[FFmpeg-devel] [PATCH v2 2/5] swscale/yuv2rgb: prepare YUV2RGBFUNC macro for multi-planar rgb
Ramiro Polla
ramiro.polla at gmail.com
Tue Aug 6 13:51:03 EEST 2024
This will be used in the upcoming yuv42{0,2}p -> gbrp unscaled
colorspace converters.
There is no difference in performance.
---
libswscale/yuv2rgb.c | 83 +++++++++++++++++++++++++-------------------
1 file changed, 48 insertions(+), 35 deletions(-)
diff --git a/libswscale/yuv2rgb.c b/libswscale/yuv2rgb.c
index d77660b3a3..31d10235ef 100644
--- a/libswscale/yuv2rgb.c
+++ b/libswscale/yuv2rgb.c
@@ -124,7 +124,7 @@ const int *sws_getCoefficients(int colorspace)
dst_##l[12 * i + 8] = dst_##l[12 * i + 9] = g[Y]; \
dst_##l[12 * i + 10] = dst_##l[12 * i + 11] = r[Y];
-#define YUV2RGBFUNC(func_name, dst_type, alpha, yuv422) \
+#define YUV2RGBFUNC(func_name, dst_type, alpha, yuv422, nb_dst_planes) \
static int func_name(SwsContext *c, const uint8_t *src[], \
int srcStride[], int srcSliceY, int srcSliceH, \
uint8_t *dst[], int dstStride[]) \
@@ -137,6 +137,7 @@ const int *sws_getCoefficients(int colorspace)
(dst_type *)(dst[0] + (yd) * dstStride[0]); \
dst_type *dst_2 = \
(dst_type *)(dst[0] + (yd + 1) * dstStride[0]); \
+ dst_type av_unused *dst1_1, *dst1_2, *dst2_1, *dst2_2; \
dst_type av_unused *r, *g, *b; \
const uint8_t *py_1 = src[0] + y * srcStride[0]; \
const uint8_t *py_2 = py_1 + srcStride[0]; \
@@ -145,6 +146,12 @@ const int *sws_getCoefficients(int colorspace)
const uint8_t av_unused *pu_2, *pv_2; \
const uint8_t av_unused *pa_1, *pa_2; \
unsigned int h_size = c->dstW >> 3; \
+ if (nb_dst_planes > 1) { \
+ dst1_1 = (dst_type *)(dst[1] + (yd) * dstStride[1]); \
+ dst1_2 = (dst_type *)(dst[1] + (yd + 1) * dstStride[1]); \
+ dst2_1 = (dst_type *)(dst[2] + (yd) * dstStride[2]); \
+ dst2_2 = (dst_type *)(dst[2] + (yd + 1) * dstStride[2]); \
+ } \
if (yuv422) { \
pu_2 = pu_1 + srcStride[1]; \
pv_2 = pv_1 + srcStride[2]; \
@@ -156,7 +163,7 @@ const int *sws_getCoefficients(int colorspace)
while (h_size--) { \
int av_unused U, V, Y; \
-#define ENDYUV2RGBLINE(dst_delta, ss, alpha, yuv422) \
+#define ENDYUV2RGBLINE(dst_delta, ss, alpha, yuv422, nb_dst_planes) \
pu_1 += 4 >> ss; \
pv_1 += 4 >> ss; \
if (yuv422) { \
@@ -171,6 +178,12 @@ const int *sws_getCoefficients(int colorspace)
} \
dst_1 += dst_delta >> ss; \
dst_2 += dst_delta >> ss; \
+ if (nb_dst_planes > 1) { \
+ dst1_1 += dst_delta >> ss; \
+ dst1_2 += dst_delta >> ss; \
+ dst2_1 += dst_delta >> ss; \
+ dst2_2 += dst_delta >> ss; \
+ } \
} \
if (c->dstW & (4 >> ss)) { \
int av_unused Y, U, V; \
@@ -181,8 +194,8 @@ const int *sws_getCoefficients(int colorspace)
return srcSliceH; \
}
-#define YUV420FUNC(func_name, dst_type, alpha, abase, PUTFUNC, dst_delta) \
- YUV2RGBFUNC(func_name, dst_type, alpha, 0) \
+#define YUV420FUNC(func_name, dst_type, alpha, abase, PUTFUNC, dst_delta, nb_dst_planes) \
+ YUV2RGBFUNC(func_name, dst_type, alpha, 0, nb_dst_planes) \
LOADCHROMA(1, 0); \
PUTFUNC(1, 0, abase); \
PUTFUNC(2, 0, abase); \
@@ -198,7 +211,7 @@ const int *sws_getCoefficients(int colorspace)
LOADCHROMA(1, 3); \
PUTFUNC(2, 3, abase); \
PUTFUNC(1, 3, abase); \
- ENDYUV2RGBLINE(dst_delta, 0, alpha, 0) \
+ ENDYUV2RGBLINE(dst_delta, 0, alpha, 0, nb_dst_planes) \
LOADCHROMA(1, 0); \
PUTFUNC(1, 0, abase); \
PUTFUNC(2, 0, abase); \
@@ -206,14 +219,14 @@ const int *sws_getCoefficients(int colorspace)
LOADCHROMA(1, 1); \
PUTFUNC(2, 1, abase); \
PUTFUNC(1, 1, abase); \
- ENDYUV2RGBLINE(dst_delta, 1, alpha, 0) \
+ ENDYUV2RGBLINE(dst_delta, 1, alpha, 0, nb_dst_planes) \
LOADCHROMA(1, 0); \
PUTFUNC(1, 0, abase); \
PUTFUNC(2, 0, abase); \
ENDYUV2RGBFUNC()
-#define YUV422FUNC(func_name, dst_type, alpha, abase, PUTFUNC, dst_delta) \
- YUV2RGBFUNC(func_name, dst_type, alpha, 1) \
+#define YUV422FUNC(func_name, dst_type, alpha, abase, PUTFUNC, dst_delta, nb_dst_planes) \
+ YUV2RGBFUNC(func_name, dst_type, alpha, 1, nb_dst_planes) \
LOADCHROMA(1, 0); \
PUTFUNC(1, 0, abase); \
\
@@ -237,7 +250,7 @@ const int *sws_getCoefficients(int colorspace)
\
LOADCHROMA(1, 3); \
PUTFUNC(1, 3, abase); \
- ENDYUV2RGBLINE(dst_delta, 0, alpha, 1) \
+ ENDYUV2RGBLINE(dst_delta, 0, alpha, 1, nb_dst_planes) \
LOADCHROMA(1, 0); \
PUTFUNC(1, 0, abase); \
\
@@ -249,7 +262,7 @@ const int *sws_getCoefficients(int colorspace)
\
LOADCHROMA(1, 1); \
PUTFUNC(1, 1, abase); \
- ENDYUV2RGBLINE(dst_delta, 1, alpha, 1) \
+ ENDYUV2RGBLINE(dst_delta, 1, alpha, 1, nb_dst_planes) \
LOADCHROMA(1, 0); \
PUTFUNC(1, 0, abase); \
\
@@ -258,7 +271,7 @@ const int *sws_getCoefficients(int colorspace)
ENDYUV2RGBFUNC()
#define YUV420FUNC_DITHER(func_name, dst_type, LOADDITHER, PUTFUNC, dst_delta) \
- YUV2RGBFUNC(func_name, dst_type, 0, 0) \
+ YUV2RGBFUNC(func_name, dst_type, 0, 0, 1) \
LOADDITHER \
\
LOADCHROMA(1, 0); \
@@ -276,7 +289,7 @@ const int *sws_getCoefficients(int colorspace)
LOADCHROMA(1, 3); \
PUTFUNC(2, 3, 6 + 8); \
PUTFUNC(1, 3, 6); \
- ENDYUV2RGBLINE(dst_delta, 0, 0, 0) \
+ ENDYUV2RGBLINE(dst_delta, 0, 0, 0, 1) \
LOADDITHER \
\
LOADCHROMA(1, 0); \
@@ -286,7 +299,7 @@ const int *sws_getCoefficients(int colorspace)
LOADCHROMA(1, 1); \
PUTFUNC(2, 1, 2 + 8); \
PUTFUNC(1, 1, 2); \
- ENDYUV2RGBLINE(dst_delta, 1, 0, 0) \
+ ENDYUV2RGBLINE(dst_delta, 1, 0, 0, 1) \
LOADDITHER \
\
LOADCHROMA(1, 0); \
@@ -295,7 +308,7 @@ const int *sws_getCoefficients(int colorspace)
ENDYUV2RGBFUNC()
#define YUV422FUNC_DITHER(func_name, dst_type, LOADDITHER, PUTFUNC, dst_delta) \
- YUV2RGBFUNC(func_name, dst_type, 0, 1) \
+ YUV2RGBFUNC(func_name, dst_type, 0, 1, 1) \
LOADDITHER \
\
LOADCHROMA(1, 0); \
@@ -321,7 +334,7 @@ const int *sws_getCoefficients(int colorspace)
\
LOADCHROMA(1, 3); \
PUTFUNC(1, 3, 6); \
- ENDYUV2RGBLINE(dst_delta, 0, 0, 1) \
+ ENDYUV2RGBLINE(dst_delta, 0, 0, 1, 1) \
LOADDITHER \
\
LOADCHROMA(1, 0); \
@@ -335,7 +348,7 @@ const int *sws_getCoefficients(int colorspace)
\
LOADCHROMA(1, 1); \
PUTFUNC(1, 1, 2); \
- ENDYUV2RGBLINE(dst_delta, 1, 0, 1) \
+ ENDYUV2RGBLINE(dst_delta, 1, 0, 1, 1) \
LOADDITHER \
\
LOADCHROMA(1, 0); \
@@ -431,7 +444,7 @@ const int *sws_getCoefficients(int colorspace)
g[Y + d64[1 + o]] + \
b[Y + d128[1 + o]];
-YUV2RGBFUNC(yuv2rgb_c_1_ordered_dither, uint8_t, 0, 0)
+YUV2RGBFUNC(yuv2rgb_c_1_ordered_dither, uint8_t, 0, 0, 1)
const uint8_t *d128 = ff_dither_8x8_220[yd & 7];
char out_1 = 0, out_2 = 0;
g = c->table_gU[128 + YUVRGB_TABLE_HEADROOM] + c->table_gV[128 + YUVRGB_TABLE_HEADROOM];
@@ -494,18 +507,18 @@ YUV2RGBFUNC(yuv2rgb_c_1_ordered_dither, uint8_t, 0, 0)
ENDYUV2RGBFUNC()
// YUV420
-YUV420FUNC(yuv2rgb_c_48, uint8_t, 0, 0, PUTRGB48, 48)
-YUV420FUNC(yuv2rgb_c_bgr48, uint8_t, 0, 0, PUTBGR48, 48)
-YUV420FUNC(yuv2rgb_c_32, uint32_t, 0, 0, PUTRGB, 8)
+YUV420FUNC(yuv2rgb_c_48, uint8_t, 0, 0, PUTRGB48, 48, 1)
+YUV420FUNC(yuv2rgb_c_bgr48, uint8_t, 0, 0, PUTBGR48, 48, 1)
+YUV420FUNC(yuv2rgb_c_32, uint32_t, 0, 0, PUTRGB, 8, 1)
#if HAVE_BIGENDIAN
-YUV420FUNC(yuva2argb_c, uint32_t, 1, 24, PUTRGBA, 8)
-YUV420FUNC(yuva2rgba_c, uint32_t, 1, 0, PUTRGBA, 8)
+YUV420FUNC(yuva2argb_c, uint32_t, 1, 24, PUTRGBA, 8, 1)
+YUV420FUNC(yuva2rgba_c, uint32_t, 1, 0, PUTRGBA, 8, 1)
#else
-YUV420FUNC(yuva2rgba_c, uint32_t, 1, 24, PUTRGBA, 8)
-YUV420FUNC(yuva2argb_c, uint32_t, 1, 0, PUTRGBA, 8)
+YUV420FUNC(yuva2rgba_c, uint32_t, 1, 24, PUTRGBA, 8, 1)
+YUV420FUNC(yuva2argb_c, uint32_t, 1, 0, PUTRGBA, 8, 1)
#endif
-YUV420FUNC(yuv2rgb_c_24_rgb, uint8_t, 0, 0, PUTRGB24, 24)
-YUV420FUNC(yuv2rgb_c_24_bgr, uint8_t, 0, 0, PUTBGR24, 24)
+YUV420FUNC(yuv2rgb_c_24_rgb, uint8_t, 0, 0, PUTRGB24, 24, 1)
+YUV420FUNC(yuv2rgb_c_24_bgr, uint8_t, 0, 0, PUTBGR24, 24, 1)
YUV420FUNC_DITHER(yuv2rgb_c_16_ordered_dither, uint16_t, LOADDITHER16, PUTRGB16, 8)
YUV420FUNC_DITHER(yuv2rgb_c_15_ordered_dither, uint16_t, LOADDITHER15, PUTRGB15, 8)
YUV420FUNC_DITHER(yuv2rgb_c_12_ordered_dither, uint16_t, LOADDITHER12, PUTRGB12, 8)
@@ -514,18 +527,18 @@ YUV420FUNC_DITHER(yuv2rgb_c_4_ordered_dither, uint8_t, LOADDITHER4D, PUTRGB4D
YUV420FUNC_DITHER(yuv2rgb_c_4b_ordered_dither, uint8_t, LOADDITHER4DB, PUTRGB4DB, 8)
// YUV422
-YUV422FUNC(yuv422p_rgb48_c, uint8_t, 0, 0, PUTRGB48, 48)
-YUV422FUNC(yuv422p_bgr48_c, uint8_t, 0, 0, PUTBGR48, 48)
-YUV422FUNC(yuv422p_rgb32_c, uint32_t, 0, 0, PUTRGB, 8)
+YUV422FUNC(yuv422p_rgb48_c, uint8_t, 0, 0, PUTRGB48, 48, 1)
+YUV422FUNC(yuv422p_bgr48_c, uint8_t, 0, 0, PUTBGR48, 48, 1)
+YUV422FUNC(yuv422p_rgb32_c, uint32_t, 0, 0, PUTRGB, 8, 1)
#if HAVE_BIGENDIAN
-YUV422FUNC(yuva422p_argb_c, uint32_t, 1, 24, PUTRGBA, 8)
-YUV422FUNC(yuva422p_rgba_c, uint32_t, 1, 0, PUTRGBA, 8)
+YUV422FUNC(yuva422p_argb_c, uint32_t, 1, 24, PUTRGBA, 8, 1)
+YUV422FUNC(yuva422p_rgba_c, uint32_t, 1, 0, PUTRGBA, 8, 1)
#else
-YUV422FUNC(yuva422p_rgba_c, uint32_t, 1, 24, PUTRGBA, 8)
-YUV422FUNC(yuva422p_argb_c, uint32_t, 1, 0, PUTRGBA, 8)
+YUV422FUNC(yuva422p_rgba_c, uint32_t, 1, 24, PUTRGBA, 8, 1)
+YUV422FUNC(yuva422p_argb_c, uint32_t, 1, 0, PUTRGBA, 8, 1)
#endif
-YUV422FUNC(yuv422p_rgb24_c, uint8_t, 0, 0, PUTRGB24, 24)
-YUV422FUNC(yuv422p_bgr24_c, uint8_t, 0, 0, PUTBGR24, 24)
+YUV422FUNC(yuv422p_rgb24_c, uint8_t, 0, 0, PUTRGB24, 24, 1)
+YUV422FUNC(yuv422p_bgr24_c, uint8_t, 0, 0, PUTBGR24, 24, 1)
YUV422FUNC_DITHER(yuv422p_bgr16, uint16_t, LOADDITHER16, PUTRGB16, 8)
YUV422FUNC_DITHER(yuv422p_bgr15, uint16_t, LOADDITHER15, PUTRGB15, 8)
YUV422FUNC_DITHER(yuv422p_bgr12, uint16_t, LOADDITHER12, PUTRGB12, 8)
--
2.30.2
More information about the ffmpeg-devel
mailing list