From: James Almer <jamrial@gmail.com> To: ffmpeg-devel@ffmpeg.org Subject: Re: [FFmpeg-devel] [PATCH 7/9] avcodec/nvenc: add MV-HEVC encoding support Date: Thu, 30 Jan 2025 17:17:55 -0300 Message-ID: <4284c116-7dcc-4e39-a31e-2309348f371f@gmail.com> (raw) In-Reply-To: <20250130194124.21836-8-timo@rothenpieler.org> [-- Attachment #1.1.1: Type: text/plain, Size: 6914 bytes --] On 1/30/2025 4:40 PM, Timo Rothenpieler wrote: > From: Diego de Souza <ddesouza@nvidia.com> > > Added support for MV-HEVC encoding for stereoscopic videos (2 views > only). Compatible with the framepack filter when using the > AV_STEREO3D_FRAMESEQUENCE format. > > Signed-off-by: Diego de Souza <ddesouza@nvidia.com> > --- > libavcodec/nvenc.c | 49 +++++++++++++++++++++++++++++++++++++++++ > libavcodec/nvenc.h | 4 ++++ > libavcodec/nvenc_hevc.c | 6 +++++ > 3 files changed, 59 insertions(+) > > diff --git a/libavcodec/nvenc.c b/libavcodec/nvenc.c > index 3403fa8996..a528f1ce93 100644 > --- a/libavcodec/nvenc.c > +++ b/libavcodec/nvenc.c > @@ -35,6 +35,7 @@ > #include "libavutil/mem.h" > #include "libavutil/pixdesc.h" > #include "libavutil/mathematics.h" > +#include "libavutil/stereo3d.h" > #include "atsc_a53.h" > #include "codec_desc.h" > #include "encode.h" > @@ -656,6 +657,14 @@ static int nvenc_check_capabilities(AVCodecContext *avctx) > > ctx->support_dyn_bitrate = nvenc_check_cap(avctx, NV_ENC_CAPS_SUPPORT_DYN_BITRATE_CHANGE); > > +#ifdef NVENC_HAVE_MVHEVC > + ret = nvenc_check_cap(avctx, NV_ENC_CAPS_SUPPORT_MVHEVC_ENCODE); > + if(ctx->multiview && ret <= 0) { > + av_log(avctx, AV_LOG_WARNING, "Multiview not supported by the device\n"); > + return AVERROR(ENOSYS); > + } > +#endif > + > return 0; > } > > @@ -1544,6 +1553,11 @@ static av_cold int nvenc_setup_hevc_config(AVCodecContext *avctx) > } > #endif > > +#ifdef NVENC_HAVE_MVHEVC > + hevc->enableMVHEVC = ctx->multiview; > + hevc->outputHevc3DReferenceDisplayInfo = ctx->display_sei; > +#endif > + > return 0; > } > > @@ -2469,6 +2483,9 @@ static int nvenc_set_timestamp(AVCodecContext *avctx, > > // This can be more than necessary, but we don't know the real reorder delay. > delay = FFMAX(ctx->encode_config.frameIntervalP - 1, 0); > +#ifdef NVENC_HAVE_MVHEVC > + delay *= ctx->multiview + 1; > +#endif > if (ctx->output_frame_num >= delay) { > pkt->dts = timestamp_queue_dequeue(ctx->timestamp_list); > ctx->output_frame_num++; > @@ -2875,6 +2892,9 @@ static int nvenc_send_frame(AVCodecContext *avctx, const AVFrame *frame) > int res, res2; > int sei_count = 0; > int i; > +#ifdef NVENC_HAVE_MVHEVC > + HEVC_3D_REFERENCE_DISPLAY_INFO ref_disp_info = { 0 }; > +#endif > > NvencContext *ctx = avctx->priv_data; > NvencDynLoadFunctions *dl_fn = &ctx->nvenc_dload_funcs; > @@ -2952,6 +2972,35 @@ static int nvenc_send_frame(AVCodecContext *avctx, const AVFrame *frame) > if (res < 0) > return res; > > +#ifdef NVENC_HAVE_MVHEVC > + if (ctx->multiview) > + { > + const AVFrameSideData *sd = av_frame_get_side_data(frame, AV_FRAME_DATA_STEREO3D); > + if (sd) > + { > + const AVStereo3D *stereo = (const AVStereo3D*)sd->data; > + if (stereo->type == AV_STEREO3D_FRAMESEQUENCE) { No brackets for single line blocks. > + ctx->next_view_id = (stereo->view == AV_STEREO3D_VIEW_LEFT) ? 0 : 1; > + } else { > + sd = NULL; > + av_log(avctx, AV_LOG_ERROR, "Stereo format %d not supported! Only AV_STEREO3D_FRAMESEQUENCE is supported!\n", stereo->type); > + } > + } > + > + pic_params.codecPicParams.hevcPicParams.viewId = ctx->next_view_id; > + if (!sd) > + ctx->next_view_id ^= 1; > + > + if (ctx->display_sei) > + { > + ref_disp_info.precRefDisplayWidth = 31; > + ref_disp_info.leftViewId[0] = 0; > + ref_disp_info.rightViewId[0] = 1; There's AV_FRAME_DATA_VIEW_ID, which the native hevc decoder uses to export the view ids when the 3D reference display information SEI message is present in the bitstream. It should be used here too. > + pic_params.codecPicParams.hevcPicParams.p3DReferenceDisplayInfo = &ref_disp_info; > + } > + } > +#endif > + > nv_status = p_nvenc->nvEncEncodePicture(ctx->nvencoder, &pic_params); > > for (i = 0; i < sei_count; i++) > diff --git a/libavcodec/nvenc.h b/libavcodec/nvenc.h > index 6f7f8d4e7f..77aef3f188 100644 > --- a/libavcodec/nvenc.h > +++ b/libavcodec/nvenc.h > @@ -99,6 +99,7 @@ typedef void ID3D11Device; > #define NVENC_HAVE_422_SUPPORT > #define NVENC_HAVE_AV1_UHQ_TUNING > #define NVENC_HAVE_H264_AND_AV1_TEMPORAL_FILTER > +#define NVENC_HAVE_MVHEVC > #endif > > typedef struct NvencSurface > @@ -245,6 +246,7 @@ typedef struct NvencContext > void *nvencoder; > > uint32_t frame_idx_counter; > + uint32_t next_view_id; > > int preset; > int profile; > @@ -291,6 +293,7 @@ typedef struct NvencContext > int single_slice_intra_refresh; > int constrained_encoding; > int udu_sei; > + int display_sei; > int timing_info; > int highbitdepth; > int max_slice_size; > @@ -299,6 +302,7 @@ typedef struct NvencContext > int lookahead_level; > int unidir_b; > int split_encode_mode; > + int multiview; > } NvencContext; > > int ff_nvenc_encode_init(AVCodecContext *avctx); > diff --git a/libavcodec/nvenc_hevc.c b/libavcodec/nvenc_hevc.c > index 5696e14dd4..f821eaee50 100644 > --- a/libavcodec/nvenc_hevc.c > +++ b/libavcodec/nvenc_hevc.c > @@ -194,6 +194,12 @@ static const AVOption options[] = { > OFFSET(extra_sei), AV_OPT_TYPE_BOOL, { .i64 = 1 }, 0, 1, VE }, > { "udu_sei", "Pass on user data unregistered SEI if available", > OFFSET(udu_sei), AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, VE }, > +#ifdef NVENC_HAVE_MVHEVC > + { "display_sei", "Pass on 3D reference display information SEI message for MV-HEVC", > + OFFSET(display_sei), AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, VE }, Shouldn't this SEI message always be present for stereoscopic MV-HEVC? > + { "mvhevc", "Set to 1 to enable MV-HEVC. This feature currently supports only 2 views", This should instead be a check for the AV_PROFILE_HEVC_MULTIVIEW_MAIN profile. > + OFFSET(multiview), AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, VE }, > +#endif > { "intra-refresh","Use Periodic Intra Refresh instead of IDR frames", > OFFSET(intra_refresh),AV_OPT_TYPE_BOOL, { .i64 = 0 }, 0, 1, VE }, > { "single-slice-intra-refresh", "Use single slice intra refresh", [-- Attachment #1.2: OpenPGP digital signature --] [-- Type: application/pgp-signature, Size: 495 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".
next prev parent reply other threads:[~2025-01-30 20:18 UTC|newest] Thread overview: 17+ messages / expand[flat|nested] mbox.gz Atom feed top 2025-01-30 19:40 [FFmpeg-devel] [PATCH 0/9] Nvidia Video Codec SDK 13.0 support Timo Rothenpieler 2025-01-30 19:40 ` [FFmpeg-devel] [PATCH 1/9] avutil/hwcontext_cuda: add 4:2:2 pixel format support Timo Rothenpieler 2025-01-30 19:40 ` [FFmpeg-devel] [PATCH 2/9] avcodec/nvdec: add 4:2:2 decoding and 10-bit support Timo Rothenpieler 2025-01-30 19:40 ` [FFmpeg-devel] [PATCH 3/9] avcodec/cuviddec: add HEVC/H.264 4:2:2 and H.264 " Timo Rothenpieler 2025-01-30 19:40 ` [FFmpeg-devel] [PATCH 4/9] avcodec/nvenc: add 4:2:2 encoding " Timo Rothenpieler 2025-01-30 19:40 ` [FFmpeg-devel] [PATCH 5/9] avcodec/nvenc: add UHQ to AV1 for NVENC Timo Rothenpieler 2025-01-30 19:40 ` [FFmpeg-devel] [PATCH 6/9] avcodec/nvenc: add Temporal Filtering for AV1 and H.264 in NVENC Timo Rothenpieler 2025-01-30 19:40 ` [FFmpeg-devel] [PATCH 7/9] avcodec/nvenc: add MV-HEVC encoding support Timo Rothenpieler 2025-01-30 20:17 ` James Almer [this message] 2025-01-30 23:23 ` [FFmpeg-devel] [PATCH v2 " Timo Rothenpieler 2025-01-31 3:26 ` James Almer 2025-01-30 19:40 ` [FFmpeg-devel] [PATCH 8/9] avcodec/nvenc: use encoder level options for qmin/qmax Timo Rothenpieler 2025-01-31 6:16 ` Zhao Zhili 2025-01-31 13:20 ` Timo Rothenpieler 2025-01-30 19:40 ` [FFmpeg-devel] [PATCH 9/9] avcodec/nvenc: finalize SDK 13.0 support Timo Rothenpieler 2025-02-01 21:15 ` [FFmpeg-devel] [PATCH 0/9] Nvidia Video Codec " Timo Rothenpieler 2025-02-02 19:05 ` Timo Rothenpieler
Reply instructions: You may reply publicly to this message via plain-text email using any one of the following methods: * Save the following mbox file, import it into your mail client, and reply-to-all from there: mbox Avoid top-posting and favor interleaved quoting: https://en.wikipedia.org/wiki/Posting_style#Interleaved_style * Reply using the --to, --cc, and --in-reply-to switches of git-send-email(1): git send-email \ --in-reply-to=4284c116-7dcc-4e39-a31e-2309348f371f@gmail.com \ --to=jamrial@gmail.com \ --cc=ffmpeg-devel@ffmpeg.org \ /path/to/YOUR_REPLY https://kernel.org/pub/software/scm/git/docs/git-send-email.html * If your mail client supports setting the In-Reply-To header via mailto: links, try the mailto: link
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