,7 +375,7 @@ static int vaapi_encode_h264_init_sequence_params(AVCodecContext *avctx)
     sps->bit_depth_chroma_minus8 = bit_depth - 8;
 
     sps->log2_max_frame_num_minus4 = 4;
-    sps->pic_order_cnt_type        = ctx->max_b_depth ? 0 : 2;
+    sps->pic_order_cnt_type        = base_ctx->max_b_depth ? 0 : 2;
     if (sps->pic_order_cnt_type == 0) {
         sps->log2_max_pic_order_cnt_lsb_minus4 = 4;
     }
@@ -501,8 +502,8 @@ static int vaapi_encode_h264_init_sequence_params(AVCodecContext *avctx)
     sps->vui.motion_vectors_over_pic_boundaries_flag = 1;
     sps->vui.log2_max_mv_length_horizontal = 15;
     sps->vui.log2_max_mv_length_vertical   = 15;
-    sps->vui.max_num_reorder_frames        = ctx->max_b_depth;
-    sps->vui.max_dec_frame_buffering       = ctx->max_b_depth + 1;
+    sps->vui.max_num_reorder_frames        = base_ctx->max_b_depth;
+    sps->vui.max_dec_frame_buffering       = base_ctx->max_b_depth + 1;
 
     pps->nal_unit_header.nal_ref_idc = 3;
     pps->nal_unit_header.nal_unit_type = H264_NAL_PPS;
@@ -535,9 +536,9 @@ static int vaapi_encode_h264_init_sequence_params(AVCodecContext *avctx)
     *vseq = (VAEncSequenceParameterBufferH264) {
         .seq_parameter_set_id = sps->seq_parameter_set_id,
         .level_idc        = sps->level_idc,
-        .intra_period     = ctx->gop_size,
-        .intra_idr_period = ctx->gop_size,
-        .ip_period        = ctx->b_per_p + 1,
+        .intra_period     = base_ctx->gop_size,
+        .intra_idr_period = base_ctx->gop_size,
+        .ip_period        = base_ctx->b_per_p + 1,
 
         .bits_per_second       = ctx->va_bit_rate,
         .max_num_ref_frames    = sps->max_num_ref_frames,
@@ -619,14 +620,15 @@ static int vaapi_encode_h264_init_sequence_params(AVCodecContext *avctx)
 }
 
 static int vaapi_encode_h264_init_picture_params(AVCodecContext *avctx,
-                                                 VAAPIEncodePicture *pic)
+                                                 VAAPIEncodePicture *vaapi_pic)
 {
-    VAAPIEncodeContext               *ctx = avctx->priv_data;
+    HWBaseEncodeContext         *base_ctx = avctx->priv_data;
     VAAPIEncodeH264Context          *priv = avctx->priv_data;
+    const HWBaseEncodePicture        *pic = &vaapi_pic->base;
     VAAPIEncodeH264Picture          *hpic = pic->priv_data;
-    VAAPIEncodePicture              *prev = pic->prev;
+    HWBaseEncodePicture             *prev = pic->prev;
     VAAPIEncodeH264Picture         *hprev = prev ? prev->priv_data : NULL;
-    VAEncPictureParameterBufferH264 *vpic = pic->codec_picture_params;
+    VAEncPictureParameterBufferH264 *vpic = vaapi_pic->codec_picture_params;
     int i, j = 0;
 
     if (pic->type == PICTURE_TYPE_IDR) {
@@ -662,7 +664,7 @@ static int vaapi_encode_h264_init_picture_params(AVCodecContext *avctx,
         hpic->pic_order_cnt *= 2;
     }
 
-    hpic->dpb_delay     = pic->display_order - pic->encode_order + ctx->max_b_depth;
+    hpic->dpb_delay     = pic->display_order - pic->encode_order + base_ctx->max_b_depth;
     hpic->cpb_delay     = pic->encode_order - hpic->last_idr_frame;
 
     if (priv->aud) {
@@ -699,7 +701,7 @@ static int vaapi_encode_h264_init_picture_params(AVCodecContext *avctx,
         priv->sei_recovery_point = (H264RawSEIRecoveryPoint) {
             .recovery_frame_cnt = 0,
             .exact_match_flag   = 1,
-            .broken_link_flag   = ctx->b_per_p > 0,
+            .broken_link_flag   = base_ctx->b_per_p > 0,
         };
 
         priv->sei_needed |= SEI_RECOVERY_POINT;
@@ -722,7 +724,7 @@ static int vaapi_encode_h264_init_picture_params(AVCodecContext *avctx,
     }
 
     vpic->CurrPic = (VAPictureH264) {
-        .picture_id          = pic->recon_surface,
+        .picture_id          = vaapi_pic->recon_surface,
         .frame_idx           = hpic->frame_num,
         .flags               = 0,
         .TopFieldOrderCnt    = hpic->pic_order_cnt,
@@ -730,14 +732,14 @@ static int vaapi_encode_h264_init_picture_params(AVCodecContext *avctx,
     };
     for (int k = 0; k < MAX_REFERENCE_LIST_NUM; k++) {
         for (i = 0; i < pic->nb_refs[k]; i++) {
-            VAAPIEncodePicture      *ref = pic->refs[k][i];
+            HWBaseEncodePicture    *ref = pic->refs[k][i];
             VAAPIEncodeH264Picture *href;
 
             av_assert0(ref && ref->encode_order < pic->encode_order);
             href = ref->priv_data;
 
             vpic->ReferenceFrames[j++] = (VAPictureH264) {
-                .picture_id          = ref->recon_surface,
+                .picture_id          = ((VAAPIEncodePicture *)ref)->recon_surface,
                 .frame_idx           = href->frame_num,
                 .flags               = VA_PICTURE_H264_SHORT_TERM_REFERENCE,
                 .TopFieldOrderCnt    = href->pic_order_cnt,
@@ -753,7 +755,7 @@ static int vaapi_encode_h264_init_picture_params(AVCodecContext *avctx,
         };
     }
 
-    vpic->coded_buf = pic->output_buffer;
+    vpic->coded_buf = vaapi_pic->output_buffer;
 
     vpic->frame_num = hpic->frame_num;
 
@@ -764,12 +766,13 @@ static int vaapi_encode_h264_init_picture_params(AVCodecContext *avctx,
 }
 
 static void vaapi_encode_h264_default_ref_pic_list(AVCodecContext *avctx,
-                                                   VAAPIEncodePicture *pic,
+                                                   VAAPIEncodePicture *vaapi_pic,
                                                    VAAPIEncodePicture **rpl0,
                                                    VAAPIEncodePicture **rpl1,
                                                    int *rpl_size)
 {
-    VAAPIEncodePicture *prev;
+    HWBaseEncodePicture *pic = &vaapi_pic->base;
+    HWBaseEncodePicture *prev;
     VAAPIEncodeH264Picture *hp, *hn, *hc;
     int i, j, n = 0;
 
@@ -783,17 +786,17 @@ static void vaapi_encode_h264_default_ref_pic_list(AVCodecContext *avctx,
 
         if (pic->type == PICTURE_TYPE_P) {
             for (j = n; j > 0; j--) {
-                hc = rpl0[j - 1]->priv_data;
+                hc = rpl0[j - 1]->base.priv_data;
                 av_assert0(hc->frame_num != hn->frame_num);
                 if (hc->frame_num > hn->frame_num)
                     break;
                 rpl0[j] = rpl0[j - 1];
             }
-            rpl0[j] = prev->dpb[i];
+            rpl0[j] = (VAAPIEncodePicture *)prev->dpb[i];
 
         } else if (pic->type == PICTURE_TYPE_B) {
             for (j = n; j > 0; j--) {
-                hc = rpl0[j - 1]->priv_data;
+                hc = rpl0[j - 1]->base.priv_data;
                 av_assert0(hc->pic_order_cnt != hp->pic_order_cnt);
                 if (hc->pic_order_cnt < hp->pic_order_cnt) {
                     if (hn->pic_order_cnt > hp->pic_order_cnt ||
@@ -805,10 +808,10 @@ static void vaapi_encode_h264_default_ref_pic_list(AVCodecContext *avctx,
                 }
                 rpl0[j] = rpl0[j - 1];
             }
-            rpl0[j] = prev->dpb[i];
+            rpl0[j] = (VAAPIEncodePicture *)prev->dpb[i];
 
             for (j = n; j > 0; j--) {
-                hc = rpl1[j - 1]->priv_data;
+                hc = rpl1[j - 1]->base.priv_data;
                 av_assert0(hc->pic_order_cnt != hp->pic_order_cnt);
                 if (hc->pic_order_cnt > hp->pic_order_cnt) {
                     if (hn->pic_order_cnt < hp->pic_order_cnt ||
@@ -820,7 +823,7 @@ static void vaapi_encode_h264_default_ref_pic_list(AVCodecContext *avctx,
                 }
                 rpl1[j] = rpl1[j - 1];
             }
-            rpl1[j] = prev->dpb[i];
+            rpl1[j] = (VAAPIEncodePicture *)prev->dpb[i];
         }
 
         ++n;
@@ -840,7 +843,7 @@ static void vaapi_encode_h264_default_ref_pic_list(AVCodecContext *avctx,
         av_log(avctx, AV_LOG_DEBUG, "Default RefPicList0 for fn=%d/poc=%d:",
                hp->frame_num, hp->pic_order_cnt);
         for (i = 0; i < n; i++) {
-            hn = rpl0[i]->priv_data;
+            hn = rpl0[i]->base.priv_data;
             av_log(avctx, AV_LOG_DEBUG, "  fn=%d/poc=%d",
                    hn->frame_num, hn->pic_order_cnt);
         }
@@ -850,7 +853,7 @@ static void vaapi_encode_h264_default_ref_pic_list(AVCodecContext *avctx,
         av_log(avctx, AV_LOG_DEBUG, "Default RefPicList1 for fn=%d/poc=%d:",
                hp->frame_num, hp->pic_order_cnt);
         for (i = 0; i < n; i++) {
-            hn = rpl1[i]->priv_data;
+            hn = rpl1[i]->base.priv_data;
             av_log(avctx, AV_LOG_DEBUG, "  fn=%d/poc=%d",
                    hn->frame_num, hn->pic_order_cnt);
         }
@@ -861,16 +864,17 @@ static void vaapi_encode_h264_default_ref_pic_list(AVCodecContext *avctx,
 }
 
 static int vaapi_encode_h264_init_slice_params(AVCodecContext *avctx,
-                                               VAAPIEncodePicture *pic,
+                                               VAAPIEncodePicture *vaapi_pic,
                                                VAAPIEncodeSlice *slice)
 {
     VAAPIEncodeH264Context          *priv = avctx->priv_data;
+    const HWBaseEncodePicture        *pic = &vaapi_pic->base;
     VAAPIEncodeH264Picture          *hpic = pic->priv_data;
-    VAAPIEncodePicture              *prev = pic->prev;
+    HWBaseEncodePicture             *prev = pic->prev;
     H264RawSPS                       *sps = &priv->raw_sps;
     H264RawPPS                       *pps = &priv->raw_pps;
     H264RawSliceHeader                *sh = &priv->raw_slice.header;
-    VAEncPictureParameterBufferH264 *vpic = pic->codec_picture_params;
+    VAEncPictureParameterBufferH264 *vpic = vaapi_pic->codec_picture_params;
     VAEncSliceParameterBufferH264 *vslice = slice->codec_slice_params;
     int i, j;
 
@@ -903,7 +907,7 @@ static int vaapi_encode_h264_init_slice_params(AVCodecContext *avctx,
         sh->slice_qp_delta = priv->fixed_qp_idr - (pps->pic_init_qp_minus26 + 26);
 
     if (pic->is_reference && pic->type != PICTURE_TYPE_IDR) {
-        VAAPIEncodePicture *discard_list[MAX_DPB_SIZE];
+        HWBaseEncodePicture *discard_list[MAX_DPB_SIZE];
         int discard = 0, keep = 0;
 
         // Discard everything which is in the DPB of the previous frame but
@@ -944,14 +948,14 @@ static int vaapi_encode_h264_init_slice_params(AVCodecContext *avctx,
         VAAPIEncodeH264Picture *href;
         int n;
 
-        vaapi_encode_h264_default_ref_pic_list(avctx, pic,
+        vaapi_encode_h264_default_ref_pic_list(avctx, vaapi_pic,
                                                def_l0, def_l1, &n);
 
         if (pic->type == PICTURE_TYPE_P) {
             int need_rplm = 0;
             for (i = 0; i < pic->nb_refs[0]; i++) {
                 av_assert0(pic->refs[0][i]);
-                if (pic->refs[0][i] != def_l0[i])
+                if (pic->refs[0][i] != (HWBaseEncodePicture *)def_l0[i])
                     need_rplm = 1;
             }
 
@@ -982,7 +986,7 @@ static int vaapi_encode_h264_init_slice_params(AVCodecContext *avctx,
                 av_assert0(pic->refs[0][i]);
                 href = pic->refs[0][i]->priv_data;
                 av_assert0(href->pic_order_cnt < hpic->pic_order_cnt);
-                if (pic->refs[0][i] != def_l0[n0])
+                if (pic->refs[0][i] != (HWBaseEncodePicture *)def_l0[n0])
                     need_rplm_l0 = 1;
                 ++n0;
             }
@@ -991,7 +995,7 @@ static int vaapi_encode_h264_init_slice_params(AVCodecContext *avctx,
                 av_assert0(pic->refs[1][i]);
                 href = pic->refs[1][i]->priv_data;
                 av_assert0(href->pic_order_cnt > hpic->pic_order_cnt);
-                if (pic->refs[1][i] != def_l1[n1])
+                if (pic->refs[1][i] != (HWBaseEncodePicture *)def_l1[n1])
                     need_rplm_l1 = 1;
                 ++n1;
             }
@@ -1380,7 +1384,7 @@ const FFCodec ff_h264_vaapi_encoder = {
     .p.id           = AV_CODEC_ID_H264,
     .priv_data_size = sizeof(VAAPIEncodeH264Context),
     .init           = &vaapi_encode_h264_init,
-    FF_CODEC_RECEIVE_PACKET_CB(&ff_vaapi_encode_receive_packet),
+    FF_CODEC_RECEIVE_PACKET_CB(&ff_hw_base_encode_receive_packet),
     .close          = &vaapi_encode_h264_close,
     .p.priv_class   = &vaapi_encode_h264_class,
     .p.capabilities = AV_CODEC_CAP_DELAY | AV_CODEC_CAP_HARDWARE |
diff --git a/libavcodec/vaapi_encode_h265.c b/libavcodec/vaapi_encode_h265.c
index 77bd5e31af..1f9d9c56a4 100644
--- a/libavcodec/vaapi_encode_h265.c
+++ b/libavcodec/vaapi_encode_h265.c
@@ -259,6 +259,7 @@ fail:
 
 static int vaapi_encode_h265_init_sequence_params(AVCodecContext *avctx)
 {
+    HWBaseEncodeContext          *base_ctx = avctx->priv_data;
     VAAPIEncodeContext                *ctx = avctx->priv_data;
     VAAPIEncodeH265Context           *priv = avctx->priv_data;
     H265RawVPS                        *vps = &priv->raw_vps;
@@ -340,7 +341,7 @@ static int vaapi_encode_h265_init_sequence_params(AVCodecContext *avctx)
     ptl->general_max_420chroma_constraint_flag  = chroma_format <= 1;
     ptl->general_max_monochrome_constraint_flag = chroma_format == 0;
 
-    ptl->general_intra_constraint_flag = ctx->gop_size == 1;
+    ptl->general_intra_constraint_flag = base_ctx->gop_size == 1;
     ptl->general_one_picture_only_constraint_flag = 0;
 
     ptl->general_lower_bit_rate_constraint_flag = 1;
@@ -353,7 +354,7 @@ static int vaapi_encode_h265_init_sequence_params(AVCodecContext *avctx)
         level = ff_h265_guess_level(ptl, avctx->bit_rate,
                                     ctx->surface_width, ctx->surface_height,
                                     ctx->nb_slices, ctx->tile_rows, ctx->tile_cols,
-                                    (ctx->b_per_p > 0) + 1);
+                                    (base_ctx->b_per_p > 0) + 1);
         if (level) {
             av_log(avctx, AV_LOG_VERBOSE, "Using level %s.\n", level->name);
             ptl->general_level_idc = level->level_idc;
@@ -367,8 +368,8 @@ static int vaapi_encode_h265_init_sequence_params(AVCodecContext *avctx)
     }
 
     vps->vps_sub_layer_ordering_info_present_flag = 0;
-    vps->vps_max_dec_pic_buffering_minus1[0]      = ctx->max_b_depth + 1;
-    vps->vps_max_num_reorder_pics[0]              = ctx->max_b_depth;
+    vps->vps_max_dec_pic_buffering_minus1[0]      = base_ctx->max_b_depth + 1;
+    vps->vps_max_num_reorder_pics[0]              = base_ctx->max_b_depth;
     vps->vps_max_latency_increase_plus1[0]        = 0;
 
     vps->vps_max_layer_id             = 0;
@@ -642,9 +643,9 @@ static int vaapi_encode_h265_init_sequence_params(AVCodecContext *avctx)
         .general_level_idc   = vps->profile_tier_level.general_level_idc,
         .general_tier_flag   = vps->profile_tier_level.general_tier_flag,
 
-        .intra_period     = ctx->gop_size,
-        .intra_idr_period = ctx->gop_size,
-        .ip_period        = ctx->b_per_p + 1,
+        .intra_period     = base_ctx->gop_size,
+        .intra_idr_period = base_ctx->gop_size,
+        .ip_period        = base_ctx->b_per_p + 1,
         .bits_per_second  = ctx->va_bit_rate,
 
         .pic_width_in_luma_samples  = sps->pic_width_in_luma_samples,
@@ -755,14 +756,15 @@ static int vaapi_encode_h265_init_sequence_params(AVCodecContext *avctx)
 }
 
 static int vaapi_encode_h265_init_picture_params(AVCodecContext *avctx,
-                                                 VAAPIEncodePicture *pic)
+                                                 VAAPIEncodePicture *vaapi_pic)
 {
-    VAAPIEncodeContext               *ctx = avctx->priv_data;
+    HWBaseEncodeContext         *base_ctx = avctx->priv_data;
     VAAPIEncodeH265Context          *priv = avctx->priv_data;
+    HWBaseEncodePicture              *pic = &vaapi_pic->base;
     VAAPIEncodeH265Picture          *hpic = pic->priv_data;
-    VAAPIEncodePicture              *prev = pic->prev;
+    HWBaseEncodePicture             *prev = pic->prev;
     VAAPIEncodeH265Picture         *hprev = prev ? prev->priv_data : NULL;
-    VAEncPictureParameterBufferHEVC *vpic = pic->codec_picture_params;
+    VAEncPictureParameterBufferHEVC *vpic = vaapi_pic->codec_picture_params;
     int i, j = 0;
 
     if (pic->type == PICTURE_TYPE_IDR) {
@@ -787,13 +789,13 @@ static int vaapi_encode_h265_init_picture_params(AVCodecContext *avctx,
             hpic->slice_type     = HEVC_SLICE_P;
             hpic->pic_type       = 1;
         } else {
-            VAAPIEncodePicture *irap_ref;
+            HWBaseEncodePicture *irap_ref;
             av_assert0(pic->refs[0][0] && pic->refs[1][0]);
             for (irap_ref = pic; irap_ref; irap_ref = irap_ref->refs[1][0]) {
                 if (irap_ref->type == PICTURE_TYPE_I)
                     break;
             }
-            if (pic->b_depth == ctx->max_b_depth) {
+            if (pic->b_depth == base_ctx->max_b_depth) {
                 hpic->slice_nal_unit = irap_ref ? HEVC_NAL_RASL_N
                                                 : HEVC_NAL_TRAIL_N;
             } else {
@@ -909,21 +911,21 @@ static int vaapi_encode_h265_init_picture_params(AVCodecContext *avctx,
     }
 
     vpic->decoded_curr_pic = (VAPictureHEVC) {
-        .picture_id    = pic->recon_surface,
+        .picture_id    = vaapi_pic->recon_surface,
         .pic_order_cnt = hpic->pic_order_cnt,
         .flags         = 0,
     };
 
     for (int k = 0; k < MAX_REFERENCE_LIST_NUM; k++) {
         for (i = 0; i < pic->nb_refs[k]; i++) {
-            VAAPIEncodePicture      *ref = pic->refs[k][i];
+            HWBaseEncodePicture    *ref = pic->refs[k][i];
             VAAPIEncodeH265Picture *href;
 
             av_assert0(ref && ref->encode_order < pic->encode_order);
             href = ref->priv_data;
 
             vpic->reference_frames[j++] = (VAPictureHEVC) {
-                .picture_id    = ref->recon_surface,
+                .picture_id    = ((VAAPIEncodePicture *)ref)->recon_surface,
                 .pic_order_cnt = href->pic_order_cnt,
                 .flags = (ref->display_order < pic->display_order ?
                           VA_PICTURE_HEVC_RPS_ST_CURR_BEFORE : 0) |
@@ -940,7 +942,7 @@ static int vaapi_encode_h265_init_picture_params(AVCodecContext *avctx,
         };
     }
 
-    vpic->coded_buf = pic->output_buffer;
+    vpic->coded_buf = vaapi_pic->output_buffer;
 
     vpic->nal_unit_type = hpic->slice_nal_unit;
 
@@ -970,16 +972,17 @@ static int vaapi_encode_h265_init_picture_params(AVCodecContext *avctx,
 }
 
 static int vaapi_encode_h265_init_slice_params(AVCodecContext *avctx,
-                                               VAAPIEncodePicture *pic,
+                                               VAAPIEncodePicture *vaapi_pic,
                                                VAAPIEncodeSlice *slice)
 {
-    VAAPIEncodeContext                *ctx = avctx->priv_data;
+    HWBaseEncodeContext          *base_ctx = avctx->priv_data;
     VAAPIEncodeH265Context           *priv = avctx->priv_data;
+    const HWBaseEncodePicture         *pic = &vaapi_pic->base;
     VAAPIEncodeH265Picture           *hpic = pic->priv_data;
     const H265RawSPS                  *sps = &priv->raw_sps;
     const H265RawPPS                  *pps = &priv->raw_pps;
     H265RawSliceHeader                 *sh = &priv->raw_slice.header;
-    VAEncPictureParameterBufferHEVC  *vpic = pic->codec_picture_params;
+    VAEncPictureParameterBufferHEVC  *vpic = vaapi_pic->codec_picture_params;
     VAEncSliceParameterBufferHEVC  *vslice = slice->codec_slice_params;
     int i;
 
@@ -996,7 +999,7 @@ static int vaapi_encode_h265_init_slice_params(AVCodecContext *avctx,
 
     sh->slice_type = hpic->slice_type;
 
-    if (sh->slice_type == HEVC_SLICE_P && ctx->p_to_gpb)
+    if (sh->slice_type == HEVC_SLICE_P && base_ctx->p_to_gpb)
         sh->slice_type = HEVC_SLICE_B;
 
     sh->slice_pic_order_cnt_lsb = hpic->pic_order_cnt &
@@ -1140,7 +1143,7 @@ static int vaapi_encode_h265_init_slice_params(AVCodecContext *avctx,
         .slice_tc_offset_div2   = sh->slice_tc_offset_div2,
 
         .slice_fields.bits = {
-            .last_slice_of_pic_flag       = slice->index == pic->nb_slices - 1,
+            .last_slice_of_pic_flag       = slice->index == vaapi_pic->nb_slices - 1,
             .dependent_slice_segment_flag = sh->dependent_slice_segment_flag,
             .colour_plane_id              = sh->colour_plane_id,
             .slice_temporal_mvp_enabled_flag =
@@ -1171,7 +1174,7 @@ static int vaapi_encode_h265_init_slice_params(AVCodecContext *avctx,
         av_assert0(pic->type == PICTURE_TYPE_P ||
                    pic->type == PICTURE_TYPE_B);
         vslice->ref_pic_list0[0] = vpic->reference_frames[0];
-        if (ctx->p_to_gpb && pic->type == PICTURE_TYPE_P)
+        if (base_ctx->p_to_gpb && pic->type == PICTURE_TYPE_P)
             // Reference for GPB B-frame, L0 == L1
             vslice->ref_pic_list1[0] = vpic->reference_frames[0];
     }
@@ -1181,7 +1184,7 @@ static int vaapi_encode_h265_init_slice_params(AVCodecContext *avctx,
         vslice->ref_pic_list1[0] = vpic->reference_frames[1];
     }
 
-    if (pic->type == PICTURE_TYPE_P && ctx->p_to_gpb) {
+    if (pic->type == PICTURE_TYPE_P && base_ctx->p_to_gpb) {
         vslice->slice_type = HEVC_SLICE_B;
         for (i = 0; i < FF_ARRAY_ELEMS(vslice->ref_pic_list0); i++) {
             vslice->ref_pic_list1[i].picture_id = vslice->ref_pic_list0[i].picture_id;
@@ -1494,7 +1497,7 @@ const FFCodec ff_hevc_vaapi_encoder = {
     .p.id           = AV_CODEC_ID_HEVC,
     .priv_data_size = sizeof(VAAPIEncodeH265Context),
     .init           = &vaapi_encode_h265_init,
-    FF_CODEC_RECEIVE_PACKET_CB(&ff_vaapi_encode_receive_packet),
+    FF_CODEC_RECEIVE_PACKET_CB(&ff_hw_base_encode_receive_packet),
     .close          = &vaapi_encode_h265_close,
     .p.priv_class   = &vaapi_encode_h265_class,
     .p.capabilities = AV_CODEC_CAP_DELAY | AV_CODEC_CAP_HARDWARE |
diff --git a/libavcodec/vaapi_encode_mjpeg.c b/libavcodec/vaapi_encode_mjpeg.c
index fb5c0d34c6..6726006774 100644
--- a/libavcodec/vaapi_encode_mjpeg.c
+++ b/libavcodec/vaapi_encode_mjpeg.c
@@ -220,12 +220,13 @@ static int vaapi_encode_mjpeg_write_extra_buffer(AVCodecContext *avctx,
 }
 
 static int vaapi_encode_mjpeg_init_picture_params(AVCodecContext *avctx,
-                                                  VAAPIEncodePicture *pic)
+                                                  VAAPIEncodePicture *vaapi_pic)
 {
     VAAPIEncodeMJPEGContext         *priv = avctx->priv_data;
+    const HWBaseEncodePicture        *pic = &vaapi_pic->base;
     JPEGRawFrameHeader                *fh = &priv->frame_header;
     JPEGRawScanHeader                 *sh = &priv->scan.header;
-    VAEncPictureParameterBufferJPEG *vpic = pic->codec_picture_params;
+    VAEncPictureParameterBufferJPEG *vpic = vaapi_pic->codec_picture_params;
     const AVPixFmtDescriptor *desc;
     const uint8_t components_rgb[3] = { 'R', 'G', 'B' };
     const uint8_t components_yuv[3] = {  1,   2,   3  };
@@ -377,8 +378,8 @@ static int vaapi_encode_mjpeg_init_picture_params(AVCodecContext *avctx,
 
 
     *vpic = (VAEncPictureParameterBufferJPEG) {
-        .reconstructed_picture = pic->recon_surface,
-        .coded_buf             = pic->output_buffer,
+        .reconstructed_picture = vaapi_pic->recon_surface,
+        .coded_buf             = vaapi_pic->output_buffer,
 
         .picture_width  = fh->X,
         .picture_height = fh->Y,
@@ -406,7 +407,7 @@ static int vaapi_encode_mjpeg_init_picture_params(AVCodecContext *avctx,
         vpic->quantiser_table_selector[i] = fh->Tq[i];
     }
 
-    pic->nb_slices = 1;
+    vaapi_pic->nb_slices = 1;
 
     return 0;
 }
@@ -572,7 +573,7 @@ const FFCodec ff_mjpeg_vaapi_encoder = {
     .p.id           = AV_CODEC_ID_MJPEG,
     .priv_data_size = sizeof(VAAPIEncodeMJPEGContext),
     .init           = &vaapi_encode_mjpeg_init,
-    FF_CODEC_RECEIVE_PACKET_CB(&ff_vaapi_encode_receive_packet),
+    FF_CODEC_RECEIVE_PACKET_CB(&ff_hw_base_encode_receive_packet),
     .close          = &vaapi_encode_mjpeg_close,
     .p.priv_class   = &vaapi_encode_mjpeg_class,
     .p.capabilities = AV_CODEC_CAP_HARDWARE | AV_CODEC_CAP_DR1 |
diff --git a/libavcodec/vaapi_encode_mpeg2.c b/libavcodec/vaapi_encode_mpeg2.c
index d0980c52b0..c9d3b60762 100644
--- a/libavcodec/vaapi_encode_mpeg2.c
+++ b/libavcodec/vaapi_encode_mpeg2.c
@@ -166,6 +166,7 @@ fail:
 
 static int vaapi_encode_mpeg2_init_sequence_params(AVCodecContext *avctx)
 {
+    HWBaseEncodeContext           *base_ctx = avctx->priv_data;
     VAAPIEncodeContext                 *ctx = avctx->priv_data;
     VAAPIEncodeMPEG2Context           *priv = avctx->priv_data;
     MPEG2RawSequenceHeader              *sh = &priv->sequence_header;
@@ -281,7 +282,7 @@ static int vaapi_encode_mpeg2_init_sequence_params(AVCodecContext *avctx)
 
     se->bit_rate_extension        = priv->bit_rate >> 18;
     se->vbv_buffer_size_extension = priv->vbv_buffer_size >> 10;
-    se->low_delay                 = ctx->b_per_p == 0;
+    se->low_delay                 = base_ctx->b_per_p == 0;
 
     se->frame_rate_extension_n = ext_n;
     se->frame_rate_extension_d = ext_d;
@@ -353,8 +354,8 @@ static int vaapi_encode_mpeg2_init_sequence_params(AVCodecContext *avctx)
 
 
     *vseq = (VAEncSequenceParameterBufferMPEG2) {
-        .intra_period = ctx->gop_size,
-        .ip_period    = ctx->b_per_p + 1,
+        .intra_period = base_ctx->gop_size,
+        .ip_period    = base_ctx->b_per_p + 1,
 
         .picture_width  = avctx->width,
         .picture_height = avctx->height,
@@ -417,12 +418,13 @@ static int vaapi_encode_mpeg2_init_sequence_params(AVCodecContext *avctx)
 }
 
 static int vaapi_encode_mpeg2_init_picture_params(AVCodecContext *avctx,
-                                                 VAAPIEncodePicture *pic)
+                                                  VAAPIEncodePicture *vaapi_pic)
 {
     VAAPIEncodeMPEG2Context          *priv = avctx->priv_data;
+    const HWBaseEncodePicture         *pic = &vaapi_pic->base;
     MPEG2RawPictureHeader              *ph = &priv->picture_header;
     MPEG2RawPictureCodingExtension    *pce = &priv->picture_coding_extension.data.picture_coding;
-    VAEncPictureParameterBufferMPEG2 *vpic = pic->codec_picture_params;
+    VAEncPictureParameterBufferMPEG2 *vpic = vaapi_pic->codec_picture_params;
 
     if (pic->type == PICTURE_TYPE_IDR || pic->type == PICTURE_TYPE_I) {
         ph->temporal_reference  = 0;
@@ -448,8 +450,8 @@ static int vaapi_encode_mpeg2_init_picture_params(AVCodecContext *avctx,
         pce->f_code[1][1] = 15;
     }
 
-    vpic->reconstructed_picture = pic->recon_surface;
-    vpic->coded_buf             = pic->output_buffer;
+    vpic->reconstructed_picture = vaapi_pic->recon_surface;
+    vpic->coded_buf             = vaapi_pic->output_buffer;
 
     switch (pic->type) {
     case PICTURE_TYPE_IDR:
@@ -458,12 +460,12 @@ static int vaapi_encode_mpeg2_init_picture_params(AVCodecContext *avctx,
         break;
     case PICTURE_TYPE_P:
         vpic->picture_type = VAEncPictureTypePredictive;
-        vpic->forward_reference_picture = pic->refs[0][0]->recon_surface;
+        vpic->forward_reference_picture = ((VAAPIEncodePicture *)pic->refs[0][0])->recon_surface;
         break;
     case PICTURE_TYPE_B:
         vpic->picture_type = VAEncPictureTypeBidirectional;
-        vpic->forward_reference_picture  = pic->refs[0][0]->recon_surface;
-        vpic->backward_reference_picture = pic->refs[1][0]->recon_surface;
+        vpic->forward_reference_picture  = ((VAAPIEncodePicture *)pic->refs[0][0])->recon_surface;
+        vpic->backward_reference_picture = ((VAAPIEncodePicture *)pic->refs[1][0])->recon_surface;
         break;
     default:
         av_assert0(0 && "invalid picture type");
@@ -479,11 +481,12 @@ static int vaapi_encode_mpeg2_init_picture_params(AVCodecContext *avctx,
 }
 
 static int vaapi_encode_mpeg2_init_slice_params(AVCodecContext *avctx,
-                                               VAAPIEncodePicture *pic,
-                                               VAAPIEncodeSlice *slice)
+                                                VAAPIEncodePicture *vaapi_pic,
+                                                VAAPIEncodeSlice *slice)
 {
-    VAAPIEncodeMPEG2Context            *priv = avctx->priv_data;
-    VAEncSliceParameterBufferMPEG2   *vslice = slice->codec_slice_params;
+    const HWBaseEncodePicture         *pic = &vaapi_pic->base;
+    VAAPIEncodeMPEG2Context          *priv = avctx->priv_data;
+    VAEncSliceParameterBufferMPEG2 *vslice = slice->codec_slice_params;
     int qp;
 
     vslice->macroblock_address = slice->block_start;
@@ -695,7 +698,7 @@ const FFCodec ff_mpeg2_vaapi_encoder = {
     .p.id           = AV_CODEC_ID_MPEG2VIDEO,
     .priv_data_size = sizeof(VAAPIEncodeMPEG2Context),
     .init           = &vaapi_encode_mpeg2_init,
-    FF_CODEC_RECEIVE_PACKET_CB(&ff_vaapi_encode_receive_packet),
+    FF_CODEC_RECEIVE_PACKET_CB(&ff_hw_base_encode_receive_packet),
     .close          = &vaapi_encode_mpeg2_close,
     .p.priv_class   = &vaapi_encode_mpeg2_class,
     .p.capabilities = AV_CODEC_CAP_DELAY | AV_CODEC_CAP_HARDWARE |
diff --git a/libavcodec/vaapi_encode_vp8.c b/libavcodec/vaapi_encode_vp8.c
index 4e284f86e2..c533536ceb 100644
--- a/libavcodec/vaapi_encode_vp8.c
+++ b/libavcodec/vaapi_encode_vp8.c
@@ -52,6 +52,7 @@ typedef struct VAAPIEncodeVP8Context {
 
 static int vaapi_encode_vp8_init_sequence_params(AVCodecContext *avctx)
 {
+    HWBaseEncodeContext         *base_ctx = avctx->priv_data;
     VAAPIEncodeContext               *ctx = avctx->priv_data;
     VAEncSequenceParameterBufferVP8 *vseq = ctx->codec_sequence_params;
 
@@ -66,22 +67,23 @@ static int vaapi_encode_vp8_init_sequence_params(AVCodecContext *avctx)
 
     if (!(ctx->va_rc_mode & VA_RC_CQP)) {
         vseq->bits_per_second = ctx->va_bit_rate;
-        vseq->intra_period    = ctx->gop_size;
+        vseq->intra_period    = base_ctx->gop_size;
     }
 
     return 0;
 }
 
 static int vaapi_encode_vp8_init_picture_params(AVCodecContext *avctx,
-                                                VAAPIEncodePicture *pic)
+                                                VAAPIEncodePicture *vaapi_pic)
 {
+    const HWBaseEncodePicture       *pic = &vaapi_pic->base;
     VAAPIEncodeVP8Context          *priv = avctx->priv_data;
-    VAEncPictureParameterBufferVP8 *vpic = pic->codec_picture_params;
+    VAEncPictureParameterBufferVP8 *vpic = vaapi_pic->codec_picture_params;
     int i;
 
-    vpic->reconstructed_frame = pic->recon_surface;
+    vpic->reconstructed_frame = vaapi_pic->recon_surface;
 
-    vpic->coded_buf = pic->output_buffer;
+    vpic->coded_buf = vaapi_pic->output_buffer;
 
     switch (pic->type) {
     case PICTURE_TYPE_IDR:
@@ -101,7 +103,7 @@ static int vaapi_encode_vp8_init_picture_params(AVCodecContext *avctx,
         vpic->ref_last_frame =
         vpic->ref_gf_frame   =
         vpic->ref_arf_frame  =
-            pic->refs[0][0]->recon_surface;
+            ((VAAPIEncodePicture *)pic->refs[0][0])->recon_surface;
         break;
     default:
         av_assert0(0 && "invalid picture type");
@@ -145,7 +147,7 @@ static int vaapi_encode_vp8_write_quant_table(AVCodecContext *avctx,
 
     memset(&quant, 0, sizeof(quant));
 
-    if (pic->type == PICTURE_TYPE_P)
+    if (pic->base.type == PICTURE_TYPE_P)
         q = priv->q_index_p;
     else
         q = priv->q_index_i;
@@ -250,7 +252,7 @@ const FFCodec ff_vp8_vaapi_encoder = {
     .p.id           = AV_CODEC_ID_VP8,
     .priv_data_size = sizeof(VAAPIEncodeVP8Context),
     .init           = &vaapi_encode_vp8_init,
-    FF_CODEC_RECEIVE_PACKET_CB(&ff_vaapi_encode_receive_packet),
+    FF_CODEC_RECEIVE_PACKET_CB(&ff_hw_base_encode_receive_packet),
     .close          = &ff_vaapi_encode_close,
     .p.priv_class   = &vaapi_encode_vp8_class,
     .p.capabilities = AV_CODEC_CAP_DELAY | AV_CODEC_CAP_HARDWARE |
diff --git a/libavcodec/vaapi_encode_vp9.c b/libavcodec/vaapi_encode_vp9.c
index 88f951652c..36dafea2c0 100644
--- a/libavcodec/vaapi_encode_vp9.c
+++ b/libavcodec/vaapi_encode_vp9.c
@@ -53,6 +53,7 @@ typedef struct VAAPIEncodeVP9Context {
 
 static int vaapi_encode_vp9_init_sequence_params(AVCodecContext *avctx)
 {
+    HWBaseEncodeContext         *base_ctx = avctx->priv_data;
     VAAPIEncodeContext               *ctx = avctx->priv_data;
     VAEncSequenceParameterBufferVP9 *vseq = ctx->codec_sequence_params;
     VAEncPictureParameterBufferVP9  *vpic = ctx->codec_picture_params;
@@ -64,7 +65,7 @@ static int vaapi_encode_vp9_init_sequence_params(AVCodecContext *avctx)
 
     if (!(ctx->va_rc_mode & VA_RC_CQP)) {
         vseq->bits_per_second = ctx->va_bit_rate;
-        vseq->intra_period    = ctx->gop_size;
+        vseq->intra_period    = base_ctx->gop_size;
     }
 
     vpic->frame_width_src  = avctx->width;
@@ -76,17 +77,18 @@ static int vaapi_encode_vp9_init_sequence_params(AVCodecContext *avctx)
 }
 
 static int vaapi_encode_vp9_init_picture_params(AVCodecContext *avctx,
-                                                VAAPIEncodePicture *pic)
+                                                VAAPIEncodePicture *vaapi_pic)
 {
-    VAAPIEncodeContext              *ctx = avctx->priv_data;
+    HWBaseEncodeContext        *base_ctx = avctx->priv_data;
     VAAPIEncodeVP9Context          *priv = avctx->priv_data;
+    const HWBaseEncodePicture       *pic = &vaapi_pic->base;
     VAAPIEncodeVP9Picture          *hpic = pic->priv_data;
-    VAEncPictureParameterBufferVP9 *vpic = pic->codec_picture_params;
+    VAEncPictureParameterBufferVP9 *vpic = vaapi_pic->codec_picture_params;
     int i;
     int num_tile_columns;
 
-    vpic->reconstructed_frame = pic->recon_surface;
-    vpic->coded_buf = pic->output_buffer;
+    vpic->reconstructed_frame = vaapi_pic->recon_surface;
+    vpic->coded_buf = vaapi_pic->output_buffer;
 
     // Maximum width of a tile in units of superblocks is MAX_TILE_WIDTH_B64(64)
     // So the number of tile columns is related to the width of the picture.
@@ -107,7 +109,7 @@ static int vaapi_encode_vp9_init_picture_params(AVCodecContext *avctx,
             VAAPIEncodeVP9Picture *href = pic->refs[0][0]->priv_data;
             av_assert0(href->slot == 0 || href->slot == 1);
 
-            if (ctx->max_b_depth > 0) {
+            if (base_ctx->max_b_depth > 0) {
                 hpic->slot = !href->slot;
                 vpic->refresh_frame_flags = 1 << hpic->slot | 0xfc;
             } else {
@@ -127,7 +129,7 @@ static int vaapi_encode_vp9_init_picture_params(AVCodecContext *avctx,
             av_assert0(href0->slot < pic->b_depth + 1 &&
                        href1->slot < pic->b_depth + 1);
 
-            if (pic->b_depth == ctx->max_b_depth) {
+            if (pic->b_depth == base_ctx->max_b_depth) {
                 // Unreferenced frame.
                 vpic->refresh_frame_flags = 0x00;
                 hpic->slot = 8;
@@ -159,11 +161,11 @@ static int vaapi_encode_vp9_init_picture_params(AVCodecContext *avctx,
 
     for (i = 0; i < MAX_REFERENCE_LIST_NUM; i++) {
         for (int j = 0; j < pic->nb_refs[i]; j++) {
-            VAAPIEncodePicture *ref_pic = pic->refs[i][j];
+            HWBaseEncodePicture *ref_pic = pic->refs[i][j];
             int slot;
             slot = ((VAAPIEncodeVP9Picture*)ref_pic->priv_data)->slot;
             av_assert0(vpic->reference_frames[slot] == VA_INVALID_SURFACE);
-            vpic->reference_frames[slot] = ref_pic->recon_surface;
+            vpic->reference_frames[slot] = ((VAAPIEncodePicture *)ref_pic)->recon_surface;
         }
     }
 
@@ -307,7 +309,7 @@ const FFCodec ff_vp9_vaapi_encoder = {
     .p.id           = AV_CODEC_ID_VP9,
     .priv_data_size = sizeof(VAAPIEncodeVP9Context),
     .init           = &vaapi_encode_vp9_init,
-    FF_CODEC_RECEIVE_PACKET_CB(&ff_vaapi_encode_receive_packet),
+    FF_CODEC_RECEIVE_PACKET_CB(&ff_hw_base_encode_receive_packet),
     .close          = &ff_vaapi_encode_close,
     .p.priv_class   = &vaapi_encode_vp9_class,
     .p.capabilities = AV_CODEC_CAP_DELAY | AV_CODEC_CAP_HARDWARE |
-- 
2.41.0.windows.1

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

^ permalink raw reply	[flat|nested] 18+ messages in thread

* [FFmpeg-devel] [PATCH v9 06/13] avcodec/vaapi_encode: extract the init and close function to base layer
  2024-05-20 14:52 [FFmpeg-devel] [PATCH v9 01/13] avcodec/vaapi_encode: introduce a base layer for vaapi encode tong1.wu-at-intel.com
                   ` (3 preceding siblings ...)
  2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 05/13] avcodec/vaapi_encode: move the dpb logic from VAAPI to base layer tong1.wu-at-intel.com
@ 2024-05-20 14:52 ` tong1.wu-at-intel.com
  2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 07/13] avcodec/vaapi_encode: extract gop configuration and two options " tong1.wu-at-intel.com
                   ` (7 subsequent siblings)
  12 siblings, 0 replies; 18+ messages in thread
From: tong1.wu-at-intel.com @ 2024-05-20 14:52 UTC (permalink / raw)
  To: ffmpeg-devel; +Cc: Tong Wu

From: Tong Wu <tong1.wu@intel.com>

Related parameters such as device context, frame context are also moved
to base layer.

Signed-off-by: Tong Wu <tong1.wu@intel.com>
---
 libavcodec/hw_base_encode.c     | 49 ++++++++++++++++++
 libavcodec/hw_base_encode.h     | 17 +++++++
 libavcodec/vaapi_encode.c       | 90 +++++++++++----------------------
 libavcodec/vaapi_encode.h       | 10 ----
 libavcodec/vaapi_encode_av1.c   |  2 +-
 libavcodec/vaapi_encode_h264.c  |  2 +-
 libavcodec/vaapi_encode_h265.c  |  2 +-
 libavcodec/vaapi_encode_mjpeg.c |  6 ++-
 8 files changed, 102 insertions(+), 76 deletions(-)

diff --git a/libavcodec/hw_base_encode.c b/libavcodec/hw_base_encode.c
index ec7178d2dc..f2b6ef4a3a 100644
--- a/libavcodec/hw_base_encode.c
+++ b/libavcodec/hw_base_encode.c
@@ -595,3 +595,52 @@ end:
 
     return 0;
 }
+
+int ff_hw_base_encode_init(AVCodecContext *avctx)
+{
+    HWBaseEncodeContext *ctx = avctx->priv_data;
+
+    ctx->frame = av_frame_alloc();
+    if (!ctx->frame)
+        return AVERROR(ENOMEM);
+
+    if (!avctx->hw_frames_ctx) {
+        av_log(avctx, AV_LOG_ERROR, "A hardware frames reference is "
+               "required to associate the encoding device.\n");
+        return AVERROR(EINVAL);
+    }
+
+    ctx->input_frames_ref = av_buffer_ref(avctx->hw_frames_ctx);
+    if (!ctx->input_frames_ref)
+        return AVERROR(ENOMEM);
+
+    ctx->input_frames = (AVHWFramesContext *)ctx->input_frames_ref->data;
+
+    ctx->device_ref = av_buffer_ref(ctx->input_frames->device_ref);
+    if (!ctx->device_ref)
+        return AVERROR(ENOMEM);
+
+    ctx->device = (AVHWDeviceContext *)ctx->device_ref->data;
+
+    ctx->tail_pkt = av_packet_alloc();
+    if (!ctx->tail_pkt)
+        return AVERROR(ENOMEM);
+
+    return 0;
+}
+
+int ff_hw_base_encode_close(AVCodecContext *avctx)
+{
+    HWBaseEncodeContext *ctx = avctx->priv_data;
+
+    av_fifo_freep2(&ctx->encode_fifo);
+
+    av_frame_free(&ctx->frame);
+    av_packet_free(&ctx->tail_pkt);
+
+    av_buffer_unref(&ctx->device_ref);
+    av_buffer_unref(&ctx->input_frames_ref);
+    av_buffer_unref(&ctx->recon_frames_ref);
+
+    return 0;
+}
diff --git a/libavcodec/hw_base_encode.h b/libavcodec/hw_base_encode.h
index 2667ae61cd..7e20f57b9c 100644
--- a/libavcodec/hw_base_encode.h
+++ b/libavcodec/hw_base_encode.h
@@ -19,6 +19,7 @@
 #ifndef AVCODEC_HW_BASE_ENCODE_H
 #define AVCODEC_HW_BASE_ENCODE_H
 
+#include "libavutil/hwcontext.h"
 #include "libavutil/fifo.h"
 
 #define MAX_DPB_SIZE 16
@@ -118,6 +119,18 @@ typedef struct HWBaseEncodeContext {
     // Hardware-specific hooks.
     const struct HWEncodePictureOperation *op;
 
+    // The hardware device context.
+    AVBufferRef    *device_ref;
+    AVHWDeviceContext *device;
+
+    // The hardware frame context containing the input frames.
+    AVBufferRef    *input_frames_ref;
+    AVHWFramesContext *input_frames;
+
+    // The hardware frame context containing the reconstructed frames.
+    AVBufferRef    *recon_frames_ref;
+    AVHWFramesContext *recon_frames;
+
     // Current encoding window, in display (input) order.
     HWBaseEncodePicture *pic_start, *pic_end;
     // The next picture to use as the previous reference picture in
@@ -184,6 +197,10 @@ typedef struct HWBaseEncodeContext {
 
 int ff_hw_base_encode_receive_packet(AVCodecContext *avctx, AVPacket *pkt);
 
+int ff_hw_base_encode_init(AVCodecContext *avctx);
+
+int ff_hw_base_encode_close(AVCodecContext *avctx);
+
 #define HW_BASE_ENCODE_COMMON_OPTIONS \
     { "async_depth", "Maximum processing parallelism. " \
       "Increase this to improve single channel performance.", \
diff --git a/libavcodec/vaapi_encode.c b/libavcodec/vaapi_encode.c
index c4bb93c520..c3ab0fc192 100644
--- a/libavcodec/vaapi_encode.c
+++ b/libavcodec/vaapi_encode.c
@@ -314,7 +314,7 @@ static int vaapi_encode_issue(AVCodecContext *avctx,
 
     av_log(avctx, AV_LOG_DEBUG, "Input surface is %#x.\n", pic->input_surface);
 
-    err = av_hwframe_get_buffer(ctx->recon_frames_ref, base_pic->recon_image, 0);
+    err = av_hwframe_get_buffer(base_ctx->recon_frames_ref, base_pic->recon_image, 0);
     if (err < 0) {
         err = AVERROR(ENOMEM);
         goto fail;
@@ -996,9 +996,10 @@ static const VAEntrypoint vaapi_encode_entrypoints_low_power[] = {
 
 static av_cold int vaapi_encode_profile_entrypoint(AVCodecContext *avctx)
 {
-    VAAPIEncodeContext      *ctx = avctx->priv_data;
-    VAProfile    *va_profiles    = NULL;
-    VAEntrypoint *va_entrypoints = NULL;
+    HWBaseEncodeContext *base_ctx = avctx->priv_data;
+    VAAPIEncodeContext       *ctx = avctx->priv_data;
+    VAProfile     *va_profiles    = NULL;
+    VAEntrypoint  *va_entrypoints = NULL;
     VAStatus vas;
     const VAEntrypoint *usable_entrypoints;
     const VAAPIEncodeProfile *profile;
@@ -1021,10 +1022,10 @@ static av_cold int vaapi_encode_profile_entrypoint(AVCodecContext *avctx)
         usable_entrypoints = vaapi_encode_entrypoints_normal;
     }
 
-    desc = av_pix_fmt_desc_get(ctx->input_frames->sw_format);
+    desc = av_pix_fmt_desc_get(base_ctx->input_frames->sw_format);
     if (!desc) {
         av_log(avctx, AV_LOG_ERROR, "Invalid input pixfmt (%d).\n",
-               ctx->input_frames->sw_format);
+               base_ctx->input_frames->sw_format);
         return AVERROR(EINVAL);
     }
     depth = desc->comp[0].depth;
@@ -2131,20 +2132,21 @@ static int vaapi_encode_alloc_output_buffer(FFRefStructOpaque opaque, void *obj)
 
 static av_cold int vaapi_encode_create_recon_frames(AVCodecContext *avctx)
 {
-    VAAPIEncodeContext *ctx = avctx->priv_data;
+    HWBaseEncodeContext *base_ctx = avctx->priv_data;
+    VAAPIEncodeContext       *ctx = avctx->priv_data;
     AVVAAPIHWConfig *hwconfig = NULL;
     AVHWFramesConstraints *constraints = NULL;
     enum AVPixelFormat recon_format;
     int err, i;
 
-    hwconfig = av_hwdevice_hwconfig_alloc(ctx->device_ref);
+    hwconfig = av_hwdevice_hwconfig_alloc(base_ctx->device_ref);
     if (!hwconfig) {
         err = AVERROR(ENOMEM);
         goto fail;
     }
     hwconfig->config_id = ctx->va_config;
 
-    constraints = av_hwdevice_get_hwframe_constraints(ctx->device_ref,
+    constraints = av_hwdevice_get_hwframe_constraints(base_ctx->device_ref,
                                                       hwconfig);
     if (!constraints) {
         err = AVERROR(ENOMEM);
@@ -2157,9 +2159,9 @@ static av_cold int vaapi_encode_create_recon_frames(AVCodecContext *avctx)
     recon_format = AV_PIX_FMT_NONE;
     if (constraints->valid_sw_formats) {
         for (i = 0; constraints->valid_sw_formats[i] != AV_PIX_FMT_NONE; i++) {
-            if (ctx->input_frames->sw_format ==
+            if (base_ctx->input_frames->sw_format ==
                 constraints->valid_sw_formats[i]) {
-                recon_format = ctx->input_frames->sw_format;
+                recon_format = base_ctx->input_frames->sw_format;
                 break;
             }
         }
@@ -2170,7 +2172,7 @@ static av_cold int vaapi_encode_create_recon_frames(AVCodecContext *avctx)
         }
     } else {
         // No idea what to use; copy input format.
-        recon_format = ctx->input_frames->sw_format;
+        recon_format = base_ctx->input_frames->sw_format;
     }
     av_log(avctx, AV_LOG_DEBUG, "Using %s as format of "
            "reconstructed frames.\n", av_get_pix_fmt_name(recon_format));
@@ -2191,19 +2193,19 @@ static av_cold int vaapi_encode_create_recon_frames(AVCodecContext *avctx)
     av_freep(&hwconfig);
     av_hwframe_constraints_free(&constraints);
 
-    ctx->recon_frames_ref = av_hwframe_ctx_alloc(ctx->device_ref);
-    if (!ctx->recon_frames_ref) {
+    base_ctx->recon_frames_ref = av_hwframe_ctx_alloc(base_ctx->device_ref);
+    if (!base_ctx->recon_frames_ref) {
         err = AVERROR(ENOMEM);
         goto fail;
     }
-    ctx->recon_frames = (AVHWFramesContext*)ctx->recon_frames_ref->data;
+    base_ctx->recon_frames = (AVHWFramesContext*)base_ctx->recon_frames_ref->data;
 
-    ctx->recon_frames->format    = AV_PIX_FMT_VAAPI;
-    ctx->recon_frames->sw_format = recon_format;
-    ctx->recon_frames->width     = ctx->surface_width;
-    ctx->recon_frames->height    = ctx->surface_height;
+    base_ctx->recon_frames->format    = AV_PIX_FMT_VAAPI;
+    base_ctx->recon_frames->sw_format = recon_format;
+    base_ctx->recon_frames->width     = ctx->surface_width;
+    base_ctx->recon_frames->height    = ctx->surface_height;
 
-    err = av_hwframe_ctx_init(ctx->recon_frames_ref);
+    err = av_hwframe_ctx_init(base_ctx->recon_frames_ref);
     if (err < 0) {
         av_log(avctx, AV_LOG_ERROR, "Failed to initialise reconstructed "
                "frame context: %d.\n", err);
@@ -2235,44 +2237,16 @@ av_cold int ff_vaapi_encode_init(AVCodecContext *avctx)
     VAStatus vas;
     int err;
 
+    err = ff_hw_base_encode_init(avctx);
+    if (err < 0)
+        goto fail;
+
     ctx->va_config  = VA_INVALID_ID;
     ctx->va_context = VA_INVALID_ID;
 
     base_ctx->op = &vaapi_op;
 
-    /* If you add something that can fail above this av_frame_alloc(),
-     * modify ff_vaapi_encode_close() accordingly. */
-    base_ctx->frame = av_frame_alloc();
-    if (!base_ctx->frame) {
-        return AVERROR(ENOMEM);
-    }
-
-    if (!avctx->hw_frames_ctx) {
-        av_log(avctx, AV_LOG_ERROR, "A hardware frames reference is "
-               "required to associate the encoding device.\n");
-        return AVERROR(EINVAL);
-    }
-
-    ctx->input_frames_ref = av_buffer_ref(avctx->hw_frames_ctx);
-    if (!ctx->input_frames_ref) {
-        err = AVERROR(ENOMEM);
-        goto fail;
-    }
-    ctx->input_frames = (AVHWFramesContext*)ctx->input_frames_ref->data;
-
-    ctx->device_ref = av_buffer_ref(ctx->input_frames->device_ref);
-    if (!ctx->device_ref) {
-        err = AVERROR(ENOMEM);
-        goto fail;
-    }
-    ctx->device = (AVHWDeviceContext*)ctx->device_ref->data;
-    ctx->hwctx = ctx->device->hwctx;
-
-    base_ctx->tail_pkt = av_packet_alloc();
-    if (!base_ctx->tail_pkt) {
-        err = AVERROR(ENOMEM);
-        goto fail;
-    }
+    ctx->hwctx = base_ctx->device->hwctx;
 
     err = vaapi_encode_profile_entrypoint(avctx);
     if (err < 0)
@@ -2339,7 +2313,7 @@ av_cold int ff_vaapi_encode_init(AVCodecContext *avctx)
     if (err < 0)
         goto fail;
 
-    recon_hwctx = ctx->recon_frames->hwctx;
+    recon_hwctx = base_ctx->recon_frames->hwctx;
     vas = vaCreateContext(ctx->hwctx->display, ctx->va_config,
                           ctx->surface_width, ctx->surface_height,
                           VA_PROGRESSIVE,
@@ -2467,16 +2441,10 @@ av_cold int ff_vaapi_encode_close(AVCodecContext *avctx)
         ctx->va_config = VA_INVALID_ID;
     }
 
-    av_frame_free(&base_ctx->frame);
-    av_packet_free(&base_ctx->tail_pkt);
-
     av_freep(&ctx->codec_sequence_params);
     av_freep(&ctx->codec_picture_params);
-    av_fifo_freep2(&base_ctx->encode_fifo);
 
-    av_buffer_unref(&ctx->recon_frames_ref);
-    av_buffer_unref(&ctx->input_frames_ref);
-    av_buffer_unref(&ctx->device_ref);
+    ff_hw_base_encode_close(avctx);
 
     return 0;
 }
diff --git a/libavcodec/vaapi_encode.h b/libavcodec/vaapi_encode.h
index 13ccad8e47..c25eab1cb7 100644
--- a/libavcodec/vaapi_encode.h
+++ b/libavcodec/vaapi_encode.h
@@ -219,18 +219,8 @@ typedef struct VAAPIEncodeContext {
     VAConfigID      va_config;
     VAContextID     va_context;
 
-    AVBufferRef    *device_ref;
-    AVHWDeviceContext *device;
     AVVAAPIDeviceContext *hwctx;
 
-    // The hardware frame context containing the input frames.
-    AVBufferRef    *input_frames_ref;
-    AVHWFramesContext *input_frames;
-
-    // The hardware frame context containing the reconstructed frames.
-    AVBufferRef    *recon_frames_ref;
-    AVHWFramesContext *recon_frames;
-
     // Pool of (reusable) bitstream output buffers.
     struct FFRefStructPool *output_buffer_pool;
 
diff --git a/libavcodec/vaapi_encode_av1.c b/libavcodec/vaapi_encode_av1.c
index f12fe54cff..a15b74f409 100644
--- a/libavcodec/vaapi_encode_av1.c
+++ b/libavcodec/vaapi_encode_av1.c
@@ -373,7 +373,7 @@ static int vaapi_encode_av1_init_sequence_params(AVCodecContext *avctx)
     memset(sh_obu, 0, sizeof(*sh_obu));
     sh_obu->header.obu_type = AV1_OBU_SEQUENCE_HEADER;
 
-    desc = av_pix_fmt_desc_get(priv->common.input_frames->sw_format);
+    desc = av_pix_fmt_desc_get(base_ctx->input_frames->sw_format);
     av_assert0(desc);
 
     sh->seq_profile  = avctx->profile;
diff --git a/libavcodec/vaapi_encode_h264.c b/libavcodec/vaapi_encode_h264.c
index 9182483762..d41b9649e8 100644
--- a/libavcodec/vaapi_encode_h264.c
+++ b/libavcodec/vaapi_encode_h264.c
@@ -308,7 +308,7 @@ static int vaapi_encode_h264_init_sequence_params(AVCodecContext *avctx)
     memset(sps, 0, sizeof(*sps));
     memset(pps, 0, sizeof(*pps));
 
-    desc = av_pix_fmt_desc_get(priv->common.input_frames->sw_format);
+    desc = av_pix_fmt_desc_get(base_ctx->input_frames->sw_format);
     av_assert0(desc);
     if (desc->nb_components == 1 || desc->log2_chroma_w != 1 || desc->log2_chroma_h != 1) {
         av_log(avctx, AV_LOG_ERROR, "Chroma format of input pixel format "
diff --git a/libavcodec/vaapi_encode_h265.c b/libavcodec/vaapi_encode_h265.c
index 1f9d9c56a4..ec48773d95 100644
--- a/libavcodec/vaapi_encode_h265.c
+++ b/libavcodec/vaapi_encode_h265.c
@@ -278,7 +278,7 @@ static int vaapi_encode_h265_init_sequence_params(AVCodecContext *avctx)
     memset(pps, 0, sizeof(*pps));
 
 
-    desc = av_pix_fmt_desc_get(priv->common.input_frames->sw_format);
+    desc = av_pix_fmt_desc_get(base_ctx->input_frames->sw_format);
     av_assert0(desc);
     if (desc->nb_components == 1) {
         chroma_format = 0;
diff --git a/libavcodec/vaapi_encode_mjpeg.c b/libavcodec/vaapi_encode_mjpeg.c
index 6726006774..24acc88a21 100644
--- a/libavcodec/vaapi_encode_mjpeg.c
+++ b/libavcodec/vaapi_encode_mjpeg.c
@@ -222,6 +222,7 @@ static int vaapi_encode_mjpeg_write_extra_buffer(AVCodecContext *avctx,
 static int vaapi_encode_mjpeg_init_picture_params(AVCodecContext *avctx,
                                                   VAAPIEncodePicture *vaapi_pic)
 {
+    HWBaseEncodeContext         *base_ctx = avctx->priv_data;
     VAAPIEncodeMJPEGContext         *priv = avctx->priv_data;
     const HWBaseEncodePicture        *pic = &vaapi_pic->base;
     JPEGRawFrameHeader                *fh = &priv->frame_header;
@@ -235,7 +236,7 @@ static int vaapi_encode_mjpeg_init_picture_params(AVCodecContext *avctx,
 
     av_assert0(pic->type == PICTURE_TYPE_IDR);
 
-    desc = av_pix_fmt_desc_get(priv->common.input_frames->sw_format);
+    desc = av_pix_fmt_desc_get(base_ctx->input_frames->sw_format);
     av_assert0(desc);
     if (desc->flags & AV_PIX_FMT_FLAG_RGB)
         components = components_rgb;
@@ -437,10 +438,11 @@ static int vaapi_encode_mjpeg_init_slice_params(AVCodecContext *avctx,
 
 static av_cold int vaapi_encode_mjpeg_get_encoder_caps(AVCodecContext *avctx)
 {
+    HWBaseEncodeContext *base_ctx = avctx->priv_data;
     VAAPIEncodeContext *ctx = avctx->priv_data;
     const AVPixFmtDescriptor *desc;
 
-    desc = av_pix_fmt_desc_get(ctx->input_frames->sw_format);
+    desc = av_pix_fmt_desc_get(base_ctx->input_frames->sw_format);
     av_assert0(desc);
 
     ctx->surface_width  = FFALIGN(avctx->width,  8 << desc->log2_chroma_w);
-- 
2.41.0.windows.1

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

^ permalink raw reply	[flat|nested] 18+ messages in thread

* [FFmpeg-devel] [PATCH v9 07/13] avcodec/vaapi_encode: extract gop configuration and two options to base layer
  2024-05-20 14:52 [FFmpeg-devel] [PATCH v9 01/13] avcodec/vaapi_encode: introduce a base layer for vaapi encode tong1.wu-at-intel.com
                   ` (4 preceding siblings ...)
  2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 06/13] avcodec/vaapi_encode: extract the init and close function " tong1.wu-at-intel.com
@ 2024-05-20 14:52 ` tong1.wu-at-intel.com
  2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 08/13] avcodec/vaapi_encode: extract set_output_property " tong1.wu-at-intel.com
                   ` (6 subsequent siblings)
  12 siblings, 0 replies; 18+ messages in thread
From: tong1.wu-at-intel.com @ 2024-05-20 14:52 UTC (permalink / raw)
  To: ffmpeg-devel; +Cc: Tong Wu

From: Tong Wu <tong1.wu@intel.com>

idr_interval and desired_b_depth are moved to HW_BASE_ENCODE_COMMON_OPTIONS.

Signed-off-by: Tong Wu <tong1.wu@intel.com>
---
 libavcodec/hw_base_encode.c | 54 +++++++++++++++++++++++++++++++++++++
 libavcodec/hw_base_encode.h | 19 +++++++++++++
 libavcodec/vaapi_encode.c   | 52 +++--------------------------------
 libavcodec/vaapi_encode.h   | 16 -----------
 4 files changed, 77 insertions(+), 64 deletions(-)

diff --git a/libavcodec/hw_base_encode.c b/libavcodec/hw_base_encode.c
index f2b6ef4a3a..dfe20c4e67 100644
--- a/libavcodec/hw_base_encode.c
+++ b/libavcodec/hw_base_encode.c
@@ -596,6 +596,60 @@ end:
     return 0;
 }
 
+int ff_hw_base_init_gop_structure(AVCodecContext *avctx, uint32_t ref_l0, uint32_t ref_l1,
+                                  int flags, int prediction_pre_only)
+{
+    HWBaseEncodeContext *ctx = avctx->priv_data;
+
+    if (flags & FLAG_INTRA_ONLY || avctx->gop_size <= 1) {
+        av_log(avctx, AV_LOG_VERBOSE, "Using intra frames only.\n");
+        ctx->gop_size = 1;
+    } else if (ref_l0 < 1) {
+        av_log(avctx, AV_LOG_ERROR, "Driver does not support any "
+               "reference frames.\n");
+        return AVERROR(EINVAL);
+    } else if (!(flags & FLAG_B_PICTURES) || ref_l1 < 1 ||
+               avctx->max_b_frames < 1 || prediction_pre_only) {
+        if (ctx->p_to_gpb)
+           av_log(avctx, AV_LOG_VERBOSE, "Using intra and B-frames "
+                  "(supported references: %d / %d).\n",
+                  ref_l0, ref_l1);
+        else
+            av_log(avctx, AV_LOG_VERBOSE, "Using intra and P-frames "
+                   "(supported references: %d / %d).\n", ref_l0, ref_l1);
+        ctx->gop_size = avctx->gop_size;
+        ctx->p_per_i  = INT_MAX;
+        ctx->b_per_p  = 0;
+    } else {
+       if (ctx->p_to_gpb)
+           av_log(avctx, AV_LOG_VERBOSE, "Using intra and B-frames "
+                  "(supported references: %d / %d).\n",
+                  ref_l0, ref_l1);
+       else
+           av_log(avctx, AV_LOG_VERBOSE, "Using intra, P- and B-frames "
+                  "(supported references: %d / %d).\n", ref_l0, ref_l1);
+        ctx->gop_size = avctx->gop_size;
+        ctx->p_per_i  = INT_MAX;
+        ctx->b_per_p  = avctx->max_b_frames;
+        if (flags & FLAG_B_PICTURE_REFERENCES) {
+            ctx->max_b_depth = FFMIN(ctx->desired_b_depth,
+                                     av_log2(ctx->b_per_p) + 1);
+        } else {
+            ctx->max_b_depth = 1;
+        }
+    }
+
+    if (flags & FLAG_NON_IDR_KEY_PICTURES) {
+        ctx->closed_gop  = !!(avctx->flags & AV_CODEC_FLAG_CLOSED_GOP);
+        ctx->gop_per_idr = ctx->idr_interval + 1;
+    } else {
+        ctx->closed_gop  = 1;
+        ctx->gop_per_idr = 1;
+    }
+
+    return 0;
+}
+
 int ff_hw_base_encode_init(AVCodecContext *avctx)
 {
     HWBaseEncodeContext *ctx = avctx->priv_data;
diff --git a/libavcodec/hw_base_encode.h b/libavcodec/hw_base_encode.h
index 7e20f57b9c..07936bf9bb 100644
--- a/libavcodec/hw_base_encode.h
+++ b/libavcodec/hw_base_encode.h
@@ -119,6 +119,14 @@ typedef struct HWBaseEncodeContext {
     // Hardware-specific hooks.
     const struct HWEncodePictureOperation *op;
 
+    // Global options.
+
+    // Number of I frames between IDR frames.
+    int             idr_interval;
+
+    // Desired B frame reference depth.
+    int             desired_b_depth;
+
     // The hardware device context.
     AVBufferRef    *device_ref;
     AVHWDeviceContext *device;
@@ -197,11 +205,22 @@ typedef struct HWBaseEncodeContext {
 
 int ff_hw_base_encode_receive_packet(AVCodecContext *avctx, AVPacket *pkt);
 
+int ff_hw_base_init_gop_structure(AVCodecContext *avctx, uint32_t ref_l0, uint32_t ref_l1,
+                                  int flags, int prediction_pre_only);
+
 int ff_hw_base_encode_init(AVCodecContext *avctx);
 
 int ff_hw_base_encode_close(AVCodecContext *avctx);
 
 #define HW_BASE_ENCODE_COMMON_OPTIONS \
+    { "idr_interval", \
+      "Distance (in I-frames) between key frames", \
+      OFFSET(common.base.idr_interval), AV_OPT_TYPE_INT, \
+      { .i64 = 0 }, 0, INT_MAX, FLAGS }, \
+    { "b_depth", \
+      "Maximum B-frame reference depth", \
+      OFFSET(common.base.desired_b_depth), AV_OPT_TYPE_INT, \
+      { .i64 = 1 }, 1, INT_MAX, FLAGS }, \
     { "async_depth", "Maximum processing parallelism. " \
       "Increase this to improve single channel performance.", \
       OFFSET(common.base.async_depth), AV_OPT_TYPE_INT, \
diff --git a/libavcodec/vaapi_encode.c b/libavcodec/vaapi_encode.c
index c3ab0fc192..4db64b686b 100644
--- a/libavcodec/vaapi_encode.c
+++ b/libavcodec/vaapi_encode.c
@@ -1638,7 +1638,7 @@ static av_cold int vaapi_encode_init_gop_structure(AVCodecContext *avctx)
     VAStatus vas;
     VAConfigAttrib attr = { VAConfigAttribEncMaxRefFrames };
     uint32_t ref_l0, ref_l1;
-    int prediction_pre_only;
+    int prediction_pre_only, err;
 
     vas = vaGetConfigAttributes(ctx->hwctx->display,
                                 ctx->va_profile,
@@ -1702,53 +1702,9 @@ static av_cold int vaapi_encode_init_gop_structure(AVCodecContext *avctx)
     }
 #endif
 
-    if (ctx->codec->flags & FLAG_INTRA_ONLY ||
-        avctx->gop_size <= 1) {
-        av_log(avctx, AV_LOG_VERBOSE, "Using intra frames only.\n");
-        base_ctx->gop_size = 1;
-    } else if (ref_l0 < 1) {
-        av_log(avctx, AV_LOG_ERROR, "Driver does not support any "
-               "reference frames.\n");
-        return AVERROR(EINVAL);
-    } else if (!(ctx->codec->flags & FLAG_B_PICTURES) ||
-               ref_l1 < 1 || avctx->max_b_frames < 1 ||
-               prediction_pre_only) {
-        if (base_ctx->p_to_gpb)
-           av_log(avctx, AV_LOG_VERBOSE, "Using intra and B-frames "
-                  "(supported references: %d / %d).\n",
-                  ref_l0, ref_l1);
-        else
-            av_log(avctx, AV_LOG_VERBOSE, "Using intra and P-frames "
-                   "(supported references: %d / %d).\n", ref_l0, ref_l1);
-        base_ctx->gop_size = avctx->gop_size;
-        base_ctx->p_per_i  = INT_MAX;
-        base_ctx->b_per_p  = 0;
-    } else {
-       if (base_ctx->p_to_gpb)
-           av_log(avctx, AV_LOG_VERBOSE, "Using intra and B-frames "
-                  "(supported references: %d / %d).\n",
-                  ref_l0, ref_l1);
-       else
-           av_log(avctx, AV_LOG_VERBOSE, "Using intra, P- and B-frames "
-                  "(supported references: %d / %d).\n", ref_l0, ref_l1);
-        base_ctx->gop_size = avctx->gop_size;
-        base_ctx->p_per_i  = INT_MAX;
-        base_ctx->b_per_p  = avctx->max_b_frames;
-        if (ctx->codec->flags & FLAG_B_PICTURE_REFERENCES) {
-            base_ctx->max_b_depth = FFMIN(ctx->desired_b_depth,
-                                          av_log2(base_ctx->b_per_p) + 1);
-        } else {
-            base_ctx->max_b_depth = 1;
-        }
-    }
-
-    if (ctx->codec->flags & FLAG_NON_IDR_KEY_PICTURES) {
-        base_ctx->closed_gop  = !!(avctx->flags & AV_CODEC_FLAG_CLOSED_GOP);
-        base_ctx->gop_per_idr = ctx->idr_interval + 1;
-    } else {
-        base_ctx->closed_gop  = 1;
-        base_ctx->gop_per_idr = 1;
-    }
+    err = ff_hw_base_init_gop_structure(avctx, ref_l0, ref_l1, ctx->codec->flags, prediction_pre_only);
+    if (err < 0)
+        return err;
 
     return 0;
 }
diff --git a/libavcodec/vaapi_encode.h b/libavcodec/vaapi_encode.h
index c25eab1cb7..76fb645d71 100644
--- a/libavcodec/vaapi_encode.h
+++ b/libavcodec/vaapi_encode.h
@@ -151,17 +151,9 @@ typedef struct VAAPIEncodeContext {
     // Codec-specific hooks.
     const struct VAAPIEncodeType *codec;
 
-    // Global options.
-
     // Use low power encoding mode.
     int             low_power;
 
-    // Number of I frames between IDR frames.
-    int             idr_interval;
-
-    // Desired B frame reference depth.
-    int             desired_b_depth;
-
     // Max Frame Size
     int             max_frame_size;
 
@@ -371,14 +363,6 @@ int ff_vaapi_encode_close(AVCodecContext *avctx);
       "may not support all encoding features)", \
       OFFSET(common.low_power), AV_OPT_TYPE_BOOL, \
       { .i64 = 0 }, 0, 1, FLAGS }, \
-    { "idr_interval", \
-      "Distance (in I-frames) between IDR frames", \
-      OFFSET(common.idr_interval), AV_OPT_TYPE_INT, \
-      { .i64 = 0 }, 0, INT_MAX, FLAGS }, \
-    { "b_depth", \
-      "Maximum B-frame reference depth", \
-      OFFSET(common.desired_b_depth), AV_OPT_TYPE_INT, \
-      { .i64 = 1 }, 1, INT_MAX, FLAGS }, \
     { "max_frame_size", \
       "Maximum frame size (in bytes)",\
       OFFSET(common.max_frame_size), AV_OPT_TYPE_INT, \
-- 
2.41.0.windows.1

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

^ permalink raw reply	[flat|nested] 18+ messages in thread

* [FFmpeg-devel] [PATCH v9 08/13] avcodec/vaapi_encode: extract set_output_property to base layer
  2024-05-20 14:52 [FFmpeg-devel] [PATCH v9 01/13] avcodec/vaapi_encode: introduce a base layer for vaapi encode tong1.wu-at-intel.com
                   ` (5 preceding siblings ...)
  2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 07/13] avcodec/vaapi_encode: extract gop configuration and two options " tong1.wu-at-intel.com
@ 2024-05-20 14:52 ` tong1.wu-at-intel.com
  2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 09/13] avcodec/vaapi_encode: extract a get_recon_format function " tong1.wu-at-intel.com
                   ` (5 subsequent siblings)
  12 siblings, 0 replies; 18+ messages in thread
From: tong1.wu-at-intel.com @ 2024-05-20 14:52 UTC (permalink / raw)
  To: ffmpeg-devel; +Cc: Tong Wu

From: Tong Wu <tong1.wu@intel.com>

Signed-off-by: Tong Wu <tong1.wu@intel.com>
---
 libavcodec/hw_base_encode.c | 40 +++++++++++++++++++++++++++++++++
 libavcodec/hw_base_encode.h |  3 +++
 libavcodec/vaapi_encode.c   | 44 ++-----------------------------------
 3 files changed, 45 insertions(+), 42 deletions(-)

diff --git a/libavcodec/hw_base_encode.c b/libavcodec/hw_base_encode.c
index dfe20c4e67..df820c2f83 100644
--- a/libavcodec/hw_base_encode.c
+++ b/libavcodec/hw_base_encode.c
@@ -491,6 +491,46 @@ fail:
     return err;
 }
 
+int ff_hw_base_encode_set_output_property(AVCodecContext *avctx,
+                                          HWBaseEncodePicture *pic,
+                                          AVPacket *pkt, int flag_no_delay)
+{
+    HWBaseEncodeContext *ctx = avctx->priv_data;
+
+    if (pic->type == PICTURE_TYPE_IDR)
+        pkt->flags |= AV_PKT_FLAG_KEY;
+
+    pkt->pts = pic->pts;
+    pkt->duration = pic->duration;
+
+    // for no-delay encoders this is handled in generic codec
+    if (avctx->codec->capabilities & AV_CODEC_CAP_DELAY &&
+        avctx->flags & AV_CODEC_FLAG_COPY_OPAQUE) {
+        pkt->opaque          = pic->opaque;
+        pkt->opaque_ref      = pic->opaque_ref;
+        pic->opaque_ref = NULL;
+    }
+
+    if (flag_no_delay) {
+        pkt->dts = pkt->pts;
+        return 0;
+    }
+
+    if (ctx->output_delay == 0) {
+        pkt->dts = pkt->pts;
+    } else if (pic->encode_order < ctx->decode_delay) {
+        if (ctx->ts_ring[pic->encode_order] < INT64_MIN + ctx->dts_pts_diff)
+            pkt->dts = INT64_MIN;
+        else
+            pkt->dts = ctx->ts_ring[pic->encode_order] - ctx->dts_pts_diff;
+    } else {
+        pkt->dts = ctx->ts_ring[(pic->encode_order - ctx->decode_delay) %
+                                (3 * ctx->output_delay + ctx->async_depth)];
+    }
+
+    return 0;
+}
+
 int ff_hw_base_encode_receive_packet(AVCodecContext *avctx, AVPacket *pkt)
 {
     HWBaseEncodeContext *ctx = avctx->priv_data;
diff --git a/libavcodec/hw_base_encode.h b/libavcodec/hw_base_encode.h
index 07936bf9bb..d363819329 100644
--- a/libavcodec/hw_base_encode.h
+++ b/libavcodec/hw_base_encode.h
@@ -203,6 +203,9 @@ typedef struct HWBaseEncodeContext {
     AVPacket        *tail_pkt;
 } HWBaseEncodeContext;
 
+int ff_hw_base_encode_set_output_property(AVCodecContext *avctx, HWBaseEncodePicture *pic,
+                                          AVPacket *pkt, int flag_no_delay);
+
 int ff_hw_base_encode_receive_packet(AVCodecContext *avctx, AVPacket *pkt);
 
 int ff_hw_base_init_gop_structure(AVCodecContext *avctx, uint32_t ref_l0, uint32_t ref_l1,
diff --git a/libavcodec/vaapi_encode.c b/libavcodec/vaapi_encode.c
index 4db64b686b..6ab1b633ed 100644
--- a/libavcodec/vaapi_encode.c
+++ b/libavcodec/vaapi_encode.c
@@ -660,47 +660,6 @@ fail_at_end:
     return err;
 }
 
-static int vaapi_encode_set_output_property(AVCodecContext *avctx,
-                                            HWBaseEncodePicture *pic,
-                                            AVPacket *pkt)
-{
-    HWBaseEncodeContext *base_ctx = avctx->priv_data;
-    VAAPIEncodeContext       *ctx = avctx->priv_data;
-
-    if (pic->type == PICTURE_TYPE_IDR)
-        pkt->flags |= AV_PKT_FLAG_KEY;
-
-    pkt->pts = pic->pts;
-    pkt->duration = pic->duration;
-
-    // for no-delay encoders this is handled in generic codec
-    if (avctx->codec->capabilities & AV_CODEC_CAP_DELAY &&
-        avctx->flags & AV_CODEC_FLAG_COPY_OPAQUE) {
-        pkt->opaque     = pic->opaque;
-        pkt->opaque_ref = pic->opaque_ref;
-        pic->opaque_ref = NULL;
-    }
-
-    if (ctx->codec->flags & FLAG_TIMESTAMP_NO_DELAY) {
-        pkt->dts = pkt->pts;
-        return 0;
-    }
-
-    if (base_ctx->output_delay == 0) {
-        pkt->dts = pkt->pts;
-    } else if (pic->encode_order < base_ctx->decode_delay) {
-        if (base_ctx->ts_ring[pic->encode_order] < INT64_MIN + base_ctx->dts_pts_diff)
-            pkt->dts = INT64_MIN;
-        else
-            pkt->dts = base_ctx->ts_ring[pic->encode_order] - base_ctx->dts_pts_diff;
-    } else {
-        pkt->dts = base_ctx->ts_ring[(pic->encode_order - base_ctx->decode_delay) %
-                                     (3 * base_ctx->output_delay + base_ctx->async_depth)];
-    }
-
-    return 0;
-}
-
 static int vaapi_encode_get_coded_buffer_size(AVCodecContext *avctx, VABufferID buf_id)
 {
     VAAPIEncodeContext *ctx = avctx->priv_data;
@@ -852,7 +811,8 @@ static int vaapi_encode_output(AVCodecContext *avctx,
     av_log(avctx, AV_LOG_DEBUG, "Output read for pic %"PRId64"/%"PRId64".\n",
            base_pic->display_order, base_pic->encode_order);
 
-    vaapi_encode_set_output_property(avctx, (HWBaseEncodePicture*)pic, pkt_ptr);
+    ff_hw_base_encode_set_output_property(avctx, (HWBaseEncodePicture*)base_pic, pkt_ptr,
+                                          ctx->codec->flags & FLAG_TIMESTAMP_NO_DELAY);
 
 end:
     ff_refstruct_unref(&pic->output_buffer_ref);
-- 
2.41.0.windows.1

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

^ permalink raw reply	[flat|nested] 18+ messages in thread

* [FFmpeg-devel] [PATCH v9 09/13] avcodec/vaapi_encode: extract a get_recon_format function to base layer
  2024-05-20 14:52 [FFmpeg-devel] [PATCH v9 01/13] avcodec/vaapi_encode: introduce a base layer for vaapi encode tong1.wu-at-intel.com
                   ` (6 preceding siblings ...)
  2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 08/13] avcodec/vaapi_encode: extract set_output_property " tong1.wu-at-intel.com
@ 2024-05-20 14:52 ` tong1.wu-at-intel.com
  2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 10/13] avcodec/vaapi_encode: extract a free funtion " tong1.wu-at-intel.com
                   ` (4 subsequent siblings)
  12 siblings, 0 replies; 18+ messages in thread
From: tong1.wu-at-intel.com @ 2024-05-20 14:52 UTC (permalink / raw)
  To: ffmpeg-devel; +Cc: Tong Wu

From: Tong Wu <tong1.wu@intel.com>

Surface size and block size parameters are also moved to base layer.

Signed-off-by: Tong Wu <tong1.wu@intel.com>
---
 libavcodec/hw_base_encode.c     | 58 +++++++++++++++++++++++
 libavcodec/hw_base_encode.h     | 12 +++++
 libavcodec/vaapi_encode.c       | 81 ++++++++-------------------------
 libavcodec/vaapi_encode.h       | 10 ----
 libavcodec/vaapi_encode_av1.c   | 10 ++--
 libavcodec/vaapi_encode_h264.c  | 11 +++--
 libavcodec/vaapi_encode_h265.c  | 25 +++++-----
 libavcodec/vaapi_encode_mjpeg.c |  5 +-
 libavcodec/vaapi_encode_vp9.c   |  6 +--
 9 files changed, 118 insertions(+), 100 deletions(-)

diff --git a/libavcodec/hw_base_encode.c b/libavcodec/hw_base_encode.c
index df820c2f83..42c40cb48f 100644
--- a/libavcodec/hw_base_encode.c
+++ b/libavcodec/hw_base_encode.c
@@ -690,6 +690,64 @@ int ff_hw_base_init_gop_structure(AVCodecContext *avctx, uint32_t ref_l0, uint32
     return 0;
 }
 
+int ff_hw_base_get_recon_format(AVCodecContext *avctx, const void *hwconfig, enum AVPixelFormat *fmt)
+{
+    HWBaseEncodeContext *ctx = avctx->priv_data;
+    AVHWFramesConstraints *constraints = NULL;
+    enum AVPixelFormat recon_format;
+    int err, i;
+
+    constraints = av_hwdevice_get_hwframe_constraints(ctx->device_ref,
+                                                      hwconfig);
+    if (!constraints) {
+        err = AVERROR(ENOMEM);
+        goto fail;
+    }
+
+    // Probably we can use the input surface format as the surface format
+    // of the reconstructed frames.  If not, we just pick the first (only?)
+    // format in the valid list and hope that it all works.
+    recon_format = AV_PIX_FMT_NONE;
+    if (constraints->valid_sw_formats) {
+        for (i = 0; constraints->valid_sw_formats[i] != AV_PIX_FMT_NONE; i++) {
+            if (ctx->input_frames->sw_format ==
+                constraints->valid_sw_formats[i]) {
+                recon_format = ctx->input_frames->sw_format;
+                break;
+            }
+        }
+        if (recon_format == AV_PIX_FMT_NONE) {
+            // No match.  Just use the first in the supported list and
+            // hope for the best.
+            recon_format = constraints->valid_sw_formats[0];
+        }
+    } else {
+        // No idea what to use; copy input format.
+        recon_format = ctx->input_frames->sw_format;
+    }
+    av_log(avctx, AV_LOG_DEBUG, "Using %s as format of "
+           "reconstructed frames.\n", av_get_pix_fmt_name(recon_format));
+
+    if (ctx->surface_width  < constraints->min_width  ||
+        ctx->surface_height < constraints->min_height ||
+        ctx->surface_width  > constraints->max_width ||
+        ctx->surface_height > constraints->max_height) {
+        av_log(avctx, AV_LOG_ERROR, "Hardware does not support encoding at "
+               "size %dx%d (constraints: width %d-%d height %d-%d).\n",
+               ctx->surface_width, ctx->surface_height,
+               constraints->min_width,  constraints->max_width,
+               constraints->min_height, constraints->max_height);
+        err = AVERROR(EINVAL);
+        goto fail;
+    }
+
+    *fmt = recon_format;
+    err = 0;
+fail:
+    av_hwframe_constraints_free(&constraints);
+    return err;
+}
+
 int ff_hw_base_encode_init(AVCodecContext *avctx)
 {
     HWBaseEncodeContext *ctx = avctx->priv_data;
diff --git a/libavcodec/hw_base_encode.h b/libavcodec/hw_base_encode.h
index d363819329..7d40da039c 100644
--- a/libavcodec/hw_base_encode.h
+++ b/libavcodec/hw_base_encode.h
@@ -127,6 +127,16 @@ typedef struct HWBaseEncodeContext {
     // Desired B frame reference depth.
     int             desired_b_depth;
 
+    // The required size of surfaces.  This is probably the input
+    // size (AVCodecContext.width|height) aligned up to whatever
+    // block size is required by the codec.
+    int             surface_width;
+    int             surface_height;
+
+    // The block size for slice calculations.
+    int             slice_block_width;
+    int             slice_block_height;
+
     // The hardware device context.
     AVBufferRef    *device_ref;
     AVHWDeviceContext *device;
@@ -211,6 +221,8 @@ int ff_hw_base_encode_receive_packet(AVCodecContext *avctx, AVPacket *pkt);
 int ff_hw_base_init_gop_structure(AVCodecContext *avctx, uint32_t ref_l0, uint32_t ref_l1,
                                   int flags, int prediction_pre_only);
 
+int ff_hw_base_get_recon_format(AVCodecContext *avctx, const void *hwconfig, enum AVPixelFormat *fmt);
+
 int ff_hw_base_encode_init(AVCodecContext *avctx);
 
 int ff_hw_base_encode_close(AVCodecContext *avctx);
diff --git a/libavcodec/vaapi_encode.c b/libavcodec/vaapi_encode.c
index 6ab1b633ed..98b8c82da3 100644
--- a/libavcodec/vaapi_encode.c
+++ b/libavcodec/vaapi_encode.c
@@ -1777,6 +1777,7 @@ static av_cold int vaapi_encode_init_tile_slice_structure(AVCodecContext *avctx,
 
 static av_cold int vaapi_encode_init_slice_structure(AVCodecContext *avctx)
 {
+    HWBaseEncodeContext *base_ctx = avctx->priv_data;
     VAAPIEncodeContext *ctx = avctx->priv_data;
     VAConfigAttrib attr[3] = { { VAConfigAttribEncMaxSlices },
                                { VAConfigAttribEncSliceStructure },
@@ -1796,12 +1797,12 @@ static av_cold int vaapi_encode_init_slice_structure(AVCodecContext *avctx)
         return 0;
     }
 
-    av_assert0(ctx->slice_block_height > 0 && ctx->slice_block_width > 0);
+    av_assert0(base_ctx->slice_block_height > 0 && base_ctx->slice_block_width > 0);
 
-    ctx->slice_block_rows = (avctx->height + ctx->slice_block_height - 1) /
-                             ctx->slice_block_height;
-    ctx->slice_block_cols = (avctx->width  + ctx->slice_block_width  - 1) /
-                             ctx->slice_block_width;
+    ctx->slice_block_rows = (avctx->height + base_ctx->slice_block_height - 1) /
+                             base_ctx->slice_block_height;
+    ctx->slice_block_cols = (avctx->width  + base_ctx->slice_block_width  - 1) /
+                             base_ctx->slice_block_width;
 
     if (avctx->slices <= 1 && !ctx->tile_rows && !ctx->tile_cols) {
         ctx->nb_slices  = 1;
@@ -2023,7 +2024,8 @@ static void vaapi_encode_free_output_buffer(FFRefStructOpaque opaque,
 static int vaapi_encode_alloc_output_buffer(FFRefStructOpaque opaque, void *obj)
 {
     AVCodecContext   *avctx = opaque.nc;
-    VAAPIEncodeContext *ctx = avctx->priv_data;
+    HWBaseEncodeContext *base_ctx = avctx->priv_data;
+    VAAPIEncodeContext       *ctx = avctx->priv_data;
     VABufferID *buffer_id = obj;
     VAStatus vas;
 
@@ -2033,7 +2035,7 @@ static int vaapi_encode_alloc_output_buffer(FFRefStructOpaque opaque, void *obj)
     // bound on that.
     vas = vaCreateBuffer(ctx->hwctx->display, ctx->va_context,
                          VAEncCodedBufferType,
-                         3 * ctx->surface_width * ctx->surface_height +
+                         3 * base_ctx->surface_width * base_ctx->surface_height +
                          (1 << 16), 1, 0, buffer_id);
     if (vas != VA_STATUS_SUCCESS) {
         av_log(avctx, AV_LOG_ERROR, "Failed to create bitstream "
@@ -2051,9 +2053,8 @@ static av_cold int vaapi_encode_create_recon_frames(AVCodecContext *avctx)
     HWBaseEncodeContext *base_ctx = avctx->priv_data;
     VAAPIEncodeContext       *ctx = avctx->priv_data;
     AVVAAPIHWConfig *hwconfig = NULL;
-    AVHWFramesConstraints *constraints = NULL;
     enum AVPixelFormat recon_format;
-    int err, i;
+    int err;
 
     hwconfig = av_hwdevice_hwconfig_alloc(base_ctx->device_ref);
     if (!hwconfig) {
@@ -2062,52 +2063,9 @@ static av_cold int vaapi_encode_create_recon_frames(AVCodecContext *avctx)
     }
     hwconfig->config_id = ctx->va_config;
 
-    constraints = av_hwdevice_get_hwframe_constraints(base_ctx->device_ref,
-                                                      hwconfig);
-    if (!constraints) {
-        err = AVERROR(ENOMEM);
-        goto fail;
-    }
-
-    // Probably we can use the input surface format as the surface format
-    // of the reconstructed frames.  If not, we just pick the first (only?)
-    // format in the valid list and hope that it all works.
-    recon_format = AV_PIX_FMT_NONE;
-    if (constraints->valid_sw_formats) {
-        for (i = 0; constraints->valid_sw_formats[i] != AV_PIX_FMT_NONE; i++) {
-            if (base_ctx->input_frames->sw_format ==
-                constraints->valid_sw_formats[i]) {
-                recon_format = base_ctx->input_frames->sw_format;
-                break;
-            }
-        }
-        if (recon_format == AV_PIX_FMT_NONE) {
-            // No match.  Just use the first in the supported list and
-            // hope for the best.
-            recon_format = constraints->valid_sw_formats[0];
-        }
-    } else {
-        // No idea what to use; copy input format.
-        recon_format = base_ctx->input_frames->sw_format;
-    }
-    av_log(avctx, AV_LOG_DEBUG, "Using %s as format of "
-           "reconstructed frames.\n", av_get_pix_fmt_name(recon_format));
-
-    if (ctx->surface_width  < constraints->min_width  ||
-        ctx->surface_height < constraints->min_height ||
-        ctx->surface_width  > constraints->max_width ||
-        ctx->surface_height > constraints->max_height) {
-        av_log(avctx, AV_LOG_ERROR, "Hardware does not support encoding at "
-               "size %dx%d (constraints: width %d-%d height %d-%d).\n",
-               ctx->surface_width, ctx->surface_height,
-               constraints->min_width,  constraints->max_width,
-               constraints->min_height, constraints->max_height);
-        err = AVERROR(EINVAL);
+    err = ff_hw_base_get_recon_format(avctx, (const void*)hwconfig, &recon_format);
+    if (err < 0)
         goto fail;
-    }
-
-    av_freep(&hwconfig);
-    av_hwframe_constraints_free(&constraints);
 
     base_ctx->recon_frames_ref = av_hwframe_ctx_alloc(base_ctx->device_ref);
     if (!base_ctx->recon_frames_ref) {
@@ -2118,8 +2076,8 @@ static av_cold int vaapi_encode_create_recon_frames(AVCodecContext *avctx)
 
     base_ctx->recon_frames->format    = AV_PIX_FMT_VAAPI;
     base_ctx->recon_frames->sw_format = recon_format;
-    base_ctx->recon_frames->width     = ctx->surface_width;
-    base_ctx->recon_frames->height    = ctx->surface_height;
+    base_ctx->recon_frames->width     = base_ctx->surface_width;
+    base_ctx->recon_frames->height    = base_ctx->surface_height;
 
     err = av_hwframe_ctx_init(base_ctx->recon_frames_ref);
     if (err < 0) {
@@ -2131,7 +2089,6 @@ static av_cold int vaapi_encode_create_recon_frames(AVCodecContext *avctx)
     err = 0;
   fail:
     av_freep(&hwconfig);
-    av_hwframe_constraints_free(&constraints);
     return err;
 }
 
@@ -2174,11 +2131,11 @@ av_cold int ff_vaapi_encode_init(AVCodecContext *avctx)
             goto fail;
     } else {
         // Assume 16x16 blocks.
-        ctx->surface_width  = FFALIGN(avctx->width,  16);
-        ctx->surface_height = FFALIGN(avctx->height, 16);
+        base_ctx->surface_width  = FFALIGN(avctx->width,  16);
+        base_ctx->surface_height = FFALIGN(avctx->height, 16);
         if (ctx->codec->flags & FLAG_SLICE_CONTROL) {
-            ctx->slice_block_width  = 16;
-            ctx->slice_block_height = 16;
+            base_ctx->slice_block_width  = 16;
+            base_ctx->slice_block_height = 16;
         }
     }
 
@@ -2231,7 +2188,7 @@ av_cold int ff_vaapi_encode_init(AVCodecContext *avctx)
 
     recon_hwctx = base_ctx->recon_frames->hwctx;
     vas = vaCreateContext(ctx->hwctx->display, ctx->va_config,
-                          ctx->surface_width, ctx->surface_height,
+                          base_ctx->surface_width, base_ctx->surface_height,
                           VA_PROGRESSIVE,
                           recon_hwctx->surface_ids,
                           recon_hwctx->nb_surfaces,
diff --git a/libavcodec/vaapi_encode.h b/libavcodec/vaapi_encode.h
index 76fb645d71..8c7568e9ae 100644
--- a/libavcodec/vaapi_encode.h
+++ b/libavcodec/vaapi_encode.h
@@ -171,16 +171,6 @@ typedef struct VAAPIEncodeContext {
     // Desired packed headers.
     unsigned int    desired_packed_headers;
 
-    // The required size of surfaces.  This is probably the input
-    // size (AVCodecContext.width|height) aligned up to whatever
-    // block size is required by the codec.
-    int             surface_width;
-    int             surface_height;
-
-    // The block size for slice calculations.
-    int             slice_block_width;
-    int             slice_block_height;
-
     // Everything above this point must be set before calling
     // ff_vaapi_encode_init().
 
diff --git a/libavcodec/vaapi_encode_av1.c b/libavcodec/vaapi_encode_av1.c
index a15b74f409..a767d22d37 100644
--- a/libavcodec/vaapi_encode_av1.c
+++ b/libavcodec/vaapi_encode_av1.c
@@ -112,12 +112,12 @@ static void vaapi_encode_av1_trace_write_log(void *ctx,
 
 static av_cold int vaapi_encode_av1_get_encoder_caps(AVCodecContext *avctx)
 {
-    VAAPIEncodeContext *ctx = avctx->priv_data;
-    VAAPIEncodeAV1Context *priv = avctx->priv_data;
+    HWBaseEncodeContext *base_ctx = avctx->priv_data;
+    VAAPIEncodeAV1Context   *priv = avctx->priv_data;
 
     // Surfaces must be aligned to superblock boundaries.
-    ctx->surface_width  = FFALIGN(avctx->width,  priv->use_128x128_superblock ? 128 : 64);
-    ctx->surface_height = FFALIGN(avctx->height, priv->use_128x128_superblock ? 128 : 64);
+    base_ctx->surface_width  = FFALIGN(avctx->width,  priv->use_128x128_superblock ? 128 : 64);
+    base_ctx->surface_height = FFALIGN(avctx->height, priv->use_128x128_superblock ? 128 : 64);
 
     return 0;
 }
@@ -425,7 +425,7 @@ static int vaapi_encode_av1_init_sequence_params(AVCodecContext *avctx)
             framerate = 0;
 
         level = ff_av1_guess_level(avctx->bit_rate, priv->tier,
-                                   ctx->surface_width, ctx->surface_height,
+                                   base_ctx->surface_width, base_ctx->surface_height,
                                    priv->tile_rows * priv->tile_cols,
                                    priv->tile_cols, framerate);
         if (level) {
diff --git a/libavcodec/vaapi_encode_h264.c b/libavcodec/vaapi_encode_h264.c
index d41b9649e8..f69bbeebc1 100644
--- a/libavcodec/vaapi_encode_h264.c
+++ b/libavcodec/vaapi_encode_h264.c
@@ -1205,8 +1205,9 @@ static const VAAPIEncodeType vaapi_encode_type_h264 = {
 
 static av_cold int vaapi_encode_h264_init(AVCodecContext *avctx)
 {
-    VAAPIEncodeContext      *ctx = avctx->priv_data;
-    VAAPIEncodeH264Context *priv = avctx->priv_data;
+    HWBaseEncodeContext *base_ctx = avctx->priv_data;
+    VAAPIEncodeContext       *ctx = avctx->priv_data;
+    VAAPIEncodeH264Context  *priv = avctx->priv_data;
 
     ctx->codec = &vaapi_encode_type_h264;
 
@@ -1254,10 +1255,10 @@ static av_cold int vaapi_encode_h264_init(AVCodecContext *avctx)
         VA_ENC_PACKED_HEADER_SLICE    | // Slice headers.
         VA_ENC_PACKED_HEADER_MISC;      // SEI.
 
-    ctx->surface_width  = FFALIGN(avctx->width,  16);
-    ctx->surface_height = FFALIGN(avctx->height, 16);
+    base_ctx->surface_width  = FFALIGN(avctx->width,  16);
+    base_ctx->surface_height = FFALIGN(avctx->height, 16);
 
-    ctx->slice_block_height = ctx->slice_block_width = 16;
+    base_ctx->slice_block_height = base_ctx->slice_block_width = 16;
 
     if (priv->qp > 0)
         ctx->explicit_qp = priv->qp;
diff --git a/libavcodec/vaapi_encode_h265.c b/libavcodec/vaapi_encode_h265.c
index ec48773d95..ff153f778a 100644
--- a/libavcodec/vaapi_encode_h265.c
+++ b/libavcodec/vaapi_encode_h265.c
@@ -352,7 +352,7 @@ static int vaapi_encode_h265_init_sequence_params(AVCodecContext *avctx)
         const H265LevelDescriptor *level;
 
         level = ff_h265_guess_level(ptl, avctx->bit_rate,
-                                    ctx->surface_width, ctx->surface_height,
+                                    base_ctx->surface_width, base_ctx->surface_height,
                                     ctx->nb_slices, ctx->tile_rows, ctx->tile_cols,
                                     (base_ctx->b_per_p > 0) + 1);
         if (level) {
@@ -410,18 +410,18 @@ static int vaapi_encode_h265_init_sequence_params(AVCodecContext *avctx)
     sps->chroma_format_idc          = chroma_format;
     sps->separate_colour_plane_flag = 0;
 
-    sps->pic_width_in_luma_samples  = ctx->surface_width;
-    sps->pic_height_in_luma_samples = ctx->surface_height;
+    sps->pic_width_in_luma_samples  = base_ctx->surface_width;
+    sps->pic_height_in_luma_samples = base_ctx->surface_height;
 
-    if (avctx->width  != ctx->surface_width ||
-        avctx->height != ctx->surface_height) {
+    if (avctx->width  != base_ctx->surface_width ||
+        avctx->height != base_ctx->surface_height) {
         sps->conformance_window_flag = 1;
         sps->conf_win_left_offset   = 0;
         sps->conf_win_right_offset  =
-            (ctx->surface_width - avctx->width) >> desc->log2_chroma_w;
+            (base_ctx->surface_width - avctx->width) >> desc->log2_chroma_w;
         sps->conf_win_top_offset    = 0;
         sps->conf_win_bottom_offset =
-            (ctx->surface_height - avctx->height) >> desc->log2_chroma_h;
+            (base_ctx->surface_height - avctx->height) >> desc->log2_chroma_h;
     } else {
         sps->conformance_window_flag = 0;
     }
@@ -1197,11 +1197,12 @@ static int vaapi_encode_h265_init_slice_params(AVCodecContext *avctx,
 
 static av_cold int vaapi_encode_h265_get_encoder_caps(AVCodecContext *avctx)
 {
-    VAAPIEncodeContext      *ctx = avctx->priv_data;
-    VAAPIEncodeH265Context *priv = avctx->priv_data;
+    HWBaseEncodeContext *base_ctx = avctx->priv_data;
+    VAAPIEncodeH265Context  *priv = avctx->priv_data;
 
 #if VA_CHECK_VERSION(1, 13, 0)
     {
+        VAAPIEncodeContext *ctx = avctx->priv_data;
         VAConfigAttribValEncHEVCBlockSizes block_size;
         VAConfigAttrib attr;
         VAStatus vas;
@@ -1249,10 +1250,10 @@ static av_cold int vaapi_encode_h265_get_encoder_caps(AVCodecContext *avctx)
            "min CB size %dx%d.\n", priv->ctu_size, priv->ctu_size,
            priv->min_cb_size, priv->min_cb_size);
 
-    ctx->surface_width  = FFALIGN(avctx->width,  priv->min_cb_size);
-    ctx->surface_height = FFALIGN(avctx->height, priv->min_cb_size);
+    base_ctx->surface_width  = FFALIGN(avctx->width,  priv->min_cb_size);
+    base_ctx->surface_height = FFALIGN(avctx->height, priv->min_cb_size);
 
-    ctx->slice_block_width = ctx->slice_block_height = priv->ctu_size;
+    base_ctx->slice_block_width = base_ctx->slice_block_height = priv->ctu_size;
 
     return 0;
 }
diff --git a/libavcodec/vaapi_encode_mjpeg.c b/libavcodec/vaapi_encode_mjpeg.c
index 24acc88a21..6ed58d7118 100644
--- a/libavcodec/vaapi_encode_mjpeg.c
+++ b/libavcodec/vaapi_encode_mjpeg.c
@@ -439,14 +439,13 @@ static int vaapi_encode_mjpeg_init_slice_params(AVCodecContext *avctx,
 static av_cold int vaapi_encode_mjpeg_get_encoder_caps(AVCodecContext *avctx)
 {
     HWBaseEncodeContext *base_ctx = avctx->priv_data;
-    VAAPIEncodeContext *ctx = avctx->priv_data;
     const AVPixFmtDescriptor *desc;
 
     desc = av_pix_fmt_desc_get(base_ctx->input_frames->sw_format);
     av_assert0(desc);
 
-    ctx->surface_width  = FFALIGN(avctx->width,  8 << desc->log2_chroma_w);
-    ctx->surface_height = FFALIGN(avctx->height, 8 << desc->log2_chroma_h);
+    base_ctx->surface_width  = FFALIGN(avctx->width,  8 << desc->log2_chroma_w);
+    base_ctx->surface_height = FFALIGN(avctx->height, 8 << desc->log2_chroma_h);
 
     return 0;
 }
diff --git a/libavcodec/vaapi_encode_vp9.c b/libavcodec/vaapi_encode_vp9.c
index 36dafea2c0..5e1bf3a7c4 100644
--- a/libavcodec/vaapi_encode_vp9.c
+++ b/libavcodec/vaapi_encode_vp9.c
@@ -190,11 +190,11 @@ static int vaapi_encode_vp9_init_picture_params(AVCodecContext *avctx,
 
 static av_cold int vaapi_encode_vp9_get_encoder_caps(AVCodecContext *avctx)
 {
-    VAAPIEncodeContext *ctx = avctx->priv_data;
+    HWBaseEncodeContext *base_ctx = avctx->priv_data;
 
     // Surfaces must be aligned to 64x64 superblock boundaries.
-    ctx->surface_width  = FFALIGN(avctx->width,  64);
-    ctx->surface_height = FFALIGN(avctx->height, 64);
+    base_ctx->surface_width  = FFALIGN(avctx->width,  64);
+    base_ctx->surface_height = FFALIGN(avctx->height, 64);
 
     return 0;
 }
-- 
2.41.0.windows.1

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

^ permalink raw reply	[flat|nested] 18+ messages in thread

* [FFmpeg-devel] [PATCH v9 10/13] avcodec/vaapi_encode: extract a free funtion to base layer
  2024-05-20 14:52 [FFmpeg-devel] [PATCH v9 01/13] avcodec/vaapi_encode: introduce a base layer for vaapi encode tong1.wu-at-intel.com
                   ` (7 preceding siblings ...)
  2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 09/13] avcodec/vaapi_encode: extract a get_recon_format function " tong1.wu-at-intel.com
@ 2024-05-20 14:52 ` tong1.wu-at-intel.com
  2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 11/13] avutil/hwcontext_d3d12va: add Flags for resource creation tong1.wu-at-intel.com
                   ` (3 subsequent siblings)
  12 siblings, 0 replies; 18+ messages in thread
From: tong1.wu-at-intel.com @ 2024-05-20 14:52 UTC (permalink / raw)
  To: ffmpeg-devel; +Cc: Tong Wu

From: Tong Wu <tong1.wu@intel.com>

Signed-off-by: Tong Wu <tong1.wu@intel.com>
---
 libavcodec/hw_base_encode.c | 11 +++++++++++
 libavcodec/hw_base_encode.h |  2 ++
 libavcodec/vaapi_encode.c   |  6 +-----
 3 files changed, 14 insertions(+), 5 deletions(-)

diff --git a/libavcodec/hw_base_encode.c b/libavcodec/hw_base_encode.c
index 42c40cb48f..f743d119cd 100644
--- a/libavcodec/hw_base_encode.c
+++ b/libavcodec/hw_base_encode.c
@@ -748,6 +748,17 @@ fail:
     return err;
 }
 
+int ff_hw_base_encode_free(AVCodecContext *avctx, HWBaseEncodePicture *pic)
+{
+    av_frame_free(&pic->input_image);
+    av_frame_free(&pic->recon_image);
+
+    av_buffer_unref(&pic->opaque_ref);
+    av_freep(&pic->priv_data);
+
+    return 0;
+}
+
 int ff_hw_base_encode_init(AVCodecContext *avctx)
 {
     HWBaseEncodeContext *ctx = avctx->priv_data;
diff --git a/libavcodec/hw_base_encode.h b/libavcodec/hw_base_encode.h
index 7d40da039c..76a39e2d97 100644
--- a/libavcodec/hw_base_encode.h
+++ b/libavcodec/hw_base_encode.h
@@ -223,6 +223,8 @@ int ff_hw_base_init_gop_structure(AVCodecContext *avctx, uint32_t ref_l0, uint32
 
 int ff_hw_base_get_recon_format(AVCodecContext *avctx, const void *hwconfig, enum AVPixelFormat *fmt);
 
+int ff_hw_base_encode_free(AVCodecContext *avctx, HWBaseEncodePicture *pic);
+
 int ff_hw_base_encode_init(AVCodecContext *avctx);
 
 int ff_hw_base_encode_close(AVCodecContext *avctx);
diff --git a/libavcodec/vaapi_encode.c b/libavcodec/vaapi_encode.c
index 98b8c82da3..e89d6e01af 100644
--- a/libavcodec/vaapi_encode.c
+++ b/libavcodec/vaapi_encode.c
@@ -878,17 +878,13 @@ static int vaapi_encode_free(AVCodecContext *avctx,
             av_freep(&pic->slices[i].codec_slice_params);
     }
 
-    av_frame_free(&base_pic->input_image);
-    av_frame_free(&base_pic->recon_image);
-
-    av_buffer_unref(&base_pic->opaque_ref);
+    ff_hw_base_encode_free(avctx, base_pic);
 
     av_freep(&pic->param_buffers);
     av_freep(&pic->slices);
     // Output buffer should already be destroyed.
     av_assert0(pic->output_buffer == VA_INVALID_ID);
 
-    av_freep(&base_pic->priv_data);
     av_freep(&pic->codec_picture_params);
     av_freep(&pic->roi);
 
-- 
2.41.0.windows.1

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

^ permalink raw reply	[flat|nested] 18+ messages in thread

* [FFmpeg-devel] [PATCH v9 11/13] avutil/hwcontext_d3d12va: add Flags for resource creation
  2024-05-20 14:52 [FFmpeg-devel] [PATCH v9 01/13] avcodec/vaapi_encode: introduce a base layer for vaapi encode tong1.wu-at-intel.com
                   ` (8 preceding siblings ...)
  2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 10/13] avcodec/vaapi_encode: extract a free funtion " tong1.wu-at-intel.com
@ 2024-05-20 14:52 ` tong1.wu-at-intel.com
  2024-05-21 16:32   ` Andrew Sayers
  2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 12/13] avcodec: add D3D12VA hardware HEVC encoder tong1.wu-at-intel.com
                   ` (2 subsequent siblings)
  12 siblings, 1 reply; 18+ messages in thread
From: tong1.wu-at-intel.com @ 2024-05-20 14:52 UTC (permalink / raw)
  To: ffmpeg-devel; +Cc: Tong Wu

From: Tong Wu <tong1.wu@intel.com>

Flags field is added to support diffferent resource creation.

Signed-off-by: Tong Wu <tong1.wu@intel.com>
---
 doc/APIchanges                | 3 +++
 libavutil/hwcontext_d3d12va.c | 2 +-
 libavutil/hwcontext_d3d12va.h | 8 ++++++++
 libavutil/version.h           | 2 +-
 4 files changed, 13 insertions(+), 2 deletions(-)

diff --git a/doc/APIchanges b/doc/APIchanges
index 269fd36559..808ba02f2d 100644
--- a/doc/APIchanges
+++ b/doc/APIchanges
@@ -2,6 +2,9 @@ The last version increases of all libraries were on 2024-03-07
 
 API changes, most recent first:
 
+2024-01-xx - xxxxxxxxxx - lavu 59.20.100 - hwcontext_d3d12va.h
+ Add AVD3D12VAFramesContext.flags
+
 2024-05-xx - xxxxxxxxxx - lavu 59.19.100 - hwcontext_qsv.h
   Add AVQSVFramesContext.info
 
diff --git a/libavutil/hwcontext_d3d12va.c b/libavutil/hwcontext_d3d12va.c
index cfc016315d..6507cf69c1 100644
--- a/libavutil/hwcontext_d3d12va.c
+++ b/libavutil/hwcontext_d3d12va.c
@@ -247,7 +247,7 @@ static AVBufferRef *d3d12va_pool_alloc(void *opaque, size_t size)
         .Format           = hwctx->format,
         .SampleDesc       = {.Count = 1, .Quality = 0 },
         .Layout           = D3D12_TEXTURE_LAYOUT_UNKNOWN,
-        .Flags            = D3D12_RESOURCE_FLAG_NONE,
+        .Flags            = hwctx->flags,
     };
 
     frame = av_mallocz(sizeof(AVD3D12VAFrame));
diff --git a/libavutil/hwcontext_d3d12va.h b/libavutil/hwcontext_d3d12va.h
index ff06e6f2ef..608dbac97f 100644
--- a/libavutil/hwcontext_d3d12va.h
+++ b/libavutil/hwcontext_d3d12va.h
@@ -129,6 +129,14 @@ typedef struct AVD3D12VAFramesContext {
      * If unset, will be automatically set.
      */
     DXGI_FORMAT format;
+
+    /**
+     * This field is used to specify options for working with resources.
+     * If unset, this will be D3D12_RESOURCE_FLAG_NONE.
+     *
+     * @see: https://learn.microsoft.com/en-us/windows/win32/api/d3d12/ne-d3d12-d3d12_resource_flags.
+     */
+    D3D12_RESOURCE_FLAGS flags;
 } AVD3D12VAFramesContext;
 
 #endif /* AVUTIL_HWCONTEXT_D3D12VA_H */
diff --git a/libavutil/version.h b/libavutil/version.h
index 3221c4c592..9c7146c228 100644
--- a/libavutil/version.h
+++ b/libavutil/version.h
@@ -79,7 +79,7 @@
  */
 
 #define LIBAVUTIL_VERSION_MAJOR  59
-#define LIBAVUTIL_VERSION_MINOR  19
+#define LIBAVUTIL_VERSION_MINOR  20
 #define LIBAVUTIL_VERSION_MICRO 100
 
 #define LIBAVUTIL_VERSION_INT   AV_VERSION_INT(LIBAVUTIL_VERSION_MAJOR, \
-- 
2.41.0.windows.1

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

^ permalink raw reply	[flat|nested] 18+ messages in thread

* [FFmpeg-devel] [PATCH v9 12/13] avcodec: add D3D12VA hardware HEVC encoder
  2024-05-20 14:52 [FFmpeg-devel] [PATCH v9 01/13] avcodec/vaapi_encode: introduce a base layer for vaapi encode tong1.wu-at-intel.com
                   ` (9 preceding siblings ...)
  2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 11/13] avutil/hwcontext_d3d12va: add Flags for resource creation tong1.wu-at-intel.com
@ 2024-05-20 14:52 ` tong1.wu-at-intel.com
  2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 13/13] Changelog: add D3D12VA HEVC encoder changelog tong1.wu-at-intel.com
  2024-05-23 16:11 ` [FFmpeg-devel] [PATCH v9 01/13] avcodec/vaapi_encode: introduce a base layer for vaapi encode Lynne via ffmpeg-devel
  12 siblings, 0 replies; 18+ messages in thread
From: tong1.wu-at-intel.com @ 2024-05-20 14:52 UTC (permalink / raw)
  To: ffmpeg-devel; +Cc: Tong Wu

From: Tong Wu <tong1.wu@intel.com>

This implementation is based on D3D12 Video Encoding Spec:
https://microsoft.github.io/DirectX-Specs/d3d/D3D12VideoEncoding.html

Sample command line for transcoding:
ffmpeg.exe -hwaccel d3d12va -hwaccel_output_format d3d12 -i input.mp4
-c:v hevc_d3d12va output.mp4

Signed-off-by: Tong Wu <tong1.wu@intel.com>
---
 configure                        |    6 +
 libavcodec/Makefile              |    5 +-
 libavcodec/allcodecs.c           |    1 +
 libavcodec/d3d12va_encode.c      | 1558 ++++++++++++++++++++++++++++++
 libavcodec/d3d12va_encode.h      |  334 +++++++
 libavcodec/d3d12va_encode_hevc.c | 1007 +++++++++++++++++++
 6 files changed, 2910 insertions(+), 1 deletion(-)
 create mode 100644 libavcodec/d3d12va_encode.c
 create mode 100644 libavcodec/d3d12va_encode.h
 create mode 100644 libavcodec/d3d12va_encode_hevc.c

diff --git a/configure b/configure
index b16722d83d..127d68e60c 100755
--- a/configure
+++ b/configure
@@ -2551,6 +2551,7 @@ CONFIG_EXTRA="
     cbs_mpeg2
     cbs_vp8
     cbs_vp9
+    d3d12va_encode
     deflate_wrapper
     dirac_parse
     dnn
@@ -3287,6 +3288,7 @@ wmv3_vaapi_hwaccel_select="vc1_vaapi_hwaccel"
 wmv3_vdpau_hwaccel_select="vc1_vdpau_hwaccel"
 
 # hardware-accelerated codecs
+d3d12va_encode_deps="d3d12va ID3D12VideoEncoder d3d12_encoder_feature"
 mediafoundation_deps="mftransform_h MFCreateAlignedMemoryBuffer"
 omx_deps="libdl pthreads"
 omx_rpi_select="omx"
@@ -3354,6 +3356,7 @@ h264_v4l2m2m_encoder_deps="v4l2_m2m h264_v4l2_m2m"
 hevc_amf_encoder_deps="amf"
 hevc_cuvid_decoder_deps="cuvid"
 hevc_cuvid_decoder_select="hevc_mp4toannexb_bsf"
+hevc_d3d12va_encoder_select="cbs_h265 d3d12va_encode"
 hevc_mediacodec_decoder_deps="mediacodec"
 hevc_mediacodec_decoder_select="hevc_mp4toannexb_bsf hevc_parser"
 hevc_mediacodec_encoder_deps="mediacodec"
@@ -6725,6 +6728,9 @@ check_type "windows.h d3d11.h" "ID3D11VideoDecoder"
 check_type "windows.h d3d11.h" "ID3D11VideoContext"
 check_type "windows.h d3d12.h" "ID3D12Device"
 check_type "windows.h d3d12video.h" "ID3D12VideoDecoder"
+check_type "windows.h d3d12video.h" "ID3D12VideoEncoder"
+test_code cc "windows.h d3d12video.h" "D3D12_FEATURE_VIDEO feature = D3D12_FEATURE_VIDEO_ENCODER_CODEC" && \
+test_code cc "windows.h d3d12video.h" "D3D12_FEATURE_DATA_VIDEO_ENCODER_RESOURCE_REQUIREMENTS req" && enable d3d12_encoder_feature
 check_type "windows.h" "DPI_AWARENESS_CONTEXT" -D_WIN32_WINNT=0x0A00
 check_type "d3d9.h dxva2api.h" DXVA2_ConfigPictureDecode -D_WIN32_WINNT=0x0602
 check_func_headers mfapi.h MFCreateAlignedMemoryBuffer -lmfplat
diff --git a/libavcodec/Makefile b/libavcodec/Makefile
index 998f6b7e12..6c4500ce6d 100644
--- a/libavcodec/Makefile
+++ b/libavcodec/Makefile
@@ -86,6 +86,7 @@ OBJS-$(CONFIG_CBS_JPEG)                += cbs_jpeg.o
 OBJS-$(CONFIG_CBS_MPEG2)               += cbs_mpeg2.o
 OBJS-$(CONFIG_CBS_VP8)                 += cbs_vp8.o vp8data.o
 OBJS-$(CONFIG_CBS_VP9)                 += cbs_vp9.o
+OBJS-$(CONFIG_D3D12VA_ENCODE)          += d3d12va_encode.o hw_base_encode.o
 OBJS-$(CONFIG_DEFLATE_WRAPPER)         += zlib_wrapper.o
 OBJS-$(CONFIG_DOVI_RPUDEC)             += dovi_rpu.o dovi_rpudec.o
 OBJS-$(CONFIG_DOVI_RPUENC)             += dovi_rpu.o dovi_rpuenc.o
@@ -436,6 +437,8 @@ OBJS-$(CONFIG_HEVC_DECODER)            += hevcdec.o hevc_mvs.o \
                                           h274.o aom_film_grain.o
 OBJS-$(CONFIG_HEVC_AMF_ENCODER)        += amfenc_hevc.o
 OBJS-$(CONFIG_HEVC_CUVID_DECODER)      += cuviddec.o
+OBJS-$(CONFIG_HEVC_D3D12VA_ENCODER)    += d3d12va_encode_hevc.o h265_profile_level.o \
+                                          h2645data.o
 OBJS-$(CONFIG_HEVC_MEDIACODEC_DECODER) += mediacodecdec.o
 OBJS-$(CONFIG_HEVC_MEDIACODEC_ENCODER) += mediacodecenc.o
 OBJS-$(CONFIG_HEVC_MF_ENCODER)         += mfenc.o mf_utils.o
@@ -1265,7 +1268,7 @@ SKIPHEADERS                            += %_tablegen.h                  \
 
 SKIPHEADERS-$(CONFIG_AMF)              += amfenc.h
 SKIPHEADERS-$(CONFIG_D3D11VA)          += d3d11va.h dxva2_internal.h
-SKIPHEADERS-$(CONFIG_D3D12VA)          += d3d12va_decode.h
+SKIPHEADERS-$(CONFIG_D3D12VA)          += d3d12va_decode.h d3d12va_encode.h
 SKIPHEADERS-$(CONFIG_DXVA2)            += dxva2.h dxva2_internal.h
 SKIPHEADERS-$(CONFIG_JNI)              += ffjni.h
 SKIPHEADERS-$(CONFIG_LCMS2)            += fflcms2.h
diff --git a/libavcodec/allcodecs.c b/libavcodec/allcodecs.c
index b102a8069e..463ffbbd08 100644
--- a/libavcodec/allcodecs.c
+++ b/libavcodec/allcodecs.c
@@ -848,6 +848,7 @@ extern const FFCodec ff_h264_vaapi_encoder;
 extern const FFCodec ff_h264_videotoolbox_encoder;
 extern const FFCodec ff_hevc_amf_encoder;
 extern const FFCodec ff_hevc_cuvid_decoder;
+extern const FFCodec ff_hevc_d3d12va_encoder;
 extern const FFCodec ff_hevc_mediacodec_decoder;
 extern const FFCodec ff_hevc_mediacodec_encoder;
 extern const FFCodec ff_hevc_mf_encoder;
diff --git a/libavcodec/d3d12va_encode.c b/libavcodec/d3d12va_encode.c
new file mode 100644
index 0000000000..a20e6cc961
--- /dev/null
+++ b/libavcodec/d3d12va_encode.c
@@ -0,0 +1,1558 @@
+/*
+ * Direct3D 12 HW acceleration video encoder
+ *
+ * Copyright (c) 2024 Intel Corporation
+ *
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#include "libavutil/avassert.h"
+#include "libavutil/common.h"
+#include "libavutil/internal.h"
+#include "libavutil/log.h"
+#include "libavutil/mem.h"
+#include "libavutil/pixdesc.h"
+#include "libavutil/hwcontext_d3d12va_internal.h"
+#include "libavutil/hwcontext_d3d12va.h"
+
+#include "avcodec.h"
+#include "d3d12va_encode.h"
+#include "encode.h"
+
+const AVCodecHWConfigInternal *const ff_d3d12va_encode_hw_configs[] = {
+    HW_CONFIG_ENCODER_FRAMES(D3D12, D3D12VA),
+    NULL,
+};
+
+static int d3d12va_fence_completion(AVD3D12VASyncContext *psync_ctx)
+{
+    uint64_t completion = ID3D12Fence_GetCompletedValue(psync_ctx->fence);
+    if (completion < psync_ctx->fence_value) {
+        if (FAILED(ID3D12Fence_SetEventOnCompletion(psync_ctx->fence, psync_ctx->fence_value, psync_ctx->event)))
+            return AVERROR(EINVAL);
+
+        WaitForSingleObjectEx(psync_ctx->event, INFINITE, FALSE);
+    }
+
+    return 0;
+}
+
+static int d3d12va_sync_with_gpu(AVCodecContext *avctx)
+{
+    D3D12VAEncodeContext *ctx = avctx->priv_data;
+
+    DX_CHECK(ID3D12CommandQueue_Signal(ctx->command_queue, ctx->sync_ctx.fence, ++ctx->sync_ctx.fence_value));
+    return d3d12va_fence_completion(&ctx->sync_ctx);
+
+fail:
+    return AVERROR(EINVAL);
+}
+
+typedef struct CommandAllocator {
+    ID3D12CommandAllocator *command_allocator;
+    uint64_t fence_value;
+} CommandAllocator;
+
+static int d3d12va_get_valid_command_allocator(AVCodecContext *avctx, ID3D12CommandAllocator **ppAllocator)
+{
+    HRESULT hr;
+    D3D12VAEncodeContext *ctx = avctx->priv_data;
+    CommandAllocator allocator;
+
+    if (av_fifo_peek(ctx->allocator_queue, &allocator, 1, 0) >= 0) {
+        uint64_t completion = ID3D12Fence_GetCompletedValue(ctx->sync_ctx.fence);
+        if (completion >= allocator.fence_value) {
+            *ppAllocator = allocator.command_allocator;
+            av_fifo_read(ctx->allocator_queue, &allocator, 1);
+            return 0;
+        }
+    }
+
+    hr = ID3D12Device_CreateCommandAllocator(ctx->hwctx->device, D3D12_COMMAND_LIST_TYPE_VIDEO_ENCODE,
+                                             &IID_ID3D12CommandAllocator, (void **)ppAllocator);
+    if (FAILED(hr)) {
+        av_log(avctx, AV_LOG_ERROR, "Failed to create a new command allocator!\n");
+        return AVERROR(EINVAL);
+    }
+
+    return 0;
+}
+
+static int d3d12va_discard_command_allocator(AVCodecContext *avctx, ID3D12CommandAllocator *pAllocator, uint64_t fence_value)
+{
+    D3D12VAEncodeContext *ctx = avctx->priv_data;
+
+    CommandAllocator allocator = {
+        .command_allocator = pAllocator,
+        .fence_value = fence_value,
+    };
+
+    av_fifo_write(ctx->allocator_queue, &allocator, 1);
+
+    return 0;
+}
+
+static int d3d12va_encode_wait(AVCodecContext *avctx,
+                               D3D12VAEncodePicture *pic)
+{
+    D3D12VAEncodeContext *ctx     = avctx->priv_data;
+    HWBaseEncodePicture *base_pic = &pic->base;
+    uint64_t completion;
+
+    av_assert0(base_pic->encode_issued);
+
+    if (base_pic->encode_complete) {
+        // Already waited for this picture.
+        return 0;
+    }
+
+    completion = ID3D12Fence_GetCompletedValue(ctx->sync_ctx.fence);
+    if (completion < pic->fence_value) {
+        if (FAILED(ID3D12Fence_SetEventOnCompletion(ctx->sync_ctx.fence, pic->fence_value,
+                                                    ctx->sync_ctx.event)))
+            return AVERROR(EINVAL);
+
+        WaitForSingleObjectEx(ctx->sync_ctx.event, INFINITE, FALSE);
+    }
+
+    av_log(avctx, AV_LOG_DEBUG, "Sync to pic %"PRId64"/%"PRId64" "
+           "(input surface %p).\n", base_pic->display_order,
+           base_pic->encode_order, pic->input_surface->texture);
+
+    av_frame_free(&base_pic->input_image);
+
+    base_pic->encode_complete = 1;
+    return 0;
+}
+
+static int d3d12va_encode_create_metadata_buffers(AVCodecContext *avctx,
+                                                  D3D12VAEncodePicture *pic)
+{
+    D3D12VAEncodeContext *ctx = avctx->priv_data;
+    int width = sizeof(D3D12_VIDEO_ENCODER_OUTPUT_METADATA) + sizeof(D3D12_VIDEO_ENCODER_FRAME_SUBREGION_METADATA);
+    D3D12_HEAP_PROPERTIES encoded_meta_props = { .Type = D3D12_HEAP_TYPE_DEFAULT }, resolved_meta_props;
+    D3D12_HEAP_TYPE resolved_heap_type = D3D12_HEAP_TYPE_READBACK;
+    HRESULT hr;
+
+    D3D12_RESOURCE_DESC meta_desc = {
+        .Dimension        = D3D12_RESOURCE_DIMENSION_BUFFER,
+        .Alignment        = 0,
+        .Width            = ctx->req.MaxEncoderOutputMetadataBufferSize,
+        .Height           = 1,
+        .DepthOrArraySize = 1,
+        .MipLevels        = 1,
+        .Format           = DXGI_FORMAT_UNKNOWN,
+        .SampleDesc       = { .Count = 1, .Quality = 0 },
+        .Layout           = D3D12_TEXTURE_LAYOUT_ROW_MAJOR,
+        .Flags            = D3D12_RESOURCE_FLAG_NONE,
+    };
+
+    hr = ID3D12Device_CreateCommittedResource(ctx->hwctx->device, &encoded_meta_props, D3D12_HEAP_FLAG_NONE,
+                                              &meta_desc, D3D12_RESOURCE_STATE_COMMON, NULL,
+                                              &IID_ID3D12Resource, (void **)&pic->encoded_metadata);
+    if (FAILED(hr)) {
+        av_log(avctx, AV_LOG_ERROR, "Failed to create metadata buffer.\n");
+        return AVERROR_UNKNOWN;
+    }
+
+    ctx->hwctx->device->lpVtbl->GetCustomHeapProperties(ctx->hwctx->device, &resolved_meta_props, 0, resolved_heap_type);
+
+    meta_desc.Width = width;
+
+    hr = ID3D12Device_CreateCommittedResource(ctx->hwctx->device, &resolved_meta_props, D3D12_HEAP_FLAG_NONE,
+                                              &meta_desc, D3D12_RESOURCE_STATE_COMMON, NULL,
+                                              &IID_ID3D12Resource, (void **)&pic->resolved_metadata);
+
+    if (FAILED(hr)) {
+        av_log(avctx, AV_LOG_ERROR, "Failed to create output metadata buffer.\n");
+        return AVERROR_UNKNOWN;
+    }
+
+    return 0;
+}
+
+static int d3d12va_encode_issue(AVCodecContext *avctx,
+                                const HWBaseEncodePicture *base_pic)
+{
+    HWBaseEncodeContext *base_ctx = avctx->priv_data;
+    D3D12VAEncodeContext     *ctx = avctx->priv_data;
+    AVD3D12VAFramesContext *frames_hwctx = base_ctx->input_frames->hwctx;
+    D3D12VAEncodePicture *pic = (D3D12VAEncodePicture *)base_pic;
+    int err, i, j;
+    HRESULT hr;
+    char data[MAX_PARAM_BUFFER_SIZE];
+    void *ptr;
+    size_t bit_len;
+    ID3D12CommandAllocator *command_allocator = NULL;
+    ID3D12VideoEncodeCommandList2 *cmd_list = ctx->command_list;
+    D3D12_RESOURCE_BARRIER barriers[32] = { 0 };
+    D3D12_VIDEO_ENCODE_REFERENCE_FRAMES d3d12_refs = { 0 };
+
+    D3D12_VIDEO_ENCODER_ENCODEFRAME_INPUT_ARGUMENTS input_args = {
+        .SequenceControlDesc = {
+            .Flags = D3D12_VIDEO_ENCODER_SEQUENCE_CONTROL_FLAG_NONE,
+            .IntraRefreshConfig = { 0 },
+            .RateControl = ctx->rc,
+            .PictureTargetResolution = ctx->resolution,
+            .SelectedLayoutMode = D3D12_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_MODE_FULL_FRAME,
+            .FrameSubregionsLayoutData = { 0 },
+            .CodecGopSequence = ctx->gop,
+        },
+        .pInputFrame = pic->input_surface->texture,
+        .InputFrameSubresource = 0,
+    };
+
+    D3D12_VIDEO_ENCODER_ENCODEFRAME_OUTPUT_ARGUMENTS output_args = { 0 };
+
+    D3D12_VIDEO_ENCODER_RESOLVE_METADATA_INPUT_ARGUMENTS input_metadata = {
+        .EncoderCodec = ctx->codec->d3d12_codec,
+        .EncoderProfile = ctx->profile->d3d12_profile,
+        .EncoderInputFormat = frames_hwctx->format,
+        .EncodedPictureEffectiveResolution = ctx->resolution,
+    };
+
+    D3D12_VIDEO_ENCODER_RESOLVE_METADATA_OUTPUT_ARGUMENTS output_metadata = { 0 };
+
+    memset(data, 0, sizeof(data));
+
+    av_log(avctx, AV_LOG_DEBUG, "Issuing encode for pic %"PRId64"/%"PRId64" "
+           "as type %s.\n", base_pic->display_order, base_pic->encode_order,
+           ff_hw_base_encode_get_pictype_name(base_pic->type));
+    if (base_pic->nb_refs[0] == 0 && base_pic->nb_refs[1] == 0) {
+        av_log(avctx, AV_LOG_DEBUG, "No reference pictures.\n");
+    } else {
+        av_log(avctx, AV_LOG_DEBUG, "L0 refers to");
+        for (i = 0; i < base_pic->nb_refs[0]; i++) {
+            av_log(avctx, AV_LOG_DEBUG, " %"PRId64"/%"PRId64,
+                   base_pic->refs[0][i]->display_order, base_pic->refs[0][i]->encode_order);
+        }
+        av_log(avctx, AV_LOG_DEBUG, ".\n");
+
+        if (base_pic->nb_refs[1]) {
+            av_log(avctx, AV_LOG_DEBUG, "L1 refers to");
+            for (i = 0; i < base_pic->nb_refs[1]; i++) {
+                av_log(avctx, AV_LOG_DEBUG, " %"PRId64"/%"PRId64,
+                       base_pic->refs[1][i]->display_order, base_pic->refs[1][i]->encode_order);
+            }
+            av_log(avctx, AV_LOG_DEBUG, ".\n");
+        }
+    }
+
+    av_assert0(!base_pic->encode_issued);
+    for (i = 0; i < base_pic->nb_refs[0]; i++) {
+        av_assert0(base_pic->refs[0][i]);
+        av_assert0(base_pic->refs[0][i]->encode_issued);
+    }
+    for (i = 0; i < base_pic->nb_refs[1]; i++) {
+        av_assert0(base_pic->refs[1][i]);
+        av_assert0(base_pic->refs[1][i]->encode_issued);
+    }
+
+    av_log(avctx, AV_LOG_DEBUG, "Input surface is %p.\n", pic->input_surface->texture);
+
+    err = av_hwframe_get_buffer(base_ctx->recon_frames_ref, base_pic->recon_image, 0);
+    if (err < 0) {
+        err = AVERROR(ENOMEM);
+        goto fail;
+    }
+
+    pic->recon_surface = (AVD3D12VAFrame *)base_pic->recon_image->data[0];
+    av_log(avctx, AV_LOG_DEBUG, "Recon surface is %p.\n",
+           pic->recon_surface->texture);
+
+    pic->output_buffer_ref = av_buffer_pool_get(ctx->output_buffer_pool);
+    if (!pic->output_buffer_ref) {
+        err = AVERROR(ENOMEM);
+        goto fail;
+    }
+    pic->output_buffer = (ID3D12Resource *)pic->output_buffer_ref->data;
+    av_log(avctx, AV_LOG_DEBUG, "Output buffer is %p.\n",
+           pic->output_buffer);
+
+    err = d3d12va_encode_create_metadata_buffers(avctx, pic);
+    if (err < 0)
+        goto fail;
+
+    if (ctx->codec->init_picture_params) {
+        err = ctx->codec->init_picture_params(avctx, pic);
+        if (err < 0) {
+            av_log(avctx, AV_LOG_ERROR, "Failed to initialise picture "
+                   "parameters: %d.\n", err);
+            goto fail;
+        }
+    }
+
+    if (base_pic->type == PICTURE_TYPE_IDR) {
+        if (ctx->codec->write_sequence_header) {
+            bit_len = 8 * sizeof(data);
+            err = ctx->codec->write_sequence_header(avctx, data, &bit_len);
+            if (err < 0) {
+                av_log(avctx, AV_LOG_ERROR, "Failed to write per-sequence "
+                       "header: %d.\n", err);
+                goto fail;
+            }
+        }
+
+        pic->header_size = (int)bit_len / 8;
+        pic->header_size = pic->header_size % ctx->req.CompressedBitstreamBufferAccessAlignment ?
+                           FFALIGN(pic->header_size, ctx->req.CompressedBitstreamBufferAccessAlignment) :
+                           pic->header_size;
+
+        hr = ID3D12Resource_Map(pic->output_buffer, 0, NULL, (void **)&ptr);
+        if (FAILED(hr)) {
+            err = AVERROR_UNKNOWN;
+            goto fail;
+        }
+
+        memcpy(ptr, data, pic->header_size);
+        ID3D12Resource_Unmap(pic->output_buffer, 0, NULL);
+    }
+
+    d3d12_refs.NumTexture2Ds = base_pic->nb_refs[0] + base_pic->nb_refs[1];
+    if (d3d12_refs.NumTexture2Ds) {
+        d3d12_refs.ppTexture2Ds = av_calloc(d3d12_refs.NumTexture2Ds,
+                                            sizeof(*d3d12_refs.ppTexture2Ds));
+        if (!d3d12_refs.ppTexture2Ds) {
+            err = AVERROR(ENOMEM);
+            goto fail;
+        }
+
+        i = 0;
+        for (j = 0; j < base_pic->nb_refs[0]; j++)
+            d3d12_refs.ppTexture2Ds[i++] = ((D3D12VAEncodePicture *)base_pic->refs[0][j])->recon_surface->texture;
+        for (j = 0; j < base_pic->nb_refs[1]; j++)
+            d3d12_refs.ppTexture2Ds[i++] = ((D3D12VAEncodePicture *)base_pic->refs[1][j])->recon_surface->texture;
+    }
+
+    input_args.PictureControlDesc.IntraRefreshFrameIndex  = 0;
+    if (base_pic->is_reference)
+        input_args.PictureControlDesc.Flags |= D3D12_VIDEO_ENCODER_PICTURE_CONTROL_FLAG_USED_AS_REFERENCE_PICTURE;
+
+    input_args.PictureControlDesc.PictureControlCodecData = pic->pic_ctl;
+    input_args.PictureControlDesc.ReferenceFrames         = d3d12_refs;
+    input_args.CurrentFrameBitstreamMetadataSize          = pic->header_size;
+
+    output_args.Bitstream.pBuffer                                    = pic->output_buffer;
+    output_args.Bitstream.FrameStartOffset                           = pic->header_size;
+    output_args.ReconstructedPicture.pReconstructedPicture           = pic->recon_surface->texture;
+    output_args.ReconstructedPicture.ReconstructedPictureSubresource = 0;
+    output_args.EncoderOutputMetadata.pBuffer                        = pic->encoded_metadata;
+    output_args.EncoderOutputMetadata.Offset                         = 0;
+
+    input_metadata.HWLayoutMetadata.pBuffer = pic->encoded_metadata;
+    input_metadata.HWLayoutMetadata.Offset  = 0;
+
+    output_metadata.ResolvedLayoutMetadata.pBuffer = pic->resolved_metadata;
+    output_metadata.ResolvedLayoutMetadata.Offset  = 0;
+
+    err = d3d12va_get_valid_command_allocator(avctx, &command_allocator);
+    if (err < 0)
+        goto fail;
+
+    hr = ID3D12CommandAllocator_Reset(command_allocator);
+    if (FAILED(hr)) {
+        err = AVERROR_UNKNOWN;
+        goto fail;
+    }
+
+    hr = ID3D12VideoEncodeCommandList2_Reset(cmd_list, command_allocator);
+    if (FAILED(hr)) {
+        err = AVERROR_UNKNOWN;
+        goto fail;
+    }
+
+#define TRANSITION_BARRIER(res, before, after)                      \
+    (D3D12_RESOURCE_BARRIER) {                                      \
+        .Type  = D3D12_RESOURCE_BARRIER_TYPE_TRANSITION,            \
+        .Flags = D3D12_RESOURCE_BARRIER_FLAG_NONE,                  \
+        .Transition = {                                             \
+            .pResource   = res,                                     \
+            .Subresource = D3D12_RESOURCE_BARRIER_ALL_SUBRESOURCES, \
+            .StateBefore = before,                                  \
+            .StateAfter  = after,                                   \
+        },                                                          \
+    }
+
+    barriers[0] = TRANSITION_BARRIER(pic->input_surface->texture,
+                                     D3D12_RESOURCE_STATE_COMMON,
+                                     D3D12_RESOURCE_STATE_VIDEO_ENCODE_READ);
+    barriers[1] = TRANSITION_BARRIER(pic->output_buffer,
+                                     D3D12_RESOURCE_STATE_COMMON,
+                                     D3D12_RESOURCE_STATE_VIDEO_ENCODE_WRITE);
+    barriers[2] = TRANSITION_BARRIER(pic->recon_surface->texture,
+                                     D3D12_RESOURCE_STATE_COMMON,
+                                     D3D12_RESOURCE_STATE_VIDEO_ENCODE_WRITE);
+    barriers[3] = TRANSITION_BARRIER(pic->encoded_metadata,
+                                     D3D12_RESOURCE_STATE_COMMON,
+                                     D3D12_RESOURCE_STATE_VIDEO_ENCODE_WRITE);
+    barriers[4] = TRANSITION_BARRIER(pic->resolved_metadata,
+                                     D3D12_RESOURCE_STATE_COMMON,
+                                     D3D12_RESOURCE_STATE_VIDEO_ENCODE_WRITE);
+
+    ID3D12VideoEncodeCommandList2_ResourceBarrier(cmd_list, 5, barriers);
+
+    if (d3d12_refs.NumTexture2Ds) {
+        D3D12_RESOURCE_BARRIER refs_barriers[3];
+
+        for (i = 0; i < d3d12_refs.NumTexture2Ds; i++)
+            refs_barriers[i] = TRANSITION_BARRIER(d3d12_refs.ppTexture2Ds[i],
+                                                  D3D12_RESOURCE_STATE_COMMON,
+                                                  D3D12_RESOURCE_STATE_VIDEO_ENCODE_READ);
+
+        ID3D12VideoEncodeCommandList2_ResourceBarrier(cmd_list, d3d12_refs.NumTexture2Ds,
+                                                      refs_barriers);
+    }
+
+    ID3D12VideoEncodeCommandList2_EncodeFrame(cmd_list, ctx->encoder, ctx->encoder_heap,
+                                              &input_args, &output_args);
+
+    barriers[3] = TRANSITION_BARRIER(pic->encoded_metadata,
+                                     D3D12_RESOURCE_STATE_VIDEO_ENCODE_WRITE,
+                                     D3D12_RESOURCE_STATE_VIDEO_ENCODE_READ);
+
+    ID3D12VideoEncodeCommandList2_ResourceBarrier(cmd_list, 1, &barriers[3]);
+
+    ID3D12VideoEncodeCommandList2_ResolveEncoderOutputMetadata(cmd_list, &input_metadata, &output_metadata);
+
+    if (d3d12_refs.NumTexture2Ds) {
+        D3D12_RESOURCE_BARRIER refs_barriers[3];
+
+        for (i = 0; i < d3d12_refs.NumTexture2Ds; i++)
+            refs_barriers[i] = TRANSITION_BARRIER(d3d12_refs.ppTexture2Ds[i],
+                                                  D3D12_RESOURCE_STATE_VIDEO_ENCODE_READ,
+                                                  D3D12_RESOURCE_STATE_COMMON);
+
+        ID3D12VideoEncodeCommandList2_ResourceBarrier(cmd_list, d3d12_refs.NumTexture2Ds,
+                                                      refs_barriers);
+    }
+
+    barriers[0] = TRANSITION_BARRIER(pic->input_surface->texture,
+                                     D3D12_RESOURCE_STATE_VIDEO_ENCODE_READ,
+                                     D3D12_RESOURCE_STATE_COMMON);
+    barriers[1] = TRANSITION_BARRIER(pic->output_buffer,
+                                     D3D12_RESOURCE_STATE_VIDEO_ENCODE_WRITE,
+                                     D3D12_RESOURCE_STATE_COMMON);
+    barriers[2] = TRANSITION_BARRIER(pic->recon_surface->texture,
+                                     D3D12_RESOURCE_STATE_VIDEO_ENCODE_WRITE,
+                                     D3D12_RESOURCE_STATE_COMMON);
+    barriers[3] = TRANSITION_BARRIER(pic->encoded_metadata,
+                                     D3D12_RESOURCE_STATE_VIDEO_ENCODE_READ,
+                                     D3D12_RESOURCE_STATE_COMMON);
+    barriers[4] = TRANSITION_BARRIER(pic->resolved_metadata,
+                                     D3D12_RESOURCE_STATE_VIDEO_ENCODE_WRITE,
+                                     D3D12_RESOURCE_STATE_COMMON);
+
+    ID3D12VideoEncodeCommandList2_ResourceBarrier(cmd_list, 5, barriers);
+
+    hr = ID3D12VideoEncodeCommandList2_Close(cmd_list);
+    if (FAILED(hr)) {
+        err = AVERROR_UNKNOWN;
+        goto fail;
+    }
+
+    hr = ID3D12CommandQueue_Wait(ctx->command_queue, pic->input_surface->sync_ctx.fence,
+                                 pic->input_surface->sync_ctx.fence_value);
+    if (FAILED(hr)) {
+        err = AVERROR_UNKNOWN;
+        goto fail;
+    }
+
+    ID3D12CommandQueue_ExecuteCommandLists(ctx->command_queue, 1, (ID3D12CommandList **)&ctx->command_list);
+
+    hr = ID3D12CommandQueue_Signal(ctx->command_queue, pic->input_surface->sync_ctx.fence,
+                                   ++pic->input_surface->sync_ctx.fence_value);
+    if (FAILED(hr)) {
+        err = AVERROR_UNKNOWN;
+        goto fail;
+    }
+
+    hr = ID3D12CommandQueue_Signal(ctx->command_queue, ctx->sync_ctx.fence, ++ctx->sync_ctx.fence_value);
+    if (FAILED(hr)) {
+        err = AVERROR_UNKNOWN;
+        goto fail;
+    }
+
+    err = d3d12va_discard_command_allocator(avctx, command_allocator, ctx->sync_ctx.fence_value);
+    if (err < 0)
+        goto fail;
+
+    pic->fence_value = ctx->sync_ctx.fence_value;
+
+    if (d3d12_refs.ppTexture2Ds)
+        av_freep(&d3d12_refs.ppTexture2Ds);
+
+    return 0;
+
+fail:
+    if (command_allocator)
+        d3d12va_discard_command_allocator(avctx, command_allocator, ctx->sync_ctx.fence_value);
+
+    if (d3d12_refs.ppTexture2Ds)
+        av_freep(&d3d12_refs.ppTexture2Ds);
+
+    if (ctx->codec->free_picture_params)
+        ctx->codec->free_picture_params(pic);
+
+    av_buffer_unref(&pic->output_buffer_ref);
+    pic->output_buffer = NULL;
+    D3D12_OBJECT_RELEASE(pic->encoded_metadata);
+    D3D12_OBJECT_RELEASE(pic->resolved_metadata);
+    return err;
+}
+
+static int d3d12va_encode_discard(AVCodecContext *avctx,
+                                  D3D12VAEncodePicture *pic)
+{
+    HWBaseEncodePicture *base_pic = &pic->base;
+    d3d12va_encode_wait(avctx, pic);
+
+    if (pic->output_buffer_ref) {
+        av_log(avctx, AV_LOG_DEBUG, "Discard output for pic "
+               "%"PRId64"/%"PRId64".\n",
+               base_pic->display_order, base_pic->encode_order);
+
+        av_buffer_unref(&pic->output_buffer_ref);
+        pic->output_buffer = NULL;
+    }
+
+    D3D12_OBJECT_RELEASE(pic->encoded_metadata);
+    D3D12_OBJECT_RELEASE(pic->resolved_metadata);
+
+    return 0;
+}
+
+static int d3d12va_encode_free_rc_params(AVCodecContext *avctx)
+{
+    D3D12VAEncodeContext *ctx = avctx->priv_data;
+
+    switch (ctx->rc.Mode)
+    {
+    case D3D12_VIDEO_ENCODER_RATE_CONTROL_MODE_CQP:
+        av_freep(&ctx->rc.ConfigParams.pConfiguration_CQP);
+        break;
+    case D3D12_VIDEO_ENCODER_RATE_CONTROL_MODE_CBR:
+        av_freep(&ctx->rc.ConfigParams.pConfiguration_CBR);
+        break;
+    case D3D12_VIDEO_ENCODER_RATE_CONTROL_MODE_VBR:
+        av_freep(&ctx->rc.ConfigParams.pConfiguration_VBR);
+        break;
+    case D3D12_VIDEO_ENCODER_RATE_CONTROL_MODE_QVBR:
+        av_freep(&ctx->rc.ConfigParams.pConfiguration_QVBR);
+        break;
+    default:
+        break;
+    }
+
+    return 0;
+}
+
+static HWBaseEncodePicture *d3d12va_encode_alloc(AVCodecContext *avctx,
+                                                 const AVFrame *frame)
+{
+    D3D12VAEncodeContext *ctx = avctx->priv_data;
+    D3D12VAEncodePicture *pic;
+
+    pic = av_mallocz(sizeof(*pic));
+    if (!pic)
+        return NULL;
+
+    if (ctx->codec->picture_priv_data_size > 0) {
+        pic->base.priv_data = av_mallocz(ctx->codec->picture_priv_data_size);
+        if (!pic->base.priv_data) {
+            av_freep(&pic);
+            return NULL;
+        }
+    }
+
+    pic->input_surface = (AVD3D12VAFrame *)frame->data[0];
+
+    return &pic->base;
+}
+
+static int d3d12va_encode_free(AVCodecContext *avctx,
+                               HWBaseEncodePicture *base_pic)
+{
+    D3D12VAEncodeContext *ctx = avctx->priv_data;
+    D3D12VAEncodePicture *pic = (D3D12VAEncodePicture *)base_pic;
+
+    if (base_pic->encode_issued)
+        d3d12va_encode_discard(avctx, pic);
+
+    if (ctx->codec->free_picture_params)
+        ctx->codec->free_picture_params(pic);
+
+    ff_hw_base_encode_free(avctx, base_pic);
+
+    av_free(pic);
+
+    return 0;
+}
+
+static int d3d12va_encode_get_buffer_size(AVCodecContext *avctx,
+                                          D3D12VAEncodePicture *pic, size_t *size)
+{
+    D3D12_VIDEO_ENCODER_OUTPUT_METADATA *meta = NULL;
+    uint8_t *data;
+    HRESULT hr;
+    int err;
+
+    hr = ID3D12Resource_Map(pic->resolved_metadata, 0, NULL, (void **)&data);
+    if (FAILED(hr)) {
+        err = AVERROR_UNKNOWN;
+        return err;
+    }
+
+    meta = (D3D12_VIDEO_ENCODER_OUTPUT_METADATA *)data;
+
+    if (meta->EncodeErrorFlags != D3D12_VIDEO_ENCODER_ENCODE_ERROR_FLAG_NO_ERROR) {
+        av_log(avctx, AV_LOG_ERROR, "Encode failed %"PRIu64"\n", meta->EncodeErrorFlags);
+        err = AVERROR(EINVAL);
+        return err;
+    }
+
+    if (meta->EncodedBitstreamWrittenBytesCount == 0) {
+        av_log(avctx, AV_LOG_ERROR, "No bytes were written to encoded bitstream\n");
+        err = AVERROR(EINVAL);
+        return err;
+    }
+
+    *size = meta->EncodedBitstreamWrittenBytesCount;
+
+    ID3D12Resource_Unmap(pic->resolved_metadata, 0, NULL);
+
+    return 0;
+}
+
+static int d3d12va_encode_get_coded_data(AVCodecContext *avctx,
+                                         D3D12VAEncodePicture *pic, AVPacket *pkt)
+{
+    int err;
+    uint8_t *ptr, *mapped_data;
+    size_t total_size = 0;
+    HRESULT hr;
+
+    err = d3d12va_encode_get_buffer_size(avctx, pic, &total_size);
+    if (err < 0)
+        goto end;
+
+    total_size += pic->header_size;
+    av_log(avctx, AV_LOG_DEBUG, "Output buffer size %"PRId64"\n", total_size);
+
+    hr = ID3D12Resource_Map(pic->output_buffer, 0, NULL, (void **)&mapped_data);
+    if (FAILED(hr)) {
+        err = AVERROR_UNKNOWN;
+        goto end;
+    }
+
+    err = ff_get_encode_buffer(avctx, pkt, total_size, 0);
+    if (err < 0)
+        goto end;
+    ptr = pkt->data;
+
+    memcpy(ptr, mapped_data, total_size);
+
+    ID3D12Resource_Unmap(pic->output_buffer, 0, NULL);
+
+end:
+    av_buffer_unref(&pic->output_buffer_ref);
+    pic->output_buffer = NULL;
+    return err;
+}
+
+static int d3d12va_encode_output(AVCodecContext *avctx,
+                                 const HWBaseEncodePicture *base_pic, AVPacket *pkt)
+{
+    D3D12VAEncodePicture *pic = (D3D12VAEncodePicture *)base_pic;
+    AVPacket *pkt_ptr = pkt;
+    int err;
+
+    err = d3d12va_encode_wait(avctx, pic);
+    if (err < 0)
+        return err;
+
+    err = d3d12va_encode_get_coded_data(avctx, pic, pkt);
+    if (err < 0)
+        return err;
+
+    av_log(avctx, AV_LOG_DEBUG, "Output read for pic %"PRId64"/%"PRId64".\n",
+           base_pic->display_order, base_pic->encode_order);
+
+    ff_hw_base_encode_set_output_property(avctx, (HWBaseEncodePicture *)base_pic, pkt_ptr, 0);
+
+    return 0;
+}
+
+static int d3d12va_encode_set_profile(AVCodecContext *avctx)
+{
+    HWBaseEncodeContext *base_ctx = avctx->priv_data;
+    D3D12VAEncodeContext *ctx     = avctx->priv_data;
+    const D3D12VAEncodeProfile *profile;
+    const AVPixFmtDescriptor *desc;
+    int i, depth;
+
+    desc = av_pix_fmt_desc_get(base_ctx->input_frames->sw_format);
+    if (!desc) {
+        av_log(avctx, AV_LOG_ERROR, "Invalid input pixfmt (%d).\n",
+               base_ctx->input_frames->sw_format);
+        return AVERROR(EINVAL);
+    }
+
+    depth = desc->comp[0].depth;
+    for (i = 1; i < desc->nb_components; i++) {
+        if (desc->comp[i].depth != depth) {
+            av_log(avctx, AV_LOG_ERROR, "Invalid input pixfmt (%s).\n",
+                   desc->name);
+            return AVERROR(EINVAL);
+        }
+    }
+    av_log(avctx, AV_LOG_VERBOSE, "Input surface format is %s.\n",
+           desc->name);
+
+    av_assert0(ctx->codec->profiles);
+    for (i = 0; (ctx->codec->profiles[i].av_profile !=
+                 AV_PROFILE_UNKNOWN); i++) {
+        profile = &ctx->codec->profiles[i];
+        if (depth               != profile->depth ||
+            desc->nb_components != profile->nb_components)
+            continue;
+        if (desc->nb_components > 1 &&
+            (desc->log2_chroma_w != profile->log2_chroma_w ||
+             desc->log2_chroma_h != profile->log2_chroma_h))
+            continue;
+        if (avctx->profile != profile->av_profile &&
+            avctx->profile != AV_PROFILE_UNKNOWN)
+            continue;
+
+        ctx->profile = profile;
+        break;
+    }
+    if (!ctx->profile) {
+        av_log(avctx, AV_LOG_ERROR, "No usable encoding profile found.\n");
+        return AVERROR(ENOSYS);
+    }
+
+    avctx->profile = profile->av_profile;
+    return 0;
+}
+
+static const D3D12VAEncodeRCMode d3d12va_encode_rc_modes[] = {
+    //                   Bitrate   Quality
+    //                      | Maxrate | HRD/VBV
+    { 0 }, //               |    |    |    |
+    { RC_MODE_CQP,  "CQP",  0,   0,   1,   0, D3D12_VIDEO_ENCODER_RATE_CONTROL_MODE_CQP },
+    { RC_MODE_CBR,  "CBR",  1,   0,   0,   1, D3D12_VIDEO_ENCODER_RATE_CONTROL_MODE_CBR },
+    { RC_MODE_VBR,  "VBR",  1,   1,   0,   1, D3D12_VIDEO_ENCODER_RATE_CONTROL_MODE_VBR },
+    { RC_MODE_QVBR, "QVBR", 1,   1,   1,   1, D3D12_VIDEO_ENCODER_RATE_CONTROL_MODE_QVBR },
+};
+
+static int check_rate_control_support(AVCodecContext *avctx, const D3D12VAEncodeRCMode *rc_mode)
+{
+    HRESULT hr;
+    D3D12VAEncodeContext *ctx = avctx->priv_data;
+    D3D12_FEATURE_DATA_VIDEO_ENCODER_RATE_CONTROL_MODE d3d12_rc_mode = {
+        .Codec = ctx->codec->d3d12_codec,
+    };
+
+    if (!rc_mode->d3d12_mode)
+        return 0;
+
+    d3d12_rc_mode.IsSupported = 0;
+    d3d12_rc_mode.RateControlMode = rc_mode->d3d12_mode;
+
+    hr = ID3D12VideoDevice3_CheckFeatureSupport(ctx->video_device3,
+                                                D3D12_FEATURE_VIDEO_ENCODER_RATE_CONTROL_MODE,
+                                                &d3d12_rc_mode, sizeof(d3d12_rc_mode));
+    if (FAILED(hr)) {
+        av_log(avctx, AV_LOG_ERROR, "Failed to check rate control support.\n");
+        return 0;
+    }
+
+    return d3d12_rc_mode.IsSupported;
+}
+
+static int d3d12va_encode_init_rate_control(AVCodecContext *avctx)
+{
+    D3D12VAEncodeContext *ctx = avctx->priv_data;
+    int64_t rc_target_bitrate;
+    int64_t rc_peak_bitrate;
+    int     rc_quality;
+    int64_t hrd_buffer_size;
+    int64_t hrd_initial_buffer_fullness;
+    int fr_num, fr_den;
+    const D3D12VAEncodeRCMode *rc_mode;
+
+    // Rate control mode selection:
+    // * If the user has set a mode explicitly with the rc_mode option,
+    //   use it and fail if it is not available.
+    // * If an explicit QP option has been set, use CQP.
+    // * If the codec is CQ-only, use CQP.
+    // * If the QSCALE avcodec option is set, use CQP.
+    // * If bitrate and quality are both set, try QVBR.
+    // * If quality is set, try CQP.
+    // * If bitrate and maxrate are set and have the same value, try CBR.
+    // * If a bitrate is set, try VBR, then CBR.
+    // * If no bitrate is set, try CQP.
+
+#define TRY_RC_MODE(mode, fail) do { \
+        rc_mode = &d3d12va_encode_rc_modes[mode]; \
+        if (!(rc_mode->d3d12_mode && check_rate_control_support(avctx, rc_mode))) { \
+            if (fail) { \
+                av_log(avctx, AV_LOG_ERROR, "Driver does not support %s " \
+                       "RC mode.\n", rc_mode->name); \
+                return AVERROR(EINVAL); \
+            } \
+            av_log(avctx, AV_LOG_DEBUG, "Driver does not support %s " \
+                   "RC mode.\n", rc_mode->name); \
+            rc_mode = NULL; \
+        } else { \
+            goto rc_mode_found; \
+        } \
+    } while (0)
+
+    if (ctx->explicit_rc_mode)
+        TRY_RC_MODE(ctx->explicit_rc_mode, 1);
+
+    if (ctx->explicit_qp)
+        TRY_RC_MODE(RC_MODE_CQP, 1);
+
+    if (ctx->codec->flags & FLAG_CONSTANT_QUALITY_ONLY)
+        TRY_RC_MODE(RC_MODE_CQP, 1);
+
+    if (avctx->flags & AV_CODEC_FLAG_QSCALE)
+        TRY_RC_MODE(RC_MODE_CQP, 1);
+
+    if (avctx->bit_rate > 0 && avctx->global_quality > 0)
+        TRY_RC_MODE(RC_MODE_QVBR, 0);
+
+    if (avctx->global_quality > 0) {
+        TRY_RC_MODE(RC_MODE_CQP, 0);
+    }
+
+    if (avctx->bit_rate > 0 && avctx->rc_max_rate == avctx->bit_rate)
+        TRY_RC_MODE(RC_MODE_CBR, 0);
+
+    if (avctx->bit_rate > 0) {
+        TRY_RC_MODE(RC_MODE_VBR, 0);
+        TRY_RC_MODE(RC_MODE_CBR, 0);
+    } else {
+        TRY_RC_MODE(RC_MODE_CQP, 0);
+    }
+
+    av_log(avctx, AV_LOG_ERROR, "Driver does not support any "
+           "RC mode compatible with selected options.\n");
+    return AVERROR(EINVAL);
+
+rc_mode_found:
+    if (rc_mode->bitrate) {
+        if (avctx->bit_rate <= 0) {
+            av_log(avctx, AV_LOG_ERROR, "Bitrate must be set for %s "
+                   "RC mode.\n", rc_mode->name);
+            return AVERROR(EINVAL);
+        }
+
+        if (rc_mode->maxrate) {
+            if (avctx->rc_max_rate > 0) {
+                if (avctx->rc_max_rate < avctx->bit_rate) {
+                    av_log(avctx, AV_LOG_ERROR, "Invalid bitrate settings: "
+                           "bitrate (%"PRId64") must not be greater than "
+                           "maxrate (%"PRId64").\n", avctx->bit_rate,
+                           avctx->rc_max_rate);
+                    return AVERROR(EINVAL);
+                }
+                rc_target_bitrate = avctx->bit_rate;
+                rc_peak_bitrate   = avctx->rc_max_rate;
+            } else {
+                // We only have a target bitrate, but this mode requires
+                // that a maximum rate be supplied as well.  Since the
+                // user does not want this to be a constraint, arbitrarily
+                // pick a maximum rate of double the target rate.
+                rc_target_bitrate = avctx->bit_rate;
+                rc_peak_bitrate   = 2 * avctx->bit_rate;
+            }
+        } else {
+            if (avctx->rc_max_rate > avctx->bit_rate) {
+                av_log(avctx, AV_LOG_WARNING, "Max bitrate is ignored "
+                       "in %s RC mode.\n", rc_mode->name);
+            }
+            rc_target_bitrate = avctx->bit_rate;
+            rc_peak_bitrate   = 0;
+        }
+    } else {
+        rc_target_bitrate = 0;
+        rc_peak_bitrate   = 0;
+    }
+
+    if (rc_mode->quality) {
+        if (ctx->explicit_qp) {
+            rc_quality = ctx->explicit_qp;
+        } else if (avctx->global_quality > 0) {
+            if (avctx->flags & AV_CODEC_FLAG_QSCALE)
+                rc_quality = avctx->global_quality / FF_QP2LAMBDA;
+            else
+                rc_quality = avctx->global_quality;
+        } else {
+            rc_quality = ctx->codec->default_quality;
+            av_log(avctx, AV_LOG_WARNING, "No quality level set; "
+                   "using default (%d).\n", rc_quality);
+        }
+    } else {
+        rc_quality = 0;
+    }
+
+    if (rc_mode->hrd) {
+        if (avctx->rc_buffer_size)
+            hrd_buffer_size = avctx->rc_buffer_size;
+        else if (avctx->rc_max_rate > 0)
+            hrd_buffer_size = avctx->rc_max_rate;
+        else
+            hrd_buffer_size = avctx->bit_rate;
+        if (avctx->rc_initial_buffer_occupancy) {
+            if (avctx->rc_initial_buffer_occupancy > hrd_buffer_size) {
+                av_log(avctx, AV_LOG_ERROR, "Invalid RC buffer settings: "
+                       "must have initial buffer size (%d) <= "
+                       "buffer size (%"PRId64").\n",
+                       avctx->rc_initial_buffer_occupancy, hrd_buffer_size);
+                return AVERROR(EINVAL);
+            }
+            hrd_initial_buffer_fullness = avctx->rc_initial_buffer_occupancy;
+        } else {
+            hrd_initial_buffer_fullness = hrd_buffer_size * 3 / 4;
+        }
+    } else {
+        if (avctx->rc_buffer_size || avctx->rc_initial_buffer_occupancy) {
+            av_log(avctx, AV_LOG_WARNING, "Buffering settings are ignored "
+                   "in %s RC mode.\n", rc_mode->name);
+        }
+
+        hrd_buffer_size             = 0;
+        hrd_initial_buffer_fullness = 0;
+    }
+
+    if (rc_target_bitrate          > UINT32_MAX ||
+        hrd_buffer_size             > UINT32_MAX ||
+        hrd_initial_buffer_fullness > UINT32_MAX) {
+        av_log(avctx, AV_LOG_ERROR, "RC parameters of 2^32 or "
+               "greater are not supported by D3D12.\n");
+        return AVERROR(EINVAL);
+    }
+
+    ctx->rc_quality = rc_quality;
+
+    av_log(avctx, AV_LOG_VERBOSE, "RC mode: %s.\n", rc_mode->name);
+
+    if (rc_mode->quality)
+        av_log(avctx, AV_LOG_VERBOSE, "RC quality: %d.\n", rc_quality);
+
+    if (rc_mode->hrd) {
+        av_log(avctx, AV_LOG_VERBOSE, "RC buffer: %"PRId64" bits, "
+               "initial fullness %"PRId64" bits.\n",
+               hrd_buffer_size, hrd_initial_buffer_fullness);
+    }
+
+    if (avctx->framerate.num > 0 && avctx->framerate.den > 0)
+        av_reduce(&fr_num, &fr_den,
+                  avctx->framerate.num, avctx->framerate.den, 65535);
+    else
+        av_reduce(&fr_num, &fr_den,
+                  avctx->time_base.den, avctx->time_base.num, 65535);
+
+    av_log(avctx, AV_LOG_VERBOSE, "RC framerate: %d/%d (%.2f fps).\n",
+           fr_num, fr_den, (double)fr_num / fr_den);
+
+    ctx->rc.Flags                       = D3D12_VIDEO_ENCODER_RATE_CONTROL_FLAG_NONE;
+    ctx->rc.TargetFrameRate.Numerator   = fr_num;
+    ctx->rc.TargetFrameRate.Denominator = fr_den;
+    ctx->rc.Mode                        = rc_mode->d3d12_mode;
+
+    switch (rc_mode->mode) {
+        case RC_MODE_CQP:
+            // cqp ConfigParams will be updated in ctx->codec->configure.
+            break;
+
+        case RC_MODE_CBR:
+            D3D12_VIDEO_ENCODER_RATE_CONTROL_CBR *cbr_ctl;
+
+            ctx->rc.ConfigParams.DataSize = sizeof(D3D12_VIDEO_ENCODER_RATE_CONTROL_CBR);
+            cbr_ctl = av_mallocz(ctx->rc.ConfigParams.DataSize);
+            if (!cbr_ctl)
+                return AVERROR(ENOMEM);
+
+            cbr_ctl->TargetBitRate      = rc_target_bitrate;
+            cbr_ctl->VBVCapacity        = hrd_buffer_size;
+            cbr_ctl->InitialVBVFullness = hrd_initial_buffer_fullness;
+            ctx->rc.Flags |= D3D12_VIDEO_ENCODER_RATE_CONTROL_FLAG_ENABLE_VBV_SIZES;
+
+            if (avctx->qmin > 0 || avctx->qmax > 0) {
+                cbr_ctl->MinQP = avctx->qmin;
+                cbr_ctl->MaxQP = avctx->qmax;
+                ctx->rc.Flags |= D3D12_VIDEO_ENCODER_RATE_CONTROL_FLAG_ENABLE_QP_RANGE;
+            }
+
+            ctx->rc.ConfigParams.pConfiguration_CBR = cbr_ctl;
+            break;
+
+        case RC_MODE_VBR:
+            D3D12_VIDEO_ENCODER_RATE_CONTROL_VBR *vbr_ctl;
+
+            ctx->rc.ConfigParams.DataSize = sizeof(D3D12_VIDEO_ENCODER_RATE_CONTROL_VBR);
+            vbr_ctl = av_mallocz(ctx->rc.ConfigParams.DataSize);
+            if (!vbr_ctl)
+                return AVERROR(ENOMEM);
+
+            vbr_ctl->TargetAvgBitRate   = rc_target_bitrate;
+            vbr_ctl->PeakBitRate        = rc_peak_bitrate;
+            vbr_ctl->VBVCapacity        = hrd_buffer_size;
+            vbr_ctl->InitialVBVFullness = hrd_initial_buffer_fullness;
+            ctx->rc.Flags |= D3D12_VIDEO_ENCODER_RATE_CONTROL_FLAG_ENABLE_VBV_SIZES;
+
+            if (avctx->qmin > 0 || avctx->qmax > 0) {
+                vbr_ctl->MinQP = avctx->qmin;
+                vbr_ctl->MaxQP = avctx->qmax;
+                ctx->rc.Flags |= D3D12_VIDEO_ENCODER_RATE_CONTROL_FLAG_ENABLE_QP_RANGE;
+            }
+
+            ctx->rc.ConfigParams.pConfiguration_VBR = vbr_ctl;
+            break;
+
+        case RC_MODE_QVBR:
+            D3D12_VIDEO_ENCODER_RATE_CONTROL_QVBR *qvbr_ctl;
+
+            ctx->rc.ConfigParams.DataSize = sizeof(D3D12_VIDEO_ENCODER_RATE_CONTROL_QVBR);
+            qvbr_ctl = av_mallocz(ctx->rc.ConfigParams.DataSize);
+            if (!qvbr_ctl)
+                return AVERROR(ENOMEM);
+
+            qvbr_ctl->TargetAvgBitRate      = rc_target_bitrate;
+            qvbr_ctl->PeakBitRate           = rc_peak_bitrate;
+            qvbr_ctl->ConstantQualityTarget = rc_quality;
+
+            if (avctx->qmin > 0 || avctx->qmax > 0) {
+                qvbr_ctl->MinQP = avctx->qmin;
+                qvbr_ctl->MaxQP = avctx->qmax;
+                ctx->rc.Flags |= D3D12_VIDEO_ENCODER_RATE_CONTROL_FLAG_ENABLE_QP_RANGE;
+            }
+
+            ctx->rc.ConfigParams.pConfiguration_QVBR = qvbr_ctl;
+            break;
+
+        default:
+            break;
+    }
+    return 0;
+}
+
+static int d3d12va_encode_init_gop_structure(AVCodecContext *avctx)
+{
+    HWBaseEncodeContext *base_ctx = avctx->priv_data;
+    D3D12VAEncodeContext     *ctx = avctx->priv_data;
+    uint32_t ref_l0, ref_l1;
+    int err;
+    HRESULT hr;
+    D3D12_FEATURE_DATA_VIDEO_ENCODER_CODEC_PICTURE_CONTROL_SUPPORT support;
+    union {
+        D3D12_VIDEO_ENCODER_CODEC_PICTURE_CONTROL_SUPPORT_H264 h264;
+        D3D12_VIDEO_ENCODER_CODEC_PICTURE_CONTROL_SUPPORT_HEVC hevc;
+    } codec_support;
+
+    support.NodeIndex = 0;
+    support.Codec     = ctx->codec->d3d12_codec;
+    support.Profile   = ctx->profile->d3d12_profile;
+
+    switch (ctx->codec->d3d12_codec) {
+        case D3D12_VIDEO_ENCODER_CODEC_H264:
+            support.PictureSupport.DataSize = sizeof(codec_support.h264);
+            support.PictureSupport.pH264Support = &codec_support.h264;
+            break;
+
+        case D3D12_VIDEO_ENCODER_CODEC_HEVC:
+            support.PictureSupport.DataSize = sizeof(codec_support.hevc);
+            support.PictureSupport.pHEVCSupport = &codec_support.hevc;
+            break;
+
+        default:
+            av_assert0(0);
+    }
+
+    hr = ID3D12VideoDevice3_CheckFeatureSupport(ctx->video_device3, D3D12_FEATURE_VIDEO_ENCODER_CODEC_PICTURE_CONTROL_SUPPORT,
+                                                &support, sizeof(support));
+    if (FAILED(hr))
+        return AVERROR(EINVAL);
+
+    if (support.IsSupported) {
+        switch (ctx->codec->d3d12_codec) {
+            case D3D12_VIDEO_ENCODER_CODEC_H264:
+                ref_l0 = FFMIN(support.PictureSupport.pH264Support->MaxL0ReferencesForP,
+                               support.PictureSupport.pH264Support->MaxL1ReferencesForB);
+                ref_l1 = support.PictureSupport.pH264Support->MaxL1ReferencesForB;
+                break;
+
+            case D3D12_VIDEO_ENCODER_CODEC_HEVC:
+                ref_l0 = FFMIN(support.PictureSupport.pHEVCSupport->MaxL0ReferencesForP,
+                               support.PictureSupport.pHEVCSupport->MaxL1ReferencesForB);
+                ref_l1 = support.PictureSupport.pHEVCSupport->MaxL1ReferencesForB;
+                break;
+
+            default:
+                av_assert0(0);
+        }
+    } else {
+        ref_l0 = ref_l1 = 0;
+    }
+
+    if (ref_l0 > 0 && ref_l1 > 0 && ctx->bi_not_empty) {
+        base_ctx->p_to_gpb = 1;
+        av_log(avctx, AV_LOG_VERBOSE, "Driver does not support P-frames, "
+               "replacing them with B-frames.\n");
+    }
+
+    err = ff_hw_base_init_gop_structure(avctx, ref_l0, ref_l1, ctx->codec->flags, 0);
+    if (err < 0)
+        return err;
+
+    return 0;
+}
+
+static int d3d12va_create_encoder(AVCodecContext *avctx)
+{
+    HWBaseEncodeContext    *base_ctx     = avctx->priv_data;
+    D3D12VAEncodeContext   *ctx          = avctx->priv_data;
+    AVD3D12VAFramesContext *frames_hwctx = base_ctx->input_frames->hwctx;
+    HRESULT hr;
+
+    D3D12_VIDEO_ENCODER_DESC desc = {
+        .NodeMask                     = 0,
+        .Flags                        = D3D12_VIDEO_ENCODER_FLAG_NONE,
+        .EncodeCodec                  = ctx->codec->d3d12_codec,
+        .EncodeProfile                = ctx->profile->d3d12_profile,
+        .InputFormat                  = frames_hwctx->format,
+        .CodecConfiguration           = ctx->codec_conf,
+        .MaxMotionEstimationPrecision = D3D12_VIDEO_ENCODER_MOTION_ESTIMATION_PRECISION_MODE_MAXIMUM,
+    };
+
+    hr = ID3D12VideoDevice3_CreateVideoEncoder(ctx->video_device3, &desc, &IID_ID3D12VideoEncoder,
+                                               (void **)&ctx->encoder);
+    if (FAILED(hr)) {
+        av_log(avctx, AV_LOG_ERROR, "Failed to create encoder.\n");
+        return AVERROR(EINVAL);
+    }
+
+    return 0;
+}
+
+static int d3d12va_create_encoder_heap(AVCodecContext* avctx)
+{
+    D3D12VAEncodeContext *ctx = avctx->priv_data;
+    HRESULT hr;
+
+    D3D12_VIDEO_ENCODER_HEAP_DESC desc = {
+        .NodeMask             = 0,
+        .Flags                = D3D12_VIDEO_ENCODER_FLAG_NONE,
+        .EncodeCodec          = ctx->codec->d3d12_codec,
+        .EncodeProfile        = ctx->profile->d3d12_profile,
+        .EncodeLevel          = ctx->level,
+        .ResolutionsListCount = 1,
+        .pResolutionList      = &ctx->resolution,
+    };
+
+    hr = ID3D12VideoDevice3_CreateVideoEncoderHeap(ctx->video_device3, &desc,
+                                                   &IID_ID3D12VideoEncoderHeap, (void **)&ctx->encoder_heap);
+    if (FAILED(hr)) {
+        av_log(avctx, AV_LOG_ERROR, "Failed to create encoder heap.\n");
+        return AVERROR(EINVAL);
+    }
+
+    return 0;
+}
+
+static void d3d12va_encode_free_buffer(void *opaque, uint8_t *data)
+{
+    ID3D12Resource *pResource;
+
+    pResource = (ID3D12Resource *)data;
+    D3D12_OBJECT_RELEASE(pResource);
+}
+
+static AVBufferRef *d3d12va_encode_alloc_output_buffer(void *opaque, size_t size)
+{
+    AVCodecContext     *avctx = opaque;
+    HWBaseEncodeContext *base_ctx = avctx->priv_data;
+    D3D12VAEncodeContext     *ctx = avctx->priv_data;
+    ID3D12Resource *pResource = NULL;
+    HRESULT hr;
+    AVBufferRef *ref;
+    D3D12_HEAP_PROPERTIES heap_props;
+    D3D12_HEAP_TYPE heap_type = D3D12_HEAP_TYPE_READBACK;
+
+    D3D12_RESOURCE_DESC desc = {
+        .Dimension        = D3D12_RESOURCE_DIMENSION_BUFFER,
+        .Alignment        = 0,
+        .Width            = FFALIGN(3 * base_ctx->surface_width * base_ctx->surface_height + (1 << 16),
+                                    D3D12_TEXTURE_DATA_PLACEMENT_ALIGNMENT),
+        .Height           = 1,
+        .DepthOrArraySize = 1,
+        .MipLevels        = 1,
+        .Format           = DXGI_FORMAT_UNKNOWN,
+        .SampleDesc       = { .Count = 1, .Quality = 0 },
+        .Layout           = D3D12_TEXTURE_LAYOUT_ROW_MAJOR,
+        .Flags            = D3D12_RESOURCE_FLAG_NONE,
+    };
+
+    ctx->hwctx->device->lpVtbl->GetCustomHeapProperties(ctx->hwctx->device, &heap_props, 0, heap_type);
+
+    hr = ID3D12Device_CreateCommittedResource(ctx->hwctx->device, &heap_props, D3D12_HEAP_FLAG_NONE,
+                                              &desc, D3D12_RESOURCE_STATE_COMMON, NULL, &IID_ID3D12Resource,
+                                              (void **)&pResource);
+
+    if (FAILED(hr)) {
+        av_log(avctx, AV_LOG_ERROR, "Failed to create d3d12 buffer.\n");
+        return NULL;
+    }
+
+    ref = av_buffer_create((uint8_t *)(uintptr_t)pResource,
+                           sizeof(pResource),
+                           &d3d12va_encode_free_buffer,
+                           avctx, AV_BUFFER_FLAG_READONLY);
+    if (!ref) {
+        D3D12_OBJECT_RELEASE(pResource);
+        return NULL;
+    }
+
+    return ref;
+}
+
+static int d3d12va_encode_prepare_output_buffers(AVCodecContext *avctx)
+{
+    HWBaseEncodeContext *base_ctx      = avctx->priv_data;
+    D3D12VAEncodeContext *ctx          = avctx->priv_data;
+    AVD3D12VAFramesContext *frames_ctx = base_ctx->input_frames->hwctx;
+    HRESULT hr;
+
+    ctx->req.NodeIndex               = 0;
+    ctx->req.Codec                   = ctx->codec->d3d12_codec;
+    ctx->req.Profile                 = ctx->profile->d3d12_profile;
+    ctx->req.InputFormat             = frames_ctx->format;
+    ctx->req.PictureTargetResolution = ctx->resolution;
+
+    hr = ID3D12VideoDevice3_CheckFeatureSupport(ctx->video_device3,
+                                                D3D12_FEATURE_VIDEO_ENCODER_RESOURCE_REQUIREMENTS,
+                                                &ctx->req, sizeof(ctx->req));
+    if (FAILED(hr)) {
+        av_log(avctx, AV_LOG_ERROR, "Failed to check encoder resource requirements support.\n");
+        return AVERROR(EINVAL);
+    }
+
+    if (!ctx->req.IsSupported) {
+        av_log(avctx, AV_LOG_ERROR, "Encoder resource requirements unsupported.\n");
+        return AVERROR(EINVAL);
+    }
+
+    ctx->output_buffer_pool = av_buffer_pool_init2(sizeof(ID3D12Resource *), avctx,
+                                                   &d3d12va_encode_alloc_output_buffer, NULL);
+    if (!ctx->output_buffer_pool)
+        return AVERROR(ENOMEM);
+
+    return 0;
+}
+
+static int d3d12va_encode_create_command_objects(AVCodecContext *avctx)
+{
+    D3D12VAEncodeContext *ctx = avctx->priv_data;
+    ID3D12CommandAllocator *command_allocator = NULL;
+    int err;
+    HRESULT hr;
+
+    D3D12_COMMAND_QUEUE_DESC queue_desc = {
+        .Type     = D3D12_COMMAND_LIST_TYPE_VIDEO_ENCODE,
+        .Priority = 0,
+        .Flags    = D3D12_COMMAND_QUEUE_FLAG_NONE,
+        .NodeMask = 0,
+    };
+
+    ctx->allocator_queue = av_fifo_alloc2(D3D12VA_VIDEO_ENC_ASYNC_DEPTH,
+                                          sizeof(CommandAllocator), AV_FIFO_FLAG_AUTO_GROW);
+    if (!ctx->allocator_queue)
+        return AVERROR(ENOMEM);
+
+    hr = ID3D12Device_CreateFence(ctx->hwctx->device, 0, D3D12_FENCE_FLAG_NONE,
+                                  &IID_ID3D12Fence, (void **)&ctx->sync_ctx.fence);
+    if (FAILED(hr)) {
+        av_log(avctx, AV_LOG_ERROR, "Failed to create fence(%lx)\n", (long)hr);
+        err = AVERROR_UNKNOWN;
+        goto fail;
+    }
+
+    ctx->sync_ctx.event = CreateEvent(NULL, FALSE, FALSE, NULL);
+    if (!ctx->sync_ctx.event)
+        goto fail;
+
+    err = d3d12va_get_valid_command_allocator(avctx, &command_allocator);
+    if (err < 0)
+        goto fail;
+
+    hr = ID3D12Device_CreateCommandQueue(ctx->hwctx->device, &queue_desc,
+                                         &IID_ID3D12CommandQueue, (void **)&ctx->command_queue);
+    if (FAILED(hr)) {
+        av_log(avctx, AV_LOG_ERROR, "Failed to create command queue(%lx)\n", (long)hr);
+        err = AVERROR_UNKNOWN;
+        goto fail;
+    }
+
+    hr = ID3D12Device_CreateCommandList(ctx->hwctx->device, 0, queue_desc.Type,
+                                        command_allocator, NULL, &IID_ID3D12CommandList,
+                                        (void **)&ctx->command_list);
+    if (FAILED(hr)) {
+        av_log(avctx, AV_LOG_ERROR, "Failed to create command list(%lx)\n", (long)hr);
+        err = AVERROR_UNKNOWN;
+        goto fail;
+    }
+
+    hr = ID3D12VideoEncodeCommandList2_Close(ctx->command_list);
+    if (FAILED(hr)) {
+        av_log(avctx, AV_LOG_ERROR, "Failed to close the command list(%lx)\n", (long)hr);
+        err = AVERROR_UNKNOWN;
+        goto fail;
+    }
+
+    ID3D12CommandQueue_ExecuteCommandLists(ctx->command_queue, 1, (ID3D12CommandList **)&ctx->command_list);
+
+    err = d3d12va_sync_with_gpu(avctx);
+    if (err < 0)
+        goto fail;
+
+    err = d3d12va_discard_command_allocator(avctx, command_allocator, ctx->sync_ctx.fence_value);
+    if (err < 0)
+        goto fail;
+
+    return 0;
+
+fail:
+    D3D12_OBJECT_RELEASE(command_allocator);
+    return err;
+}
+
+static int d3d12va_encode_create_recon_frames(AVCodecContext *avctx)
+{
+    HWBaseEncodeContext *base_ctx = avctx->priv_data;
+    AVD3D12VAFramesContext *hwctx;
+    enum AVPixelFormat recon_format;
+    int err;
+
+    err = ff_hw_base_get_recon_format(avctx, NULL, &recon_format);
+    if (err < 0)
+        return err;
+
+    base_ctx->recon_frames_ref = av_hwframe_ctx_alloc(base_ctx->device_ref);
+    if (!base_ctx->recon_frames_ref)
+        return AVERROR(ENOMEM);
+
+    base_ctx->recon_frames = (AVHWFramesContext *)base_ctx->recon_frames_ref->data;
+    hwctx = (AVD3D12VAFramesContext *)base_ctx->recon_frames->hwctx;
+
+    base_ctx->recon_frames->format    = AV_PIX_FMT_D3D12;
+    base_ctx->recon_frames->sw_format = recon_format;
+    base_ctx->recon_frames->width     = base_ctx->surface_width;
+    base_ctx->recon_frames->height    = base_ctx->surface_height;
+
+    hwctx->flags = D3D12_RESOURCE_FLAG_VIDEO_ENCODE_REFERENCE_ONLY |
+                   D3D12_RESOURCE_FLAG_DENY_SHADER_RESOURCE;
+
+    err = av_hwframe_ctx_init(base_ctx->recon_frames_ref);
+    if (err < 0) {
+        av_log(avctx, AV_LOG_ERROR, "Failed to initialise reconstructed "
+               "frame context: %d.\n", err);
+        return err;
+    }
+
+    return 0;
+}
+
+static const HWEncodePictureOperation d3d12va_type = {
+    .alloc  = &d3d12va_encode_alloc,
+
+    .issue  = &d3d12va_encode_issue,
+
+    .output = &d3d12va_encode_output,
+
+    .free   = &d3d12va_encode_free,
+};
+
+int ff_d3d12va_encode_init(AVCodecContext *avctx)
+{
+    HWBaseEncodeContext *base_ctx = avctx->priv_data;
+    D3D12VAEncodeContext     *ctx = avctx->priv_data;
+    D3D12_FEATURE_DATA_VIDEO_FEATURE_AREA_SUPPORT support = { 0 };
+    int err;
+    HRESULT hr;
+
+    err = ff_hw_base_encode_init(avctx);
+    if (err < 0)
+        goto fail;
+
+    base_ctx->op = &d3d12va_type;
+
+    ctx->hwctx = base_ctx->device->hwctx;
+
+    ctx->resolution.Width  = base_ctx->input_frames->width;
+    ctx->resolution.Height = base_ctx->input_frames->height;
+
+    hr = ID3D12Device_QueryInterface(ctx->hwctx->device, &IID_ID3D12Device3, (void **)&ctx->device3);
+    if (FAILED(hr)) {
+        av_log(avctx, AV_LOG_ERROR, "ID3D12Device3 interface is not supported.\n");
+        err = AVERROR_UNKNOWN;
+        goto fail;
+    }
+
+    hr = ID3D12Device3_QueryInterface(ctx->device3, &IID_ID3D12VideoDevice3, (void **)&ctx->video_device3);
+    if (FAILED(hr)) {
+        av_log(avctx, AV_LOG_ERROR, "ID3D12VideoDevice3 interface is not supported.\n");
+        err = AVERROR_UNKNOWN;
+        goto fail;
+    }
+
+    if (FAILED(ID3D12VideoDevice3_CheckFeatureSupport(ctx->video_device3, D3D12_FEATURE_VIDEO_FEATURE_AREA_SUPPORT,
+                                                      &support, sizeof(support))) && !support.VideoEncodeSupport) {
+        av_log(avctx, AV_LOG_ERROR, "D3D12 video device has no video encoder support.\n");
+        err = AVERROR(EINVAL);
+        goto fail;
+    }
+
+    err = d3d12va_encode_set_profile(avctx);
+    if (err < 0)
+        goto fail;
+
+    err = d3d12va_encode_init_rate_control(avctx);
+    if (err < 0)
+        goto fail;
+
+    if (ctx->codec->get_encoder_caps) {
+        err = ctx->codec->get_encoder_caps(avctx);
+        if (err < 0)
+            goto fail;
+    }
+
+    err = d3d12va_encode_init_gop_structure(avctx);
+    if (err < 0)
+        goto fail;
+
+    if (!(ctx->codec->flags & FLAG_SLICE_CONTROL) && avctx->slices > 0) {
+        av_log(avctx, AV_LOG_WARNING, "Multiple slices were requested "
+               "but this codec does not support controlling slices.\n");
+    }
+
+    err = d3d12va_encode_create_command_objects(avctx);
+    if (err < 0)
+        goto fail;
+
+    err = d3d12va_encode_create_recon_frames(avctx);
+    if (err < 0)
+        goto fail;
+
+    err = d3d12va_encode_prepare_output_buffers(avctx);
+    if (err < 0)
+        goto fail;
+
+    if (ctx->codec->configure) {
+        err = ctx->codec->configure(avctx);
+        if (err < 0)
+            goto fail;
+    }
+
+    if (ctx->codec->init_sequence_params) {
+        err = ctx->codec->init_sequence_params(avctx);
+        if (err < 0) {
+            av_log(avctx, AV_LOG_ERROR, "Codec sequence initialisation "
+                   "failed: %d.\n", err);
+            goto fail;
+        }
+    }
+
+    if (ctx->codec->set_level) {
+        err = ctx->codec->set_level(avctx);
+        if (err < 0)
+            goto fail;
+    }
+
+    base_ctx->output_delay = base_ctx->b_per_p;
+    base_ctx->decode_delay = base_ctx->max_b_depth;
+
+    err = d3d12va_create_encoder(avctx);
+    if (err < 0)
+        goto fail;
+
+    err = d3d12va_create_encoder_heap(avctx);
+    if (err < 0)
+        goto fail;
+
+    base_ctx->async_encode = 1;
+    base_ctx->encode_fifo = av_fifo_alloc2(base_ctx->async_depth,
+                                           sizeof(D3D12VAEncodePicture *), 0);
+    if (!base_ctx->encode_fifo)
+        return AVERROR(ENOMEM);
+
+    return 0;
+
+fail:
+    return err;
+}
+
+int ff_d3d12va_encode_close(AVCodecContext *avctx)
+{
+    int num_allocator = 0;
+    HWBaseEncodeContext *base_ctx = avctx->priv_data;
+    D3D12VAEncodeContext     *ctx = avctx->priv_data;
+    HWBaseEncodePicture *pic, *next;
+    CommandAllocator allocator;
+
+    if (!base_ctx->frame)
+        return 0;
+
+    for (pic = base_ctx->pic_start; pic; pic = next) {
+        next = pic->next;
+        d3d12va_encode_free(avctx, pic);
+    }
+
+    d3d12va_encode_free_rc_params(avctx);
+
+    av_buffer_pool_uninit(&ctx->output_buffer_pool);
+
+    D3D12_OBJECT_RELEASE(ctx->command_list);
+    D3D12_OBJECT_RELEASE(ctx->command_queue);
+
+    if (ctx->allocator_queue) {
+        while (av_fifo_read(ctx->allocator_queue, &allocator, 1) >= 0) {
+            num_allocator++;
+            D3D12_OBJECT_RELEASE(allocator.command_allocator);
+        }
+
+        av_log(avctx, AV_LOG_VERBOSE, "Total number of command allocators reused: %d\n", num_allocator);
+    }
+
+    av_fifo_freep2(&ctx->allocator_queue);
+
+    D3D12_OBJECT_RELEASE(ctx->sync_ctx.fence);
+    if (ctx->sync_ctx.event)
+        CloseHandle(ctx->sync_ctx.event);
+
+    D3D12_OBJECT_RELEASE(ctx->encoder_heap);
+    D3D12_OBJECT_RELEASE(ctx->encoder);
+    D3D12_OBJECT_RELEASE(ctx->video_device3);
+    D3D12_OBJECT_RELEASE(ctx->device3);
+
+    ff_hw_base_encode_close(avctx);
+
+    return 0;
+}
diff --git a/libavcodec/d3d12va_encode.h b/libavcodec/d3d12va_encode.h
new file mode 100644
index 0000000000..510c36a5eb
--- /dev/null
+++ b/libavcodec/d3d12va_encode.h
@@ -0,0 +1,334 @@
+/*
+ * Direct3D 12 HW acceleration video encoder
+ *
+ * Copyright (c) 2024 Intel Corporation
+ *
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+
+#ifndef AVCODEC_D3D12VA_ENCODE_H
+#define AVCODEC_D3D12VA_ENCODE_H
+
+#include "libavutil/fifo.h"
+#include "libavutil/hwcontext.h"
+#include "libavutil/hwcontext_d3d12va_internal.h"
+#include "libavutil/hwcontext_d3d12va.h"
+#include "avcodec.h"
+#include "internal.h"
+#include "hwconfig.h"
+#include "hw_base_encode.h"
+
+struct D3D12VAEncodeType;
+
+extern const AVCodecHWConfigInternal *const ff_d3d12va_encode_hw_configs[];
+
+#define MAX_PARAM_BUFFER_SIZE 4096
+#define D3D12VA_VIDEO_ENC_ASYNC_DEPTH 8
+
+typedef struct D3D12VAEncodePicture {
+    HWBaseEncodePicture base;
+
+    int             header_size;
+
+    AVD3D12VAFrame *input_surface;
+    AVD3D12VAFrame *recon_surface;
+
+    AVBufferRef    *output_buffer_ref;
+    ID3D12Resource *output_buffer;
+
+    ID3D12Resource *encoded_metadata;
+    ID3D12Resource *resolved_metadata;
+
+    D3D12_VIDEO_ENCODER_PICTURE_CONTROL_CODEC_DATA pic_ctl;
+
+    int             fence_value;
+} D3D12VAEncodePicture;
+
+typedef struct D3D12VAEncodeProfile {
+    /**
+     * lavc profile value (AV_PROFILE_*).
+     */
+    int       av_profile;
+
+    /**
+     * Supported bit depth.
+     */
+    int       depth;
+
+    /**
+     * Number of components.
+     */
+    int       nb_components;
+
+    /**
+     * Chroma subsampling in width dimension.
+     */
+    int       log2_chroma_w;
+
+    /**
+     * Chroma subsampling in height dimension.
+     */
+    int       log2_chroma_h;
+
+    /**
+     * D3D12 profile value.
+     */
+    D3D12_VIDEO_ENCODER_PROFILE_DESC d3d12_profile;
+} D3D12VAEncodeProfile;
+
+enum {
+    RC_MODE_AUTO,
+    RC_MODE_CQP,
+    RC_MODE_CBR,
+    RC_MODE_VBR,
+    RC_MODE_QVBR,
+    RC_MODE_MAX = RC_MODE_QVBR,
+};
+
+
+typedef struct D3D12VAEncodeRCMode {
+    /**
+     * Mode from above enum (RC_MODE_*).
+     */
+    int mode;
+
+    /**
+     * Name.
+     *
+     */
+    const char *name;
+
+    /**
+     * Uses bitrate parameters.
+     *
+     */
+    int bitrate;
+
+    /**
+     * Supports maxrate distinct from bitrate.
+     *
+     */
+    int maxrate;
+
+    /**
+     * Uses quality value.
+     *
+     */
+    int quality;
+
+    /**
+     * Supports HRD/VBV parameters.
+     *
+     */
+    int hrd;
+
+    /**
+     * D3D12 mode value.
+     */
+    D3D12_VIDEO_ENCODER_RATE_CONTROL_MODE d3d12_mode;
+} D3D12VAEncodeRCMode;
+
+typedef struct D3D12VAEncodeContext {
+    HWBaseEncodeContext base;
+
+    /**
+     * Codec-specific hooks.
+     */
+    const struct D3D12VAEncodeType *codec;
+
+    /**
+     * Explicitly set RC mode (otherwise attempt to pick from
+     * available modes).
+     */
+    int explicit_rc_mode;
+
+    /**
+     * Explicitly-set QP, for use with the "qp" options.
+     * (Forces CQP mode when set, overriding everything else.)
+     */
+    int explicit_qp;
+
+    /**
+     * RC quality level - meaning depends on codec and RC mode.
+     * In CQP mode this sets the fixed quantiser value.
+     */
+    int rc_quality;
+
+    /**
+     * Chosen encoding profile details.
+     */
+    const D3D12VAEncodeProfile *profile;
+
+    AVD3D12VADeviceContext *hwctx;
+
+    /**
+     * ID3D12Device3 interface.
+     */
+    ID3D12Device3 *device3;
+
+    /**
+     * ID3D12VideoDevice3 interface.
+     */
+    ID3D12VideoDevice3 *video_device3;
+
+    /**
+     * Pool of (reusable) bitstream output buffers.
+     */
+    AVBufferPool *output_buffer_pool;
+
+    /**
+     * D3D12 video encoder.
+     */
+    AVBufferRef *encoder_ref;
+
+    ID3D12VideoEncoder *encoder;
+
+    /**
+     * D3D12 video encoder heap.
+     */
+    ID3D12VideoEncoderHeap *encoder_heap;
+
+    /**
+     * A cached queue for reusing the D3D12 command allocators.
+     *
+     * @see https://learn.microsoft.com/en-us/windows/win32/direct3d12/recording-command-lists-and-bundles#id3d12commandallocator
+     */
+    AVFifo *allocator_queue;
+
+    /**
+     * D3D12 command queue.
+     */
+    ID3D12CommandQueue *command_queue;
+
+    /**
+     * D3D12 video encode command list.
+     */
+    ID3D12VideoEncodeCommandList2 *command_list;
+
+    /**
+     * The sync context used to sync command queue.
+     */
+    AVD3D12VASyncContext sync_ctx;
+
+    /**
+     * The bi_not_empty feature.
+     */
+    int bi_not_empty;
+
+    /**
+     * D3D12_FEATURE structures.
+     */
+    D3D12_FEATURE_DATA_VIDEO_ENCODER_RESOURCE_REQUIREMENTS req;
+
+    D3D12_FEATURE_DATA_VIDEO_ENCODER_RESOLUTION_SUPPORT_LIMITS res_limits;
+
+    /**
+     * D3D12_VIDEO_ENCODER structures.
+     */
+    D3D12_VIDEO_ENCODER_PICTURE_RESOLUTION_DESC resolution;
+
+    D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION codec_conf;
+
+    D3D12_VIDEO_ENCODER_RATE_CONTROL rc;
+
+    D3D12_VIDEO_ENCODER_SEQUENCE_GOP_STRUCTURE gop;
+
+    D3D12_VIDEO_ENCODER_LEVEL_SETTING level;
+} D3D12VAEncodeContext;
+
+typedef struct D3D12VAEncodeType {
+    /**
+     * List of supported profiles.
+     */
+   const D3D12VAEncodeProfile *profiles;
+
+    /**
+     * D3D12 codec name.
+     */
+    D3D12_VIDEO_ENCODER_CODEC d3d12_codec;
+
+    /**
+     * Codec feature flags.
+     */
+    int flags;
+
+    /**
+     * Default quality for this codec - used as quantiser or RC quality
+     * factor depending on RC mode.
+     */
+    int default_quality;
+
+    /**
+     * Query codec configuration and determine encode parameters like
+     * block sizes for surface alignment and slices. If not set, assume
+     * that all blocks are 16x16 and that surfaces should be aligned to match
+     * this.
+     */
+    int (*get_encoder_caps)(AVCodecContext *avctx);
+
+    /**
+     * Perform any extra codec-specific configuration.
+     */
+    int (*configure)(AVCodecContext *avctx);
+
+    /**
+     * Set codec-specific level setting.
+     */
+    int (*set_level)(AVCodecContext *avctx);
+
+    /**
+     * The size of any private data structure associated with each
+     * picture (can be zero if not required).
+     */
+    size_t picture_priv_data_size;
+
+    /**
+     * Fill the corresponding parameters.
+     */
+    int (*init_sequence_params)(AVCodecContext *avctx);
+
+    int (*init_picture_params)(AVCodecContext *avctx,
+                               D3D12VAEncodePicture *pic);
+
+    void (*free_picture_params)(D3D12VAEncodePicture *pic);
+
+    /**
+     * Write the packed header data to the provided buffer.
+     */
+    int (*write_sequence_header)(AVCodecContext *avctx,
+                                 char *data, size_t *data_len);
+} D3D12VAEncodeType;
+
+int ff_d3d12va_encode_init(AVCodecContext *avctx);
+int ff_d3d12va_encode_close(AVCodecContext *avctx);
+
+#define D3D12VA_ENCODE_RC_MODE(name, desc) \
+    { #name, desc, 0, AV_OPT_TYPE_CONST, { .i64 = RC_MODE_ ## name }, \
+      0, 0, FLAGS, .unit = "rc_mode" }
+#define D3D12VA_ENCODE_RC_OPTIONS \
+    { "rc_mode",\
+      "Set rate control mode", \
+      OFFSET(common.explicit_rc_mode), AV_OPT_TYPE_INT, \
+      { .i64 = RC_MODE_AUTO }, RC_MODE_AUTO, RC_MODE_MAX, FLAGS, .unit = "rc_mode" }, \
+    { "auto", "Choose mode automatically based on other parameters", \
+      0, AV_OPT_TYPE_CONST, { .i64 = RC_MODE_AUTO }, 0, 0, FLAGS, .unit = "rc_mode" }, \
+    D3D12VA_ENCODE_RC_MODE(CQP,  "Constant-quality"), \
+    D3D12VA_ENCODE_RC_MODE(CBR,  "Constant-bitrate"), \
+    D3D12VA_ENCODE_RC_MODE(VBR,  "Variable-bitrate"), \
+    D3D12VA_ENCODE_RC_MODE(QVBR, "Quality-defined variable-bitrate")
+
+#endif /* AVCODEC_D3D12VA_ENCODE_H */
diff --git a/libavcodec/d3d12va_encode_hevc.c b/libavcodec/d3d12va_encode_hevc.c
new file mode 100644
index 0000000000..7f7d98db17
--- /dev/null
+++ b/libavcodec/d3d12va_encode_hevc.c
@@ -0,0 +1,1007 @@
+/*
+ * Direct3D 12 HW acceleration video encoder
+ *
+ * Copyright (c) 2024 Intel Corporation
+ *
+ * This file is part of FFmpeg.
+ *
+ * FFmpeg is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU Lesser General Public
+ * License as published by the Free Software Foundation; either
+ * version 2.1 of the License, or (at your option) any later version.
+ *
+ * FFmpeg is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
+ * Lesser General Public License for more details.
+ *
+ * You should have received a copy of the GNU Lesser General Public
+ * License along with FFmpeg; if not, write to the Free Software
+ * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
+ */
+#include "libavutil/opt.h"
+#include "libavutil/common.h"
+#include "libavutil/mem.h"
+#include "libavutil/pixdesc.h"
+#include "libavutil/hwcontext_d3d12va_internal.h"
+
+#include "avcodec.h"
+#include "cbs.h"
+#include "cbs_h265.h"
+#include "h2645data.h"
+#include "h265_profile_level.h"
+#include "codec_internal.h"
+#include "d3d12va_encode.h"
+
+typedef struct D3D12VAEncodeHEVCPicture {
+    int pic_order_cnt;
+    int64_t last_idr_frame;
+} D3D12VAEncodeHEVCPicture;
+
+typedef struct D3D12VAEncodeHEVCContext {
+    D3D12VAEncodeContext common;
+
+    // User options.
+    int qp;
+    int profile;
+    int tier;
+    int level;
+
+    // Writer structures.
+    H265RawVPS   raw_vps;
+    H265RawSPS   raw_sps;
+    H265RawPPS   raw_pps;
+
+    CodedBitstreamContext *cbc;
+    CodedBitstreamFragment current_access_unit;
+} D3D12VAEncodeHEVCContext;
+
+typedef struct D3D12VAEncodeHEVCLevel {
+    int level;
+    D3D12_VIDEO_ENCODER_LEVELS_HEVC d3d12_level;
+} D3D12VAEncodeHEVCLevel;
+
+static const D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_SUPPORT_HEVC hevc_config_support_sets[] =
+{
+    {
+        D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_SUPPORT_HEVC_FLAG_NONE,
+        D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_CUSIZE_8x8,
+        D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_CUSIZE_32x32,
+        D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_TUSIZE_4x4,
+        D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_TUSIZE_32x32,
+        3,
+        3,
+    },
+    {
+        D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_SUPPORT_HEVC_FLAG_NONE,
+        D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_CUSIZE_8x8,
+        D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_CUSIZE_32x32,
+        D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_TUSIZE_4x4,
+        D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_TUSIZE_32x32,
+        0,
+        0,
+    },
+    {
+        D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_SUPPORT_HEVC_FLAG_NONE,
+        D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_CUSIZE_8x8,
+        D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_CUSIZE_32x32,
+        D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_TUSIZE_4x4,
+        D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_TUSIZE_32x32,
+        2,
+        2,
+    },
+    {
+        D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_SUPPORT_HEVC_FLAG_NONE,
+        D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_CUSIZE_8x8,
+        D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_CUSIZE_64x64,
+        D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_TUSIZE_4x4,
+        D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_TUSIZE_32x32,
+        2,
+        2,
+    },
+    {
+        D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_SUPPORT_HEVC_FLAG_NONE,
+        D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_CUSIZE_8x8,
+        D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_CUSIZE_64x64,
+        D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_TUSIZE_4x4,
+        D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_TUSIZE_32x32,
+        4,
+        4,
+    },
+};
+
+static const D3D12VAEncodeHEVCLevel hevc_levels[] = {
+    { 30,  D3D12_VIDEO_ENCODER_LEVELS_HEVC_1  },
+    { 60,  D3D12_VIDEO_ENCODER_LEVELS_HEVC_2  },
+    { 63,  D3D12_VIDEO_ENCODER_LEVELS_HEVC_21 },
+    { 90,  D3D12_VIDEO_ENCODER_LEVELS_HEVC_3  },
+    { 93,  D3D12_VIDEO_ENCODER_LEVELS_HEVC_31 },
+    { 120, D3D12_VIDEO_ENCODER_LEVELS_HEVC_4  },
+    { 123, D3D12_VIDEO_ENCODER_LEVELS_HEVC_41 },
+    { 150, D3D12_VIDEO_ENCODER_LEVELS_HEVC_5  },
+    { 153, D3D12_VIDEO_ENCODER_LEVELS_HEVC_51 },
+    { 156, D3D12_VIDEO_ENCODER_LEVELS_HEVC_52 },
+    { 180, D3D12_VIDEO_ENCODER_LEVELS_HEVC_6  },
+    { 183, D3D12_VIDEO_ENCODER_LEVELS_HEVC_61 },
+    { 186, D3D12_VIDEO_ENCODER_LEVELS_HEVC_62 },
+};
+
+static const D3D12_VIDEO_ENCODER_PROFILE_HEVC profile_main   = D3D12_VIDEO_ENCODER_PROFILE_HEVC_MAIN;
+static const D3D12_VIDEO_ENCODER_PROFILE_HEVC profile_main10 = D3D12_VIDEO_ENCODER_PROFILE_HEVC_MAIN10;
+
+#define D3D_PROFILE_DESC(name) \
+    { sizeof(D3D12_VIDEO_ENCODER_PROFILE_HEVC), { .pHEVCProfile = (D3D12_VIDEO_ENCODER_PROFILE_HEVC *)&profile_ ## name } }
+static const D3D12VAEncodeProfile d3d12va_encode_hevc_profiles[] = {
+    { AV_PROFILE_HEVC_MAIN,     8, 3, 1, 1, D3D_PROFILE_DESC(main)   },
+    { AV_PROFILE_HEVC_MAIN_10, 10, 3, 1, 1, D3D_PROFILE_DESC(main10) },
+    { AV_PROFILE_UNKNOWN },
+};
+
+static uint8_t d3d12va_encode_hevc_map_cusize(D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_CUSIZE cusize)
+{
+    switch (cusize) {
+        case D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_CUSIZE_8x8:   return 8;
+        case D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_CUSIZE_16x16: return 16;
+        case D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_CUSIZE_32x32: return 32;
+        case D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_CUSIZE_64x64: return 64;
+        default: av_assert0(0);
+    }
+    return 0;
+}
+
+static uint8_t d3d12va_encode_hevc_map_tusize(D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_TUSIZE tusize)
+{
+    switch (tusize) {
+        case D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_TUSIZE_4x4:   return 4;
+        case D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_TUSIZE_8x8:   return 8;
+        case D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_TUSIZE_16x16: return 16;
+        case D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_TUSIZE_32x32: return 32;
+        default: av_assert0(0);
+    }
+    return 0;
+}
+
+static int d3d12va_encode_hevc_write_access_unit(AVCodecContext *avctx,
+                                                 char *data, size_t *data_len,
+                                                 CodedBitstreamFragment *au)
+{
+    D3D12VAEncodeHEVCContext *priv = avctx->priv_data;
+    int err;
+
+    err = ff_cbs_write_fragment_data(priv->cbc, au);
+    if (err < 0) {
+        av_log(avctx, AV_LOG_ERROR, "Failed to write packed header.\n");
+        return err;
+    }
+
+    if (*data_len < 8 * au->data_size - au->data_bit_padding) {
+        av_log(avctx, AV_LOG_ERROR, "Access unit too large: "
+               "%zu < %zu.\n", *data_len,
+               8 * au->data_size - au->data_bit_padding);
+        return AVERROR(ENOSPC);
+    }
+
+    memcpy(data, au->data, au->data_size);
+    *data_len = 8 * au->data_size - au->data_bit_padding;
+
+    return 0;
+}
+
+static int d3d12va_encode_hevc_add_nal(AVCodecContext *avctx,
+                                       CodedBitstreamFragment *au,
+                                       void *nal_unit)
+{
+    H265RawNALUnitHeader *header = nal_unit;
+    int err;
+
+    err = ff_cbs_insert_unit_content(au, -1,
+                                     header->nal_unit_type, nal_unit, NULL);
+    if (err < 0) {
+        av_log(avctx, AV_LOG_ERROR, "Failed to add NAL unit: "
+               "type = %d.\n", header->nal_unit_type);
+        return err;
+    }
+
+    return 0;
+}
+
+static int d3d12va_encode_hevc_write_sequence_header(AVCodecContext *avctx,
+                                                     char *data, size_t *data_len)
+{
+    D3D12VAEncodeHEVCContext *priv = avctx->priv_data;
+    CodedBitstreamFragment   *au   = &priv->current_access_unit;
+    int err;
+
+    err = d3d12va_encode_hevc_add_nal(avctx, au, &priv->raw_vps);
+    if (err < 0)
+        goto fail;
+
+    err = d3d12va_encode_hevc_add_nal(avctx, au, &priv->raw_sps);
+    if (err < 0)
+        goto fail;
+
+    err = d3d12va_encode_hevc_add_nal(avctx, au, &priv->raw_pps);
+    if (err < 0)
+        goto fail;
+
+    err = d3d12va_encode_hevc_write_access_unit(avctx, data, data_len, au);
+fail:
+    ff_cbs_fragment_reset(au);
+    return err;
+
+}
+
+static int d3d12va_encode_hevc_init_sequence_params(AVCodecContext *avctx)
+{
+    HWBaseEncodeContext  *base_ctx = avctx->priv_data;
+    D3D12VAEncodeContext     *ctx  = avctx->priv_data;
+    D3D12VAEncodeHEVCContext *priv = avctx->priv_data;
+    AVD3D12VAFramesContext  *hwctx = base_ctx->input_frames->hwctx;
+    H265RawVPS               *vps  = &priv->raw_vps;
+    H265RawSPS               *sps  = &priv->raw_sps;
+    H265RawPPS               *pps  = &priv->raw_pps;
+    H265RawProfileTierLevel  *ptl  = &vps->profile_tier_level;
+    H265RawVUI               *vui  = &sps->vui;
+    D3D12_VIDEO_ENCODER_PROFILE_HEVC profile = D3D12_VIDEO_ENCODER_PROFILE_HEVC_MAIN;
+    D3D12_VIDEO_ENCODER_LEVEL_TIER_CONSTRAINTS_HEVC level = { 0 };
+    const AVPixFmtDescriptor *desc;
+    uint8_t min_cu_size, max_cu_size, min_tu_size, max_tu_size;
+    int chroma_format, bit_depth;
+    HRESULT hr;
+    int i;
+
+    D3D12_FEATURE_DATA_VIDEO_ENCODER_SUPPORT support = {
+        .NodeIndex                        = 0,
+        .Codec                            = D3D12_VIDEO_ENCODER_CODEC_HEVC,
+        .InputFormat                      = hwctx->format,
+        .RateControl                      = ctx->rc,
+        .IntraRefresh                     = D3D12_VIDEO_ENCODER_INTRA_REFRESH_MODE_NONE,
+        .SubregionFrameEncoding           = D3D12_VIDEO_ENCODER_FRAME_SUBREGION_LAYOUT_MODE_FULL_FRAME,
+        .ResolutionsListCount             = 1,
+        .pResolutionList                  = &ctx->resolution,
+        .CodecGopSequence                 = ctx->gop,
+        .MaxReferenceFramesInDPB          = MAX_DPB_SIZE - 1,
+        .CodecConfiguration               = ctx->codec_conf,
+        .SuggestedProfile.DataSize        = sizeof(D3D12_VIDEO_ENCODER_PROFILE_HEVC),
+        .SuggestedProfile.pHEVCProfile    = &profile,
+        .SuggestedLevel.DataSize          = sizeof(D3D12_VIDEO_ENCODER_LEVEL_TIER_CONSTRAINTS_HEVC),
+        .SuggestedLevel.pHEVCLevelSetting = &level,
+        .pResolutionDependentSupport      = &ctx->res_limits,
+     };
+
+    hr = ID3D12VideoDevice3_CheckFeatureSupport(ctx->video_device3, D3D12_FEATURE_VIDEO_ENCODER_SUPPORT,
+                                                &support, sizeof(support));
+
+    if (FAILED(hr)) {
+        av_log(avctx, AV_LOG_ERROR, "Failed to check encoder support(%lx).\n", (long)hr);
+        return AVERROR(EINVAL);
+    }
+
+    if (!(support.SupportFlags & D3D12_VIDEO_ENCODER_SUPPORT_FLAG_GENERAL_SUPPORT_OK)) {
+        av_log(avctx, AV_LOG_ERROR, "Driver does not support some request features. %#x\n",
+               support.ValidationFlags);
+        return AVERROR(EINVAL);
+    }
+
+    if (support.SupportFlags & D3D12_VIDEO_ENCODER_SUPPORT_FLAG_RECONSTRUCTED_FRAMES_REQUIRE_TEXTURE_ARRAYS) {
+        av_log(avctx, AV_LOG_ERROR, "D3D12 video encode on this device requires texture array support, "
+               "but it's not implemented.\n");
+        return AVERROR_PATCHWELCOME;
+    }
+
+    memset(vps, 0, sizeof(*vps));
+    memset(sps, 0, sizeof(*sps));
+    memset(pps, 0, sizeof(*pps));
+
+    desc = av_pix_fmt_desc_get(base_ctx->input_frames->sw_format);
+    av_assert0(desc);
+    if (desc->nb_components == 1) {
+        chroma_format = 0;
+    } else {
+        if (desc->log2_chroma_w == 1 && desc->log2_chroma_h == 1) {
+            chroma_format = 1;
+        } else if (desc->log2_chroma_w == 1 && desc->log2_chroma_h == 0) {
+            chroma_format = 2;
+        } else if (desc->log2_chroma_w == 0 && desc->log2_chroma_h == 0) {
+            chroma_format = 3;
+        } else {
+            av_log(avctx, AV_LOG_ERROR, "Chroma format of input pixel format "
+                   "%s is not supported.\n", desc->name);
+            return AVERROR(EINVAL);
+        }
+    }
+    bit_depth = desc->comp[0].depth;
+
+    min_cu_size = d3d12va_encode_hevc_map_cusize(ctx->codec_conf.pHEVCConfig->MinLumaCodingUnitSize);
+    max_cu_size = d3d12va_encode_hevc_map_cusize(ctx->codec_conf.pHEVCConfig->MaxLumaCodingUnitSize);
+    min_tu_size = d3d12va_encode_hevc_map_tusize(ctx->codec_conf.pHEVCConfig->MinLumaTransformUnitSize);
+    max_tu_size = d3d12va_encode_hevc_map_tusize(ctx->codec_conf.pHEVCConfig->MaxLumaTransformUnitSize);
+
+    // VPS
+
+    vps->nal_unit_header = (H265RawNALUnitHeader) {
+        .nal_unit_type         = HEVC_NAL_VPS,
+        .nuh_layer_id          = 0,
+        .nuh_temporal_id_plus1 = 1,
+    };
+
+    vps->vps_video_parameter_set_id = 0;
+
+    vps->vps_base_layer_internal_flag  = 1;
+    vps->vps_base_layer_available_flag = 1;
+    vps->vps_max_layers_minus1         = 0;
+    vps->vps_max_sub_layers_minus1     = 0;
+    vps->vps_temporal_id_nesting_flag  = 1;
+
+    ptl->general_profile_space = 0;
+    ptl->general_profile_idc   = avctx->profile;
+    ptl->general_tier_flag     = priv->tier;
+
+    ptl->general_profile_compatibility_flag[ptl->general_profile_idc] = 1;
+
+    ptl->general_progressive_source_flag    = 1;
+    ptl->general_interlaced_source_flag     = 0;
+    ptl->general_non_packed_constraint_flag = 1;
+    ptl->general_frame_only_constraint_flag = 1;
+
+    ptl->general_max_14bit_constraint_flag = bit_depth <= 14;
+    ptl->general_max_12bit_constraint_flag = bit_depth <= 12;
+    ptl->general_max_10bit_constraint_flag = bit_depth <= 10;
+    ptl->general_max_8bit_constraint_flag  = bit_depth ==  8;
+
+    ptl->general_max_422chroma_constraint_flag  = chroma_format <= 2;
+    ptl->general_max_420chroma_constraint_flag  = chroma_format <= 1;
+    ptl->general_max_monochrome_constraint_flag = chroma_format == 0;
+
+    ptl->general_intra_constraint_flag = base_ctx->gop_size == 1;
+    ptl->general_one_picture_only_constraint_flag = 0;
+
+    ptl->general_lower_bit_rate_constraint_flag = 1;
+
+    if (avctx->level != FF_LEVEL_UNKNOWN) {
+        ptl->general_level_idc = avctx->level;
+    } else {
+        const H265LevelDescriptor *level;
+
+        level = ff_h265_guess_level(ptl, avctx->bit_rate,
+                                    base_ctx->surface_width, base_ctx->surface_height,
+                                    1, 1, 1, (base_ctx->b_per_p > 0) + 1);
+        if (level) {
+            av_log(avctx, AV_LOG_VERBOSE, "Using level %s.\n", level->name);
+            ptl->general_level_idc = level->level_idc;
+        } else {
+            av_log(avctx, AV_LOG_VERBOSE, "Stream will not conform to "
+                   "any normal level; using level 8.5.\n");
+            ptl->general_level_idc = 255;
+            // The tier flag must be set in level 8.5.
+            ptl->general_tier_flag = 1;
+        }
+        avctx->level = ptl->general_level_idc;
+    }
+
+    vps->vps_sub_layer_ordering_info_present_flag = 0;
+    vps->vps_max_dec_pic_buffering_minus1[0]      = base_ctx->max_b_depth + 1;
+    vps->vps_max_num_reorder_pics[0]              = base_ctx->max_b_depth;
+    vps->vps_max_latency_increase_plus1[0]        = 0;
+
+    vps->vps_max_layer_id             = 0;
+    vps->vps_num_layer_sets_minus1    = 0;
+    vps->layer_id_included_flag[0][0] = 1;
+
+    vps->vps_timing_info_present_flag = 1;
+    if (avctx->framerate.num > 0 && avctx->framerate.den > 0) {
+        vps->vps_num_units_in_tick  = avctx->framerate.den;
+        vps->vps_time_scale         = avctx->framerate.num;
+        vps->vps_poc_proportional_to_timing_flag = 1;
+        vps->vps_num_ticks_poc_diff_one_minus1   = 0;
+    } else {
+        vps->vps_num_units_in_tick  = avctx->time_base.num;
+        vps->vps_time_scale         = avctx->time_base.den;
+        vps->vps_poc_proportional_to_timing_flag = 0;
+    }
+    vps->vps_num_hrd_parameters = 0;
+
+    // SPS
+
+    sps->nal_unit_header = (H265RawNALUnitHeader) {
+        .nal_unit_type         = HEVC_NAL_SPS,
+        .nuh_layer_id          = 0,
+        .nuh_temporal_id_plus1 = 1,
+    };
+
+    sps->sps_video_parameter_set_id = vps->vps_video_parameter_set_id;
+
+    sps->sps_max_sub_layers_minus1    = vps->vps_max_sub_layers_minus1;
+    sps->sps_temporal_id_nesting_flag = vps->vps_temporal_id_nesting_flag;
+
+    sps->profile_tier_level = vps->profile_tier_level;
+
+    sps->sps_seq_parameter_set_id = 0;
+
+    sps->chroma_format_idc          = chroma_format;
+    sps->separate_colour_plane_flag = 0;
+
+    av_assert0(ctx->res_limits.SubregionBlockPixelsSize % min_cu_size == 0);
+
+    sps->pic_width_in_luma_samples  = FFALIGN(base_ctx->surface_width,
+                                              ctx->res_limits.SubregionBlockPixelsSize);
+    sps->pic_height_in_luma_samples = FFALIGN(base_ctx->surface_height,
+                                              ctx->res_limits.SubregionBlockPixelsSize);
+
+    if (avctx->width  != sps->pic_width_in_luma_samples ||
+        avctx->height != sps->pic_height_in_luma_samples) {
+        sps->conformance_window_flag = 1;
+        sps->conf_win_left_offset   = 0;
+        sps->conf_win_right_offset  =
+            (sps->pic_width_in_luma_samples - avctx->width) >> desc->log2_chroma_w;
+        sps->conf_win_top_offset    = 0;
+        sps->conf_win_bottom_offset =
+            (sps->pic_height_in_luma_samples - avctx->height) >> desc->log2_chroma_h;
+    } else {
+        sps->conformance_window_flag = 0;
+    }
+
+    sps->bit_depth_luma_minus8   = bit_depth - 8;
+    sps->bit_depth_chroma_minus8 = bit_depth - 8;
+
+    sps->log2_max_pic_order_cnt_lsb_minus4 = ctx->gop.pHEVCGroupOfPictures->log2_max_pic_order_cnt_lsb_minus4;
+
+    sps->sps_sub_layer_ordering_info_present_flag =
+        vps->vps_sub_layer_ordering_info_present_flag;
+    for (i = 0; i <= sps->sps_max_sub_layers_minus1; i++) {
+        sps->sps_max_dec_pic_buffering_minus1[i] =
+            vps->vps_max_dec_pic_buffering_minus1[i];
+        sps->sps_max_num_reorder_pics[i] =
+            vps->vps_max_num_reorder_pics[i];
+        sps->sps_max_latency_increase_plus1[i] =
+            vps->vps_max_latency_increase_plus1[i];
+    }
+
+    sps->log2_min_luma_coding_block_size_minus3      = (uint8_t)(av_log2(min_cu_size) - 3);
+    sps->log2_diff_max_min_luma_coding_block_size    = (uint8_t)(av_log2(max_cu_size) - av_log2(min_cu_size));
+    sps->log2_min_luma_transform_block_size_minus2   = (uint8_t)(av_log2(min_tu_size) - 2);
+    sps->log2_diff_max_min_luma_transform_block_size = (uint8_t)(av_log2(max_tu_size) - av_log2(min_tu_size));
+
+    sps->max_transform_hierarchy_depth_inter = ctx->codec_conf.pHEVCConfig->max_transform_hierarchy_depth_inter;
+    sps->max_transform_hierarchy_depth_intra = ctx->codec_conf.pHEVCConfig->max_transform_hierarchy_depth_intra;
+
+    sps->amp_enabled_flag = !!(ctx->codec_conf.pHEVCConfig->ConfigurationFlags &
+                               D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_FLAG_USE_ASYMETRIC_MOTION_PARTITION);
+    sps->sample_adaptive_offset_enabled_flag = !!(ctx->codec_conf.pHEVCConfig->ConfigurationFlags &
+                                                  D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_FLAG_ENABLE_SAO_FILTER);
+    sps->sps_temporal_mvp_enabled_flag = 0;
+    sps->pcm_enabled_flag = 0;
+
+    sps->vui_parameters_present_flag = 1;
+
+    if (avctx->sample_aspect_ratio.num != 0 &&
+        avctx->sample_aspect_ratio.den != 0) {
+        int num, den, i;
+        av_reduce(&num, &den, avctx->sample_aspect_ratio.num,
+                  avctx->sample_aspect_ratio.den, 65535);
+        for (i = 0; i < FF_ARRAY_ELEMS(ff_h2645_pixel_aspect); i++) {
+            if (num == ff_h2645_pixel_aspect[i].num &&
+                den == ff_h2645_pixel_aspect[i].den) {
+                vui->aspect_ratio_idc = i;
+                break;
+            }
+        }
+        if (i >= FF_ARRAY_ELEMS(ff_h2645_pixel_aspect)) {
+            vui->aspect_ratio_idc = 255;
+            vui->sar_width  = num;
+            vui->sar_height = den;
+        }
+        vui->aspect_ratio_info_present_flag = 1;
+    }
+
+    // Unspecified video format, from table E-2.
+    vui->video_format             = 5;
+    vui->video_full_range_flag    =
+        avctx->color_range == AVCOL_RANGE_JPEG;
+    vui->colour_primaries         = avctx->color_primaries;
+    vui->transfer_characteristics = avctx->color_trc;
+    vui->matrix_coefficients      = avctx->colorspace;
+    if (avctx->color_primaries != AVCOL_PRI_UNSPECIFIED ||
+        avctx->color_trc       != AVCOL_TRC_UNSPECIFIED ||
+        avctx->colorspace      != AVCOL_SPC_UNSPECIFIED)
+        vui->colour_description_present_flag = 1;
+    if (avctx->color_range     != AVCOL_RANGE_UNSPECIFIED ||
+        vui->colour_description_present_flag)
+        vui->video_signal_type_present_flag = 1;
+
+    if (avctx->chroma_sample_location != AVCHROMA_LOC_UNSPECIFIED) {
+        vui->chroma_loc_info_present_flag = 1;
+        vui->chroma_sample_loc_type_top_field    =
+        vui->chroma_sample_loc_type_bottom_field =
+            avctx->chroma_sample_location - 1;
+    }
+
+    vui->vui_timing_info_present_flag        = 1;
+    vui->vui_num_units_in_tick               = vps->vps_num_units_in_tick;
+    vui->vui_time_scale                      = vps->vps_time_scale;
+    vui->vui_poc_proportional_to_timing_flag = vps->vps_poc_proportional_to_timing_flag;
+    vui->vui_num_ticks_poc_diff_one_minus1   = vps->vps_num_ticks_poc_diff_one_minus1;
+    vui->vui_hrd_parameters_present_flag     = 0;
+
+    vui->bitstream_restriction_flag    = 1;
+    vui->motion_vectors_over_pic_boundaries_flag = 1;
+    vui->restricted_ref_pic_lists_flag = 1;
+    vui->max_bytes_per_pic_denom       = 0;
+    vui->max_bits_per_min_cu_denom     = 0;
+    vui->log2_max_mv_length_horizontal = 15;
+    vui->log2_max_mv_length_vertical   = 15;
+
+    // PPS
+
+    pps->nal_unit_header = (H265RawNALUnitHeader) {
+        .nal_unit_type         = HEVC_NAL_PPS,
+        .nuh_layer_id          = 0,
+        .nuh_temporal_id_plus1 = 1,
+    };
+
+    pps->pps_pic_parameter_set_id = 0;
+    pps->pps_seq_parameter_set_id = sps->sps_seq_parameter_set_id;
+
+    pps->cabac_init_present_flag = 1;
+
+    pps->num_ref_idx_l0_default_active_minus1 = 0;
+    pps->num_ref_idx_l1_default_active_minus1 = 0;
+
+    pps->init_qp_minus26 = 0;
+
+    pps->transform_skip_enabled_flag = !!(ctx->codec_conf.pHEVCConfig->ConfigurationFlags &
+                                          D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_FLAG_ENABLE_TRANSFORM_SKIPPING);
+
+    // cu_qp_delta always required to be 1 in https://github.com/microsoft/DirectX-Specs/blob/master/d3d/D3D12VideoEncoding.md
+    pps->cu_qp_delta_enabled_flag = 1;
+
+    pps->diff_cu_qp_delta_depth   = 0;
+
+    pps->pps_slice_chroma_qp_offsets_present_flag = 1;
+
+    pps->tiles_enabled_flag = 0; // no tiling in D3D12
+
+    pps->pps_loop_filter_across_slices_enabled_flag = !(ctx->codec_conf.pHEVCConfig->ConfigurationFlags &
+                                                        D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_FLAG_DISABLE_LOOP_FILTER_ACROSS_SLICES);
+    pps->deblocking_filter_control_present_flag = 1;
+
+    return 0;
+}
+
+static int d3d12va_encode_hevc_get_encoder_caps(AVCodecContext *avctx)
+{
+    int i;
+    HRESULT hr;
+    uint8_t min_cu_size, max_cu_size;
+    HWBaseEncodeContext *base_ctx = avctx->priv_data;
+    D3D12VAEncodeContext     *ctx = avctx->priv_data;
+    D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC *config;
+    D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_SUPPORT_HEVC hevc_caps;
+
+    D3D12_FEATURE_DATA_VIDEO_ENCODER_CODEC_CONFIGURATION_SUPPORT codec_caps = {
+        .NodeIndex                   = 0,
+        .Codec                       = D3D12_VIDEO_ENCODER_CODEC_HEVC,
+        .Profile                     = ctx->profile->d3d12_profile,
+        .CodecSupportLimits.DataSize = sizeof(D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_SUPPORT_HEVC),
+    };
+
+    for (i = 0; i < FF_ARRAY_ELEMS(hevc_config_support_sets); i++) {
+        hevc_caps = hevc_config_support_sets[i];
+        codec_caps.CodecSupportLimits.pHEVCSupport = &hevc_caps;
+        hr = ID3D12VideoDevice3_CheckFeatureSupport(ctx->video_device3, D3D12_FEATURE_VIDEO_ENCODER_CODEC_CONFIGURATION_SUPPORT,
+                                                    &codec_caps, sizeof(codec_caps));
+        if (SUCCEEDED(hr) && codec_caps.IsSupported)
+            break;
+    }
+
+    if (i == FF_ARRAY_ELEMS(hevc_config_support_sets)) {
+        av_log(avctx, AV_LOG_ERROR, "Unsupported codec configuration\n");
+        return AVERROR(EINVAL);
+    }
+
+    ctx->codec_conf.DataSize = sizeof(D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC);
+    ctx->codec_conf.pHEVCConfig = av_mallocz(ctx->codec_conf.DataSize);
+    if (!ctx->codec_conf.pHEVCConfig)
+        return AVERROR(ENOMEM);
+
+    config = ctx->codec_conf.pHEVCConfig;
+
+    config->ConfigurationFlags                  = D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_FLAG_NONE;
+    config->MinLumaCodingUnitSize               = hevc_caps.MinLumaCodingUnitSize;
+    config->MaxLumaCodingUnitSize               = hevc_caps.MaxLumaCodingUnitSize;
+    config->MinLumaTransformUnitSize            = hevc_caps.MinLumaTransformUnitSize;
+    config->MaxLumaTransformUnitSize            = hevc_caps.MaxLumaTransformUnitSize;
+    config->max_transform_hierarchy_depth_inter = hevc_caps.max_transform_hierarchy_depth_inter;
+    config->max_transform_hierarchy_depth_intra = hevc_caps.max_transform_hierarchy_depth_intra;
+
+    if (hevc_caps.SupportFlags & D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_SUPPORT_HEVC_FLAG_ASYMETRIC_MOTION_PARTITION_SUPPORT ||
+        hevc_caps.SupportFlags & D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_SUPPORT_HEVC_FLAG_ASYMETRIC_MOTION_PARTITION_REQUIRED)
+        config->ConfigurationFlags |= D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_FLAG_USE_ASYMETRIC_MOTION_PARTITION;
+
+    if (hevc_caps.SupportFlags & D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_SUPPORT_HEVC_FLAG_SAO_FILTER_SUPPORT)
+        config->ConfigurationFlags |= D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_FLAG_ENABLE_SAO_FILTER;
+
+    if (hevc_caps.SupportFlags & D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_SUPPORT_HEVC_FLAG_DISABLING_LOOP_FILTER_ACROSS_SLICES_SUPPORT)
+        config->ConfigurationFlags |= D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_FLAG_DISABLE_LOOP_FILTER_ACROSS_SLICES;
+
+    if (hevc_caps.SupportFlags & D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_SUPPORT_HEVC_FLAG_TRANSFORM_SKIP_SUPPORT)
+        config->ConfigurationFlags |= D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_HEVC_FLAG_ENABLE_TRANSFORM_SKIPPING;
+
+    if (hevc_caps.SupportFlags & D3D12_VIDEO_ENCODER_CODEC_CONFIGURATION_SUPPORT_HEVC_FLAG_P_FRAMES_IMPLEMENTED_AS_LOW_DELAY_B_FRAMES)
+        ctx->bi_not_empty = 1;
+
+    // block sizes
+    min_cu_size = d3d12va_encode_hevc_map_cusize(hevc_caps.MinLumaCodingUnitSize);
+    max_cu_size = d3d12va_encode_hevc_map_cusize(hevc_caps.MaxLumaCodingUnitSize);
+
+    av_log(avctx, AV_LOG_VERBOSE, "Using CTU size %dx%d, "
+           "min CB size %dx%d.\n", max_cu_size, max_cu_size,
+           min_cu_size, min_cu_size);
+
+    base_ctx->surface_width  = FFALIGN(avctx->width,  min_cu_size);
+    base_ctx->surface_height = FFALIGN(avctx->height, min_cu_size);
+
+    return 0;
+}
+
+static int d3d12va_encode_hevc_configure(AVCodecContext *avctx)
+{
+    HWBaseEncodeContext  *base_ctx = avctx->priv_data;
+    D3D12VAEncodeContext      *ctx = avctx->priv_data;
+    D3D12VAEncodeHEVCContext *priv = avctx->priv_data;
+    int fixed_qp_idr, fixed_qp_p, fixed_qp_b;
+    int err;
+
+    err = ff_cbs_init(&priv->cbc, AV_CODEC_ID_HEVC, avctx);
+    if (err < 0)
+        return err;
+
+    // Rate control
+    if (ctx->rc.Mode == D3D12_VIDEO_ENCODER_RATE_CONTROL_MODE_CQP) {
+        D3D12_VIDEO_ENCODER_RATE_CONTROL_CQP *cqp_ctl;
+        fixed_qp_p = av_clip(ctx->rc_quality, 1, 51);
+        if (avctx->i_quant_factor > 0.0)
+            fixed_qp_idr = av_clip((avctx->i_quant_factor * fixed_qp_p +
+                                    avctx->i_quant_offset) + 0.5, 1, 51);
+        else
+            fixed_qp_idr = fixed_qp_p;
+        if (avctx->b_quant_factor > 0.0)
+            fixed_qp_b = av_clip((avctx->b_quant_factor * fixed_qp_p +
+                                  avctx->b_quant_offset) + 0.5, 1, 51);
+        else
+            fixed_qp_b = fixed_qp_p;
+
+        av_log(avctx, AV_LOG_DEBUG, "Using fixed QP = "
+               "%d / %d / %d for IDR- / P- / B-frames.\n",
+               fixed_qp_idr, fixed_qp_p, fixed_qp_b);
+
+        ctx->rc.ConfigParams.DataSize = sizeof(D3D12_VIDEO_ENCODER_RATE_CONTROL_CQP);
+        cqp_ctl = av_mallocz(ctx->rc.ConfigParams.DataSize);
+        if (!cqp_ctl)
+            return AVERROR(ENOMEM);
+
+        cqp_ctl->ConstantQP_FullIntracodedFrame                  = fixed_qp_idr;
+        cqp_ctl->ConstantQP_InterPredictedFrame_PrevRefOnly      = fixed_qp_p;
+        cqp_ctl->ConstantQP_InterPredictedFrame_BiDirectionalRef = fixed_qp_b;
+
+        ctx->rc.ConfigParams.pConfiguration_CQP = cqp_ctl;
+    }
+
+    // GOP
+    ctx->gop.DataSize = sizeof(D3D12_VIDEO_ENCODER_SEQUENCE_GOP_STRUCTURE_HEVC);
+    ctx->gop.pHEVCGroupOfPictures = av_mallocz(ctx->gop.DataSize);
+    if (!ctx->gop.pHEVCGroupOfPictures)
+        return AVERROR(ENOMEM);
+
+    ctx->gop.pHEVCGroupOfPictures->GOPLength      = base_ctx->gop_size;
+    ctx->gop.pHEVCGroupOfPictures->PPicturePeriod = base_ctx->b_per_p + 1;
+    // Power of 2
+    if (base_ctx->gop_size & base_ctx->gop_size - 1 == 0)
+        ctx->gop.pHEVCGroupOfPictures->log2_max_pic_order_cnt_lsb_minus4 =
+            FFMAX(av_log2(base_ctx->gop_size) - 4, 0);
+    else
+        ctx->gop.pHEVCGroupOfPictures->log2_max_pic_order_cnt_lsb_minus4 =
+            FFMAX(av_log2(base_ctx->gop_size) - 3, 0);
+
+    return 0;
+}
+
+static int d3d12va_encode_hevc_set_level(AVCodecContext *avctx)
+{
+    D3D12VAEncodeContext      *ctx = avctx->priv_data;
+    D3D12VAEncodeHEVCContext *priv = avctx->priv_data;
+    int i;
+
+    ctx->level.DataSize = sizeof(D3D12_VIDEO_ENCODER_LEVEL_TIER_CONSTRAINTS_HEVC);
+    ctx->level.pHEVCLevelSetting = av_mallocz(ctx->level.DataSize);
+    if (!ctx->level.pHEVCLevelSetting)
+        return AVERROR(ENOMEM);
+
+    for (i = 0; i < FF_ARRAY_ELEMS(hevc_levels); i++) {
+        if (avctx->level == hevc_levels[i].level) {
+            ctx->level.pHEVCLevelSetting->Level = hevc_levels[i].d3d12_level;
+            break;
+        }
+    }
+
+    if (i == FF_ARRAY_ELEMS(hevc_levels)) {
+        av_log(avctx, AV_LOG_ERROR, "Invalid level %d.\n", avctx->level);
+        return AVERROR(EINVAL);
+    }
+
+    ctx->level.pHEVCLevelSetting->Tier = priv->raw_vps.profile_tier_level.general_tier_flag == 0 ?
+                                         D3D12_VIDEO_ENCODER_TIER_HEVC_MAIN :
+                                         D3D12_VIDEO_ENCODER_TIER_HEVC_HIGH;
+
+    return 0;
+}
+
+static void d3d12va_encode_hevc_free_picture_params(D3D12VAEncodePicture *pic)
+{
+    if (!pic->pic_ctl.pHEVCPicData)
+        return;
+
+    av_freep(&pic->pic_ctl.pHEVCPicData->pList0ReferenceFrames);
+    av_freep(&pic->pic_ctl.pHEVCPicData->pList1ReferenceFrames);
+    av_freep(&pic->pic_ctl.pHEVCPicData->pReferenceFramesReconPictureDescriptors);
+    av_freep(&pic->pic_ctl.pHEVCPicData);
+}
+
+static int d3d12va_encode_hevc_init_picture_params(AVCodecContext *avctx,
+                                                   D3D12VAEncodePicture *pic)
+{
+    HWBaseEncodePicture                             *base_pic = &pic->base;
+    D3D12VAEncodeHEVCPicture                            *hpic = base_pic->priv_data;
+    HWBaseEncodePicture                                 *prev = base_pic->prev;
+    D3D12VAEncodeHEVCPicture                           *hprev = prev ? prev->priv_data : NULL;
+    D3D12_VIDEO_ENCODER_REFERENCE_PICTURE_DESCRIPTOR_HEVC *pd = NULL;
+    UINT                                           *ref_list0 = NULL, *ref_list1 = NULL;
+    int i, idx = 0;
+
+    pic->pic_ctl.DataSize = sizeof(D3D12_VIDEO_ENCODER_PICTURE_CONTROL_CODEC_DATA_HEVC);
+    pic->pic_ctl.pHEVCPicData = av_mallocz(pic->pic_ctl.DataSize);
+    if (!pic->pic_ctl.pHEVCPicData)
+        return AVERROR(ENOMEM);
+
+    if (base_pic->type == PICTURE_TYPE_IDR) {
+        av_assert0(base_pic->display_order == base_pic->encode_order);
+        hpic->last_idr_frame = base_pic->display_order;
+    } else {
+        av_assert0(prev);
+        hpic->last_idr_frame = hprev->last_idr_frame;
+    }
+    hpic->pic_order_cnt = base_pic->display_order - hpic->last_idr_frame;
+
+    switch(base_pic->type) {
+        case PICTURE_TYPE_IDR:
+            pic->pic_ctl.pHEVCPicData->FrameType = D3D12_VIDEO_ENCODER_FRAME_TYPE_HEVC_IDR_FRAME;
+            break;
+        case PICTURE_TYPE_I:
+            pic->pic_ctl.pHEVCPicData->FrameType = D3D12_VIDEO_ENCODER_FRAME_TYPE_HEVC_I_FRAME;
+            break;
+        case PICTURE_TYPE_P:
+            pic->pic_ctl.pHEVCPicData->FrameType = D3D12_VIDEO_ENCODER_FRAME_TYPE_HEVC_P_FRAME;
+            break;
+        case PICTURE_TYPE_B:
+            pic->pic_ctl.pHEVCPicData->FrameType = D3D12_VIDEO_ENCODER_FRAME_TYPE_HEVC_B_FRAME;
+            break;
+        default:
+            av_assert0(0 && "invalid picture type");
+    }
+
+    pic->pic_ctl.pHEVCPicData->slice_pic_parameter_set_id = 0;
+    pic->pic_ctl.pHEVCPicData->PictureOrderCountNumber    = hpic->pic_order_cnt;
+
+    if (base_pic->type == PICTURE_TYPE_P || base_pic->type == PICTURE_TYPE_B) {
+        pd = av_calloc(MAX_PICTURE_REFERENCES, sizeof(*pd));
+        if (!pd)
+            return AVERROR(ENOMEM);
+
+        ref_list0 = av_calloc(MAX_PICTURE_REFERENCES, sizeof(*ref_list0));
+        if (!ref_list0)
+            return AVERROR(ENOMEM);
+
+        pic->pic_ctl.pHEVCPicData->List0ReferenceFramesCount = base_pic->nb_refs[0];
+        for (i = 0; i < base_pic->nb_refs[0]; i++) {
+            HWBaseEncodePicture      *ref = base_pic->refs[0][i];
+            D3D12VAEncodeHEVCPicture *href;
+
+            av_assert0(ref && ref->encode_order < base_pic->encode_order);
+            href = ref->priv_data;
+
+            ref_list0[i] = idx;
+            pd[idx].ReconstructedPictureResourceIndex = idx;
+            pd[idx].IsRefUsedByCurrentPic = TRUE;
+            pd[idx].PictureOrderCountNumber = href->pic_order_cnt;
+            idx++;
+        }
+    }
+
+    if (base_pic->type == PICTURE_TYPE_B) {
+        ref_list1 = av_calloc(MAX_PICTURE_REFERENCES, sizeof(*ref_list1));
+        if (!ref_list1)
+            return AVERROR(ENOMEM);
+
+        pic->pic_ctl.pHEVCPicData->List1ReferenceFramesCount = base_pic->nb_refs[1];
+        for (i = 0; i < base_pic->nb_refs[1]; i++) {
+            HWBaseEncodePicture      *ref = base_pic->refs[1][i];
+            D3D12VAEncodeHEVCPicture *href;
+
+            av_assert0(ref && ref->encode_order < base_pic->encode_order);
+            href = ref->priv_data;
+
+            ref_list1[i] = idx;
+            pd[idx].ReconstructedPictureResourceIndex = idx;
+            pd[idx].IsRefUsedByCurrentPic = TRUE;
+            pd[idx].PictureOrderCountNumber = href->pic_order_cnt;
+            idx++;
+        }
+    }
+
+    pic->pic_ctl.pHEVCPicData->pList0ReferenceFrames = ref_list0;
+    pic->pic_ctl.pHEVCPicData->pList1ReferenceFrames = ref_list1;
+    pic->pic_ctl.pHEVCPicData->ReferenceFramesReconPictureDescriptorsCount = idx;
+    pic->pic_ctl.pHEVCPicData->pReferenceFramesReconPictureDescriptors = pd;
+
+    return 0;
+}
+
+static const D3D12VAEncodeType d3d12va_encode_type_hevc = {
+    .profiles               = d3d12va_encode_hevc_profiles,
+
+    .d3d12_codec            = D3D12_VIDEO_ENCODER_CODEC_HEVC,
+
+    .flags                  = FLAG_B_PICTURES |
+                              FLAG_B_PICTURE_REFERENCES |
+                              FLAG_NON_IDR_KEY_PICTURES,
+
+    .default_quality        = 25,
+
+    .get_encoder_caps       = &d3d12va_encode_hevc_get_encoder_caps,
+
+    .configure              = &d3d12va_encode_hevc_configure,
+
+    .set_level              = &d3d12va_encode_hevc_set_level,
+
+    .picture_priv_data_size = sizeof(D3D12VAEncodeHEVCPicture),
+
+    .init_sequence_params   = &d3d12va_encode_hevc_init_sequence_params,
+
+    .init_picture_params    = &d3d12va_encode_hevc_init_picture_params,
+
+    .free_picture_params    = &d3d12va_encode_hevc_free_picture_params,
+
+    .write_sequence_header  = &d3d12va_encode_hevc_write_sequence_header,
+};
+
+static int d3d12va_encode_hevc_init(AVCodecContext *avctx)
+{
+    D3D12VAEncodeContext      *ctx = avctx->priv_data;
+    D3D12VAEncodeHEVCContext *priv = avctx->priv_data;
+
+    ctx->codec = &d3d12va_encode_type_hevc;
+
+    if (avctx->profile == AV_PROFILE_UNKNOWN)
+        avctx->profile = priv->profile;
+    if (avctx->level == FF_LEVEL_UNKNOWN)
+        avctx->level = priv->level;
+
+    if (avctx->level != FF_LEVEL_UNKNOWN && avctx->level & ~0xff) {
+        av_log(avctx, AV_LOG_ERROR, "Invalid level %d: must fit "
+               "in 8-bit unsigned integer.\n", avctx->level);
+        return AVERROR(EINVAL);
+    }
+
+    if (priv->qp > 0)
+        ctx->explicit_qp = priv->qp;
+
+    return ff_d3d12va_encode_init(avctx);
+}
+
+static int d3d12va_encode_hevc_close(AVCodecContext *avctx)
+{
+    D3D12VAEncodeHEVCContext *priv = avctx->priv_data;
+
+    ff_cbs_fragment_free(&priv->current_access_unit);
+    ff_cbs_close(&priv->cbc);
+
+    av_freep(&priv->common.codec_conf.pHEVCConfig);
+    av_freep(&priv->common.gop.pHEVCGroupOfPictures);
+    av_freep(&priv->common.level.pHEVCLevelSetting);
+
+    return ff_d3d12va_encode_close(avctx);
+}
+
+#define OFFSET(x) offsetof(D3D12VAEncodeHEVCContext, x)
+#define FLAGS (AV_OPT_FLAG_VIDEO_PARAM | AV_OPT_FLAG_ENCODING_PARAM)
+static const AVOption d3d12va_encode_hevc_options[] = {
+    HW_BASE_ENCODE_COMMON_OPTIONS,
+    D3D12VA_ENCODE_RC_OPTIONS,
+
+    { "qp", "Constant QP (for P-frames; scaled by qfactor/qoffset for I/B)",
+      OFFSET(qp), AV_OPT_TYPE_INT, { .i64 = 0 }, 0, 52, FLAGS },
+
+    { "profile", "Set profile (general_profile_idc)",
+      OFFSET(profile), AV_OPT_TYPE_INT,
+      { .i64 = AV_PROFILE_UNKNOWN }, AV_PROFILE_UNKNOWN, 0xff, FLAGS, "profile" },
+
+#define PROFILE(name, value)  name, NULL, 0, AV_OPT_TYPE_CONST, \
+      { .i64 = value }, 0, 0, FLAGS, "profile"
+    { PROFILE("main",               AV_PROFILE_HEVC_MAIN) },
+    { PROFILE("main10",             AV_PROFILE_HEVC_MAIN_10) },
+#undef PROFILE
+
+    { "tier", "Set tier (general_tier_flag)",
+      OFFSET(tier), AV_OPT_TYPE_INT,
+      { .i64 = 0 }, 0, 1, FLAGS, "tier" },
+    { "main", NULL, 0, AV_OPT_TYPE_CONST,
+      { .i64 = 0 }, 0, 0, FLAGS, "tier" },
+    { "high", NULL, 0, AV_OPT_TYPE_CONST,
+      { .i64 = 1 }, 0, 0, FLAGS, "tier" },
+
+    { "level", "Set level (general_level_idc)",
+      OFFSET(level), AV_OPT_TYPE_INT,
+      { .i64 = FF_LEVEL_UNKNOWN }, FF_LEVEL_UNKNOWN, 0xff, FLAGS, "level" },
+
+#define LEVEL(name, value) name, NULL, 0, AV_OPT_TYPE_CONST, \
+      { .i64 = value }, 0, 0, FLAGS, "level"
+    { LEVEL("1",    30) },
+    { LEVEL("2",    60) },
+    { LEVEL("2.1",  63) },
+    { LEVEL("3",    90) },
+    { LEVEL("3.1",  93) },
+    { LEVEL("4",   120) },
+    { LEVEL("4.1", 123) },
+    { LEVEL("5",   150) },
+    { LEVEL("5.1", 153) },
+    { LEVEL("5.2", 156) },
+    { LEVEL("6",   180) },
+    { LEVEL("6.1", 183) },
+    { LEVEL("6.2", 186) },
+#undef LEVEL
+
+    { NULL },
+};
+
+static const FFCodecDefault d3d12va_encode_hevc_defaults[] = {
+    { "b",              "0"   },
+    { "bf",             "2"   },
+    { "g",              "120" },
+    { "i_qfactor",      "1"   },
+    { "i_qoffset",      "0"   },
+    { "b_qfactor",      "1"   },
+    { "b_qoffset",      "0"   },
+    { "qmin",           "-1"  },
+    { "qmax",           "-1"  },
+    { NULL },
+};
+
+static const AVClass d3d12va_encode_hevc_class = {
+    .class_name = "hevc_d3d12va",
+    .item_name  = av_default_item_name,
+    .option     = d3d12va_encode_hevc_options,
+    .version    = LIBAVUTIL_VERSION_INT,
+};
+
+const FFCodec ff_hevc_d3d12va_encoder = {
+    .p.name         = "hevc_d3d12va",
+    CODEC_LONG_NAME("D3D12VA hevc encoder"),
+    .p.type         = AVMEDIA_TYPE_VIDEO,
+    .p.id           = AV_CODEC_ID_HEVC,
+    .priv_data_size = sizeof(D3D12VAEncodeHEVCContext),
+    .init           = &d3d12va_encode_hevc_init,
+    FF_CODEC_RECEIVE_PACKET_CB(&ff_hw_base_encode_receive_packet),
+    .close          = &d3d12va_encode_hevc_close,
+    .p.priv_class   = &d3d12va_encode_hevc_class,
+    .p.capabilities = AV_CODEC_CAP_DELAY | AV_CODEC_CAP_HARDWARE |
+                      AV_CODEC_CAP_DR1 | AV_CODEC_CAP_ENCODER_REORDERED_OPAQUE,
+    .caps_internal  = FF_CODEC_CAP_NOT_INIT_THREADSAFE |
+                      FF_CODEC_CAP_INIT_CLEANUP,
+    .defaults       = d3d12va_encode_hevc_defaults,
+    .p.pix_fmts = (const enum AVPixelFormat[]) {
+        AV_PIX_FMT_D3D12,
+        AV_PIX_FMT_NONE,
+    },
+    .hw_configs     = ff_d3d12va_encode_hw_configs,
+    .p.wrapper_name = "d3d12va",
+};
-- 
2.41.0.windows.1

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

^ permalink raw reply	[flat|nested] 18+ messages in thread

* [FFmpeg-devel] [PATCH v9 13/13] Changelog: add D3D12VA HEVC encoder changelog
  2024-05-20 14:52 [FFmpeg-devel] [PATCH v9 01/13] avcodec/vaapi_encode: introduce a base layer for vaapi encode tong1.wu-at-intel.com
                   ` (10 preceding siblings ...)
  2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 12/13] avcodec: add D3D12VA hardware HEVC encoder tong1.wu-at-intel.com
@ 2024-05-20 14:52 ` tong1.wu-at-intel.com
  2024-05-23 16:11 ` [FFmpeg-devel] [PATCH v9 01/13] avcodec/vaapi_encode: introduce a base layer for vaapi encode Lynne via ffmpeg-devel
  12 siblings, 0 replies; 18+ messages in thread
From: tong1.wu-at-intel.com @ 2024-05-20 14:52 UTC (permalink / raw)
  To: ffmpeg-devel; +Cc: Tong Wu

From: Tong Wu <tong1.wu@intel.com>

Signed-off-by: Tong Wu <tong1.wu@intel.com>
---
 Changelog | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/Changelog b/Changelog
index dd25715d6b..c3f802617f 100644
--- a/Changelog
+++ b/Changelog
@@ -10,7 +10,7 @@ version <next>:
 - vf_scale supports secondary ref input and framesync options
 - vf_scale2ref deprecated
 - qsv_params option added for QSV encoders
-
+- D3D12VA HEVC encoder
 
 version 7.0:
 - DXV DXT1 encoder
-- 
2.41.0.windows.1

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

^ permalink raw reply	[flat|nested] 18+ messages in thread

* Re: [FFmpeg-devel] [PATCH v9 11/13] avutil/hwcontext_d3d12va: add Flags for resource creation
  2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 11/13] avutil/hwcontext_d3d12va: add Flags for resource creation tong1.wu-at-intel.com
@ 2024-05-21 16:32   ` Andrew Sayers
  2024-05-22  1:29     ` Wu, Tong1
  0 siblings, 1 reply; 18+ messages in thread
From: Andrew Sayers @ 2024-05-21 16:32 UTC (permalink / raw)
  To: FFmpeg development discussions and patches

(Only reviewing documentation, not code)

On Mon, May 20, 2024 at 10:52:20PM +0800, tong1.wu-at-intel.com@ffmpeg.org wrote:
> From: Tong Wu <tong1.wu@intel.com>
> 
> Flags field is added to support diffferent resource creation.
> 
> Signed-off-by: Tong Wu <tong1.wu@intel.com>
> ---
>  doc/APIchanges                | 3 +++
>  libavutil/hwcontext_d3d12va.c | 2 +-
>  libavutil/hwcontext_d3d12va.h | 8 ++++++++
>  libavutil/version.h           | 2 +-
>  4 files changed, 13 insertions(+), 2 deletions(-)
> 
> diff --git a/doc/APIchanges b/doc/APIchanges
> index 269fd36559..808ba02f2d 100644
> --- a/doc/APIchanges
> +++ b/doc/APIchanges
> @@ -2,6 +2,9 @@ The last version increases of all libraries were on 2024-03-07
>  
>  API changes, most recent first:
>  
> +2024-01-xx - xxxxxxxxxx - lavu 59.20.100 - hwcontext_d3d12va.h
> + Add AVD3D12VAFramesContext.flags
> +
>  2024-05-xx - xxxxxxxxxx - lavu 59.19.100 - hwcontext_qsv.h
>    Add AVQSVFramesContext.info
>  
> diff --git a/libavutil/hwcontext_d3d12va.c b/libavutil/hwcontext_d3d12va.c
> index cfc016315d..6507cf69c1 100644
> --- a/libavutil/hwcontext_d3d12va.c
> +++ b/libavutil/hwcontext_d3d12va.c
> @@ -247,7 +247,7 @@ static AVBufferRef *d3d12va_pool_alloc(void *opaque, size_t size)
>          .Format           = hwctx->format,
>          .SampleDesc       = {.Count = 1, .Quality = 0 },
>          .Layout           = D3D12_TEXTURE_LAYOUT_UNKNOWN,
> -        .Flags            = D3D12_RESOURCE_FLAG_NONE,
> +        .Flags            = hwctx->flags,
>      };
>  
>      frame = av_mallocz(sizeof(AVD3D12VAFrame));
> diff --git a/libavutil/hwcontext_d3d12va.h b/libavutil/hwcontext_d3d12va.h
> index ff06e6f2ef..608dbac97f 100644
> --- a/libavutil/hwcontext_d3d12va.h
> +++ b/libavutil/hwcontext_d3d12va.h
> @@ -129,6 +129,14 @@ typedef struct AVD3D12VAFramesContext {
>       * If unset, will be automatically set.
>       */
>      DXGI_FORMAT format;
> +
> +    /**
> +     * This field is used to specify options for working with resources.
> +     * If unset, this will be D3D12_RESOURCE_FLAG_NONE.
> +     *
> +     * @see: https://learn.microsoft.com/en-us/windows/win32/api/d3d12/ne-d3d12-d3d12_resource_flags.
> +     */
> +    D3D12_RESOURCE_FLAGS flags;

Some nitpicks:

* "This field is used to specify" is redundant, you can save the reader
  a few seconds by starting the sentence with just "Options..."
* "@see" starts a paragraph, so the rendered documentation will look better
  without the ":"
* the full stop after the URL makes it harder to copy/paste the text -
  remove the full stop or use a [markdown link](...)
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

^ permalink raw reply	[flat|nested] 18+ messages in thread

* Re: [FFmpeg-devel] [PATCH v9 11/13] avutil/hwcontext_d3d12va: add Flags for resource creation
  2024-05-21 16:32   ` Andrew Sayers
@ 2024-05-22  1:29     ` Wu, Tong1
  0 siblings, 0 replies; 18+ messages in thread
From: Wu, Tong1 @ 2024-05-22  1:29 UTC (permalink / raw)
  To: FFmpeg development discussions and patches

>-----Original Message-----
>From: ffmpeg-devel <ffmpeg-devel-bounces@ffmpeg.org> On Behalf Of
>Andrew Sayers
>Sent: Wednesday, May 22, 2024 12:32 AM
>To: FFmpeg development discussions and patches <ffmpeg-
>devel@ffmpeg.org>
>Subject: Re: [FFmpeg-devel] [PATCH v9 11/13] avutil/hwcontext_d3d12va: add
>Flags for resource creation
>
>(Only reviewing documentation, not code)
>
>On Mon, May 20, 2024 at 10:52:20PM +0800, tong1.wu-at-
>intel.com@ffmpeg.org wrote:
>> From: Tong Wu <tong1.wu@intel.com>
>>
>> Flags field is added to support diffferent resource creation.
>>
>> Signed-off-by: Tong Wu <tong1.wu@intel.com>
>> ---
>>  doc/APIchanges                | 3 +++
>>  libavutil/hwcontext_d3d12va.c | 2 +-
>>  libavutil/hwcontext_d3d12va.h | 8 ++++++++
>>  libavutil/version.h           | 2 +-
>>  4 files changed, 13 insertions(+), 2 deletions(-)
>>
>> diff --git a/doc/APIchanges b/doc/APIchanges
>> index 269fd36559..808ba02f2d 100644
>> --- a/doc/APIchanges
>> +++ b/doc/APIchanges
>> @@ -2,6 +2,9 @@ The last version increases of all libraries were on 2024-03-
>07
>>
>>  API changes, most recent first:
>>
>> +2024-01-xx - xxxxxxxxxx - lavu 59.20.100 - hwcontext_d3d12va.h
>> + Add AVD3D12VAFramesContext.flags
>> +
>>  2024-05-xx - xxxxxxxxxx - lavu 59.19.100 - hwcontext_qsv.h
>>    Add AVQSVFramesContext.info
>>
>> diff --git a/libavutil/hwcontext_d3d12va.c b/libavutil/hwcontext_d3d12va.c
>> index cfc016315d..6507cf69c1 100644
>> --- a/libavutil/hwcontext_d3d12va.c
>> +++ b/libavutil/hwcontext_d3d12va.c
>> @@ -247,7 +247,7 @@ static AVBufferRef *d3d12va_pool_alloc(void
>*opaque, size_t size)
>>          .Format           = hwctx->format,
>>          .SampleDesc       = {.Count = 1, .Quality = 0 },
>>          .Layout           = D3D12_TEXTURE_LAYOUT_UNKNOWN,
>> -        .Flags            = D3D12_RESOURCE_FLAG_NONE,
>> +        .Flags            = hwctx->flags,
>>      };
>>
>>      frame = av_mallocz(sizeof(AVD3D12VAFrame));
>> diff --git a/libavutil/hwcontext_d3d12va.h b/libavutil/hwcontext_d3d12va.h
>> index ff06e6f2ef..608dbac97f 100644
>> --- a/libavutil/hwcontext_d3d12va.h
>> +++ b/libavutil/hwcontext_d3d12va.h
>> @@ -129,6 +129,14 @@ typedef struct AVD3D12VAFramesContext {
>>       * If unset, will be automatically set.
>>       */
>>      DXGI_FORMAT format;
>> +
>> +    /**
>> +     * This field is used to specify options for working with resources.
>> +     * If unset, this will be D3D12_RESOURCE_FLAG_NONE.
>> +     *
>> +     * @see: https://learn.microsoft.com/en-
>us/windows/win32/api/d3d12/ne-d3d12-d3d12_resource_flags.
>> +     */
>> +    D3D12_RESOURCE_FLAGS flags;
>
>Some nitpicks:
>
>* "This field is used to specify" is redundant, you can save the reader
>  a few seconds by starting the sentence with just "Options..."
>* "@see" starts a paragraph, so the rendered documentation will look better
>  without the ":"
>* the full stop after the URL makes it harder to copy/paste the text -
>  remove the full stop or use a [markdown link](...)

Sounds good. I've updated it with a new version.

-Tong


_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

^ permalink raw reply	[flat|nested] 18+ messages in thread

* Re: [FFmpeg-devel] [PATCH v9 01/13] avcodec/vaapi_encode: introduce a base layer for vaapi encode
  2024-05-20 14:52 [FFmpeg-devel] [PATCH v9 01/13] avcodec/vaapi_encode: introduce a base layer for vaapi encode tong1.wu-at-intel.com
                   ` (11 preceding siblings ...)
  2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 13/13] Changelog: add D3D12VA HEVC encoder changelog tong1.wu-at-intel.com
@ 2024-05-23 16:11 ` Lynne via ffmpeg-devel
  2024-05-24 15:39   ` Wu, Tong1
  12 siblings, 1 reply; 18+ messages in thread
From: Lynne via ffmpeg-devel @ 2024-05-23 16:11 UTC (permalink / raw)
  To: ffmpeg-devel; +Cc: Lynne


[-- Attachment #1.1.1.1: Type: text/plain, Size: 6576 bytes --]

On 20/05/2024 16:52, tong1.wu-at-intel.com@ffmpeg.org wrote:
> From: Tong Wu <tong1.wu@intel.com>
> 
> Since VAAPI and future D3D12VA implementation may share some common parameters,
> a base layer encode context is introduced as vaapi context's base.
> 
> Signed-off-by: Tong Wu <tong1.wu@intel.com>
> ---
>   libavcodec/hw_base_encode.h | 56 +++++++++++++++++++++++++++++++++++++
>   libavcodec/vaapi_encode.h   | 39 +++++---------------------
>   2 files changed, 63 insertions(+), 32 deletions(-)
>   create mode 100644 libavcodec/hw_base_encode.h
> 
> diff --git a/libavcodec/hw_base_encode.h b/libavcodec/hw_base_encode.h
> new file mode 100644
> index 0000000000..1996179456
> --- /dev/null
> +++ b/libavcodec/hw_base_encode.h
> @@ -0,0 +1,56 @@
> +/*
> + * This file is part of FFmpeg.
> + *
> + * FFmpeg is free software; you can redistribute it and/or
> + * modify it under the terms of the GNU Lesser General Public
> + * License as published by the Free Software Foundation; either
> + * version 2.1 of the License, or (at your option) any later version.
> + *
> + * FFmpeg is distributed in the hope that it will be useful,
> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
> + * Lesser General Public License for more details.
> + *
> + * You should have received a copy of the GNU Lesser General Public
> + * License along with FFmpeg; if not, write to the Free Software
> + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
> + */
> +
> +#ifndef AVCODEC_HW_BASE_ENCODE_H
> +#define AVCODEC_HW_BASE_ENCODE_H
> +
> +#define MAX_DPB_SIZE 16
> +#define MAX_PICTURE_REFERENCES 2
> +#define MAX_REORDER_DELAY 16
> +#define MAX_ASYNC_DEPTH 64
> +#define MAX_REFERENCE_LIST_NUM 2
> +
> +enum {
> +    PICTURE_TYPE_IDR = 0,
> +    PICTURE_TYPE_I   = 1,
> +    PICTURE_TYPE_P   = 2,
> +    PICTURE_TYPE_B   = 3,
> +};
> +
> +enum {
> +    // Codec supports controlling the subdivision of pictures into slices.
> +    FLAG_SLICE_CONTROL         = 1 << 0,
> +    // Codec only supports constant quality (no rate control).
> +    FLAG_CONSTANT_QUALITY_ONLY = 1 << 1,
> +    // Codec is intra-only.
> +    FLAG_INTRA_ONLY            = 1 << 2,
> +    // Codec supports B-pictures.
> +    FLAG_B_PICTURES            = 1 << 3,
> +    // Codec supports referencing B-pictures.
> +    FLAG_B_PICTURE_REFERENCES  = 1 << 4,
> +    // Codec supports non-IDR key pictures (that is, key pictures do
> +    // not necessarily empty the DPB).
> +    FLAG_NON_IDR_KEY_PICTURES  = 1 << 5,
> +};
> +
> +typedef struct HWBaseEncodeContext {
> +    const AVClass *class;
> +} HWBaseEncodeContext;
> +
> +#endif /* AVCODEC_HW_BASE_ENCODE_H */
> +
> diff --git a/libavcodec/vaapi_encode.h b/libavcodec/vaapi_encode.h
> index 0eed9691ca..f5c9be8973 100644
> --- a/libavcodec/vaapi_encode.h
> +++ b/libavcodec/vaapi_encode.h
> @@ -33,34 +33,27 @@
>   
>   #include "avcodec.h"
>   #include "hwconfig.h"
> +#include "hw_base_encode.h"
>   
>   struct VAAPIEncodeType;
>   struct VAAPIEncodePicture;
>   
> +// Codec output packet without timestamp delay, which means the
> +// output packet has same PTS and DTS.
> +#define FLAG_TIMESTAMP_NO_DELAY 1 << 6
> +
>   enum {
>       MAX_CONFIG_ATTRIBUTES  = 4,
>       MAX_GLOBAL_PARAMS      = 4,
> -    MAX_DPB_SIZE           = 16,
> -    MAX_PICTURE_REFERENCES = 2,
> -    MAX_REORDER_DELAY      = 16,
>       MAX_PARAM_BUFFER_SIZE  = 1024,
>       // A.4.1: table A.6 allows at most 22 tile rows for any level.
>       MAX_TILE_ROWS          = 22,
>       // A.4.1: table A.6 allows at most 20 tile columns for any level.
>       MAX_TILE_COLS          = 20,
> -    MAX_ASYNC_DEPTH        = 64,
> -    MAX_REFERENCE_LIST_NUM = 2,
>   };
>   
>   extern const AVCodecHWConfigInternal *const ff_vaapi_encode_hw_configs[];
>   
> -enum {
> -    PICTURE_TYPE_IDR = 0,
> -    PICTURE_TYPE_I   = 1,
> -    PICTURE_TYPE_P   = 2,
> -    PICTURE_TYPE_B   = 3,
> -};
> -
>   typedef struct VAAPIEncodeSlice {
>       int             index;
>       int             row_start;
> @@ -193,7 +186,8 @@ typedef struct VAAPIEncodeRCMode {
>   } VAAPIEncodeRCMode;
>   
>   typedef struct VAAPIEncodeContext {
> -    const AVClass *class;
> +    // Base context.
> +    HWBaseEncodeContext base;
>   
>       // Codec-specific hooks.
>       const struct VAAPIEncodeType *codec;
> @@ -397,25 +391,6 @@ typedef struct VAAPIEncodeContext {
>       AVPacket        *tail_pkt;
>   } VAAPIEncodeContext;
>   
> -enum {
> -    // Codec supports controlling the subdivision of pictures into slices.
> -    FLAG_SLICE_CONTROL         = 1 << 0,
> -    // Codec only supports constant quality (no rate control).
> -    FLAG_CONSTANT_QUALITY_ONLY = 1 << 1,
> -    // Codec is intra-only.
> -    FLAG_INTRA_ONLY            = 1 << 2,
> -    // Codec supports B-pictures.
> -    FLAG_B_PICTURES            = 1 << 3,
> -    // Codec supports referencing B-pictures.
> -    FLAG_B_PICTURE_REFERENCES  = 1 << 4,
> -    // Codec supports non-IDR key pictures (that is, key pictures do
> -    // not necessarily empty the DPB).
> -    FLAG_NON_IDR_KEY_PICTURES  = 1 << 5,
> -    // Codec output packet without timestamp delay, which means the
> -    // output packet has same PTS and DTS.
> -    FLAG_TIMESTAMP_NO_DELAY    = 1 << 6,
> -};
> -
>   typedef struct VAAPIEncodeType {
>       // List of supported profiles and corresponding VAAPI profiles.
>       // (Must end with AV_PROFILE_UNKNOWN.)

Would you mind changing ff_hw_ functions to take in HWBaseEncodePicture 
first, and AVCodecContext only if needed?
You have a lot of functions like hw_base_encode_add_ref that don't even 
use AVCodecContext, not even to get a context.

Also, HWBaseEncodePicture should be prefixed with FF, so 
FFHWBaseEncodePicture.
PICTURE_TYPE_* and FLAG_SLICE_* should also have an FF_HW_ prefix.

Instead of defining PICTURE_TYPE_I/P/B, would it be possible to use 
AV_PICTURE_TYPE_I/P/B with an additional `bool key;` or similar, which 
would remove hardcoding of MPEGese in the API.
That would allow differentiating intra-only frames from IDR (intra-only 
keyframes).

 > static inline const char *ff_hw_base_encode_get_pictype_name(const 
int type) {
Newline missing.

I'm working on integrating this into Vulkan right now, it seems suitable 
and saves me a lot of time, thanks for working on it.

[-- Attachment #1.1.1.2: OpenPGP public key --]
[-- Type: application/pgp-keys, Size: 637 bytes --]

[-- Attachment #1.2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 236 bytes --]

[-- Attachment #2: Type: text/plain, Size: 251 bytes --]

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

^ permalink raw reply	[flat|nested] 18+ messages in thread

* Re: [FFmpeg-devel] [PATCH v9 01/13] avcodec/vaapi_encode: introduce a base layer for vaapi encode
  2024-05-23 16:11 ` [FFmpeg-devel] [PATCH v9 01/13] avcodec/vaapi_encode: introduce a base layer for vaapi encode Lynne via ffmpeg-devel
@ 2024-05-24 15:39   ` Wu, Tong1
  2024-05-24 16:35     ` Lynne via ffmpeg-devel
  0 siblings, 1 reply; 18+ messages in thread
From: Wu, Tong1 @ 2024-05-24 15:39 UTC (permalink / raw)
  To: FFmpeg development discussions and patches; +Cc: Lynne

>-----Original Message-----
>From: ffmpeg-devel <ffmpeg-devel-bounces@ffmpeg.org> On Behalf Of Lynne
>via ffmpeg-devel
>Sent: Friday, May 24, 2024 12:11 AM
>To: ffmpeg-devel@ffmpeg.org
>Cc: Lynne <dev@lynne.ee>
>Subject: Re: [FFmpeg-devel] [PATCH v9 01/13] avcodec/vaapi_encode:
>introduce a base layer for vaapi encode
>
>On 20/05/2024 16:52, tong1.wu-at-intel.com@ffmpeg.org wrote:
>> From: Tong Wu <tong1.wu@intel.com>
>>
>> Since VAAPI and future D3D12VA implementation may share some common
>parameters,
>> a base layer encode context is introduced as vaapi context's base.
>>
>> Signed-off-by: Tong Wu <tong1.wu@intel.com>
>> ---
>>   libavcodec/hw_base_encode.h | 56
>+++++++++++++++++++++++++++++++++++++
>>   libavcodec/vaapi_encode.h   | 39 +++++---------------------
>>   2 files changed, 63 insertions(+), 32 deletions(-)
>>   create mode 100644 libavcodec/hw_base_encode.h
>>
>> diff --git a/libavcodec/hw_base_encode.h b/libavcodec/hw_base_encode.h
>> new file mode 100644
>> index 0000000000..1996179456
>> --- /dev/null
>> +++ b/libavcodec/hw_base_encode.h
>> @@ -0,0 +1,56 @@
>> +/*
>> + * This file is part of FFmpeg.
>> + *
>> + * FFmpeg is free software; you can redistribute it and/or
>> + * modify it under the terms of the GNU Lesser General Public
>> + * License as published by the Free Software Foundation; either
>> + * version 2.1 of the License, or (at your option) any later version.
>> + *
>> + * FFmpeg is distributed in the hope that it will be useful,
>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
>GNU
>> + * Lesser General Public License for more details.
>> + *
>> + * You should have received a copy of the GNU Lesser General Public
>> + * License along with FFmpeg; if not, write to the Free Software
>> + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301
>USA
>> + */
>> +
>> +#ifndef AVCODEC_HW_BASE_ENCODE_H
>> +#define AVCODEC_HW_BASE_ENCODE_H
>> +
>> +#define MAX_DPB_SIZE 16
>> +#define MAX_PICTURE_REFERENCES 2
>> +#define MAX_REORDER_DELAY 16
>> +#define MAX_ASYNC_DEPTH 64
>> +#define MAX_REFERENCE_LIST_NUM 2
>> +
>> +enum {
>> +    PICTURE_TYPE_IDR = 0,
>> +    PICTURE_TYPE_I   = 1,
>> +    PICTURE_TYPE_P   = 2,
>> +    PICTURE_TYPE_B   = 3,
>> +};
>> +
>> +enum {
>> +    // Codec supports controlling the subdivision of pictures into slices.
>> +    FLAG_SLICE_CONTROL         = 1 << 0,
>> +    // Codec only supports constant quality (no rate control).
>> +    FLAG_CONSTANT_QUALITY_ONLY = 1 << 1,
>> +    // Codec is intra-only.
>> +    FLAG_INTRA_ONLY            = 1 << 2,
>> +    // Codec supports B-pictures.
>> +    FLAG_B_PICTURES            = 1 << 3,
>> +    // Codec supports referencing B-pictures.
>> +    FLAG_B_PICTURE_REFERENCES  = 1 << 4,
>> +    // Codec supports non-IDR key pictures (that is, key pictures do
>> +    // not necessarily empty the DPB).
>> +    FLAG_NON_IDR_KEY_PICTURES  = 1 << 5,
>> +};
>> +
>> +typedef struct HWBaseEncodeContext {
>> +    const AVClass *class;
>> +} HWBaseEncodeContext;
>> +
>> +#endif /* AVCODEC_HW_BASE_ENCODE_H */
>> +
>> diff --git a/libavcodec/vaapi_encode.h b/libavcodec/vaapi_encode.h
>> index 0eed9691ca..f5c9be8973 100644
>> --- a/libavcodec/vaapi_encode.h
>> +++ b/libavcodec/vaapi_encode.h
>> @@ -33,34 +33,27 @@
>>
>>   #include "avcodec.h"
>>   #include "hwconfig.h"
>> +#include "hw_base_encode.h"
>>
>>   struct VAAPIEncodeType;
>>   struct VAAPIEncodePicture;
>>
>> +// Codec output packet without timestamp delay, which means the
>> +// output packet has same PTS and DTS.
>> +#define FLAG_TIMESTAMP_NO_DELAY 1 << 6
>> +
>>   enum {
>>       MAX_CONFIG_ATTRIBUTES  = 4,
>>       MAX_GLOBAL_PARAMS      = 4,
>> -    MAX_DPB_SIZE           = 16,
>> -    MAX_PICTURE_REFERENCES = 2,
>> -    MAX_REORDER_DELAY      = 16,
>>       MAX_PARAM_BUFFER_SIZE  = 1024,
>>       // A.4.1: table A.6 allows at most 22 tile rows for any level.
>>       MAX_TILE_ROWS          = 22,
>>       // A.4.1: table A.6 allows at most 20 tile columns for any level.
>>       MAX_TILE_COLS          = 20,
>> -    MAX_ASYNC_DEPTH        = 64,
>> -    MAX_REFERENCE_LIST_NUM = 2,
>>   };
>>
>>   extern const AVCodecHWConfigInternal *const
>ff_vaapi_encode_hw_configs[];
>>
>> -enum {
>> -    PICTURE_TYPE_IDR = 0,
>> -    PICTURE_TYPE_I   = 1,
>> -    PICTURE_TYPE_P   = 2,
>> -    PICTURE_TYPE_B   = 3,
>> -};
>> -
>>   typedef struct VAAPIEncodeSlice {
>>       int             index;
>>       int             row_start;
>> @@ -193,7 +186,8 @@ typedef struct VAAPIEncodeRCMode {
>>   } VAAPIEncodeRCMode;
>>
>>   typedef struct VAAPIEncodeContext {
>> -    const AVClass *class;
>> +    // Base context.
>> +    HWBaseEncodeContext base;
>>
>>       // Codec-specific hooks.
>>       const struct VAAPIEncodeType *codec;
>> @@ -397,25 +391,6 @@ typedef struct VAAPIEncodeContext {
>>       AVPacket        *tail_pkt;
>>   } VAAPIEncodeContext;
>>
>> -enum {
>> -    // Codec supports controlling the subdivision of pictures into slices.
>> -    FLAG_SLICE_CONTROL         = 1 << 0,
>> -    // Codec only supports constant quality (no rate control).
>> -    FLAG_CONSTANT_QUALITY_ONLY = 1 << 1,
>> -    // Codec is intra-only.
>> -    FLAG_INTRA_ONLY            = 1 << 2,
>> -    // Codec supports B-pictures.
>> -    FLAG_B_PICTURES            = 1 << 3,
>> -    // Codec supports referencing B-pictures.
>> -    FLAG_B_PICTURE_REFERENCES  = 1 << 4,
>> -    // Codec supports non-IDR key pictures (that is, key pictures do
>> -    // not necessarily empty the DPB).
>> -    FLAG_NON_IDR_KEY_PICTURES  = 1 << 5,
>> -    // Codec output packet without timestamp delay, which means the
>> -    // output packet has same PTS and DTS.
>> -    FLAG_TIMESTAMP_NO_DELAY    = 1 << 6,
>> -};
>> -
>>   typedef struct VAAPIEncodeType {
>>       // List of supported profiles and corresponding VAAPI profiles.
>>       // (Must end with AV_PROFILE_UNKNOWN.)
>
>Would you mind changing ff_hw_ functions to take in HWBaseEncodePicture
>first, and AVCodecContext only if needed?
>You have a lot of functions like hw_base_encode_add_ref that don't even
>use AVCodecContext, not even to get a context.
>

I've reviewed all the functions and will remove the unnecessary AVCodecContext in the the next version.

>Also, HWBaseEncodePicture should be prefixed with FF, so
>FFHWBaseEncodePicture.
>PICTURE_TYPE_* and FLAG_SLICE_* should also have an FF_HW_ prefix.
>

Sure I'll update in the next version.

>Instead of defining PICTURE_TYPE_I/P/B, would it be possible to use
>AV_PICTURE_TYPE_I/P/B with an additional `bool key;` or similar, which
>would remove hardcoding of MPEGese in the API.
>That would allow differentiating intra-only frames from IDR (intra-only
>keyframes).

Would you mind we sending a separate patch for this? Since the hardcoding has already existed for a long time and this patch set was intended to only moves the vaapi functions as-is. Changing it in this patch set only makes it even larger and harder to review. I think we should raise another thread to discuss this.


>
> > static inline const char *ff_hw_base_encode_get_pictype_name(const
>int type) {
>Newline missing.

Sorry don’t get it.

>
>I'm working on integrating this into Vulkan right now, it seems suitable
>and saves me a lot of time, thanks for working on it.

Very glad to hear that. Looking forward to getting it merged.

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

^ permalink raw reply	[flat|nested] 18+ messages in thread

* Re: [FFmpeg-devel] [PATCH v9 01/13] avcodec/vaapi_encode: introduce a base layer for vaapi encode
  2024-05-24 15:39   ` Wu, Tong1
@ 2024-05-24 16:35     ` Lynne via ffmpeg-devel
  0 siblings, 0 replies; 18+ messages in thread
From: Lynne via ffmpeg-devel @ 2024-05-24 16:35 UTC (permalink / raw)
  To: Wu, Tong1, FFmpeg development discussions and patches; +Cc: Lynne


[-- Attachment #1.1.1.1: Type: text/plain, Size: 8379 bytes --]

On 24/05/2024 17:39, Wu, Tong1 wrote:
>> -----Original Message-----
>> From: ffmpeg-devel <ffmpeg-devel-bounces@ffmpeg.org> On Behalf Of Lynne
>> via ffmpeg-devel
>> Sent: Friday, May 24, 2024 12:11 AM
>> To: ffmpeg-devel@ffmpeg.org
>> Cc: Lynne <dev@lynne.ee>
>> Subject: Re: [FFmpeg-devel] [PATCH v9 01/13] avcodec/vaapi_encode:
>> introduce a base layer for vaapi encode
>>
>> On 20/05/2024 16:52, tong1.wu-at-intel.com@ffmpeg.org wrote:
>>> From: Tong Wu <tong1.wu@intel.com>
>>>
>>> Since VAAPI and future D3D12VA implementation may share some common
>> parameters,
>>> a base layer encode context is introduced as vaapi context's base.
>>>
>>> Signed-off-by: Tong Wu <tong1.wu@intel.com>
>>> ---
>>>    libavcodec/hw_base_encode.h | 56
>> +++++++++++++++++++++++++++++++++++++
>>>    libavcodec/vaapi_encode.h   | 39 +++++---------------------
>>>    2 files changed, 63 insertions(+), 32 deletions(-)
>>>    create mode 100644 libavcodec/hw_base_encode.h
>>>
>>> diff --git a/libavcodec/hw_base_encode.h b/libavcodec/hw_base_encode.h
>>> new file mode 100644
>>> index 0000000000..1996179456
>>> --- /dev/null
>>> +++ b/libavcodec/hw_base_encode.h
>>> @@ -0,0 +1,56 @@
>>> +/*
>>> + * This file is part of FFmpeg.
>>> + *
>>> + * FFmpeg is free software; you can redistribute it and/or
>>> + * modify it under the terms of the GNU Lesser General Public
>>> + * License as published by the Free Software Foundation; either
>>> + * version 2.1 of the License, or (at your option) any later version.
>>> + *
>>> + * FFmpeg is distributed in the hope that it will be useful,
>>> + * but WITHOUT ANY WARRANTY; without even the implied warranty of
>>> + * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
>> GNU
>>> + * Lesser General Public License for more details.
>>> + *
>>> + * You should have received a copy of the GNU Lesser General Public
>>> + * License along with FFmpeg; if not, write to the Free Software
>>> + * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301
>> USA
>>> + */
>>> +
>>> +#ifndef AVCODEC_HW_BASE_ENCODE_H
>>> +#define AVCODEC_HW_BASE_ENCODE_H
>>> +
>>> +#define MAX_DPB_SIZE 16
>>> +#define MAX_PICTURE_REFERENCES 2
>>> +#define MAX_REORDER_DELAY 16
>>> +#define MAX_ASYNC_DEPTH 64
>>> +#define MAX_REFERENCE_LIST_NUM 2
>>> +
>>> +enum {
>>> +    PICTURE_TYPE_IDR = 0,
>>> +    PICTURE_TYPE_I   = 1,
>>> +    PICTURE_TYPE_P   = 2,
>>> +    PICTURE_TYPE_B   = 3,
>>> +};
>>> +
>>> +enum {
>>> +    // Codec supports controlling the subdivision of pictures into slices.
>>> +    FLAG_SLICE_CONTROL         = 1 << 0,
>>> +    // Codec only supports constant quality (no rate control).
>>> +    FLAG_CONSTANT_QUALITY_ONLY = 1 << 1,
>>> +    // Codec is intra-only.
>>> +    FLAG_INTRA_ONLY            = 1 << 2,
>>> +    // Codec supports B-pictures.
>>> +    FLAG_B_PICTURES            = 1 << 3,
>>> +    // Codec supports referencing B-pictures.
>>> +    FLAG_B_PICTURE_REFERENCES  = 1 << 4,
>>> +    // Codec supports non-IDR key pictures (that is, key pictures do
>>> +    // not necessarily empty the DPB).
>>> +    FLAG_NON_IDR_KEY_PICTURES  = 1 << 5,
>>> +};
>>> +
>>> +typedef struct HWBaseEncodeContext {
>>> +    const AVClass *class;
>>> +} HWBaseEncodeContext;
>>> +
>>> +#endif /* AVCODEC_HW_BASE_ENCODE_H */
>>> +
>>> diff --git a/libavcodec/vaapi_encode.h b/libavcodec/vaapi_encode.h
>>> index 0eed9691ca..f5c9be8973 100644
>>> --- a/libavcodec/vaapi_encode.h
>>> +++ b/libavcodec/vaapi_encode.h
>>> @@ -33,34 +33,27 @@
>>>
>>>    #include "avcodec.h"
>>>    #include "hwconfig.h"
>>> +#include "hw_base_encode.h"
>>>
>>>    struct VAAPIEncodeType;
>>>    struct VAAPIEncodePicture;
>>>
>>> +// Codec output packet without timestamp delay, which means the
>>> +// output packet has same PTS and DTS.
>>> +#define FLAG_TIMESTAMP_NO_DELAY 1 << 6
>>> +
>>>    enum {
>>>        MAX_CONFIG_ATTRIBUTES  = 4,
>>>        MAX_GLOBAL_PARAMS      = 4,
>>> -    MAX_DPB_SIZE           = 16,
>>> -    MAX_PICTURE_REFERENCES = 2,
>>> -    MAX_REORDER_DELAY      = 16,
>>>        MAX_PARAM_BUFFER_SIZE  = 1024,
>>>        // A.4.1: table A.6 allows at most 22 tile rows for any level.
>>>        MAX_TILE_ROWS          = 22,
>>>        // A.4.1: table A.6 allows at most 20 tile columns for any level.
>>>        MAX_TILE_COLS          = 20,
>>> -    MAX_ASYNC_DEPTH        = 64,
>>> -    MAX_REFERENCE_LIST_NUM = 2,
>>>    };
>>>
>>>    extern const AVCodecHWConfigInternal *const
>> ff_vaapi_encode_hw_configs[];
>>>
>>> -enum {
>>> -    PICTURE_TYPE_IDR = 0,
>>> -    PICTURE_TYPE_I   = 1,
>>> -    PICTURE_TYPE_P   = 2,
>>> -    PICTURE_TYPE_B   = 3,
>>> -};
>>> -
>>>    typedef struct VAAPIEncodeSlice {
>>>        int             index;
>>>        int             row_start;
>>> @@ -193,7 +186,8 @@ typedef struct VAAPIEncodeRCMode {
>>>    } VAAPIEncodeRCMode;
>>>
>>>    typedef struct VAAPIEncodeContext {
>>> -    const AVClass *class;
>>> +    // Base context.
>>> +    HWBaseEncodeContext base;
>>>
>>>        // Codec-specific hooks.
>>>        const struct VAAPIEncodeType *codec;
>>> @@ -397,25 +391,6 @@ typedef struct VAAPIEncodeContext {
>>>        AVPacket        *tail_pkt;
>>>    } VAAPIEncodeContext;
>>>
>>> -enum {
>>> -    // Codec supports controlling the subdivision of pictures into slices.
>>> -    FLAG_SLICE_CONTROL         = 1 << 0,
>>> -    // Codec only supports constant quality (no rate control).
>>> -    FLAG_CONSTANT_QUALITY_ONLY = 1 << 1,
>>> -    // Codec is intra-only.
>>> -    FLAG_INTRA_ONLY            = 1 << 2,
>>> -    // Codec supports B-pictures.
>>> -    FLAG_B_PICTURES            = 1 << 3,
>>> -    // Codec supports referencing B-pictures.
>>> -    FLAG_B_PICTURE_REFERENCES  = 1 << 4,
>>> -    // Codec supports non-IDR key pictures (that is, key pictures do
>>> -    // not necessarily empty the DPB).
>>> -    FLAG_NON_IDR_KEY_PICTURES  = 1 << 5,
>>> -    // Codec output packet without timestamp delay, which means the
>>> -    // output packet has same PTS and DTS.
>>> -    FLAG_TIMESTAMP_NO_DELAY    = 1 << 6,
>>> -};
>>> -
>>>    typedef struct VAAPIEncodeType {
>>>        // List of supported profiles and corresponding VAAPI profiles.
>>>        // (Must end with AV_PROFILE_UNKNOWN.)
>>
>> Would you mind changing ff_hw_ functions to take in HWBaseEncodePicture
>> first, and AVCodecContext only if needed?
>> You have a lot of functions like hw_base_encode_add_ref that don't even
>> use AVCodecContext, not even to get a context.
>>
> 
> I've reviewed all the functions and will remove the unnecessary AVCodecContext in the the next version.

Thanks. The idea is rather than making HWBaseEncodePicture the main 
encoder context, having the possibility that encoders just include 
HWBaseEncodePicture into their own context and use it standalone, rather
than loading this structure with all fields encoders need.

>> Also, HWBaseEncodePicture should be prefixed with FF, so
>> FFHWBaseEncodePicture.
>> PICTURE_TYPE_* and FLAG_SLICE_* should also have an FF_HW_ prefix.
>>
> 
> Sure I'll update in the next version.
> 
>> Instead of defining PICTURE_TYPE_I/P/B, would it be possible to use
>> AV_PICTURE_TYPE_I/P/B with an additional `bool key;` or similar, which
>> would remove hardcoding of MPEGese in the API.
>> That would allow differentiating intra-only frames from IDR (intra-only
>> keyframes).
> 
> Would you mind we sending a separate patch for this? Since the hardcoding has already existed for a long time and this patch set was intended to only moves the vaapi functions as-is. Changing it in this patch set only makes it even larger and harder to review. I think we should raise another thread to discuss this.

Sure, that can be done after this is merged.

>>> static inline const char *ff_hw_base_encode_get_pictype_name(const
>> int type) {
>> Newline missing.
> 
> Sorry don’t get it.

- ...base_encode_get_pictype_name(const int type) {
====
+ ...base_encode_get_pictype_name(const int type)
+ {

> 
>>
>> I'm working on integrating this into Vulkan right now, it seems suitable
>> and saves me a lot of time, thanks for working on it.
> 
> Very glad to hear that. Looking forward to getting it merged.
>

[-- Attachment #1.1.1.2: OpenPGP public key --]
[-- Type: application/pgp-keys, Size: 637 bytes --]

[-- Attachment #1.2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 236 bytes --]

[-- Attachment #2: Type: text/plain, Size: 251 bytes --]

_______________________________________________
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

To unsubscribe, visit link above, or email
ffmpeg-devel-request@ffmpeg.org with subject "unsubscribe".

^ permalink raw reply	[flat|nested] 18+ messages in thread

end of thread, other threads:[~2024-05-24 16:35 UTC | newest]

Thread overview: 18+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2024-05-20 14:52 [FFmpeg-devel] [PATCH v9 01/13] avcodec/vaapi_encode: introduce a base layer for vaapi encode tong1.wu-at-intel.com
2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 02/13] avcodec/vaapi_encode: add async_depth to common options tong1.wu-at-intel.com
2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 03/13] avcodec/vaapi_encode: add picture type name to base tong1.wu-at-intel.com
2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 04/13] avcodec/vaapi_encode: move pic->input_surface initialization to encode_alloc tong1.wu-at-intel.com
2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 05/13] avcodec/vaapi_encode: move the dpb logic from VAAPI to base layer tong1.wu-at-intel.com
2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 06/13] avcodec/vaapi_encode: extract the init and close function " tong1.wu-at-intel.com
2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 07/13] avcodec/vaapi_encode: extract gop configuration and two options " tong1.wu-at-intel.com
2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 08/13] avcodec/vaapi_encode: extract set_output_property " tong1.wu-at-intel.com
2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 09/13] avcodec/vaapi_encode: extract a get_recon_format function " tong1.wu-at-intel.com
2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 10/13] avcodec/vaapi_encode: extract a free funtion " tong1.wu-at-intel.com
2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 11/13] avutil/hwcontext_d3d12va: add Flags for resource creation tong1.wu-at-intel.com
2024-05-21 16:32   ` Andrew Sayers
2024-05-22  1:29     ` Wu, Tong1
2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 12/13] avcodec: add D3D12VA hardware HEVC encoder tong1.wu-at-intel.com
2024-05-20 14:52 ` [FFmpeg-devel] [PATCH v9 13/13] Changelog: add D3D12VA HEVC encoder changelog tong1.wu-at-intel.com
2024-05-23 16:11 ` [FFmpeg-devel] [PATCH v9 01/13] avcodec/vaapi_encode: introduce a base layer for vaapi encode Lynne via ffmpeg-devel
2024-05-24 15:39   ` Wu, Tong1
2024-05-24 16:35     ` Lynne via ffmpeg-devel

Git Inbox Mirror of the ffmpeg-devel mailing list - see https://ffmpeg.org/mailman/listinfo/ffmpeg-devel

This inbox may be cloned and mirrored by anyone:

	git clone --mirror https://master.gitmailbox.com/ffmpegdev/0 ffmpegdev/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 ffmpegdev ffmpegdev/ https://master.gitmailbox.com/ffmpegdev \
		ffmpegdev@gitmailbox.com
	public-inbox-index ffmpegdev

Example config snippet for mirrors.


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git