2 * software YUV to RGB converter
4 * Copyright (C) 2009 Konstantin Shishkov
6 * 1,4,8bpp support and context / deglobalize stuff
7 * by Michael Niedermayer (michaelni@gmx.at)
9 * This file is part of FFmpeg.
11 * FFmpeg is free software; you can redistribute it and/or
12 * modify it under the terms of the GNU Lesser General Public
13 * License as published by the Free Software Foundation; either
14 * version 2.1 of the License, or (at your option) any later version.
16 * FFmpeg is distributed in the hope that it will be useful,
17 * but WITHOUT ANY WARRANTY; without even the implied warranty of
18 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
19 * Lesser General Public License for more details.
21 * You should have received a copy of the GNU Lesser General Public
22 * License along with FFmpeg; if not, write to the Free Software
23 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
34 #include "swscale_internal.h"
35 #include "libavutil/x86_cpu.h"
36 #include "libavutil/bswap.h"
38 extern const uint8_t dither_4x4_16[4][8];
39 extern const uint8_t dither_8x8_32[8][8];
40 extern const uint8_t dither_8x8_73[8][8];
41 extern const uint8_t dither_8x8_220[8][8];
43 const int32_t ff_yuv2rgb_coeffs[8][4] = {
44 {117504, 138453, 13954, 34903}, /* no sequence_display_extension */
45 {117504, 138453, 13954, 34903}, /* ITU-R Rec. 709 (1990) */
46 {104597, 132201, 25675, 53279}, /* unspecified */
47 {104597, 132201, 25675, 53279}, /* reserved */
48 {104448, 132798, 24759, 53109}, /* FCC */
49 {104597, 132201, 25675, 53279}, /* ITU-R Rec. 624-4 System B, G */
50 {104597, 132201, 25675, 53279}, /* SMPTE 170M */
51 {117579, 136230, 16907, 35559} /* SMPTE 240M (1987) */
54 const int *sws_getCoefficients(int colorspace)
56 if (colorspace > 7 || colorspace < 0)
57 colorspace = SWS_CS_DEFAULT;
58 return ff_yuv2rgb_coeffs[colorspace];
61 #define LOADCHROMA(i) \
64 r = (void *)c->table_rV[V]; \
65 g = (void *)(c->table_gU[U] + c->table_gV[V]); \
66 b = (void *)c->table_bU[U];
68 #define PUTRGB(dst,src,i) \
70 dst[2*i ] = r[Y] + g[Y] + b[Y]; \
72 dst[2*i+1] = r[Y] + g[Y] + b[Y];
74 #define PUTRGB24(dst,src,i) \
76 dst[6*i+0] = r[Y]; dst[6*i+1] = g[Y]; dst[6*i+2] = b[Y]; \
78 dst[6*i+3] = r[Y]; dst[6*i+4] = g[Y]; dst[6*i+5] = b[Y];
80 #define PUTBGR24(dst,src,i) \
82 dst[6*i+0] = b[Y]; dst[6*i+1] = g[Y]; dst[6*i+2] = r[Y]; \
84 dst[6*i+3] = b[Y]; dst[6*i+4] = g[Y]; dst[6*i+5] = r[Y];
86 #define PUTRGBA(dst,ysrc,asrc,i,s) \
88 dst[2*i ] = r[Y] + g[Y] + b[Y] + (asrc[2*i ]<<s); \
90 dst[2*i+1] = r[Y] + g[Y] + b[Y] + (asrc[2*i+1]<<s);
92 #define PUTRGB48(dst,src,i) \
94 dst[12*i+ 0] = dst[12*i+ 1] = r[Y]; \
95 dst[12*i+ 2] = dst[12*i+ 3] = g[Y]; \
96 dst[12*i+ 4] = dst[12*i+ 5] = b[Y]; \
98 dst[12*i+ 6] = dst[12*i+ 7] = r[Y]; \
99 dst[12*i+ 8] = dst[12*i+ 9] = g[Y]; \
100 dst[12*i+10] = dst[12*i+11] = b[Y];
102 #define PUTBGR48(dst,src,i) \
104 dst[12*i+ 0] = dst[12*i+ 1] = b[Y]; \
105 dst[12*i+ 2] = dst[12*i+ 3] = g[Y]; \
106 dst[12*i+ 4] = dst[12*i+ 5] = r[Y]; \
108 dst[12*i+ 6] = dst[12*i+ 7] = b[Y]; \
109 dst[12*i+ 8] = dst[12*i+ 9] = g[Y]; \
110 dst[12*i+10] = dst[12*i+11] = r[Y];
112 #define YUV2RGBFUNC(func_name, dst_type, alpha) \
113 static int func_name(SwsContext *c, const uint8_t* src[], int srcStride[], int srcSliceY, \
114 int srcSliceH, uint8_t* dst[], int dstStride[]) \
118 if (!alpha && c->srcFormat == PIX_FMT_YUV422P) {\
122 for (y=0; y<srcSliceH; y+=2) {\
123 dst_type *dst_1 = (dst_type*)(dst[0] + (y+srcSliceY )*dstStride[0]);\
124 dst_type *dst_2 = (dst_type*)(dst[0] + (y+srcSliceY+1)*dstStride[0]);\
125 dst_type av_unused *r, *b;\
127 const uint8_t *py_1 = src[0] + y*srcStride[0];\
128 const uint8_t *py_2 = py_1 + srcStride[0];\
129 const uint8_t *pu = src[1] + (y>>1)*srcStride[1];\
130 const uint8_t *pv = src[2] + (y>>1)*srcStride[2];\
131 const uint8_t av_unused *pa_1, *pa_2;\
132 unsigned int h_size = c->dstW>>3;\
134 pa_1 = src[3] + y*srcStride[3];\
135 pa_2 = pa_1 + srcStride[3];\
141 #define ENDYUV2RGBLINE(dst_delta)\
150 int av_unused Y, U, V;\
152 #define ENDYUV2RGBFUNC()\
158 #define CLOSEYUV2RGBFUNC(dst_delta)\
159 ENDYUV2RGBLINE(dst_delta)\
162 YUV2RGBFUNC(yuv2rgb_c_48, uint8_t, 0)
164 PUTRGB48(dst_1,py_1,0);
165 PUTRGB48(dst_2,py_2,0);
168 PUTRGB48(dst_2,py_2,1);
169 PUTRGB48(dst_1,py_1,1);
172 PUTRGB48(dst_1,py_1,2);
173 PUTRGB48(dst_2,py_2,2);
176 PUTRGB48(dst_2,py_2,3);
177 PUTRGB48(dst_1,py_1,3);
180 PUTRGB48(dst_1,py_1,0);
181 PUTRGB48(dst_2,py_2,0);
184 PUTRGB48(dst_2,py_2,1);
185 PUTRGB48(dst_1,py_1,1);
188 YUV2RGBFUNC(yuv2rgb_c_bgr48, uint8_t, 0)
190 PUTBGR48(dst_1,py_1,0);
191 PUTBGR48(dst_2,py_2,0);
194 PUTBGR48(dst_2,py_2,1);
195 PUTBGR48(dst_1,py_1,1);
198 PUTBGR48(dst_1,py_1,2);
199 PUTBGR48(dst_2,py_2,2);
202 PUTBGR48(dst_2,py_2,3);
203 PUTBGR48(dst_1,py_1,3);
206 PUTBGR48(dst_1,py_1,0);
207 PUTBGR48(dst_2,py_2,0);
210 PUTBGR48(dst_2,py_2,1);
211 PUTBGR48(dst_1,py_1,1);
214 YUV2RGBFUNC(yuv2rgb_c_32, uint32_t, 0)
216 PUTRGB(dst_1,py_1,0);
217 PUTRGB(dst_2,py_2,0);
220 PUTRGB(dst_2,py_2,1);
221 PUTRGB(dst_1,py_1,1);
224 PUTRGB(dst_1,py_1,2);
225 PUTRGB(dst_2,py_2,2);
228 PUTRGB(dst_2,py_2,3);
229 PUTRGB(dst_1,py_1,3);
232 PUTRGB(dst_1,py_1,0);
233 PUTRGB(dst_2,py_2,0);
236 PUTRGB(dst_2,py_2,1);
237 PUTRGB(dst_1,py_1,1);
240 YUV2RGBFUNC(yuva2rgba_c, uint32_t, 1)
242 PUTRGBA(dst_1,py_1,pa_1,0,24);
243 PUTRGBA(dst_2,py_2,pa_2,0,24);
246 PUTRGBA(dst_2,py_2,pa_1,1,24);
247 PUTRGBA(dst_1,py_1,pa_2,1,24);
250 PUTRGBA(dst_1,py_1,pa_1,2,24);
251 PUTRGBA(dst_2,py_2,pa_2,2,24);
254 PUTRGBA(dst_2,py_2,pa_1,3,24);
255 PUTRGBA(dst_1,py_1,pa_2,3,24);
260 PUTRGBA(dst_1,py_1,pa_1,0,24);
261 PUTRGBA(dst_2,py_2,pa_2,0,24);
264 PUTRGBA(dst_2,py_2,pa_1,1,24);
265 PUTRGBA(dst_1,py_1,pa_2,1,24);
268 YUV2RGBFUNC(yuva2argb_c, uint32_t, 1)
270 PUTRGBA(dst_1,py_1,pa_1,0,0);
271 PUTRGBA(dst_2,py_2,pa_2,0,0);
274 PUTRGBA(dst_2,py_2,pa_2,1,0);
275 PUTRGBA(dst_1,py_1,pa_1,1,0);
278 PUTRGBA(dst_1,py_1,pa_1,2,0);
279 PUTRGBA(dst_2,py_2,pa_2,2,0);
282 PUTRGBA(dst_2,py_2,pa_2,3,0);
283 PUTRGBA(dst_1,py_1,pa_1,3,0);
288 PUTRGBA(dst_1,py_1,pa_1,0,0);
289 PUTRGBA(dst_2,py_2,pa_2,0,0);
292 PUTRGBA(dst_2,py_2,pa_2,1,0);
293 PUTRGBA(dst_1,py_1,pa_1,1,0);
296 YUV2RGBFUNC(yuv2rgb_c_24_rgb, uint8_t, 0)
298 PUTRGB24(dst_1,py_1,0);
299 PUTRGB24(dst_2,py_2,0);
302 PUTRGB24(dst_2,py_2,1);
303 PUTRGB24(dst_1,py_1,1);
306 PUTRGB24(dst_1,py_1,2);
307 PUTRGB24(dst_2,py_2,2);
310 PUTRGB24(dst_2,py_2,3);
311 PUTRGB24(dst_1,py_1,3);
314 PUTRGB24(dst_1,py_1,0);
315 PUTRGB24(dst_2,py_2,0);
318 PUTRGB24(dst_2,py_2,1);
319 PUTRGB24(dst_1,py_1,1);
322 // only trivial mods from yuv2rgb_c_24_rgb
323 YUV2RGBFUNC(yuv2rgb_c_24_bgr, uint8_t, 0)
325 PUTBGR24(dst_1,py_1,0);
326 PUTBGR24(dst_2,py_2,0);
329 PUTBGR24(dst_2,py_2,1);
330 PUTBGR24(dst_1,py_1,1);
333 PUTBGR24(dst_1,py_1,2);
334 PUTBGR24(dst_2,py_2,2);
337 PUTBGR24(dst_2,py_2,3);
338 PUTBGR24(dst_1,py_1,3);
341 PUTBGR24(dst_1,py_1,0);
342 PUTBGR24(dst_2,py_2,0);
345 PUTBGR24(dst_2,py_2,1);
346 PUTBGR24(dst_1,py_1,1);
349 // This is exactly the same code as yuv2rgb_c_32 except for the types of
350 // r, g, b, dst_1, dst_2
351 YUV2RGBFUNC(yuv2rgb_c_16, uint16_t, 0)
353 PUTRGB(dst_1,py_1,0);
354 PUTRGB(dst_2,py_2,0);
357 PUTRGB(dst_2,py_2,1);
358 PUTRGB(dst_1,py_1,1);
361 PUTRGB(dst_1,py_1,2);
362 PUTRGB(dst_2,py_2,2);
365 PUTRGB(dst_2,py_2,3);
366 PUTRGB(dst_1,py_1,3);
369 #if 0 // Currently unused
370 // This is exactly the same code as yuv2rgb_c_32 except for the types of
371 // r, g, b, dst_1, dst_2
372 YUV2RGBFUNC(yuv2rgb_c_8, uint8_t, 0)
374 PUTRGB(dst_1,py_1,0);
375 PUTRGB(dst_2,py_2,0);
378 PUTRGB(dst_2,py_2,1);
379 PUTRGB(dst_1,py_1,1);
382 PUTRGB(dst_1,py_1,2);
383 PUTRGB(dst_2,py_2,2);
386 PUTRGB(dst_2,py_2,3);
387 PUTRGB(dst_1,py_1,3);
391 // r, g, b, dst_1, dst_2
392 YUV2RGBFUNC(yuv2rgb_c_12_ordered_dither, uint16_t, 0)
393 const uint8_t *d16 = dither_4x4_16[y&3];
394 #define PUTRGB12(dst,src,i,o) \
396 dst[2*i] = r[Y+d16[0+o]] + g[Y+d16[0+o]] + b[Y+d16[0+o]]; \
398 dst[2*i+1] = r[Y+d16[1+o]] + g[Y+d16[1+o]] + b[Y+d16[1+o]];
401 PUTRGB12(dst_1,py_1,0,0);
402 PUTRGB12(dst_2,py_2,0,0+8);
405 PUTRGB12(dst_2,py_2,1,2+8);
406 PUTRGB12(dst_1,py_1,1,2);
409 PUTRGB12(dst_1,py_1,2,4);
410 PUTRGB12(dst_2,py_2,2,4+8);
413 PUTRGB12(dst_2,py_2,3,6+8);
414 PUTRGB12(dst_1,py_1,3,6);
417 // r, g, b, dst_1, dst_2
418 YUV2RGBFUNC(yuv2rgb_c_8_ordered_dither, uint8_t, 0)
419 const uint8_t *d32 = dither_8x8_32[y&7];
420 const uint8_t *d64 = dither_8x8_73[y&7];
421 #define PUTRGB8(dst,src,i,o) \
423 dst[2*i] = r[Y+d32[0+o]] + g[Y+d32[0+o]] + b[Y+d64[0+o]]; \
425 dst[2*i+1] = r[Y+d32[1+o]] + g[Y+d32[1+o]] + b[Y+d64[1+o]];
428 PUTRGB8(dst_1,py_1,0,0);
429 PUTRGB8(dst_2,py_2,0,0+8);
432 PUTRGB8(dst_2,py_2,1,2+8);
433 PUTRGB8(dst_1,py_1,1,2);
436 PUTRGB8(dst_1,py_1,2,4);
437 PUTRGB8(dst_2,py_2,2,4+8);
440 PUTRGB8(dst_2,py_2,3,6+8);
441 PUTRGB8(dst_1,py_1,3,6);
444 #if 0 // Currently unused
445 // This is exactly the same code as yuv2rgb_c_32 except for the types of
446 // r, g, b, dst_1, dst_2
447 YUV2RGBFUNC(yuv2rgb_c_4, uint8_t, 0)
449 #define PUTRGB4(dst,src,i) \
451 acc = r[Y] + g[Y] + b[Y]; \
453 acc |= (r[Y] + g[Y] + b[Y])<<4; \
457 PUTRGB4(dst_1,py_1,0);
458 PUTRGB4(dst_2,py_2,0);
461 PUTRGB4(dst_2,py_2,1);
462 PUTRGB4(dst_1,py_1,1);
465 PUTRGB4(dst_1,py_1,2);
466 PUTRGB4(dst_2,py_2,2);
469 PUTRGB4(dst_2,py_2,3);
470 PUTRGB4(dst_1,py_1,3);
474 YUV2RGBFUNC(yuv2rgb_c_4_ordered_dither, uint8_t, 0)
475 const uint8_t *d64 = dither_8x8_73[y&7];
476 const uint8_t *d128 = dither_8x8_220[y&7];
479 #define PUTRGB4D(dst,src,i,o) \
481 acc = r[Y+d128[0+o]] + g[Y+d64[0+o]] + b[Y+d128[0+o]]; \
483 acc |= (r[Y+d128[1+o]] + g[Y+d64[1+o]] + b[Y+d128[1+o]])<<4; \
487 PUTRGB4D(dst_1,py_1,0,0);
488 PUTRGB4D(dst_2,py_2,0,0+8);
491 PUTRGB4D(dst_2,py_2,1,2+8);
492 PUTRGB4D(dst_1,py_1,1,2);
495 PUTRGB4D(dst_1,py_1,2,4);
496 PUTRGB4D(dst_2,py_2,2,4+8);
499 PUTRGB4D(dst_2,py_2,3,6+8);
500 PUTRGB4D(dst_1,py_1,3,6);
503 #if 0 // Currently unused
504 // This is exactly the same code as yuv2rgb_c_32 except for the types of
505 // r, g, b, dst_1, dst_2
506 YUV2RGBFUNC(yuv2rgb_c_4b, uint8_t, 0)
508 PUTRGB(dst_1,py_1,0);
509 PUTRGB(dst_2,py_2,0);
512 PUTRGB(dst_2,py_2,1);
513 PUTRGB(dst_1,py_1,1);
516 PUTRGB(dst_1,py_1,2);
517 PUTRGB(dst_2,py_2,2);
520 PUTRGB(dst_2,py_2,3);
521 PUTRGB(dst_1,py_1,3);
525 YUV2RGBFUNC(yuv2rgb_c_4b_ordered_dither, uint8_t, 0)
526 const uint8_t *d64 = dither_8x8_73[y&7];
527 const uint8_t *d128 = dither_8x8_220[y&7];
529 #define PUTRGB4DB(dst,src,i,o) \
531 dst[2*i] = r[Y+d128[0+o]] + g[Y+d64[0+o]] + b[Y+d128[0+o]]; \
533 dst[2*i+1] = r[Y+d128[1+o]] + g[Y+d64[1+o]] + b[Y+d128[1+o]];
536 PUTRGB4DB(dst_1,py_1,0,0);
537 PUTRGB4DB(dst_2,py_2,0,0+8);
540 PUTRGB4DB(dst_2,py_2,1,2+8);
541 PUTRGB4DB(dst_1,py_1,1,2);
544 PUTRGB4DB(dst_1,py_1,2,4);
545 PUTRGB4DB(dst_2,py_2,2,4+8);
548 PUTRGB4DB(dst_2,py_2,3,6+8);
549 PUTRGB4DB(dst_1,py_1,3,6);
552 YUV2RGBFUNC(yuv2rgb_c_1_ordered_dither, uint8_t, 0)
553 const uint8_t *d128 = dither_8x8_220[y&7];
554 char out_1 = 0, out_2 = 0;
555 g= c->table_gU[128] + c->table_gV[128];
557 #define PUTRGB1(out,src,i,o) \
559 out+= out + g[Y+d128[0+o]]; \
561 out+= out + g[Y+d128[1+o]];
563 PUTRGB1(out_1,py_1,0,0);
564 PUTRGB1(out_2,py_2,0,0+8);
566 PUTRGB1(out_2,py_2,1,2+8);
567 PUTRGB1(out_1,py_1,1,2);
569 PUTRGB1(out_1,py_1,2,4);
570 PUTRGB1(out_2,py_2,2,4+8);
572 PUTRGB1(out_2,py_2,3,6+8);
573 PUTRGB1(out_1,py_1,3,6);
579 SwsFunc ff_yuv2rgb_get_func_ptr(SwsContext *c)
583 t = ff_yuv2rgb_init_mmx(c);
586 t = ff_yuv2rgb_init_vis(c);
589 t = ff_yuv2rgb_init_mlib(c);
592 if (c->flags & SWS_CPU_CAPS_ALTIVEC)
593 t = ff_yuv2rgb_init_altivec(c);
597 if (c->flags & SWS_CPU_CAPS_BFIN)
598 t = ff_yuv2rgb_get_func_ptr_bfin(c);
604 av_log(c, AV_LOG_WARNING, "No accelerated colorspace conversion found from %s to %s.\n", sws_format_name(c->srcFormat), sws_format_name(c->dstFormat));
606 switch (c->dstFormat) {
607 case PIX_FMT_BGR48BE:
608 case PIX_FMT_BGR48LE: return yuv2rgb_c_bgr48;
609 case PIX_FMT_RGB48BE:
610 case PIX_FMT_RGB48LE: return yuv2rgb_c_48;
612 case PIX_FMT_ABGR: if (CONFIG_SWSCALE_ALPHA && c->srcFormat == PIX_FMT_YUVA420P) return yuva2argb_c;
614 case PIX_FMT_BGRA: return (CONFIG_SWSCALE_ALPHA && c->srcFormat == PIX_FMT_YUVA420P) ? yuva2rgba_c : yuv2rgb_c_32;
615 case PIX_FMT_RGB24: return yuv2rgb_c_24_rgb;
616 case PIX_FMT_BGR24: return yuv2rgb_c_24_bgr;
620 case PIX_FMT_BGR555: return yuv2rgb_c_16;
622 case PIX_FMT_BGR444: return yuv2rgb_c_12_ordered_dither;
624 case PIX_FMT_BGR8: return yuv2rgb_c_8_ordered_dither;
626 case PIX_FMT_BGR4: return yuv2rgb_c_4_ordered_dither;
627 case PIX_FMT_RGB4_BYTE:
628 case PIX_FMT_BGR4_BYTE: return yuv2rgb_c_4b_ordered_dither;
629 case PIX_FMT_MONOBLACK: return yuv2rgb_c_1_ordered_dither;
636 static void fill_table(uint8_t* table[256], const int elemsize, const int inc, void *y_tab)
640 uint8_t *y_table = y_tab;
642 y_table -= elemsize * (inc >> 9);
644 for (i = 0; i < 256; i++) {
645 table[i] = y_table + elemsize * (cb >> 16);
650 static void fill_gv_table(int table[256], const int elemsize, const int inc)
654 int off = -(inc >> 9);
656 for (i = 0; i < 256; i++) {
657 table[i] = elemsize * (off + (cb >> 16));
662 static uint16_t roundToInt16(int64_t f)
664 int r= (f + (1<<15))>>16;
665 if (r<-0x7FFF) return 0x8000;
666 else if (r> 0x7FFF) return 0x7FFF;
670 av_cold int ff_yuv2rgb_c_init_tables(SwsContext *c, const int inv_table[4], int fullRange,
671 int brightness, int contrast, int saturation)
673 const int isRgb = c->dstFormat==PIX_FMT_RGB32
674 || c->dstFormat==PIX_FMT_RGB32_1
675 || c->dstFormat==PIX_FMT_BGR24
676 || c->dstFormat==PIX_FMT_RGB565BE
677 || c->dstFormat==PIX_FMT_RGB565LE
678 || c->dstFormat==PIX_FMT_RGB555BE
679 || c->dstFormat==PIX_FMT_RGB555LE
680 || c->dstFormat==PIX_FMT_RGB444BE
681 || c->dstFormat==PIX_FMT_RGB444LE
682 || c->dstFormat==PIX_FMT_RGB8
683 || c->dstFormat==PIX_FMT_RGB4
684 || c->dstFormat==PIX_FMT_RGB4_BYTE
685 || c->dstFormat==PIX_FMT_MONOBLACK;
686 const int isNotNe = c->dstFormat==PIX_FMT_NE(RGB565LE,RGB565BE)
687 || c->dstFormat==PIX_FMT_NE(RGB555LE,RGB555BE)
688 || c->dstFormat==PIX_FMT_NE(RGB444LE,RGB444BE)
689 || c->dstFormat==PIX_FMT_NE(BGR565LE,BGR565BE)
690 || c->dstFormat==PIX_FMT_NE(BGR555LE,BGR555BE)
691 || c->dstFormat==PIX_FMT_NE(BGR444LE,BGR444BE);
692 const int bpp = c->dstFormatBpp;
696 int i, base, rbase, gbase, bbase, abase, needAlpha;
697 const int yoffs = fullRange ? 384 : 326;
699 int64_t crv = inv_table[0];
700 int64_t cbu = inv_table[1];
701 int64_t cgu = -inv_table[2];
702 int64_t cgv = -inv_table[3];
712 crv = (crv*224) / 255;
713 cbu = (cbu*224) / 255;
714 cgu = (cgu*224) / 255;
715 cgv = (cgv*224) / 255;
718 cy = (cy *contrast ) >> 16;
719 crv = (crv*contrast * saturation) >> 32;
720 cbu = (cbu*contrast * saturation) >> 32;
721 cgu = (cgu*contrast * saturation) >> 32;
722 cgv = (cgv*contrast * saturation) >> 32;
723 oy -= 256*brightness;
725 c->uOffset= 0x0400040004000400LL;
726 c->vOffset= 0x0400040004000400LL;
727 c->yCoeff= roundToInt16(cy *8192) * 0x0001000100010001ULL;
728 c->vrCoeff= roundToInt16(crv*8192) * 0x0001000100010001ULL;
729 c->ubCoeff= roundToInt16(cbu*8192) * 0x0001000100010001ULL;
730 c->vgCoeff= roundToInt16(cgv*8192) * 0x0001000100010001ULL;
731 c->ugCoeff= roundToInt16(cgu*8192) * 0x0001000100010001ULL;
732 c->yOffset= roundToInt16(oy * 8) * 0x0001000100010001ULL;
734 c->yuv2rgb_y_coeff = (int16_t)roundToInt16(cy <<13);
735 c->yuv2rgb_y_offset = (int16_t)roundToInt16(oy << 9);
736 c->yuv2rgb_v2r_coeff= (int16_t)roundToInt16(crv<<13);
737 c->yuv2rgb_v2g_coeff= (int16_t)roundToInt16(cgv<<13);
738 c->yuv2rgb_u2g_coeff= (int16_t)roundToInt16(cgu<<13);
739 c->yuv2rgb_u2b_coeff= (int16_t)roundToInt16(cbu<<13);
741 //scale coefficients by cy
742 crv = ((crv << 16) + 0x8000) / cy;
743 cbu = ((cbu << 16) + 0x8000) / cy;
744 cgu = ((cgu << 16) + 0x8000) / cy;
745 cgv = ((cgv << 16) + 0x8000) / cy;
747 av_free(c->yuvTable);
751 c->yuvTable = av_malloc(1024);
752 y_table = c->yuvTable;
753 yb = -(384<<16) - oy;
754 for (i = 0; i < 1024-110; i++) {
755 y_table[i+110] = av_clip_uint8((yb + 0x8000) >> 16) >> 7;
758 fill_table(c->table_gU, 1, cgu, y_table + yoffs);
759 fill_gv_table(c->table_gV, 1, cgv);
763 rbase = isRgb ? 3 : 0;
765 bbase = isRgb ? 0 : 3;
766 c->yuvTable = av_malloc(1024*3);
767 y_table = c->yuvTable;
768 yb = -(384<<16) - oy;
769 for (i = 0; i < 1024-110; i++) {
770 int yval = av_clip_uint8((yb + 0x8000) >> 16);
771 y_table[i+110 ] = (yval >> 7) << rbase;
772 y_table[i+ 37+1024] = ((yval + 43) / 85) << gbase;
773 y_table[i+110+2048] = (yval >> 7) << bbase;
776 fill_table(c->table_rV, 1, crv, y_table + yoffs);
777 fill_table(c->table_gU, 1, cgu, y_table + yoffs + 1024);
778 fill_table(c->table_bU, 1, cbu, y_table + yoffs + 2048);
779 fill_gv_table(c->table_gV, 1, cgv);
782 rbase = isRgb ? 5 : 0;
783 gbase = isRgb ? 2 : 3;
784 bbase = isRgb ? 0 : 6;
785 c->yuvTable = av_malloc(1024*3);
786 y_table = c->yuvTable;
787 yb = -(384<<16) - oy;
788 for (i = 0; i < 1024-38; i++) {
789 int yval = av_clip_uint8((yb + 0x8000) >> 16);
790 y_table[i+16 ] = ((yval + 18) / 36) << rbase;
791 y_table[i+16+1024] = ((yval + 18) / 36) << gbase;
792 y_table[i+37+2048] = ((yval + 43) / 85) << bbase;
795 fill_table(c->table_rV, 1, crv, y_table + yoffs);
796 fill_table(c->table_gU, 1, cgu, y_table + yoffs + 1024);
797 fill_table(c->table_bU, 1, cbu, y_table + yoffs + 2048);
798 fill_gv_table(c->table_gV, 1, cgv);
801 rbase = isRgb ? 8 : 0;
803 bbase = isRgb ? 0 : 8;
804 c->yuvTable = av_malloc(1024*3*2);
805 y_table16 = c->yuvTable;
806 yb = -(384<<16) - oy;
807 for (i = 0; i < 1024; i++) {
808 uint8_t yval = av_clip_uint8((yb + 0x8000) >> 16);
809 y_table16[i ] = (yval >> 4) << rbase;
810 y_table16[i+1024] = (yval >> 4) << gbase;
811 y_table16[i+2048] = (yval >> 4) << bbase;
815 for (i = 0; i < 1024*3; i++)
816 y_table16[i] = av_bswap16(y_table16[i]);
817 fill_table(c->table_rV, 2, crv, y_table16 + yoffs);
818 fill_table(c->table_gU, 2, cgu, y_table16 + yoffs + 1024);
819 fill_table(c->table_bU, 2, cbu, y_table16 + yoffs + 2048);
820 fill_gv_table(c->table_gV, 2, cgv);
824 rbase = isRgb ? bpp - 5 : 0;
826 bbase = isRgb ? 0 : (bpp - 5);
827 c->yuvTable = av_malloc(1024*3*2);
828 y_table16 = c->yuvTable;
829 yb = -(384<<16) - oy;
830 for (i = 0; i < 1024; i++) {
831 uint8_t yval = av_clip_uint8((yb + 0x8000) >> 16);
832 y_table16[i ] = (yval >> 3) << rbase;
833 y_table16[i+1024] = (yval >> (18 - bpp)) << gbase;
834 y_table16[i+2048] = (yval >> 3) << bbase;
838 for (i = 0; i < 1024*3; i++)
839 y_table16[i] = av_bswap16(y_table16[i]);
840 fill_table(c->table_rV, 2, crv, y_table16 + yoffs);
841 fill_table(c->table_gU, 2, cgu, y_table16 + yoffs + 1024);
842 fill_table(c->table_bU, 2, cbu, y_table16 + yoffs + 2048);
843 fill_gv_table(c->table_gV, 2, cgv);
847 c->yuvTable = av_malloc(1024);
848 y_table = c->yuvTable;
849 yb = -(384<<16) - oy;
850 for (i = 0; i < 1024; i++) {
851 y_table[i] = av_clip_uint8((yb + 0x8000) >> 16);
854 fill_table(c->table_rV, 1, crv, y_table + yoffs);
855 fill_table(c->table_gU, 1, cgu, y_table + yoffs);
856 fill_table(c->table_bU, 1, cbu, y_table + yoffs);
857 fill_gv_table(c->table_gV, 1, cgv);
860 base = (c->dstFormat == PIX_FMT_RGB32_1 || c->dstFormat == PIX_FMT_BGR32_1) ? 8 : 0;
861 rbase = base + (isRgb ? 16 : 0);
863 bbase = base + (isRgb ? 0 : 16);
864 needAlpha = CONFIG_SWSCALE_ALPHA && isALPHA(c->srcFormat);
866 abase = (base + 24) & 31;
867 c->yuvTable = av_malloc(1024*3*4);
868 y_table32 = c->yuvTable;
869 yb = -(384<<16) - oy;
870 for (i = 0; i < 1024; i++) {
871 uint8_t yval = av_clip_uint8((yb + 0x8000) >> 16);
872 y_table32[i ] = (yval << rbase) + (needAlpha ? 0 : (255 << abase));
873 y_table32[i+1024] = yval << gbase;
874 y_table32[i+2048] = yval << bbase;
877 fill_table(c->table_rV, 4, crv, y_table32 + yoffs);
878 fill_table(c->table_gU, 4, cgu, y_table32 + yoffs + 1024);
879 fill_table(c->table_bU, 4, cbu, y_table32 + yoffs + 2048);
880 fill_gv_table(c->table_gV, 4, cgv);
884 av_log(c, AV_LOG_ERROR, "%ibpp not supported by yuv2rgb\n", bpp);