[FFmpeg-devel] [PATCH 4/5] sws/output: convert 1 and 2 type yuv2rgb to tableless solution
Michael Niedermayer
michaelni at gmx.at
Sun Apr 14 12:17:21 CEST 2013
Signed-off-by: Michael Niedermayer <michaelni at gmx.at>
---
libswscale/output.c | 131 ++++++++++++++++++++++++++++++++++-----------------
1 file changed, 89 insertions(+), 42 deletions(-)
diff --git a/libswscale/output.c b/libswscale/output.c
index 8429efc..3d7c0e1 100644
--- a/libswscale/output.c
+++ b/libswscale/output.c
@@ -1212,6 +1212,10 @@ yuv2rgb_2_c_template(SwsContext *c, const int16_t *buf[2],
int yalpha, int uvalpha, int y,
enum AVPixelFormat target, int hasAlpha)
{
+ int isBGR = c->dstFormat == AV_PIX_FMT_ABGR || c->dstFormat == AV_PIX_FMT_BGRA || c->dstFormat == AV_PIX_FMT_BGR24
+ || c->dstFormat == AV_PIX_FMT_BGR565LE || c->dstFormat == AV_PIX_FMT_BGR555LE || c->dstFormat == AV_PIX_FMT_BGR444LE
+ || c->dstFormat == AV_PIX_FMT_BGR565BE || c->dstFormat == AV_PIX_FMT_BGR555BE || c->dstFormat == AV_PIX_FMT_BGR444BE
+ || c->dstFormat == AV_PIX_FMT_BGR8 || c->dstFormat == AV_PIX_FMT_BGR4 || c->dstFormat == AV_PIX_FMT_BGR4_BYTE;
const int16_t *buf0 = buf[0], *buf1 = buf[1],
*ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
*vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
@@ -1222,24 +1226,40 @@ yuv2rgb_2_c_template(SwsContext *c, const int16_t *buf[2],
int i;
for (i = 0; i < ((dstW + 1) >> 1); i++) {
- int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 19;
- int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 19;
- int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha) >> 19;
- int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha) >> 19;
- int A1, A2;
- const void *r = c->table_rV[V + YUVRGB_TABLE_HEADROOM],
- *g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]),
- *b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
+ int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 10;
+ int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 10;
+ int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha) >> 10;
+ int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha) >> 10;
+ int A1, A2, R, G, B;
if (hasAlpha) {
- A1 = (abuf0[i * 2 ] * yalpha1 + abuf1[i * 2 ] * yalpha) >> 19;
- A2 = (abuf0[i * 2 + 1] * yalpha1 + abuf1[i * 2 + 1] * yalpha) >> 19;
- A1 = av_clip_uint8(A1);
- A2 = av_clip_uint8(A2);
+ A1 = (abuf0[i * 2 ] * yalpha1 + abuf1[i * 2 ] * yalpha);
+ A2 = (abuf0[i * 2 + 1] * yalpha1 + abuf1[i * 2 + 1] * yalpha);
}
+ U += (-128 << 9);
+ V += (-128 << 9);
+ Y1 -= c->yuv2rgb_y_offset;
+ Y2 -= c->yuv2rgb_y_offset;
+ Y1 *= c->yuv2rgb_y_coeff;
+ Y2 *= c->yuv2rgb_y_coeff;
+ Y1 += 1 << 13;
+ Y2 += 1 << 13;
- yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
- r, g, b, y, target, hasAlpha);
+ R = V * c->yuv2rgb_v2r_coeff;
+ G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
+ B = U * c->yuv2rgb_u2b_coeff;
+
+ yuv2rgb_write2(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
+ R, G, B, y, target, hasAlpha, isBGR, 27);
+ }
+ if ((!isBE(c->dstFormat)) != (!HAVE_BIGENDIAN)) {
+ const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(c->dstFormat);
+ uint16_t *dest16 = (uint16_t*)dest;
+ if (6-1 > desc->comp[0].depth_minus1 && desc->comp[0].depth_minus1 > 3-1) {
+ for (i = 0; i < dstW; i++) {
+ dest16[i] = av_bswap16(dest16[i]);
+ }
+ }
}
}
@@ -1250,51 +1270,78 @@ yuv2rgb_1_c_template(SwsContext *c, const int16_t *buf0,
int uvalpha, int y, enum AVPixelFormat target,
int hasAlpha)
{
+ int isBGR = c->dstFormat == AV_PIX_FMT_ABGR || c->dstFormat == AV_PIX_FMT_BGRA || c->dstFormat == AV_PIX_FMT_BGR24
+ || c->dstFormat == AV_PIX_FMT_BGR565LE || c->dstFormat == AV_PIX_FMT_BGR555LE || c->dstFormat == AV_PIX_FMT_BGR444LE
+ || c->dstFormat == AV_PIX_FMT_BGR565BE || c->dstFormat == AV_PIX_FMT_BGR555BE || c->dstFormat == AV_PIX_FMT_BGR444BE
+ || c->dstFormat == AV_PIX_FMT_BGR8 || c->dstFormat == AV_PIX_FMT_BGR4 || c->dstFormat == AV_PIX_FMT_BGR4_BYTE;
const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
int i;
if (uvalpha < 2048) {
for (i = 0; i < ((dstW + 1) >> 1); i++) {
- int Y1 = (buf0[i * 2 ] + 64) >> 7;
- int Y2 = (buf0[i * 2 + 1] + 64) >> 7;
- int U = (ubuf0[i] + 64) >> 7;
- int V = (vbuf0[i] + 64) >> 7;
- int A1, A2;
- const void *r = c->table_rV[V + YUVRGB_TABLE_HEADROOM],
- *g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]),
- *b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
+ int Y1 = buf0[i * 2 ] << 2;
+ int Y2 = buf0[i * 2 + 1] << 2;
+ int U = ubuf0[i] << 2;
+ int V = vbuf0[i] << 2;
+ int A1, A2, R, G, B;
if (hasAlpha) {
- A1 = abuf0[i * 2 ] * 255 + 16384 >> 15;
- A2 = abuf0[i * 2 + 1] * 255 + 16384 >> 15;
- A1 = av_clip_uint8(A1);
- A2 = av_clip_uint8(A2);
+ A1 = abuf0[i * 2 ];
+ A2 = abuf0[i * 2 + 1];
}
+ U += (-128 << 9);
+ V += (-128 << 9);
+ Y1 -= c->yuv2rgb_y_offset;
+ Y2 -= c->yuv2rgb_y_offset;
+ Y1 *= c->yuv2rgb_y_coeff;
+ Y2 *= c->yuv2rgb_y_coeff;
+ Y1 += 1 << 13;
+ Y2 += 1 << 13;
- yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
- r, g, b, y, target, hasAlpha);
+ R = V * c->yuv2rgb_v2r_coeff;
+ G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
+ B = U * c->yuv2rgb_u2b_coeff;
+
+ yuv2rgb_write2(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
+ R, G, B, y, target, hasAlpha, isBGR, 15);
}
} else {
const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
for (i = 0; i < ((dstW + 1) >> 1); i++) {
- int Y1 = (buf0[i * 2 ] + 64) >> 7;
- int Y2 = (buf0[i * 2 + 1] + 64) >> 7;
- int U = (ubuf0[i] + ubuf1[i] + 128) >> 8;
- int V = (vbuf0[i] + vbuf1[i] + 128) >> 8;
- int A1, A2;
- const void *r = c->table_rV[V + YUVRGB_TABLE_HEADROOM],
- *g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]),
- *b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
+ int Y1 = buf0[i * 2 ] << 2;
+ int Y2 = buf0[i * 2 + 1] << 2;
+ int U = (ubuf0[i] + ubuf1[i]) << 1;
+ int V = (vbuf0[i] + vbuf1[i]) << 1;
+ int A1, A2, R, G, B;
if (hasAlpha) {
- A1 = (abuf0[i * 2 ] + 64) >> 7;
- A2 = (abuf0[i * 2 + 1] + 64) >> 7;
- A1 = av_clip_uint8(A1);
- A2 = av_clip_uint8(A2);
+ A1 = abuf0[i * 2 ];
+ A2 = abuf0[i * 2 + 1];
}
+ U += (-128 << 9);
+ V += (-128 << 9);
+ Y1 -= c->yuv2rgb_y_offset;
+ Y2 -= c->yuv2rgb_y_offset;
+ Y1 *= c->yuv2rgb_y_coeff;
+ Y2 *= c->yuv2rgb_y_coeff;
+ Y1 += 1 << 13;
+ Y2 += 1 << 13;
+
+ R = V * c->yuv2rgb_v2r_coeff;
+ G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
+ B = U * c->yuv2rgb_u2b_coeff;
- yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
- r, g, b, y, target, hasAlpha);
+ yuv2rgb_write2(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
+ R, G, B, y, target, hasAlpha, isBGR, 15);
+ }
+ }
+ if ((!isBE(c->dstFormat)) != (!HAVE_BIGENDIAN)) {
+ const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(c->dstFormat);
+ uint16_t *dest16 = (uint16_t*)dest;
+ if (6-1 > desc->comp[0].depth_minus1 && desc->comp[0].depth_minus1 > 3-1) {
+ for (i = 0; i < dstW; i++) {
+ dest16[i] = av_bswap16(dest16[i]);
+ }
}
}
}
--
1.7.9.5
More information about the ffmpeg-devel
mailing list