* [FFmpeg-devel] [PATCH v5 1/4] swscale/input: add rgbaf32 input support
2022-11-23 19:35 [FFmpeg-devel] [PATCH v5 0/4] swscale: rgbaf32 input/output support mindmark
@ 2022-11-23 19:35 ` mindmark
2022-11-23 19:35 ` [FFmpeg-devel] [PATCH v5 2/4] avfilter/vf_hflip: add support for packed rgb float formats mindmark
` (3 subsequent siblings)
4 siblings, 0 replies; 8+ messages in thread
From: mindmark @ 2022-11-23 19:35 UTC (permalink / raw)
To: ffmpeg-devel; +Cc: Mark Reid
From: Mark Reid <mindmark@gmail.com>
The input functions are currently matching the planar f32 functions.
They can be factorized removing multiple lrintf/av_clipf calls,
this will be addressed in a future patch.
---
libswscale/input.c | 122 +++++++++++++++++++++++++++++++++++++++++++++
libswscale/utils.c | 6 +++
2 files changed, 128 insertions(+)
diff --git a/libswscale/input.c b/libswscale/input.c
index d5676062a2..9c9eb31cde 100644
--- a/libswscale/input.c
+++ b/libswscale/input.c
@@ -1284,6 +1284,98 @@ static void rgbaf16##endian_name##ToA_c(uint8_t *_dst, const uint8_t *_src, cons
rgbaf16_funcs_endian(le, 0)
rgbaf16_funcs_endian(be, 1)
+#define rdpx(src) (is_be ? av_int2float(AV_RB32(&src)): av_int2float(AV_RL32(&src)))
+
+static av_always_inline void rgbaf32ToUV_endian(uint16_t *dstU, uint16_t *dstV, int is_be,
+ const float *src, int width,
+ int32_t *rgb2yuv, int comp)
+{
+ int32_t ru = rgb2yuv[RU_IDX], gu = rgb2yuv[GU_IDX], bu = rgb2yuv[BU_IDX];
+ int32_t rv = rgb2yuv[RV_IDX], gv = rgb2yuv[GV_IDX], bv = rgb2yuv[BV_IDX];
+ int i;
+ /*TODO: refactor these f32 conversions to only have one lrintf and av_clipf call*/
+ for (i = 0; i < width; i++) {
+ int r = lrintf(av_clipf(65535.0f * rdpx(src[i*comp+0]), 0.0f, 65535.0f));
+ int g = lrintf(av_clipf(65535.0f * rdpx(src[i*comp+1]), 0.0f, 65535.0f));
+ int b = lrintf(av_clipf(65535.0f * rdpx(src[i*comp+2]), 0.0f, 65535.0f));
+
+ dstU[i] = (ru*r + gu*g + bu*b + (0x10001<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;
+ dstV[i] = (rv*r + gv*g + bv*b + (0x10001<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;
+ }
+}
+
+static av_always_inline void rgbaf32ToY_endian(uint16_t *dst, const float *src, int is_be,
+ int width, int32_t *rgb2yuv, int comp)
+{
+ int32_t ry = rgb2yuv[RY_IDX], gy = rgb2yuv[GY_IDX], by = rgb2yuv[BY_IDX];
+ int i;
+ /*TODO: refactor these f32 conversions to only have one lrintf and av_clipf call*/
+ for (i = 0; i < width; i++) {
+ int r = lrintf(av_clipf(65535.0f * rdpx(src[i*comp+0]), 0.0f, 65535.0f));
+ int g = lrintf(av_clipf(65535.0f * rdpx(src[i*comp+1]), 0.0f, 65535.0f));
+ int b = lrintf(av_clipf(65535.0f * rdpx(src[i*comp+2]), 0.0f, 65535.0f));
+
+ dst[i] = (ry*r + gy*g + by*b + (0x2001<<(RGB2YUV_SHIFT-1))) >> RGB2YUV_SHIFT;
+ }
+}
+
+static av_always_inline void rgbaf32ToA_endian(uint16_t *dst, const float *src, int is_be,
+ int width, void *opq)
+{
+ int i;
+ for (i=0; i<width; i++) {
+ dst[i] = lrintf(av_clipf(65535.0f * rdpx(src[i*4+3]), 0.0f, 65535.0f));
+ }
+}
+
+#undef rdpx
+
+#define rgbaf32_funcs_endian(endian_name, endian) \
+static void rgbf32##endian_name##ToUV_c(uint8_t *_dstU, uint8_t *_dstV, const uint8_t *unused, \
+ const uint8_t *src1, const uint8_t *src2, \
+ int width, uint32_t *rgb2yuv, void *opq) \
+{ \
+ const float *src = (const float*)src1; \
+ uint16_t *dstU = (uint16_t*)_dstU; \
+ uint16_t *dstV = (uint16_t*)_dstV; \
+ av_assert1(src1==src2); \
+ rgbaf32ToUV_endian(dstU, dstV, endian, src, width, rgb2yuv, 3); \
+} \
+static void rgbf32##endian_name##ToY_c(uint8_t *_dst, const uint8_t *_src, const uint8_t *unused0, \
+ const uint8_t *unused1, int width, uint32_t *rgb2yuv, void *opq) \
+{ \
+ const float *src = (const float*)_src; \
+ uint16_t *dst = (uint16_t*)_dst; \
+ rgbaf32ToY_endian(dst, src, endian, width, rgb2yuv, 3); \
+} \
+static void rgbaf32##endian_name##ToUV_c(uint8_t *_dstU, uint8_t *_dstV, const uint8_t *unused, \
+ const uint8_t *src1, const uint8_t *src2, \
+ int width, uint32_t *rgb2yuv, void *opq) \
+{ \
+ const float *src = (const float*)src1; \
+ uint16_t *dstU = (uint16_t*)_dstU; \
+ uint16_t *dstV = (uint16_t*)_dstV; \
+ av_assert1(src1==src2); \
+ rgbaf32ToUV_endian(dstU, dstV, endian, src, width, rgb2yuv, 4); \
+} \
+static void rgbaf32##endian_name##ToY_c(uint8_t *_dst, const uint8_t *_src, const uint8_t *unused0, \
+ const uint8_t *unused1, int width, uint32_t *rgb2yuv, void *opq) \
+{ \
+ const float *src = (const float*)_src; \
+ uint16_t *dst = (uint16_t*)_dst; \
+ rgbaf32ToY_endian(dst, src, endian, width, rgb2yuv, 4); \
+} \
+static void rgbaf32##endian_name##ToA_c(uint8_t *_dst, const uint8_t *_src, const uint8_t *unused0, \
+ const uint8_t *unused1, int width, uint32_t *unused2, void *opq) \
+{ \
+ const float *src = (const float*)_src; \
+ uint16_t *dst = (uint16_t*)_dst; \
+ rgbaf32ToA_endian(dst, src, endian, width, opq); \
+}
+
+rgbaf32_funcs_endian(le, 0)
+rgbaf32_funcs_endian(be, 1)
+
av_cold void ff_sws_init_input_funcs(SwsContext *c)
{
enum AVPixelFormat srcFormat = c->srcFormat;
@@ -1663,6 +1755,18 @@ av_cold void ff_sws_init_input_funcs(SwsContext *c)
case AV_PIX_FMT_RGBAF16LE:
c->chrToYV12 = rgbaf16leToUV_c;
break;
+ case AV_PIX_FMT_RGBF32BE:
+ c->chrToYV12 = rgbf32beToUV_c;
+ break;
+ case AV_PIX_FMT_RGBAF32BE:
+ c->chrToYV12 = rgbaf32beToUV_c;
+ break;
+ case AV_PIX_FMT_RGBF32LE:
+ c->chrToYV12 = rgbf32leToUV_c;
+ break;
+ case AV_PIX_FMT_RGBAF32LE:
+ c->chrToYV12 = rgbaf32leToUV_c;
+ break;
}
}
@@ -1973,6 +2077,18 @@ av_cold void ff_sws_init_input_funcs(SwsContext *c)
case AV_PIX_FMT_RGBAF16LE:
c->lumToYV12 = rgbaf16leToY_c;
break;
+ case AV_PIX_FMT_RGBF32BE:
+ c->lumToYV12 = rgbf32beToY_c;
+ break;
+ case AV_PIX_FMT_RGBAF32BE:
+ c->lumToYV12 = rgbaf32beToY_c;
+ break;
+ case AV_PIX_FMT_RGBF32LE:
+ c->lumToYV12 = rgbf32leToY_c;
+ break;
+ case AV_PIX_FMT_RGBAF32LE:
+ c->lumToYV12 = rgbaf32leToY_c;
+ break;
}
if (c->needAlpha) {
if (is16BPS(srcFormat) || isNBPS(srcFormat)) {
@@ -1998,6 +2114,12 @@ av_cold void ff_sws_init_input_funcs(SwsContext *c)
case AV_PIX_FMT_RGBAF16LE:
c->alpToYV12 = rgbaf16leToA_c;
break;
+ case AV_PIX_FMT_RGBAF32BE:
+ c->alpToYV12 = rgbaf32beToA_c;
+ break;
+ case AV_PIX_FMT_RGBAF32LE:
+ c->alpToYV12 = rgbaf32leToA_c;
+ break;
case AV_PIX_FMT_YA8:
c->alpToYV12 = uyvyToY_c;
break;
diff --git a/libswscale/utils.c b/libswscale/utils.c
index 85640a143f..2c520f68d1 100644
--- a/libswscale/utils.c
+++ b/libswscale/utils.c
@@ -266,6 +266,10 @@ static const FormatEntry format_entries[] = {
[AV_PIX_FMT_VUYX] = { 1, 1 },
[AV_PIX_FMT_RGBAF16BE] = { 1, 0 },
[AV_PIX_FMT_RGBAF16LE] = { 1, 0 },
+ [AV_PIX_FMT_RGBF32BE] = { 1, 0 },
+ [AV_PIX_FMT_RGBF32LE] = { 1, 0 },
+ [AV_PIX_FMT_RGBAF32BE] = { 1, 0 },
+ [AV_PIX_FMT_RGBAF32LE] = { 1, 0 },
[AV_PIX_FMT_XV30LE] = { 1, 1 },
[AV_PIX_FMT_XV36LE] = { 1, 1 },
};
@@ -1572,6 +1576,8 @@ av_cold int sws_init_context(SwsContext *c, SwsFilter *srcFilter,
srcFormat != AV_PIX_FMT_GBRAP16BE && srcFormat != AV_PIX_FMT_GBRAP16LE &&
srcFormat != AV_PIX_FMT_GBRPF32BE && srcFormat != AV_PIX_FMT_GBRPF32LE &&
srcFormat != AV_PIX_FMT_GBRAPF32BE && srcFormat != AV_PIX_FMT_GBRAPF32LE &&
+ srcFormat != AV_PIX_FMT_RGBF32BE && srcFormat != AV_PIX_FMT_RGBF32LE &&
+ srcFormat != AV_PIX_FMT_RGBAF32BE && srcFormat != AV_PIX_FMT_RGBAF32LE &&
((dstW >> c->chrDstHSubSample) <= (srcW >> 1) ||
(flags & SWS_FAST_BILINEAR)))
c->chrSrcHSubSample = 1;
--
2.31.1.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] 8+ messages in thread* [FFmpeg-devel] [PATCH v5 2/4] avfilter/vf_hflip: add support for packed rgb float formats
2022-11-23 19:35 [FFmpeg-devel] [PATCH v5 0/4] swscale: rgbaf32 input/output support mindmark
2022-11-23 19:35 ` [FFmpeg-devel] [PATCH v5 1/4] swscale/input: add rgbaf32 input support mindmark
@ 2022-11-23 19:35 ` mindmark
2022-11-23 19:35 ` [FFmpeg-devel] [PATCH v5 3/4] avfilter/vf_transpose: " mindmark
` (2 subsequent siblings)
4 siblings, 0 replies; 8+ messages in thread
From: mindmark @ 2022-11-23 19:35 UTC (permalink / raw)
To: ffmpeg-devel; +Cc: Mark Reid
From: Mark Reid <mindmark@gmail.com>
---
libavfilter/vf_hflip_init.h | 25 +++++++++++++++++++++++++
1 file changed, 25 insertions(+)
diff --git a/libavfilter/vf_hflip_init.h b/libavfilter/vf_hflip_init.h
index d0319f463d..31173f73fc 100644
--- a/libavfilter/vf_hflip_init.h
+++ b/libavfilter/vf_hflip_init.h
@@ -86,6 +86,29 @@ static void hflip_qword_c(const uint8_t *ssrc, uint8_t *ddst, int w)
dst[j] = src[-j];
}
+static void hflip_b96_c(const uint8_t *ssrc, uint8_t *ddst, int w)
+{
+ const uint32_t *in = (const uint32_t *)ssrc;
+ uint32_t *out = (uint32_t *)ddst;
+
+ for (int j = 0; j < w; j++, out += 3, in -= 3) {
+ out[0] = in[0];
+ out[1] = in[1];
+ out[2] = in[2];
+ }
+}
+
+static void hflip_b128_c(const uint8_t *ssrc, uint8_t *ddst, int w)
+{
+ const uint64_t *in = (const uint64_t *)ssrc;
+ uint64_t *out = (uint64_t *)ddst;
+
+ for (int j = 0; j < w; j++, out += 2, in -= 2) {
+ out[0] = in[0];
+ out[1] = in[1];
+ }
+}
+
static av_unused int ff_hflip_init(FlipContext *s, int step[4], int nb_planes)
{
for (int i = 0; i < nb_planes; i++) {
@@ -97,6 +120,8 @@ static av_unused int ff_hflip_init(FlipContext *s, int step[4], int nb_planes)
case 4: s->flip_line[i] = hflip_dword_c; break;
case 6: s->flip_line[i] = hflip_b48_c; break;
case 8: s->flip_line[i] = hflip_qword_c; break;
+ case 12: s->flip_line[i] = hflip_b96_c; break;
+ case 16: s->flip_line[i] = hflip_b128_c; break;
default:
return AVERROR_BUG;
}
--
2.31.1.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] 8+ messages in thread* [FFmpeg-devel] [PATCH v5 3/4] avfilter/vf_transpose: add support for packed rgb float formats
2022-11-23 19:35 [FFmpeg-devel] [PATCH v5 0/4] swscale: rgbaf32 input/output support mindmark
2022-11-23 19:35 ` [FFmpeg-devel] [PATCH v5 1/4] swscale/input: add rgbaf32 input support mindmark
2022-11-23 19:35 ` [FFmpeg-devel] [PATCH v5 2/4] avfilter/vf_hflip: add support for packed rgb float formats mindmark
@ 2022-11-23 19:35 ` mindmark
2022-11-23 19:35 ` [FFmpeg-devel] [PATCH v5 4/4] swscale/output: add rgbaf32 output support mindmark
2022-12-04 21:48 ` [FFmpeg-devel] [PATCH v5 0/4] swscale: rgbaf32 input/output support Mark Reid
4 siblings, 0 replies; 8+ messages in thread
From: mindmark @ 2022-11-23 19:35 UTC (permalink / raw)
To: ffmpeg-devel; +Cc: Mark Reid
From: Mark Reid <mindmark@gmail.com>
---
libavfilter/vf_transpose.c | 44 ++++++++++++++++++++++++++++++++++++++
1 file changed, 44 insertions(+)
diff --git a/libavfilter/vf_transpose.c b/libavfilter/vf_transpose.c
index 469e66729f..1023d6fe82 100644
--- a/libavfilter/vf_transpose.c
+++ b/libavfilter/vf_transpose.c
@@ -174,6 +174,46 @@ static void transpose_8x8_64_c(uint8_t *src, ptrdiff_t src_linesize,
transpose_block_64_c(src, src_linesize, dst, dst_linesize, 8, 8);
}
+static inline void transpose_block_96_c(uint8_t *src, ptrdiff_t src_linesize,
+ uint8_t *dst, ptrdiff_t dst_linesize,
+ int w, int h)
+{
+ int x, y;
+ for (y = 0; y < h; y++, dst += dst_linesize, src += 12) {
+ for (x = 0; x < w; x++) {
+ *((uint32_t *)(dst+0 + 12*x)) = *((uint32_t *)(src+0 + x*src_linesize));
+ *((uint32_t *)(dst+4 + 12*x)) = *((uint32_t *)(src+4 + x*src_linesize));
+ *((uint32_t *)(dst+8 + 12*x)) = *((uint32_t *)(src+8 + x*src_linesize));
+ }
+ }
+}
+
+static void transpose_8x8_96_c(uint8_t *src, ptrdiff_t src_linesize,
+ uint8_t *dst, ptrdiff_t dst_linesize)
+{
+ transpose_block_96_c(src, src_linesize, dst, dst_linesize, 8, 8);
+}
+
+
+static inline void transpose_block_128_c(uint8_t *src, ptrdiff_t src_linesize,
+ uint8_t *dst, ptrdiff_t dst_linesize,
+ int w, int h)
+{
+ int x, y;
+ for (y = 0; y < h; y++, dst += dst_linesize, src += 16) {
+ for (x = 0; x < w; x++) {
+ *((uint64_t *)(dst+0 + 16*x)) = *((uint64_t *)(src+0 + x*src_linesize));
+ *((uint64_t *)(dst+8 + 16*x)) = *((uint64_t *)(src+8 + x*src_linesize));
+ }
+ }
+}
+
+static void transpose_8x8_128_c(uint8_t *src, ptrdiff_t src_linesize,
+ uint8_t *dst, ptrdiff_t dst_linesize)
+{
+ transpose_block_128_c(src, src_linesize, dst, dst_linesize, 8, 8);
+}
+
static int config_props_output(AVFilterLink *outlink)
{
AVFilterContext *ctx = outlink->src;
@@ -232,6 +272,10 @@ static int config_props_output(AVFilterLink *outlink)
v->transpose_8x8 = transpose_8x8_48_c; break;
case 8: v->transpose_block = transpose_block_64_c;
v->transpose_8x8 = transpose_8x8_64_c; break;
+ case 12: v->transpose_block = transpose_block_96_c;
+ v->transpose_8x8 = transpose_8x8_96_c; break;
+ case 16: v->transpose_block = transpose_block_128_c;
+ v->transpose_8x8 = transpose_8x8_128_c; break;
}
}
--
2.31.1.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] 8+ messages in thread* [FFmpeg-devel] [PATCH v5 4/4] swscale/output: add rgbaf32 output support
2022-11-23 19:35 [FFmpeg-devel] [PATCH v5 0/4] swscale: rgbaf32 input/output support mindmark
` (2 preceding siblings ...)
2022-11-23 19:35 ` [FFmpeg-devel] [PATCH v5 3/4] avfilter/vf_transpose: " mindmark
@ 2022-11-23 19:35 ` mindmark
2022-12-05 0:05 ` Michael Niedermayer
2022-12-04 21:48 ` [FFmpeg-devel] [PATCH v5 0/4] swscale: rgbaf32 input/output support Mark Reid
4 siblings, 1 reply; 8+ messages in thread
From: mindmark @ 2022-11-23 19:35 UTC (permalink / raw)
To: ffmpeg-devel; +Cc: Mark Reid
From: Mark Reid <mindmark@gmail.com>
---
libswscale/output.c | 92 ++++++++++++++++++++++++
libswscale/swscale_unscaled.c | 4 +-
libswscale/tests/floatimg_cmp.c | 4 +-
libswscale/utils.c | 16 +++--
libswscale/yuv2rgb.c | 2 +
tests/ref/fate/filter-pixdesc-rgbaf32be | 1 +
tests/ref/fate/filter-pixdesc-rgbaf32le | 1 +
tests/ref/fate/filter-pixdesc-rgbf32be | 1 +
tests/ref/fate/filter-pixdesc-rgbf32le | 1 +
tests/ref/fate/filter-pixfmts-copy | 4 ++
tests/ref/fate/filter-pixfmts-crop | 4 ++
tests/ref/fate/filter-pixfmts-field | 4 ++
tests/ref/fate/filter-pixfmts-fieldorder | 4 ++
tests/ref/fate/filter-pixfmts-hflip | 4 ++
tests/ref/fate/filter-pixfmts-il | 4 ++
tests/ref/fate/filter-pixfmts-null | 4 ++
tests/ref/fate/filter-pixfmts-scale | 4 ++
tests/ref/fate/filter-pixfmts-transpose | 4 ++
tests/ref/fate/filter-pixfmts-vflip | 4 ++
tests/ref/fate/sws-floatimg-cmp | 16 +++++
20 files changed, 170 insertions(+), 8 deletions(-)
create mode 100644 tests/ref/fate/filter-pixdesc-rgbaf32be
create mode 100644 tests/ref/fate/filter-pixdesc-rgbaf32le
create mode 100644 tests/ref/fate/filter-pixdesc-rgbf32be
create mode 100644 tests/ref/fate/filter-pixdesc-rgbf32le
diff --git a/libswscale/output.c b/libswscale/output.c
index 5c85bff971..1d86a244f9 100644
--- a/libswscale/output.c
+++ b/libswscale/output.c
@@ -2471,6 +2471,92 @@ yuv2gbrpf32_full_X_c(SwsContext *c, const int16_t *lumFilter,
}
}
+static void
+yuv2rgbaf32_full_X_c(SwsContext *c, const int16_t *lumFilter,
+ const int16_t **lumSrcx, int lumFilterSize,
+ const int16_t *chrFilter, const int16_t **chrUSrcx,
+ const int16_t **chrVSrcx, int chrFilterSize,
+ const int16_t **alpSrcx, uint8_t *dest,
+ int dstW, int y)
+{
+ const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(c->dstFormat);
+ int i;
+ int alpha = desc->flags & AV_PIX_FMT_FLAG_ALPHA;
+ int hasAlpha = alpha && alpSrcx;
+ int pixelStep = alpha ? 4 : 3;
+ uint32_t *dest32 = (uint32_t*)dest;
+ const int32_t **lumSrc = (const int32_t**)lumSrcx;
+ const int32_t **chrUSrc = (const int32_t**)chrUSrcx;
+ const int32_t **chrVSrc = (const int32_t**)chrVSrcx;
+ const int32_t **alpSrc = (const int32_t**)alpSrcx;
+ static const float float_mult = 1.0f / 65535.0f;
+ uint32_t a = av_float2int(1.0f);
+
+ for (i = 0; i < dstW; i++) {
+ int j;
+ int Y = -0x40000000;
+ int U = -(128 << 23);
+ int V = -(128 << 23);
+ int R, G, B, A;
+
+ for (j = 0; j < lumFilterSize; j++)
+ Y += lumSrc[j][i] * (unsigned)lumFilter[j];
+
+ for (j = 0; j < chrFilterSize; j++) {
+ U += chrUSrc[j][i] * (unsigned)chrFilter[j];
+ V += chrVSrc[j][i] * (unsigned)chrFilter[j];
+ }
+
+ Y >>= 14;
+ Y += 0x10000;
+ U >>= 14;
+ V >>= 14;
+
+ if (hasAlpha) {
+ A = -0x40000000;
+
+ for (j = 0; j < lumFilterSize; j++)
+ A += alpSrc[j][i] * (unsigned)lumFilter[j];
+
+ A >>= 1;
+ A += 0x20002000;
+ a = av_float2int(float_mult * (float)(av_clip_uintp2(A, 30) >> 14));
+ }
+
+ Y -= c->yuv2rgb_y_offset;
+ Y *= c->yuv2rgb_y_coeff;
+ Y += (1 << 13) - (1 << 29);
+ R = V * c->yuv2rgb_v2r_coeff;
+ G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
+ B = U * c->yuv2rgb_u2b_coeff;
+
+ R = av_clip_uintp2(((Y + R) >> 14) + (1<<15), 16);
+ G = av_clip_uintp2(((Y + G) >> 14) + (1<<15), 16);
+ B = av_clip_uintp2(((Y + B) >> 14) + (1<<15), 16);
+
+ dest32[0] = av_float2int(float_mult * (float)R);
+ dest32[1] = av_float2int(float_mult * (float)G);
+ dest32[2] = av_float2int(float_mult * (float)B);
+ if (alpha)
+ dest32[3] = a;
+
+ dest32 += pixelStep;
+ }
+ if ((!isBE(c->dstFormat)) != (!HAVE_BIGENDIAN)) {
+ dest32 = (uint32_t*)dest;
+ for (i = 0; i < dstW; i++) {
+ dest32[0] = av_bswap32(dest32[0]);
+ dest32[1] = av_bswap32(dest32[1]);
+ dest32[2] = av_bswap32(dest32[2]);
+ if (alpha)
+ dest32[3] = av_bswap32(dest32[3]);
+
+ dest32 += pixelStep;
+ }
+ }
+
+}
+
static void
yuv2ya8_1_c(SwsContext *c, const int16_t *buf0,
const int16_t *ubuf[2], const int16_t *vbuf[2],
@@ -2983,6 +3069,12 @@ av_cold void ff_sws_init_output_funcs(SwsContext *c,
}
break;
+ case AV_PIX_FMT_RGBF32LE:
+ case AV_PIX_FMT_RGBF32BE:
+ case AV_PIX_FMT_RGBAF32LE:
+ case AV_PIX_FMT_RGBAF32BE:
+ *yuv2packedX = yuv2rgbaf32_full_X_c;
+ break;
case AV_PIX_FMT_RGB24:
*yuv2packedX = yuv2rgb24_full_X_c;
*yuv2packed2 = yuv2rgb24_full_2_c;
diff --git a/libswscale/swscale_unscaled.c b/libswscale/swscale_unscaled.c
index 9af2e7ecc3..5a73cfa541 100644
--- a/libswscale/swscale_unscaled.c
+++ b/libswscale/swscale_unscaled.c
@@ -2160,7 +2160,9 @@ void ff_get_unscaled_swscale(SwsContext *c)
/* bswap 32 bits per pixel/component formats */
if (IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_GBRPF32) ||
- IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_GBRAPF32))
+ IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_GBRAPF32) ||
+ IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_RGBF32) ||
+ IS_DIFFERENT_ENDIANESS(srcFormat, dstFormat, AV_PIX_FMT_RGBAF32))
c->convert_unscaled = bswap_32bpc;
if (usePal(srcFormat) && isByteRGB(dstFormat))
diff --git a/libswscale/tests/floatimg_cmp.c b/libswscale/tests/floatimg_cmp.c
index 5c67594fb6..9559c93aac 100644
--- a/libswscale/tests/floatimg_cmp.c
+++ b/libswscale/tests/floatimg_cmp.c
@@ -54,7 +54,9 @@ static const enum AVPixelFormat pix_fmts[] = {
AV_PIX_FMT_GBRP10LE, AV_PIX_FMT_GBRAP10LE,
AV_PIX_FMT_GBRP12LE, AV_PIX_FMT_GBRAP12LE,
AV_PIX_FMT_GBRP14LE,
- AV_PIX_FMT_GBRP16LE, AV_PIX_FMT_GBRAP16LE
+ AV_PIX_FMT_GBRP16LE, AV_PIX_FMT_GBRAP16LE,
+ AV_PIX_FMT_RGBF32LE, AV_PIX_FMT_RGBAF32LE,
+ AV_PIX_FMT_RGBF32BE, AV_PIX_FMT_RGBAF32BE
};
const char *usage = "floatimg_cmp -pixel_format <pix_fmt> -size <image_size> -ref <testfile>\n";
diff --git a/libswscale/utils.c b/libswscale/utils.c
index 2c520f68d1..642f765550 100644
--- a/libswscale/utils.c
+++ b/libswscale/utils.c
@@ -266,10 +266,10 @@ static const FormatEntry format_entries[] = {
[AV_PIX_FMT_VUYX] = { 1, 1 },
[AV_PIX_FMT_RGBAF16BE] = { 1, 0 },
[AV_PIX_FMT_RGBAF16LE] = { 1, 0 },
- [AV_PIX_FMT_RGBF32BE] = { 1, 0 },
- [AV_PIX_FMT_RGBF32LE] = { 1, 0 },
- [AV_PIX_FMT_RGBAF32BE] = { 1, 0 },
- [AV_PIX_FMT_RGBAF32LE] = { 1, 0 },
+ [AV_PIX_FMT_RGBF32BE] = { 1, 1 },
+ [AV_PIX_FMT_RGBF32LE] = { 1, 1 },
+ [AV_PIX_FMT_RGBAF32BE] = { 1, 1 },
+ [AV_PIX_FMT_RGBAF32LE] = { 1, 1 },
[AV_PIX_FMT_XV30LE] = { 1, 1 },
[AV_PIX_FMT_XV36LE] = { 1, 1 },
};
@@ -1512,7 +1512,7 @@ av_cold int sws_init_context(SwsContext *c, SwsFilter *srcFilter,
}
}
}
- if (isPlanarRGB(dstFormat)) {
+ if (isPlanarRGB(dstFormat) || isFloat(dstFormat)) {
if (!(flags & SWS_FULL_CHR_H_INT)) {
av_log(c, AV_LOG_DEBUG,
"%s output is not supported with half chroma resolution, switching to full\n",
@@ -1544,7 +1544,11 @@ av_cold int sws_init_context(SwsContext *c, SwsFilter *srcFilter,
dstFormat != AV_PIX_FMT_BGR4_BYTE &&
dstFormat != AV_PIX_FMT_RGB4_BYTE &&
dstFormat != AV_PIX_FMT_BGR8 &&
- dstFormat != AV_PIX_FMT_RGB8
+ dstFormat != AV_PIX_FMT_RGB8 &&
+ dstFormat != AV_PIX_FMT_RGBF32LE &&
+ dstFormat != AV_PIX_FMT_RGBF32BE &&
+ dstFormat != AV_PIX_FMT_RGBAF32LE &&
+ dstFormat != AV_PIX_FMT_RGBAF32BE
) {
av_log(c, AV_LOG_WARNING,
"full chroma interpolation for destination format '%s' not yet implemented\n",
diff --git a/libswscale/yuv2rgb.c b/libswscale/yuv2rgb.c
index 9c3f5e23c6..7ad9d2c9dc 100644
--- a/libswscale/yuv2rgb.c
+++ b/libswscale/yuv2rgb.c
@@ -999,6 +999,8 @@ av_cold int ff_yuv2rgb_c_init_tables(SwsContext *c, const int inv_table[4],
break;
case 32:
case 64:
+ case 96:
+ case 128:
base = (c->dstFormat == AV_PIX_FMT_RGB32_1 ||
c->dstFormat == AV_PIX_FMT_BGR32_1) ? 8 : 0;
rbase = base + (isRgb ? 16 : 0);
diff --git a/tests/ref/fate/filter-pixdesc-rgbaf32be b/tests/ref/fate/filter-pixdesc-rgbaf32be
new file mode 100644
index 0000000000..def2537716
--- /dev/null
+++ b/tests/ref/fate/filter-pixdesc-rgbaf32be
@@ -0,0 +1 @@
+pixdesc-rgbaf32be 8c618ffd38084857013b0f50d2cee0dd
diff --git a/tests/ref/fate/filter-pixdesc-rgbaf32le b/tests/ref/fate/filter-pixdesc-rgbaf32le
new file mode 100644
index 0000000000..01913ecaa7
--- /dev/null
+++ b/tests/ref/fate/filter-pixdesc-rgbaf32le
@@ -0,0 +1 @@
+pixdesc-rgbaf32le bb5239a00a3ec2b1f1d569733a1e03a2
diff --git a/tests/ref/fate/filter-pixdesc-rgbf32be b/tests/ref/fate/filter-pixdesc-rgbf32be
new file mode 100644
index 0000000000..77e40f20b9
--- /dev/null
+++ b/tests/ref/fate/filter-pixdesc-rgbf32be
@@ -0,0 +1 @@
+pixdesc-rgbf32be ca6c4115dc368a192a67d06bd47f369f
diff --git a/tests/ref/fate/filter-pixdesc-rgbf32le b/tests/ref/fate/filter-pixdesc-rgbf32le
new file mode 100644
index 0000000000..c0cb4d60f0
--- /dev/null
+++ b/tests/ref/fate/filter-pixdesc-rgbf32le
@@ -0,0 +1 @@
+pixdesc-rgbf32le 290153561ddc3266b253fa075e040578
diff --git a/tests/ref/fate/filter-pixfmts-copy b/tests/ref/fate/filter-pixfmts-copy
index b28a114c7b..9aa653d12d 100644
--- a/tests/ref/fate/filter-pixfmts-copy
+++ b/tests/ref/fate/filter-pixfmts-copy
@@ -90,6 +90,10 @@ rgb8 7ac6008c84d622c2fc50581706e17576
rgba b6e1b441c365e03b5ffdf9b7b68d9a0c
rgba64be ae2ae04b5efedca3505f47c4dd6ea6ea
rgba64le b91e1d77f799eb92241a2d2d28437b15
+rgbaf32be abb244bb1af2247bbba7b4eac0357f6b
+rgbaf32le 99b2afe809649e58eea0a188fd6ab3f2
+rgbf32be e4a0b47e8ecb2e7e461915227ebd4edd
+rgbf32le 2979d6bfe509a55486e55e4de63d67d1
uyvy422 3bcf3c80047592f2211fae3260b1b65d
vuya 3d5e934651cae1ce334001cb1829ad22
vuyx 3f68ea6ec492b30d867cb5401562264e
diff --git a/tests/ref/fate/filter-pixfmts-crop b/tests/ref/fate/filter-pixfmts-crop
index bdb2536f7d..9bb54cf9f0 100644
--- a/tests/ref/fate/filter-pixfmts-crop
+++ b/tests/ref/fate/filter-pixfmts-crop
@@ -88,6 +88,10 @@ rgb8 9b364a8f112ad9459fec47a51cc03b30
rgba 9488ac85abceaf99a9309eac5a87697e
rgba64be 89910046972ab3c68e2a348302cc8ca9
rgba64le fea8ebfc869b52adf353778f29eac7a7
+rgbaf32be 48be3ace31f293fa2bf174c583c46b50
+rgbaf32le a2a358ecc5fe53a684d3de994134c4ce
+rgbf32be ad982fbaf08c908658e66cb4330257cd
+rgbf32le 14b944af62af458fa53da118721a258f
vuya 76578a705ff3a37559653c1289bd03dd
vuyx 5d2bae51a2f4892bd5f177f190cc323b
x2bgr10le 84de725b85662c362862820dc4a309aa
diff --git a/tests/ref/fate/filter-pixfmts-field b/tests/ref/fate/filter-pixfmts-field
index 4e5a798471..5c8817a5c7 100644
--- a/tests/ref/fate/filter-pixfmts-field
+++ b/tests/ref/fate/filter-pixfmts-field
@@ -90,6 +90,10 @@ rgb8 62c3b9e2a171de3d894a8eeb271c85e8
rgba ee616262ca6d67b7ecfba4b36c602ce3
rgba64be 23c8c0edaabe3eaec89ce69633fb0048
rgba64le dfdba4de4a7cac9abf08852666c341d3
+rgbaf32be f081b6c6df6094d36563b78dccdbc97a
+rgbaf32le e461aff4408751a03f8effd0e16be47c
+rgbf32be d71855f12b960ab02b787c99bc0e4cf1
+rgbf32le a4251c66110290418f15cb1e334a7f0d
uyvy422 1c49e44ab3f060e85fc4a3a9464f045e
vuya f72bcf29d75cd143d0c565f7cc49119a
vuyx 6257cd1ce11330660e9fa9c675acbdcc
diff --git a/tests/ref/fate/filter-pixfmts-fieldorder b/tests/ref/fate/filter-pixfmts-fieldorder
index bebaf07371..05b288dbe3 100644
--- a/tests/ref/fate/filter-pixfmts-fieldorder
+++ b/tests/ref/fate/filter-pixfmts-fieldorder
@@ -79,6 +79,10 @@ rgb8 6deae05ccac5c50bd0d9c9fe8e124557
rgba 1fdf872a087a32cd35b80cc7be399578
rgba64be 5598f44514d122b9a57c5c92c20bbc61
rgba64le b34e6e30621ae579519a2d91a96a0acf
+rgbaf32be 9a7c2182fd89218172f07e3b37f9728e
+rgbaf32le 940e87c0ddb7a86f9df438591b4870e6
+rgbf32be f23c5dd85e312471e3def0204051d264
+rgbf32le fec06f561663e7d01dec08c1c337f068
uyvy422 75de70e31c435dde878002d3f22b238a
vuya a3891d4168ff208948fd0b3ba0910495
vuyx d7a900e970c9a69ed41f8b220114b9fa
diff --git a/tests/ref/fate/filter-pixfmts-hflip b/tests/ref/fate/filter-pixfmts-hflip
index fd5e9723fd..02a734d8af 100644
--- a/tests/ref/fate/filter-pixfmts-hflip
+++ b/tests/ref/fate/filter-pixfmts-hflip
@@ -88,6 +88,10 @@ rgb8 68a3a575badadd9e4f90226209f11699
rgba 51961c723ea6707e0a410cd3f21f15d3
rgba64be c910444019f4cfbf4d995227af55da8d
rgba64le 0c810d8b3a6bca10321788e1cb145340
+rgbaf32be 6077603ed554db4149f82d891e1d1120
+rgbaf32le 3081376a3a9e90fd2849751dfd69094b
+rgbf32be 0e5ffacba8dc768acfa50fc590dddac6
+rgbf32le 542380fd2490e3227b70373eace3bade
vuya 7e530261e7ac4eae4fd616fd7572d0b8
vuyx 3ce9890363cad3984521293be1eb679c
x2bgr10le 827cc659f29378e00c5a7d2c0ada8f9a
diff --git a/tests/ref/fate/filter-pixfmts-il b/tests/ref/fate/filter-pixfmts-il
index ec9d809721..48701fe9a3 100644
--- a/tests/ref/fate/filter-pixfmts-il
+++ b/tests/ref/fate/filter-pixfmts-il
@@ -89,6 +89,10 @@ rgb8 93f9fa5ecf522abe13ed34f21831fdfe
rgba 625d8f4bd39c4bdbf61eb5e4713aecc9
rgba64be db70d33aa6c06f3e0a1c77bd11284261
rgba64le a8a2daae04374a27219bc1c890204007
+rgbaf32be e803073624dee4ce4c10d5d22f95b788
+rgbaf32le f87c6898ef385fb7239b16677a450326
+rgbf32be 3696160d95d41564a1a375ddeab82d70
+rgbf32le 71d9d0c83ded4c0866ea0673c1eed8d0
uyvy422 d6ee3ca43356d08c392382b24b22cda5
vuya b9deab5ba249dd608b709c09255a4932
vuyx 49cc92fcc002ec0f312017014dd68c0c
diff --git a/tests/ref/fate/filter-pixfmts-null b/tests/ref/fate/filter-pixfmts-null
index b28a114c7b..9aa653d12d 100644
--- a/tests/ref/fate/filter-pixfmts-null
+++ b/tests/ref/fate/filter-pixfmts-null
@@ -90,6 +90,10 @@ rgb8 7ac6008c84d622c2fc50581706e17576
rgba b6e1b441c365e03b5ffdf9b7b68d9a0c
rgba64be ae2ae04b5efedca3505f47c4dd6ea6ea
rgba64le b91e1d77f799eb92241a2d2d28437b15
+rgbaf32be abb244bb1af2247bbba7b4eac0357f6b
+rgbaf32le 99b2afe809649e58eea0a188fd6ab3f2
+rgbf32be e4a0b47e8ecb2e7e461915227ebd4edd
+rgbf32le 2979d6bfe509a55486e55e4de63d67d1
uyvy422 3bcf3c80047592f2211fae3260b1b65d
vuya 3d5e934651cae1ce334001cb1829ad22
vuyx 3f68ea6ec492b30d867cb5401562264e
diff --git a/tests/ref/fate/filter-pixfmts-scale b/tests/ref/fate/filter-pixfmts-scale
index 525306ec12..b22a6d639d 100644
--- a/tests/ref/fate/filter-pixfmts-scale
+++ b/tests/ref/fate/filter-pixfmts-scale
@@ -90,6 +90,10 @@ rgb8 bcdc033b4ef0979d060dbc8893d4db58
rgba 85bb5d03cea1c6e8002ced3373904336
rgba64be ee73e57923af984b31cc7795d13929da
rgba64le 783d2779adfafe3548bdb671ec0de69e
+rgbaf32be 0ccc31b373d6465ab72247ae42f395ce
+rgbaf32le 3eb06a5ba1c2341f12a30ff2539d5de0
+rgbf32be 11d0d9722abf9ea4fdce370920ff5820
+rgbf32le a9c247ff3d50e07bd7d0f777b9f82b32
uyvy422 aeb4ba4f9f003ae21f6d18089198244f
vuya ffa817e283bf6a0b6fba21b07523ccaa
vuyx ba182200e20e0c82765eba15217848d3
diff --git a/tests/ref/fate/filter-pixfmts-transpose b/tests/ref/fate/filter-pixfmts-transpose
index 24f4249639..868a784731 100644
--- a/tests/ref/fate/filter-pixfmts-transpose
+++ b/tests/ref/fate/filter-pixfmts-transpose
@@ -82,6 +82,10 @@ rgb8 c90feb30c3c9391ef5f470209d7b7a15
rgba 4d76a9542143752a4ac30f82f88f68f1
rgba64be a60041217f4c0cd796d19d3940a12a41
rgba64le ad47197774858858ae7b0c177dffa459
+rgbaf32be 83d7d66f59a8b62c40aa59c81188584d
+rgbaf32le d9cb46bfebebadaf838bb13d243feabe
+rgbf32be b090912ac60da12be5f84672d223300f
+rgbf32le 4b4e06621fb626952fc632f5049453ba
vuya 9ece18a345beb17cd19e09e443eca4bf
vuyx 4c2929cd1c6e5512f62e802f482f0ef2
x2bgr10le 4aa774b6d8f6d446a64f1f288e5c97eb
diff --git a/tests/ref/fate/filter-pixfmts-vflip b/tests/ref/fate/filter-pixfmts-vflip
index b7b0526588..b999109ff8 100644
--- a/tests/ref/fate/filter-pixfmts-vflip
+++ b/tests/ref/fate/filter-pixfmts-vflip
@@ -90,6 +90,10 @@ rgb8 7df049b6094f8a5e084d74462f6d6cde
rgba c1a5908572737f2ae1e5d8218af65f4b
rgba64be 17e6273323b5779b5f3f775f150c1011
rgba64le 48f45b10503b7dd140329c3dd0d54c98
+rgbaf32be cd7a7ce6f70ef8ca4a05d69458c6c40d
+rgbaf32le 4516303140dd7ff5bc02e4ae0c94b46d
+rgbf32be f416f773dd17dd15ca0845b47bf5ff44
+rgbf32le 16c50e897557972203c7e637080dc885
uyvy422 3a237e8376264e0cfa78f8a3fdadec8a
vuya fb849f76e56181e005c31fce75d7038c
vuyx 7a8079a97610e2c1c97aa8832b58a102
diff --git a/tests/ref/fate/sws-floatimg-cmp b/tests/ref/fate/sws-floatimg-cmp
index 251042f1c3..d524dd612f 100644
--- a/tests/ref/fate/sws-floatimg-cmp
+++ b/tests/ref/fate/sws-floatimg-cmp
@@ -118,3 +118,19 @@ gbrpf32le -> gbrap16le -> gbrpf32le
avg diff: 0.000249
min diff: 0.000000
max diff: 0.000990
+gbrpf32le -> rgbf32le -> gbrpf32le
+avg diff: 0.000249
+min diff: 0.000000
+max diff: 0.000990
+gbrpf32le -> rgbaf32le -> gbrpf32le
+avg diff: 0.000249
+min diff: 0.000000
+max diff: 0.000990
+gbrpf32le -> rgbf32be -> gbrpf32le
+avg diff: 0.000249
+min diff: 0.000000
+max diff: 0.000990
+gbrpf32le -> rgbaf32be -> gbrpf32le
+avg diff: 0.000249
+min diff: 0.000000
+max diff: 0.000990
--
2.31.1.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] 8+ messages in thread* Re: [FFmpeg-devel] [PATCH v5 4/4] swscale/output: add rgbaf32 output support
2022-11-23 19:35 ` [FFmpeg-devel] [PATCH v5 4/4] swscale/output: add rgbaf32 output support mindmark
@ 2022-12-05 0:05 ` Michael Niedermayer
2022-12-05 5:39 ` Mark Reid
0 siblings, 1 reply; 8+ messages in thread
From: Michael Niedermayer @ 2022-12-05 0:05 UTC (permalink / raw)
To: FFmpeg development discussions and patches
[-- Attachment #1.1: Type: text/plain, Size: 5455 bytes --]
On Wed, Nov 23, 2022 at 11:35:40AM -0800, mindmark@gmail.com wrote:
> From: Mark Reid <mindmark@gmail.com>
>
> ---
> libswscale/output.c | 92 ++++++++++++++++++++++++
> libswscale/swscale_unscaled.c | 4 +-
> libswscale/tests/floatimg_cmp.c | 4 +-
> libswscale/utils.c | 16 +++--
> libswscale/yuv2rgb.c | 2 +
> tests/ref/fate/filter-pixdesc-rgbaf32be | 1 +
> tests/ref/fate/filter-pixdesc-rgbaf32le | 1 +
> tests/ref/fate/filter-pixdesc-rgbf32be | 1 +
> tests/ref/fate/filter-pixdesc-rgbf32le | 1 +
> tests/ref/fate/filter-pixfmts-copy | 4 ++
> tests/ref/fate/filter-pixfmts-crop | 4 ++
> tests/ref/fate/filter-pixfmts-field | 4 ++
> tests/ref/fate/filter-pixfmts-fieldorder | 4 ++
> tests/ref/fate/filter-pixfmts-hflip | 4 ++
> tests/ref/fate/filter-pixfmts-il | 4 ++
> tests/ref/fate/filter-pixfmts-null | 4 ++
> tests/ref/fate/filter-pixfmts-scale | 4 ++
> tests/ref/fate/filter-pixfmts-transpose | 4 ++
> tests/ref/fate/filter-pixfmts-vflip | 4 ++
> tests/ref/fate/sws-floatimg-cmp | 16 +++++
> 20 files changed, 170 insertions(+), 8 deletions(-)
> create mode 100644 tests/ref/fate/filter-pixdesc-rgbaf32be
> create mode 100644 tests/ref/fate/filter-pixdesc-rgbaf32le
> create mode 100644 tests/ref/fate/filter-pixdesc-rgbf32be
> create mode 100644 tests/ref/fate/filter-pixdesc-rgbf32le
>
> diff --git a/libswscale/output.c b/libswscale/output.c
> index 5c85bff971..1d86a244f9 100644
> --- a/libswscale/output.c
> +++ b/libswscale/output.c
> @@ -2471,6 +2471,92 @@ yuv2gbrpf32_full_X_c(SwsContext *c, const int16_t *lumFilter,
> }
> }
>
> +static void
> +yuv2rgbaf32_full_X_c(SwsContext *c, const int16_t *lumFilter,
> + const int16_t **lumSrcx, int lumFilterSize,
> + const int16_t *chrFilter, const int16_t **chrUSrcx,
> + const int16_t **chrVSrcx, int chrFilterSize,
> + const int16_t **alpSrcx, uint8_t *dest,
> + int dstW, int y)
> +{
> + const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(c->dstFormat);
> + int i;
> + int alpha = desc->flags & AV_PIX_FMT_FLAG_ALPHA;
> + int hasAlpha = alpha && alpSrcx;
> + int pixelStep = alpha ? 4 : 3;
> + uint32_t *dest32 = (uint32_t*)dest;
> + const int32_t **lumSrc = (const int32_t**)lumSrcx;
> + const int32_t **chrUSrc = (const int32_t**)chrUSrcx;
> + const int32_t **chrVSrc = (const int32_t**)chrVSrcx;
> + const int32_t **alpSrc = (const int32_t**)alpSrcx;
> + static const float float_mult = 1.0f / 65535.0f;
> + uint32_t a = av_float2int(1.0f);
> +
> + for (i = 0; i < dstW; i++) {
> + int j;
> + int Y = -0x40000000;
> + int U = -(128 << 23);
> + int V = -(128 << 23);
> + int R, G, B, A;
> +
> + for (j = 0; j < lumFilterSize; j++)
> + Y += lumSrc[j][i] * (unsigned)lumFilter[j];
> +
> + for (j = 0; j < chrFilterSize; j++) {
> + U += chrUSrc[j][i] * (unsigned)chrFilter[j];
> + V += chrVSrc[j][i] * (unsigned)chrFilter[j];
> + }
> +
> + Y >>= 14;
> + Y += 0x10000;
> + U >>= 14;
> + V >>= 14;
> +
> + if (hasAlpha) {
> + A = -0x40000000;
> +
> + for (j = 0; j < lumFilterSize; j++)
> + A += alpSrc[j][i] * (unsigned)lumFilter[j];
> +
> + A >>= 1;
> + A += 0x20002000;
> + a = av_float2int(float_mult * (float)(av_clip_uintp2(A, 30) >> 14));
> + }
> +
> + Y -= c->yuv2rgb_y_offset;
> + Y *= c->yuv2rgb_y_coeff;
> + Y += (1 << 13) - (1 << 29);
> + R = V * c->yuv2rgb_v2r_coeff;
> + G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
> + B = U * c->yuv2rgb_u2b_coeff;
> +
> + R = av_clip_uintp2(((Y + R) >> 14) + (1<<15), 16);
> + G = av_clip_uintp2(((Y + G) >> 14) + (1<<15), 16);
> + B = av_clip_uintp2(((Y + B) >> 14) + (1<<15), 16);
> +
> + dest32[0] = av_float2int(float_mult * (float)R);
> + dest32[1] = av_float2int(float_mult * (float)G);
> + dest32[2] = av_float2int(float_mult * (float)B);
> + if (alpha)
> + dest32[3] = a;
why is this using uint32_t with av_float2int() and not floats straight ?
> +
> + dest32 += pixelStep;
> + }
> + if ((!isBE(c->dstFormat)) != (!HAVE_BIGENDIAN)) {
> + dest32 = (uint32_t*)dest;
> + for (i = 0; i < dstW; i++) {
> + dest32[0] = av_bswap32(dest32[0]);
> + dest32[1] = av_bswap32(dest32[1]);
> + dest32[2] = av_bswap32(dest32[2]);
> + if (alpha)
> + dest32[3] = av_bswap32(dest32[3]);
> +
> + dest32 += pixelStep;
> + }
> + }
teh code in bswapdsp seems more efficient, that should be shared and
used ideally
thx
[...]
--
Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB
Frequently ignored answer#1 FFmpeg bugs should be sent to our bugtracker. User
questions about the command line tools should be sent to the ffmpeg-user ML.
And questions about how to use libav* should be sent to the libav-user ML.
[-- Attachment #1.2: signature.asc --]
[-- Type: application/pgp-signature, Size: 195 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] 8+ messages in thread* Re: [FFmpeg-devel] [PATCH v5 4/4] swscale/output: add rgbaf32 output support
2022-12-05 0:05 ` Michael Niedermayer
@ 2022-12-05 5:39 ` Mark Reid
0 siblings, 0 replies; 8+ messages in thread
From: Mark Reid @ 2022-12-05 5:39 UTC (permalink / raw)
To: FFmpeg development discussions and patches
On Sun, Dec 4, 2022 at 4:05 PM Michael Niedermayer <michael@niedermayer.cc>
wrote:
> On Wed, Nov 23, 2022 at 11:35:40AM -0800, mindmark@gmail.com wrote:
> > From: Mark Reid <mindmark@gmail.com>
> >
> > ---
> > libswscale/output.c | 92 ++++++++++++++++++++++++
> > libswscale/swscale_unscaled.c | 4 +-
> > libswscale/tests/floatimg_cmp.c | 4 +-
> > libswscale/utils.c | 16 +++--
> > libswscale/yuv2rgb.c | 2 +
> > tests/ref/fate/filter-pixdesc-rgbaf32be | 1 +
> > tests/ref/fate/filter-pixdesc-rgbaf32le | 1 +
> > tests/ref/fate/filter-pixdesc-rgbf32be | 1 +
> > tests/ref/fate/filter-pixdesc-rgbf32le | 1 +
> > tests/ref/fate/filter-pixfmts-copy | 4 ++
> > tests/ref/fate/filter-pixfmts-crop | 4 ++
> > tests/ref/fate/filter-pixfmts-field | 4 ++
> > tests/ref/fate/filter-pixfmts-fieldorder | 4 ++
> > tests/ref/fate/filter-pixfmts-hflip | 4 ++
> > tests/ref/fate/filter-pixfmts-il | 4 ++
> > tests/ref/fate/filter-pixfmts-null | 4 ++
> > tests/ref/fate/filter-pixfmts-scale | 4 ++
> > tests/ref/fate/filter-pixfmts-transpose | 4 ++
> > tests/ref/fate/filter-pixfmts-vflip | 4 ++
> > tests/ref/fate/sws-floatimg-cmp | 16 +++++
> > 20 files changed, 170 insertions(+), 8 deletions(-)
> > create mode 100644 tests/ref/fate/filter-pixdesc-rgbaf32be
> > create mode 100644 tests/ref/fate/filter-pixdesc-rgbaf32le
> > create mode 100644 tests/ref/fate/filter-pixdesc-rgbf32be
> > create mode 100644 tests/ref/fate/filter-pixdesc-rgbf32le
> >
> > diff --git a/libswscale/output.c b/libswscale/output.c
> > index 5c85bff971..1d86a244f9 100644
> > --- a/libswscale/output.c
> > +++ b/libswscale/output.c
> > @@ -2471,6 +2471,92 @@ yuv2gbrpf32_full_X_c(SwsContext *c, const int16_t
> *lumFilter,
> > }
> > }
> >
> > +static void
> > +yuv2rgbaf32_full_X_c(SwsContext *c, const int16_t *lumFilter,
> > + const int16_t **lumSrcx, int lumFilterSize,
> > + const int16_t *chrFilter, const int16_t **chrUSrcx,
> > + const int16_t **chrVSrcx, int chrFilterSize,
> > + const int16_t **alpSrcx, uint8_t *dest,
> > + int dstW, int y)
> > +{
> > + const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(c->dstFormat);
> > + int i;
> > + int alpha = desc->flags & AV_PIX_FMT_FLAG_ALPHA;
> > + int hasAlpha = alpha && alpSrcx;
> > + int pixelStep = alpha ? 4 : 3;
> > + uint32_t *dest32 = (uint32_t*)dest;
> > + const int32_t **lumSrc = (const int32_t**)lumSrcx;
> > + const int32_t **chrUSrc = (const int32_t**)chrUSrcx;
> > + const int32_t **chrVSrc = (const int32_t**)chrVSrcx;
> > + const int32_t **alpSrc = (const int32_t**)alpSrcx;
> > + static const float float_mult = 1.0f / 65535.0f;
> > + uint32_t a = av_float2int(1.0f);
> > +
> > + for (i = 0; i < dstW; i++) {
> > + int j;
> > + int Y = -0x40000000;
> > + int U = -(128 << 23);
> > + int V = -(128 << 23);
> > + int R, G, B, A;
> > +
> > + for (j = 0; j < lumFilterSize; j++)
> > + Y += lumSrc[j][i] * (unsigned)lumFilter[j];
> > +
> > + for (j = 0; j < chrFilterSize; j++) {
> > + U += chrUSrc[j][i] * (unsigned)chrFilter[j];
> > + V += chrVSrc[j][i] * (unsigned)chrFilter[j];
> > + }
> > +
> > + Y >>= 14;
> > + Y += 0x10000;
> > + U >>= 14;
> > + V >>= 14;
> > +
> > + if (hasAlpha) {
> > + A = -0x40000000;
> > +
> > + for (j = 0; j < lumFilterSize; j++)
> > + A += alpSrc[j][i] * (unsigned)lumFilter[j];
> > +
> > + A >>= 1;
> > + A += 0x20002000;
> > + a = av_float2int(float_mult * (float)(av_clip_uintp2(A, 30)
> >> 14));
> > + }
> > +
> > + Y -= c->yuv2rgb_y_offset;
> > + Y *= c->yuv2rgb_y_coeff;
> > + Y += (1 << 13) - (1 << 29);
> > + R = V * c->yuv2rgb_v2r_coeff;
> > + G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
> > + B = U * c->yuv2rgb_u2b_coeff;
> > +
> > + R = av_clip_uintp2(((Y + R) >> 14) + (1<<15), 16);
> > + G = av_clip_uintp2(((Y + G) >> 14) + (1<<15), 16);
> > + B = av_clip_uintp2(((Y + B) >> 14) + (1<<15), 16);
> > +
> > + dest32[0] = av_float2int(float_mult * (float)R);
> > + dest32[1] = av_float2int(float_mult * (float)G);
> > + dest32[2] = av_float2int(float_mult * (float)B);
> > + if (alpha)
> > + dest32[3] = a;
>
> why is this using uint32_t with av_float2int() and not floats straight ?
>
>
It's this way because it is matching the planar f32 version, I will change
both.
>
>
> > +
> > + dest32 += pixelStep;
> > + }
> > + if ((!isBE(c->dstFormat)) != (!HAVE_BIGENDIAN)) {
> > + dest32 = (uint32_t*)dest;
> > + for (i = 0; i < dstW; i++) {
> > + dest32[0] = av_bswap32(dest32[0]);
> > + dest32[1] = av_bswap32(dest32[1]);
> > + dest32[2] = av_bswap32(dest32[2]);
> > + if (alpha)
> > + dest32[3] = av_bswap32(dest32[3]);
> > +
> > + dest32 += pixelStep;
> > + }
> > + }
>
> teh code in bswapdsp seems more efficient, that should be shared and
> used ideally
>
I just sent a patch moving the bswapdsp code from avcodec to avutil
>
> thx
>
> [...]
> --
> Michael GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB
>
> Frequently ignored answer#1 FFmpeg bugs should be sent to our bugtracker.
> User
> questions about the command line tools should be sent to the ffmpeg-user
> ML.
> And questions about how to use libav* should be sent to the libav-user ML.
> _______________________________________________
> 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".
>
_______________________________________________
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] 8+ messages in thread
* Re: [FFmpeg-devel] [PATCH v5 0/4] swscale: rgbaf32 input/output support
2022-11-23 19:35 [FFmpeg-devel] [PATCH v5 0/4] swscale: rgbaf32 input/output support mindmark
` (3 preceding siblings ...)
2022-11-23 19:35 ` [FFmpeg-devel] [PATCH v5 4/4] swscale/output: add rgbaf32 output support mindmark
@ 2022-12-04 21:48 ` Mark Reid
4 siblings, 0 replies; 8+ messages in thread
From: Mark Reid @ 2022-12-04 21:48 UTC (permalink / raw)
To: ffmpeg-devel
On Wed, Nov 23, 2022 at 11:35 AM <mindmark@gmail.com> wrote:
> From: Mark Reid <mindmark@gmail.com>
>
> This patch series adds swscale input/output support for the packed rgb
> float formats.
> A few of the filters also needed support the larger 96/128 bit packed
> pixel sizes.
>
> I also plan to eventually add lossless unscaled conversions between the
> planer and packed formats.
>
> changes since v4
> * added comment about refactoring input functions
> changes since v3
> * removed half uv path implementation
> changes since v2
> * add bias to rgbaf32 output to improve non overflowing range
> changes since v1
> * output correct alpha if src doesn't have alpha
>
>
> Mark Reid (4):
> swscale/input: add rgbaf32 input support
> avfilter/vf_hflip: add support for packed rgb float formats
> avfilter/vf_transpose: add support for packed rgb float formats
> swscale/output: add rgbaf32 output support
>
> libavfilter/vf_hflip_init.h | 25 +++++
> libavfilter/vf_transpose.c | 44 ++++++++
> libswscale/input.c | 122 +++++++++++++++++++++++
> libswscale/output.c | 92 +++++++++++++++++
> libswscale/swscale_unscaled.c | 4 +-
> libswscale/tests/floatimg_cmp.c | 4 +-
> libswscale/utils.c | 14 ++-
> libswscale/yuv2rgb.c | 2 +
> tests/ref/fate/filter-pixdesc-rgbaf32be | 1 +
> tests/ref/fate/filter-pixdesc-rgbaf32le | 1 +
> tests/ref/fate/filter-pixdesc-rgbf32be | 1 +
> tests/ref/fate/filter-pixdesc-rgbf32le | 1 +
> tests/ref/fate/filter-pixfmts-copy | 4 +
> tests/ref/fate/filter-pixfmts-crop | 4 +
> tests/ref/fate/filter-pixfmts-field | 4 +
> tests/ref/fate/filter-pixfmts-fieldorder | 4 +
> tests/ref/fate/filter-pixfmts-hflip | 4 +
> tests/ref/fate/filter-pixfmts-il | 4 +
> tests/ref/fate/filter-pixfmts-null | 4 +
> tests/ref/fate/filter-pixfmts-scale | 4 +
> tests/ref/fate/filter-pixfmts-transpose | 4 +
> tests/ref/fate/filter-pixfmts-vflip | 4 +
> tests/ref/fate/sws-floatimg-cmp | 16 +++
> 23 files changed, 363 insertions(+), 4 deletions(-)
> create mode 100644 tests/ref/fate/filter-pixdesc-rgbaf32be
> create mode 100644 tests/ref/fate/filter-pixdesc-rgbaf32le
> create mode 100644 tests/ref/fate/filter-pixdesc-rgbf32be
> create mode 100644 tests/ref/fate/filter-pixdesc-rgbf32le
>
> --
> 2.31.1.windows.1
>
>
ping
_______________________________________________
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] 8+ messages in thread