[FFmpeg-devel] [PATCH v12 4/9] avutil/swscale: add YUV444F32 and UYVYF32 pixel format and input support
Martin Schitter
ms+git at mur.at
Mon Oct 21 22:57:17 EEST 2024
---
libavutil/pixdesc.c | 48 +++++++++++++++
libavutil/pixfmt.h | 6 ++
libswscale/input.c | 100 +++++++++++++++++++++++++++++++
libswscale/utils.c | 6 +-
tests/ref/fate/imgutils | 8 +++
tests/ref/fate/sws-pixdesc-query | 10 ++++
6 files changed, 177 insertions(+), 1 deletion(-)
diff --git a/libavutil/pixdesc.c b/libavutil/pixdesc.c
index 609f07a..f69c9b2 100644
--- a/libavutil/pixdesc.c
+++ b/libavutil/pixdesc.c
@@ -539,6 +539,30 @@ static const AVPixFmtDescriptor av_pix_fmt_descriptors[AV_PIX_FMT_NB] = {
},
.flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_FLOAT,
},
+ [AV_PIX_FMT_UYVY422F32LE] = {
+ .name = "uyvy422f32le",
+ .nb_components = 3,
+ .log2_chroma_w = 1,
+ .log2_chroma_h = 0,
+ .comp = {
+ { 0, 8, 4, 0, 32 }, /* Y */
+ { 0, 16, 0, 0, 32 }, /* U */
+ { 0, 16, 8, 0, 32 }, /* V */
+ },
+ .flags = AV_PIX_FMT_FLAG_FLOAT,
+ },
+ [AV_PIX_FMT_UYVY422F32BE] = {
+ .name = "uyvy422f32be",
+ .nb_components = 3,
+ .log2_chroma_w = 1,
+ .log2_chroma_h = 0,
+ .comp = {
+ { 0, 8, 4, 0, 32 }, /* Y */
+ { 0, 16, 0, 0, 32 }, /* U */
+ { 0, 16, 8, 0, 32 }, /* V */
+ },
+ .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_FLOAT,
+ },
[AV_PIX_FMT_UYYVYY411] = {
.name = "uyyvyy411",
.nb_components = 3,
@@ -1761,6 +1785,30 @@ static const AVPixFmtDescriptor av_pix_fmt_descriptors[AV_PIX_FMT_NB] = {
},
.flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_FLOAT,
},
+ [AV_PIX_FMT_YUV444F32LE] = {
+ .name = "yuv444f32le",
+ .nb_components = 3,
+ .log2_chroma_w = 0,
+ .log2_chroma_h = 0,
+ .comp = {
+ { 0, 12, 0, 0, 32 }, /* Y */
+ { 0, 12, 4, 0, 32 }, /* U */
+ { 0, 12, 8, 0, 32 }, /* V */
+ },
+ .flags = AV_PIX_FMT_FLAG_FLOAT,
+ },
+ [AV_PIX_FMT_YUV444F32BE] = {
+ .name = "yuv444f32be",
+ .nb_components = 3,
+ .log2_chroma_w = 0,
+ .log2_chroma_h = 0,
+ .comp = {
+ { 0, 12, 0, 0, 32 }, /* Y */
+ { 0, 12, 4, 0, 32 }, /* U */
+ { 0, 12, 8, 0, 32 }, /* V */
+ },
+ .flags = AV_PIX_FMT_FLAG_BE | AV_PIX_FMT_FLAG_FLOAT,
+ },
[AV_PIX_FMT_YUV444P10LE] = {
.name = "yuv444p10le",
.nb_components = 3,
diff --git a/libavutil/pixfmt.h b/libavutil/pixfmt.h
index 67f568f..b946b8d 100644
--- a/libavutil/pixfmt.h
+++ b/libavutil/pixfmt.h
@@ -462,11 +462,15 @@ enum AVPixelFormat {
AV_PIX_FMT_YUV444_16LE, ///< packed YUV 4:4:4, 48bpp (1 Cr & Cb sample per 1x1 Y), YUVYUV..., little-endian
AV_PIX_FMT_YUV444F16BE, ///< IEEE-754 half precision packed YUV 4:4:4, 48bpp (1 Cr & Cb sample per 1x1 Y), YUVYUV..., big-endian
AV_PIX_FMT_YUV444F16LE, ///< IEEE-754 half precision packed YUV 4:4:4, 48bpp (1 Cr & Cb sample per 1x1 Y), YUVYUV..., little-endian
+ AV_PIX_FMT_YUV444F32BE, ///< IEEE-754 single precision packed YUV 4:4:4, 96bpp (1 Cr & Cb sample per 1x1 Y), YUVYUV..., big-endian
+ AV_PIX_FMT_YUV444F32LE, ///< IEEE-754 single precision packed YUV 4:4:4, 96bpp (1 Cr & Cb sample per 1x1 Y), YUVYUV..., little-endian
AV_PIX_FMT_UYVY422_16BE, ///< packed UYVU 4:2:2, 32bpp, Cb Y0 Cr Y1, big-endian
AV_PIX_FMT_UYVY422_16LE, ///< packed UYVU 4:2:2, 32bpp, Cb Y0 Cr Y1, little-endian
AV_PIX_FMT_UYVY422F16BE, ///< IEEE-754 half precision packed UYVU 4:2:2, 32bpp, Cb Y0 Cr Y1, big-endian
AV_PIX_FMT_UYVY422F16LE, ///< IEEE-754 half precision packed UYVU 4:2:2, 32bpp, Cb Y0 Cr Y1, little-endian
+ AV_PIX_FMT_UYVY422F32BE, ///< IEEE-754 single precision packed UYVU 4:2:2, 64bpp, Cb Y0 Cr Y1, big-endian
+ AV_PIX_FMT_UYVY422F32LE, ///< IEEE-754 single precision packed UYVU 4:2:2, 64bpp, Cb Y0 Cr Y1, little-endian
AV_PIX_FMT_NB ///< number of pixel formats, DO NOT USE THIS if you want to link with shared libav* because the number of formats might differ between versions
};
@@ -520,6 +524,7 @@ enum AVPixelFormat {
#define AV_PIX_FMT_YUV444P16 AV_PIX_FMT_NE(YUV444P16BE, YUV444P16LE)
#define AV_PIX_FMT_YUV444_16 AV_PIX_FMT_NE(YUV444_16BE, YUV444_16LE)
#define AV_PIX_FMT_YUV444F16 AV_PIX_FMT_NE(YUV444F16BE, YUV444F16LE)
+#define AV_PIX_FMT_YUV444F32 AV_PIX_FMT_NE(YUV444F32BE, YUV444F32LE)
#define AV_PIX_FMT_GBRP9 AV_PIX_FMT_NE(GBRP9BE , GBRP9LE)
#define AV_PIX_FMT_GBRP10 AV_PIX_FMT_NE(GBRP10BE, GBRP10LE)
@@ -555,6 +560,7 @@ enum AVPixelFormat {
#define AV_PIX_FMT_UYVY422_16 AV_PIX_FMT_NE(UYVY422_16BE, UYVY422_16LE)
#define AV_PIX_FMT_UYVY422F16 AV_PIX_FMT_NE(UYVY422F16BE, UYVY422F16LE)
+#define AV_PIX_FMT_UYVY422F32 AV_PIX_FMT_NE(UYVY422F32BE, UYVY422F32LE)
#define AV_PIX_FMT_XYZ12 AV_PIX_FMT_NE(XYZ12BE, XYZ12LE)
#define AV_PIX_FMT_NV20 AV_PIX_FMT_NE(NV20BE, NV20LE)
diff --git a/libswscale/input.c b/libswscale/input.c
index 685b2ea..f9d7c39 100644
--- a/libswscale/input.c
+++ b/libswscale/input.c
@@ -876,6 +876,44 @@ static void read_yuv444f16be_UV_c(uint8_t *dstU, uint8_t *dstV, const uint8_t *u
}
}
+static void read_yuv444f32le_Y_c(uint8_t *_dst, const uint8_t *src, const uint8_t *unused0,
+ const uint8_t *unused1, int width, uint32_t *unused2, void *opq)
+{
+ uint16_t *dst = (uint16_t *) _dst;
+ for (int i = 0; i < width; i++)
+ dst[i] = lrintf(av_clipf(65535.0f * av_int2float(AV_RL32(src + i * 12)), 0.0f, 65535.0f));
+}
+
+static void read_yuv444f32le_UV_c(uint8_t *_dstU, uint8_t *_dstV, const uint8_t *unused0, const uint8_t *src,
+ const uint8_t *unused1, int width, uint32_t *unused2, void *opq)
+{
+ uint16_t *dstU = (uint16_t *) _dstU;
+ uint16_t *dstV = (uint16_t *) _dstV;
+ for (int i = 0; i < width; i++) {
+ dstU[i] = lrintf(av_clipf(65535.0f * (av_int2float(AV_RL32(src + i * 12 + 4)) + 0.5f), 0.0f, 65535.0f));
+ dstV[i] = lrintf(av_clipf(65535.0f * (av_int2float(AV_RL32(src + i * 12 + 8)) + 0.5f), 0.0f, 65535.0f));
+ }
+}
+
+static void read_yuv444f32be_Y_c(uint8_t *_dst, const uint8_t *src, const uint8_t *unused0,
+ const uint8_t *unused1, int width, uint32_t *unused2, void *opq)
+{
+ uint16_t *dst = (uint16_t *) _dst;
+ for (int i = 0; i < width; i++)
+ dst[i] = lrintf(av_clipf(65535.0f * av_int2float(AV_RB32(src + i * 12)), 0.0f, 65535.0f));
+}
+
+static void read_yuv444f32be_UV_c(uint8_t *_dstU, uint8_t *_dstV, const uint8_t *unused0, const uint8_t *src,
+ const uint8_t *unused1, int width, uint32_t *unused2, void *opq)
+{
+ uint16_t *dstU = (uint16_t *) _dstU;
+ uint16_t *dstV = (uint16_t *) _dstV;
+ for (int i = 0; i < width; i++) {
+ dstU[i] = lrintf(av_clipf(65535.0f * (av_int2float(AV_RB32(src + i * 12 + 4)) + 0.5f), 0.0f, 65535.0f));
+ dstV[i] = lrintf(av_clipf(65535.0f * (av_int2float(AV_RB32(src + i * 12 + 8)) + 0.5f), 0.0f, 65535.0f));
+ }
+}
+
static void read_v30xle_Y_c(uint8_t *dst, const uint8_t *src, const uint8_t *unused0, const uint8_t *unused1, int width,
uint32_t *unused2, void *opq)
{
@@ -1044,6 +1082,44 @@ static void uyvyf16be_ToUV_c(uint8_t *dstU, uint8_t *dstV, const uint8_t *unused
}
}
+static void uyvyf32le_ToY_c(uint8_t *_dst, const uint8_t *src, const uint8_t *unused1, const uint8_t *unused2, int width,
+ uint32_t *unused, void *obl)
+{
+ uint16_t *dst = (uint16_t *) _dst;
+ for (int i = 0; i < width; i++)
+ dst[i] = lrintf(av_clipf(65535.0f * av_int2float(AV_RL32(src + i * 8 + 4)), 0.0f, 65535.0f));
+}
+
+static void uyvyf32le_ToUV_c(uint8_t *_dstU, uint8_t *_dstV, const uint8_t *unused0, const uint8_t *src1,
+ const uint8_t *src2, int width, uint32_t *unused, void *obl)
+{
+ uint16_t *dstU = (uint16_t *) _dstU;
+ uint16_t *dstV = (uint16_t *) _dstV;
+ for (int i = 0; i < width; i++) {
+ dstU[i] = lrintf(av_clipf(65535.0f * (av_int2float(AV_RL32(src1 + i * 16)) + 0.5f), 0.0f, 65535.0f));
+ dstV[i] = lrintf(av_clipf(65535.0f * (av_int2float(AV_RL32(src1 + i * 16 + 8)) + 0.5f), 0.0f, 65535.0f));
+ }
+}
+
+static void uyvyf32be_ToY_c(uint8_t *_dst, const uint8_t *src, const uint8_t *unused1, const uint8_t *unused2, int width,
+ uint32_t *unused, void *obl)
+{
+ uint16_t *dst = (uint16_t *) _dst;
+ for (int i = 0; i < width; i++)
+ dst[i] = lrintf(av_clipf(65535.0f * av_int2float(AV_RB32(src + i * 8 + 4)), 0.0f, 65535.0f));
+}
+
+static void uyvyf32be_ToUV_c(uint8_t *_dstU, uint8_t *_dstV, const uint8_t *unused0, const uint8_t *src1,
+ const uint8_t *src2, int width, uint32_t *unused, void *obl)
+{
+ uint16_t *dstU = (uint16_t *) _dstU;
+ uint16_t *dstV = (uint16_t *) _dstV;
+ for (int i = 0; i < width; i++) {
+ dstU[i] = lrintf(av_clipf(65535.0f * (av_int2float(AV_RB32(src1 + i * 16)) + 0.5f), 0.0f, 65535.0f));
+ dstV[i] = lrintf(av_clipf(65535.0f * (av_int2float(AV_RB32(src1 + i * 16 + 8)) + 0.5f), 0.0f, 65535.0f));
+ }
+}
+
static av_always_inline void nvXXtoUV_c(uint8_t *dst1, uint8_t *dst2,
const uint8_t *src, int width)
{
@@ -1728,6 +1804,12 @@ av_cold void ff_sws_init_input_funcs(SwsContext *c,
case AV_PIX_FMT_UYVY422F16BE:
*chrToYV12 = uyvyf16be_ToUV_c;
break;
+ case AV_PIX_FMT_UYVY422F32LE:
+ *chrToYV12 = uyvyf32le_ToUV_c;
+ break;
+ case AV_PIX_FMT_UYVY422F32BE:
+ *chrToYV12 = uyvyf32be_ToUV_c;
+ break;
case AV_PIX_FMT_VYU444:
*chrToYV12 = vyuToUV_c;
break;
@@ -1746,6 +1828,12 @@ av_cold void ff_sws_init_input_funcs(SwsContext *c,
case AV_PIX_FMT_YUV444F16BE:
*chrToYV12 = read_yuv444f16be_UV_c;
break;
+ case AV_PIX_FMT_YUV444F32LE:
+ *chrToYV12 = read_yuv444f32le_UV_c;
+ break;
+ case AV_PIX_FMT_YUV444F32BE:
+ *chrToYV12 = read_yuv444f32be_UV_c;
+ break;
case AV_PIX_FMT_NV12:
case AV_PIX_FMT_NV16:
case AV_PIX_FMT_NV24:
@@ -2357,6 +2445,12 @@ av_cold void ff_sws_init_input_funcs(SwsContext *c,
case AV_PIX_FMT_UYVY422F16BE:
*lumToYV12 = uyvyf16be_ToY_c;
break;
+ case AV_PIX_FMT_UYVY422F32LE:
+ *lumToYV12 = uyvyf32le_ToY_c;
+ break;
+ case AV_PIX_FMT_UYVY422F32BE:
+ *lumToYV12 = uyvyf32be_ToY_c;
+ break;
case AV_PIX_FMT_VYU444:
*lumToYV12 = vyuToY_c;
break;
@@ -2375,6 +2469,12 @@ av_cold void ff_sws_init_input_funcs(SwsContext *c,
case AV_PIX_FMT_YUV444F16BE:
*lumToYV12 = read_yuv444f16be_Y_c;
break;
+ case AV_PIX_FMT_YUV444F32LE:
+ *lumToYV12 = read_yuv444f32le_Y_c;
+ break;
+ case AV_PIX_FMT_YUV444F32BE:
+ *lumToYV12 = read_yuv444f32be_Y_c;
+ break;
case AV_PIX_FMT_BGR24:
*lumToYV12 = bgr24ToY_c;
break;
diff --git a/libswscale/utils.c b/libswscale/utils.c
index 4573355..5318ae8 100644
--- a/libswscale/utils.c
+++ b/libswscale/utils.c
@@ -78,6 +78,8 @@ static const FormatEntry format_entries[] = {
[AV_PIX_FMT_YUV444_16LE] = { 1, 0 },
[AV_PIX_FMT_YUV444F16BE] = { 1, 0 },
[AV_PIX_FMT_YUV444F16LE] = { 1, 0 },
+ [AV_PIX_FMT_YUV444F32BE] = { 1, 0 },
+ [AV_PIX_FMT_YUV444F32LE] = { 1, 0 },
[AV_PIX_FMT_YUV410P] = { 1, 1 },
[AV_PIX_FMT_YUV411P] = { 1, 1 },
[AV_PIX_FMT_GRAY8] = { 1, 1 },
@@ -94,6 +96,8 @@ static const FormatEntry format_entries[] = {
[AV_PIX_FMT_UYVY422_16LE]= { 1, 0 },
[AV_PIX_FMT_UYVY422F16BE]= { 1, 0 },
[AV_PIX_FMT_UYVY422F16LE]= { 1, 0 },
+ [AV_PIX_FMT_UYVY422F32BE]= { 1, 0 },
+ [AV_PIX_FMT_UYVY422F32LE]= { 1, 0 },
[AV_PIX_FMT_UYYVYY411] = { 0, 0 },
[AV_PIX_FMT_BGR8] = { 1, 1 },
[AV_PIX_FMT_BGR4] = { 0, 1 },
@@ -1717,7 +1721,7 @@ static av_cold int sws_init_single_context(SwsContext *c, SwsFilter *srcFilter,
}
// float will be converted to uint16_t
- if ((srcFormat == AV_PIX_FMT_GRAYF32BE || srcFormat == AV_PIX_FMT_GRAYF32LE) &&
+ if (isFloat(srcFormat) &&
(!unscaled || unscaled && dstFormat != srcFormat && (srcFormat != AV_PIX_FMT_GRAYF32 ||
dstFormat != AV_PIX_FMT_GRAY8))){
c->srcBpc = 16;
diff --git a/tests/ref/fate/imgutils b/tests/ref/fate/imgutils
index bebd540..f0ac9d2 100644
--- a/tests/ref/fate/imgutils
+++ b/tests/ref/fate/imgutils
@@ -285,10 +285,14 @@ yuv444_16be planes: 1, linesizes: 384 0 0 0, plane_sizes: 18432 0
yuv444_16le planes: 1, linesizes: 384 0 0 0, plane_sizes: 18432 0 0 0, plane_offsets: 0 0 0, total_size: 18432
yuv444f16be planes: 1, linesizes: 384 0 0 0, plane_sizes: 18432 0 0 0, plane_offsets: 0 0 0, total_size: 18432
yuv444f16le planes: 1, linesizes: 384 0 0 0, plane_sizes: 18432 0 0 0, plane_offsets: 0 0 0, total_size: 18432
+yuv444f32be planes: 1, linesizes: 768 0 0 0, plane_sizes: 36864 0 0 0, plane_offsets: 0 0 0, total_size: 36864
+yuv444f32le planes: 1, linesizes: 768 0 0 0, plane_sizes: 36864 0 0 0, plane_offsets: 0 0 0, total_size: 36864
uyvy422_16be planes: 1, linesizes: 256 0 0 0, plane_sizes: 12288 0 0 0, plane_offsets: 0 0 0, total_size: 12288
uyvy422_16le planes: 1, linesizes: 256 0 0 0, plane_sizes: 12288 0 0 0, plane_offsets: 0 0 0, total_size: 12288
uyvy422f16be planes: 1, linesizes: 256 0 0 0, plane_sizes: 12288 0 0 0, plane_offsets: 0 0 0, total_size: 12288
uyvy422f16le planes: 1, linesizes: 256 0 0 0, plane_sizes: 12288 0 0 0, plane_offsets: 0 0 0, total_size: 12288
+uyvy422f32be planes: 1, linesizes: 512 0 0 0, plane_sizes: 24576 0 0 0, plane_offsets: 0 0 0, total_size: 24576
+uyvy422f32le planes: 1, linesizes: 512 0 0 0, plane_sizes: 24576 0 0 0, plane_offsets: 0 0 0, total_size: 24576
image_fill_black tests
yuv420p total_size: 4608, black_unknown_crc: 0xd00f6cc6, black_tv_crc: 0xd00f6cc6, black_pc_crc: 0x234969af
@@ -521,7 +525,11 @@ yuv444_16be total_size: 18432, black_unknown_crc: 0xac983d03, black_tv_cr
yuv444_16le total_size: 18432, black_unknown_crc: 0x24c4432b, black_tv_crc: 0x24c4432b, black_pc_crc: 0xfe5a7889
yuv444f16be total_size: 18432, black_unknown_crc: 0x5cf972e1, black_tv_crc: 0x5cf972e1, black_pc_crc: 0x482033cf
yuv444f16le total_size: 18432, black_unknown_crc: 0x4a18e5bb, black_tv_crc: 0x4a18e5bb, black_pc_crc: 0x920584a7
+yuv444f32be total_size: 36864, black_unknown_crc: 0x1ea7a984, black_tv_crc: 0x1ea7a984, black_pc_crc: 0x63cf67b9
+yuv444f32le total_size: 36864, black_unknown_crc: 0x63d60c2d, black_tv_crc: 0x63d60c2d, black_pc_crc: 0x2d5f6a04
uyvy422_16be total_size: 12288, black_unknown_crc: 0xb53352d5, black_tv_crc: 0xb53352d5, black_pc_crc: 0xf8461cf3
uyvy422_16le total_size: 12288, black_unknown_crc: 0xe39de4bd, black_tv_crc: 0xe39de4bd, black_pc_crc: 0x24b29add
uyvy422f16be total_size: 12288, black_unknown_crc: 0x9c275ddf, black_tv_crc: 0x9c275ddf, black_pc_crc: 0xde92b629
uyvy422f16le total_size: 12288, black_unknown_crc: 0x2e5b76c4, black_tv_crc: 0x2e5b76c4, black_pc_crc: 0x2928eced
+uyvy422f32be total_size: 24576, black_unknown_crc: 0x04614e4a, black_tv_crc: 0x04614e4a, black_pc_crc: 0xaa67a8ce
+uyvy422f32le total_size: 24576, black_unknown_crc: 0x3bf0427e, black_tv_crc: 0x3bf0427e, black_pc_crc: 0x7636d7e3
diff --git a/tests/ref/fate/sws-pixdesc-query b/tests/ref/fate/sws-pixdesc-query
index 0d564c8..d19007f 100644
--- a/tests/ref/fate/sws-pixdesc-query
+++ b/tests/ref/fate/sws-pixdesc-query
@@ -197,6 +197,7 @@ isBE:
rgbf32be
uyvy422_16be
uyvy422f16be
+ uyvy422f32be
v30xbe
x2bgr10be
x2rgb10be
@@ -220,6 +221,7 @@ isBE:
yuv440p12be
yuv444_16be
yuv444f16be
+ yuv444f32be
yuv444p10be
yuv444p12be
yuv444p14be
@@ -272,6 +274,8 @@ isYUV:
uyvy422_16le
uyvy422f16be
uyvy422f16le
+ uyvy422f32be
+ uyvy422f32le
uyyvyy411
v30xbe
v30xle
@@ -325,6 +329,8 @@ isYUV:
yuv444_16le
yuv444f16be
yuv444f16le
+ yuv444f32be
+ yuv444f32le
yuv444p
yuv444p10be
yuv444p10le
@@ -881,6 +887,8 @@ Packed:
uyvy422_16le
uyvy422f16be
uyvy422f16le
+ uyvy422f32be
+ uyvy422f32le
uyyvyy411
v30xbe
v30xle
@@ -909,6 +917,8 @@ Packed:
yuv444_16le
yuv444f16be
yuv444f16le
+ yuv444f32be
+ yuv444f32le
yuyv422
yvyu422
--
2.45.2
More information about the ffmpeg-devel
mailing list