Traneptora/FFmpeg
1
Fork 0
forked from FFmpeg/FFmpeg
Code Pull requests Activity

swscale: 16bit planar float input support

Browse source 
Sponsored-by: Sovereign Tech Fund
Signed-off-by: Michael Niedermayer <michael@niedermayer.cc>
This commit is contained in:
Michael Niedermayer 2025-01-09 22:14:57 +01:00
parent 62c98cdd54
commit 665b0cf3bf
No known key found for this signature in database
GPG key ID: B18E8928B3948D64
3 changed files with 107 additions and 1 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

97
libswscale/input.c
View file

@ -1369,6 +1369,57 @@ rgbf32_funcs_endian(le, 0)
rgbf32_funcs_endian(be, 1)
#define rdpx(src) av_int2float(half2float(is_be ? AV_RB16(&src) : AV_RL16(&src), h2f_tbl))
#define rdpx2(src) av_int2float(half2float(is_be ? AV_RB16(src) : AV_RL16(src), h2f_tbl))
static av_always_inline void planar_rgbf16_to_a(uint8_t *dst, const uint8_t *src[4], int width, int is_be, int32_t *rgb2yuv, Half2FloatTables *h2f_tbl)
{
int i;
for (i = 0; i < width; i++) {
AV_WN16(dst + 2*i, lrintf(av_clipf(65535.0f * rdpx2(src[3] + 2*i), 0.0f, 65535.0f)));
}
}
static av_always_inline void planar_rgbf16_to_uv(uint8_t *dstU, uint8_t *dstV, const uint8_t *src[4], int width, int is_be, int32_t *rgb2yuv, Half2FloatTables *h2f_tbl)
{
int i;
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];
for (i = 0; i < width; i++) {
int g = lrintf(av_clipf(65535.0f * rdpx2(src[0] + 2*i), 0.0f, 65535.0f));
int b = lrintf(av_clipf(65535.0f * rdpx2(src[1] + 2*i), 0.0f, 65535.0f));
int r = lrintf(av_clipf(65535.0f * rdpx2(src[2] + 2*i), 0.0f, 65535.0f));
AV_WN16(dstU + 2*i, (ru*r + gu*g + bu*b + (0x10001 << (RGB2YUV_SHIFT - 1))) >> RGB2YUV_SHIFT);
AV_WN16(dstV + 2*i, (rv*r + gv*g + bv*b + (0x10001 << (RGB2YUV_SHIFT - 1))) >> RGB2YUV_SHIFT);
}
}
static av_always_inline void planar_rgbf16_to_y(uint8_t *dst, const uint8_t *src[4], int width, int is_be, int32_t *rgb2yuv, Half2FloatTables *h2f_tbl)
{
int i;
int32_t ry = rgb2yuv[RY_IDX], gy = rgb2yuv[GY_IDX], by = rgb2yuv[BY_IDX];
for (i = 0; i < width; i++) {
int g = lrintf(av_clipf(65535.0f * rdpx2(src[0] + 2*i), 0.0f, 65535.0f));
int b = lrintf(av_clipf(65535.0f * rdpx2(src[1] + 2*i), 0.0f, 65535.0f));
int r = lrintf(av_clipf(65535.0f * rdpx2(src[2] + 2*i), 0.0f, 65535.0f));
AV_WN16(dst + 2*i, (ry*r + gy*g + by*b + (0x2001 << (RGB2YUV_SHIFT - 1))) >> RGB2YUV_SHIFT);
}
}
static av_always_inline void grayf16ToY16_c(uint8_t *dst, const uint8_t *src, const uint8_t *unused1,
const uint8_t *unused2, int width, int is_be, uint32_t *unused, Half2FloatTables *h2f_tbl)
{
int i;
for (i = 0; i < width; ++i){
AV_WN16(dst + 2*i, lrintf(av_clipf(65535.0f * rdpx2(src + 2*i), 0.0f, 65535.0f)));
}
}
static av_always_inline void rgbaf16ToUV_half_endian(uint16_t *dstU, uint16_t *dstV, int is_be,
const uint16_t *src, int width,
@ -1484,6 +1535,28 @@ static av_always_inline void rgbf16ToY_endian(uint16_t *dst, const uint16_t *src
#undef rdpx
#define rgbaf16_funcs_endian(endian_name, endian) \
static void planar_rgbf16##endian_name##_to_y(uint8_t *dst, const uint8_t *src[4], \
int w, int32_t *rgb2yuv, void *opq) \
{ \
planar_rgbf16_to_y(dst, src, w, endian, rgb2yuv, opq); \
} \
static void planar_rgbf16##endian_name##_to_uv(uint8_t *dstU, uint8_t *dstV, \
const uint8_t *src[4], int w, int32_t *rgb2yuv, \
void *opq) \
{ \
planar_rgbf16_to_uv(dstU, dstV, src, w, endian, rgb2yuv, opq); \
} \
static void planar_rgbf16##endian_name##_to_a(uint8_t *dst, const uint8_t *src[4], \
int w, int32_t *rgb2yuv, void *opq) \
{ \
planar_rgbf16_to_a(dst, src, w, endian, rgb2yuv, opq); \
} \
static void grayf16##endian_name##ToY16_c(uint8_t *dst, const uint8_t *src, \
const uint8_t *unused1, const uint8_t *unused2, \
int width, uint32_t *unused, void *opq) \
{ \
grayf16ToY16_c(dst, src, unused1, unused2, width, endian, unused, opq); \
} \
static void rgbaf16##endian_name##ToUV_half_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) \
@ -1618,6 +1691,10 @@ av_cold void ff_sws_init_input_funcs(SwsInternal *c,
case AV_PIX_FMT_GBRPF32LE:
*readChrPlanar = planar_rgbf32le_to_uv;
break;
case AV_PIX_FMT_GBRAPF16LE:
case AV_PIX_FMT_GBRPF16LE:
*readChrPlanar = planar_rgbf16le_to_uv;
break;
case AV_PIX_FMT_GBRP9BE:
*readChrPlanar = planar_rgb9be_to_uv;
break;
@ -1641,6 +1718,10 @@ av_cold void ff_sws_init_input_funcs(SwsInternal *c,
case AV_PIX_FMT_GBRPF32BE:
*readChrPlanar = planar_rgbf32be_to_uv;
break;
case AV_PIX_FMT_GBRAPF16BE:
case AV_PIX_FMT_GBRPF16BE:
*readChrPlanar = planar_rgbf16be_to_uv;
break;
case AV_PIX_FMT_GBRAP:
case AV_PIX_FMT_GBRP:
*readChrPlanar = planar_rgb_to_uv;
@ -2025,6 +2106,11 @@ av_cold void ff_sws_init_input_funcs(SwsInternal *c,
case AV_PIX_FMT_GBRPF32LE:
*readLumPlanar = planar_rgbf32le_to_y;
break;
case AV_PIX_FMT_GBRAPF16LE:
*readAlpPlanar = planar_rgbf16le_to_a;
case AV_PIX_FMT_GBRPF16LE:
*readLumPlanar = planar_rgbf16le_to_y;
break;
case AV_PIX_FMT_GBRP9BE:
*readLumPlanar = planar_rgb9be_to_y;
break;
@ -2053,6 +2139,11 @@ av_cold void ff_sws_init_input_funcs(SwsInternal *c,
case AV_PIX_FMT_GBRPF32BE:
*readLumPlanar = planar_rgbf32be_to_y;
break;
case AV_PIX_FMT_GBRAPF16BE:
*readAlpPlanar = planar_rgbf16be_to_a;
case AV_PIX_FMT_GBRPF16BE:
*readLumPlanar = planar_rgbf16be_to_y;
break;
case AV_PIX_FMT_GBRAP:
*readAlpPlanar = planar_rgb_to_a;
case AV_PIX_FMT_GBRP:
@ -2309,6 +2400,12 @@ av_cold void ff_sws_init_input_funcs(SwsInternal *c,
case AV_PIX_FMT_GRAYF32BE:
*lumToYV12 = grayf32beToY16_c;
break;
case AV_PIX_FMT_GRAYF16LE:
*lumToYV12 = grayf16leToY16_c;
break;
case AV_PIX_FMT_GRAYF16BE:
*lumToYV12 = grayf16beToY16_c;
break;
case AV_PIX_FMT_Y210LE:
*lumToYV12 = y210le_Y_c;
break;

3
libswscale/swscale_unscaled.c
View file

@ -2655,11 +2655,12 @@ void ff_get_unscaled_swscale(SwsInternal *c)
c->convert_unscaled = nv24ToYuv420Wrapper;
#define isPlanarGray(x) (isGray(x) && (x) != AV_PIX_FMT_YA8 && (x) != AV_PIX_FMT_YA16LE && (x) != AV_PIX_FMT_YA16BE)
/* simple copy */
if ( srcFormat == dstFormat ||
(srcFormat == AV_PIX_FMT_YUVA420P && dstFormat == AV_PIX_FMT_YUV420P) ||
(srcFormat == AV_PIX_FMT_YUV420P && dstFormat == AV_PIX_FMT_YUVA420P) ||
(isFloat(srcFormat) == isFloat(dstFormat)) && ((isPlanarYUV(srcFormat) && isPlanarGray(dstFormat)) ||
(isFloat(srcFormat) == isFloat(dstFormat) && isFloat16(srcFormat) == isFloat16(dstFormat)) && ((isPlanarYUV(srcFormat) && isPlanarGray(dstFormat)) ||
(isPlanarYUV(dstFormat) && isPlanarGray(srcFormat)) ||
(isPlanarGray(dstFormat) && isPlanarGray(srcFormat)) ||
(isPlanarYUV(srcFormat) && isPlanarYUV(dstFormat) &&

8
libswscale/utils.c
View file

@ -219,6 +219,10 @@ static const FormatEntry format_entries[] = {
[AV_PIX_FMT_GBRPF32BE] = { 1, 1 },
[AV_PIX_FMT_GBRAPF32LE] = { 1, 1 },
[AV_PIX_FMT_GBRAPF32BE] = { 1, 1 },
[AV_PIX_FMT_GBRPF16LE] = { 1, 0 },
[AV_PIX_FMT_GBRPF16BE] = { 1, 0 },
[AV_PIX_FMT_GBRAPF16LE] = { 1, 0 },
[AV_PIX_FMT_GBRAPF16BE] = { 1, 0 },
[AV_PIX_FMT_GBRAP] = { 1, 1 },
[AV_PIX_FMT_GBRAP16LE] = { 1, 1 },
[AV_PIX_FMT_GBRAP16BE] = { 1, 1 },
@ -246,6 +250,8 @@ static const FormatEntry format_entries[] = {
[AV_PIX_FMT_P016BE] = { 1, 1 },
[AV_PIX_FMT_GRAYF32LE] = { 1, 1 },
[AV_PIX_FMT_GRAYF32BE] = { 1, 1 },
[AV_PIX_FMT_GRAYF16LE] = { 1, 0 },
[AV_PIX_FMT_GRAYF16BE] = { 1, 0 },
[AV_PIX_FMT_YUVA422P12BE] = { 1, 1 },
[AV_PIX_FMT_YUVA422P12LE] = { 1, 1 },
[AV_PIX_FMT_YUVA444P12BE] = { 1, 1 },
@ -1578,6 +1584,8 @@ av_cold int ff_sws_init_single_context(SwsContext *sws, 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_GBRPF16BE && srcFormat != AV_PIX_FMT_GBRPF16LE &&
srcFormat != AV_PIX_FMT_GBRAPF16BE && srcFormat != AV_PIX_FMT_GBRAPF16LE &&
((dstW >> c->chrDstHSubSample) <= (srcW >> 1) ||
(flags & SWS_FAST_BILINEAR)))
c->chrSrcHSubSample = 1;