2 * Copyright (C) 2001-2012 Michael Niedermayer <michaelni@gmx.at>
4 * This file is part of FFmpeg.
6 * FFmpeg is free software; you can redistribute it and/or
7 * modify it under the terms of the GNU Lesser General Public
8 * License as published by the Free Software Foundation; either
9 * version 2.1 of the License, or (at your option) any later version.
11 * FFmpeg is distributed in the hope that it will be useful,
12 * but WITHOUT ANY WARRANTY; without even the implied warranty of
13 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
14 * Lesser General Public License for more details.
16 * You should have received a copy of the GNU Lesser General Public
17 * License along with FFmpeg; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
26 #include "libavutil/attributes.h"
27 #include "libavutil/avutil.h"
28 #include "libavutil/avassert.h"
29 #include "libavutil/bswap.h"
30 #include "libavutil/cpu.h"
31 #include "libavutil/intreadwrite.h"
32 #include "libavutil/mathematics.h"
33 #include "libavutil/pixdesc.h"
37 #include "swscale_internal.h"
39 DECLARE_ALIGNED(8, const uint8_t, ff_dither_2x2_4)[][8] = {
40 { 1, 3, 1, 3, 1, 3, 1, 3, },
41 { 2, 0, 2, 0, 2, 0, 2, 0, },
42 { 1, 3, 1, 3, 1, 3, 1, 3, },
45 DECLARE_ALIGNED(8, const uint8_t, ff_dither_2x2_8)[][8] = {
46 { 6, 2, 6, 2, 6, 2, 6, 2, },
47 { 0, 4, 0, 4, 0, 4, 0, 4, },
48 { 6, 2, 6, 2, 6, 2, 6, 2, },
51 DECLARE_ALIGNED(8, const uint8_t, ff_dither_4x4_16)[][8] = {
52 { 8, 4, 11, 7, 8, 4, 11, 7, },
53 { 2, 14, 1, 13, 2, 14, 1, 13, },
54 { 10, 6, 9, 5, 10, 6, 9, 5, },
55 { 0, 12, 3, 15, 0, 12, 3, 15, },
56 { 8, 4, 11, 7, 8, 4, 11, 7, },
59 DECLARE_ALIGNED(8, const uint8_t, ff_dither_8x8_32)[][8] = {
60 { 17, 9, 23, 15, 16, 8, 22, 14, },
61 { 5, 29, 3, 27, 4, 28, 2, 26, },
62 { 21, 13, 19, 11, 20, 12, 18, 10, },
63 { 0, 24, 6, 30, 1, 25, 7, 31, },
64 { 16, 8, 22, 14, 17, 9, 23, 15, },
65 { 4, 28, 2, 26, 5, 29, 3, 27, },
66 { 20, 12, 18, 10, 21, 13, 19, 11, },
67 { 1, 25, 7, 31, 0, 24, 6, 30, },
68 { 17, 9, 23, 15, 16, 8, 22, 14, },
71 DECLARE_ALIGNED(8, const uint8_t, ff_dither_8x8_73)[][8] = {
72 { 0, 55, 14, 68, 3, 58, 17, 72, },
73 { 37, 18, 50, 32, 40, 22, 54, 35, },
74 { 9, 64, 5, 59, 13, 67, 8, 63, },
75 { 46, 27, 41, 23, 49, 31, 44, 26, },
76 { 2, 57, 16, 71, 1, 56, 15, 70, },
77 { 39, 21, 52, 34, 38, 19, 51, 33, },
78 { 11, 66, 7, 62, 10, 65, 6, 60, },
79 { 48, 30, 43, 25, 47, 29, 42, 24, },
80 { 0, 55, 14, 68, 3, 58, 17, 72, },
84 DECLARE_ALIGNED(8, const uint8_t, ff_dither_8x8_220)[][8] = {
85 {117, 62, 158, 103, 113, 58, 155, 100, },
86 { 34, 199, 21, 186, 31, 196, 17, 182, },
87 {144, 89, 131, 76, 141, 86, 127, 72, },
88 { 0, 165, 41, 206, 10, 175, 52, 217, },
89 {110, 55, 151, 96, 120, 65, 162, 107, },
90 { 28, 193, 14, 179, 38, 203, 24, 189, },
91 {138, 83, 124, 69, 148, 93, 134, 79, },
92 { 7, 172, 48, 213, 3, 168, 45, 210, },
93 {117, 62, 158, 103, 113, 58, 155, 100, },
96 // tries to correct a gamma of 1.5
97 DECLARE_ALIGNED(8, const uint8_t, ff_dither_8x8_220)[][8] = {
98 { 0, 143, 18, 200, 2, 156, 25, 215, },
99 { 78, 28, 125, 64, 89, 36, 138, 74, },
100 { 10, 180, 3, 161, 16, 195, 8, 175, },
101 {109, 51, 93, 38, 121, 60, 105, 47, },
102 { 1, 152, 23, 210, 0, 147, 20, 205, },
103 { 85, 33, 134, 71, 81, 30, 130, 67, },
104 { 14, 190, 6, 171, 12, 185, 5, 166, },
105 {117, 57, 101, 44, 113, 54, 97, 41, },
106 { 0, 143, 18, 200, 2, 156, 25, 215, },
109 // tries to correct a gamma of 2.0
110 DECLARE_ALIGNED(8, const uint8_t, ff_dither_8x8_220)[][8] = {
111 { 0, 124, 8, 193, 0, 140, 12, 213, },
112 { 55, 14, 104, 42, 66, 19, 119, 52, },
113 { 3, 168, 1, 145, 6, 187, 3, 162, },
114 { 86, 31, 70, 21, 99, 39, 82, 28, },
115 { 0, 134, 11, 206, 0, 129, 9, 200, },
116 { 62, 17, 114, 48, 58, 16, 109, 45, },
117 { 5, 181, 2, 157, 4, 175, 1, 151, },
118 { 95, 36, 78, 26, 90, 34, 74, 24, },
119 { 0, 124, 8, 193, 0, 140, 12, 213, },
122 // tries to correct a gamma of 2.5
123 DECLARE_ALIGNED(8, const uint8_t, ff_dither_8x8_220)[][8] = {
124 { 0, 107, 3, 187, 0, 125, 6, 212, },
125 { 39, 7, 86, 28, 49, 11, 102, 36, },
126 { 1, 158, 0, 131, 3, 180, 1, 151, },
127 { 68, 19, 52, 12, 81, 25, 64, 17, },
128 { 0, 119, 5, 203, 0, 113, 4, 195, },
129 { 45, 9, 96, 33, 42, 8, 91, 30, },
130 { 2, 172, 1, 144, 2, 165, 0, 137, },
131 { 77, 23, 60, 15, 72, 21, 56, 14, },
132 { 0, 107, 3, 187, 0, 125, 6, 212, },
136 #define output_pixel(pos, val, bias, signedness) \
138 AV_WB16(pos, bias + av_clip_ ## signedness ## 16(val >> shift)); \
140 AV_WL16(pos, bias + av_clip_ ## signedness ## 16(val >> shift)); \
143 static av_always_inline void
144 yuv2plane1_16_c_template(const int32_t *src, uint16_t *dest, int dstW,
145 int big_endian, int output_bits)
149 av_assert0(output_bits == 16);
151 for (i = 0; i < dstW; i++) {
152 int val = src[i] + (1 << (shift - 1));
153 output_pixel(&dest[i], val, 0, uint);
157 static av_always_inline void
158 yuv2planeX_16_c_template(const int16_t *filter, int filterSize,
159 const int32_t **src, uint16_t *dest, int dstW,
160 int big_endian, int output_bits)
164 av_assert0(output_bits == 16);
166 for (i = 0; i < dstW; i++) {
167 int val = 1 << (shift - 1);
170 /* range of val is [0,0x7FFFFFFF], so 31 bits, but with lanczos/spline
171 * filters (or anything with negative coeffs, the range can be slightly
172 * wider in both directions. To account for this overflow, we subtract
173 * a constant so it always fits in the signed range (assuming a
174 * reasonable filterSize), and re-add that at the end. */
176 for (j = 0; j < filterSize; j++)
177 val += src[j][i] * (unsigned)filter[j];
179 output_pixel(&dest[i], val, 0x8000, int);
185 #define output_pixel(pos, val) \
187 AV_WB16(pos, av_clip_uintp2(val >> shift, output_bits)); \
189 AV_WL16(pos, av_clip_uintp2(val >> shift, output_bits)); \
192 static av_always_inline void
193 yuv2plane1_10_c_template(const int16_t *src, uint16_t *dest, int dstW,
194 int big_endian, int output_bits)
197 int shift = 15 - output_bits;
199 for (i = 0; i < dstW; i++) {
200 int val = src[i] + (1 << (shift - 1));
201 output_pixel(&dest[i], val);
205 static av_always_inline void
206 yuv2planeX_10_c_template(const int16_t *filter, int filterSize,
207 const int16_t **src, uint16_t *dest, int dstW,
208 int big_endian, int output_bits)
211 int shift = 11 + 16 - output_bits;
213 for (i = 0; i < dstW; i++) {
214 int val = 1 << (shift - 1);
217 for (j = 0; j < filterSize; j++)
218 val += src[j][i] * filter[j];
220 output_pixel(&dest[i], val);
226 #define yuv2NBPS(bits, BE_LE, is_be, template_size, typeX_t) \
227 static void yuv2plane1_ ## bits ## BE_LE ## _c(const int16_t *src, \
228 uint8_t *dest, int dstW, \
229 const uint8_t *dither, int offset)\
231 yuv2plane1_ ## template_size ## _c_template((const typeX_t *) src, \
232 (uint16_t *) dest, dstW, is_be, bits); \
234 static void yuv2planeX_ ## bits ## BE_LE ## _c(const int16_t *filter, int filterSize, \
235 const int16_t **src, uint8_t *dest, int dstW, \
236 const uint8_t *dither, int offset)\
238 yuv2planeX_## template_size ## _c_template(filter, \
239 filterSize, (const typeX_t **) src, \
240 (uint16_t *) dest, dstW, is_be, bits); \
242 yuv2NBPS( 9, BE, 1, 10, int16_t)
243 yuv2NBPS( 9, LE, 0, 10, int16_t)
244 yuv2NBPS(10, BE, 1, 10, int16_t)
245 yuv2NBPS(10, LE, 0, 10, int16_t)
246 yuv2NBPS(12, BE, 1, 10, int16_t)
247 yuv2NBPS(12, LE, 0, 10, int16_t)
248 yuv2NBPS(14, BE, 1, 10, int16_t)
249 yuv2NBPS(14, LE, 0, 10, int16_t)
250 yuv2NBPS(16, BE, 1, 16, int32_t)
251 yuv2NBPS(16, LE, 0, 16, int32_t)
253 static void yuv2planeX_8_c(const int16_t *filter, int filterSize,
254 const int16_t **src, uint8_t *dest, int dstW,
255 const uint8_t *dither, int offset)
258 for (i=0; i<dstW; i++) {
259 int val = dither[(i + offset) & 7] << 12;
261 for (j=0; j<filterSize; j++)
262 val += src[j][i] * filter[j];
264 dest[i]= av_clip_uint8(val>>19);
268 static void yuv2plane1_8_c(const int16_t *src, uint8_t *dest, int dstW,
269 const uint8_t *dither, int offset)
272 for (i=0; i<dstW; i++) {
273 int val = (src[i] + dither[(i + offset) & 7]) >> 7;
274 dest[i]= av_clip_uint8(val);
278 static void yuv2nv12cX_c(SwsContext *c, const int16_t *chrFilter, int chrFilterSize,
279 const int16_t **chrUSrc, const int16_t **chrVSrc,
280 uint8_t *dest, int chrDstW)
282 enum AVPixelFormat dstFormat = c->dstFormat;
283 const uint8_t *chrDither = c->chrDither8;
286 if (dstFormat == AV_PIX_FMT_NV12)
287 for (i=0; i<chrDstW; i++) {
288 int u = chrDither[i & 7] << 12;
289 int v = chrDither[(i + 3) & 7] << 12;
291 for (j=0; j<chrFilterSize; j++) {
292 u += chrUSrc[j][i] * chrFilter[j];
293 v += chrVSrc[j][i] * chrFilter[j];
296 dest[2*i]= av_clip_uint8(u>>19);
297 dest[2*i+1]= av_clip_uint8(v>>19);
300 for (i=0; i<chrDstW; i++) {
301 int u = chrDither[i & 7] << 12;
302 int v = chrDither[(i + 3) & 7] << 12;
304 for (j=0; j<chrFilterSize; j++) {
305 u += chrUSrc[j][i] * chrFilter[j];
306 v += chrVSrc[j][i] * chrFilter[j];
309 dest[2*i]= av_clip_uint8(v>>19);
310 dest[2*i+1]= av_clip_uint8(u>>19);
315 #define output_pixel(pos, val) \
317 AV_WB16(pos, av_clip_uintp2(val >> shift, 10) << 6); \
319 AV_WL16(pos, av_clip_uintp2(val >> shift, 10) << 6); \
322 static void yuv2p010l1_c(const int16_t *src,
323 uint16_t *dest, int dstW,
329 for (i = 0; i < dstW; i++) {
330 int val = src[i] + (1 << (shift - 1));
331 output_pixel(&dest[i], val);
335 static void yuv2p010lX_c(const int16_t *filter, int filterSize,
336 const int16_t **src, uint16_t *dest, int dstW,
342 for (i = 0; i < dstW; i++) {
343 int val = 1 << (shift - 1);
345 for (j = 0; j < filterSize; j++)
346 val += src[j][i] * filter[j];
348 output_pixel(&dest[i], val);
352 static void yuv2p010cX_c(SwsContext *c, const int16_t *chrFilter, int chrFilterSize,
353 const int16_t **chrUSrc, const int16_t **chrVSrc,
354 uint8_t *dest8, int chrDstW)
356 uint16_t *dest = (uint16_t*)dest8;
358 int big_endian = c->dstFormat == AV_PIX_FMT_P010BE;
361 for (i = 0; i < chrDstW; i++) {
362 int u = 1 << (shift - 1);
363 int v = 1 << (shift - 1);
365 for (j = 0; j < chrFilterSize; j++) {
366 u += chrUSrc[j][i] * chrFilter[j];
367 v += chrVSrc[j][i] * chrFilter[j];
370 output_pixel(&dest[2*i] , u);
371 output_pixel(&dest[2*i+1], v);
375 static void yuv2p010l1_LE_c(const int16_t *src,
376 uint8_t *dest, int dstW,
377 const uint8_t *dither, int offset)
379 yuv2p010l1_c(src, (uint16_t*)dest, dstW, 0);
382 static void yuv2p010l1_BE_c(const int16_t *src,
383 uint8_t *dest, int dstW,
384 const uint8_t *dither, int offset)
386 yuv2p010l1_c(src, (uint16_t*)dest, dstW, 1);
389 static void yuv2p010lX_LE_c(const int16_t *filter, int filterSize,
390 const int16_t **src, uint8_t *dest, int dstW,
391 const uint8_t *dither, int offset)
393 yuv2p010lX_c(filter, filterSize, src, (uint16_t*)dest, dstW, 0);
396 static void yuv2p010lX_BE_c(const int16_t *filter, int filterSize,
397 const int16_t **src, uint8_t *dest, int dstW,
398 const uint8_t *dither, int offset)
400 yuv2p010lX_c(filter, filterSize, src, (uint16_t*)dest, dstW, 1);
406 #define accumulate_bit(acc, val) \
409 #define output_pixel(pos, acc) \
410 if (target == AV_PIX_FMT_MONOBLACK) { \
416 static av_always_inline void
417 yuv2mono_X_c_template(SwsContext *c, const int16_t *lumFilter,
418 const int16_t **lumSrc, int lumFilterSize,
419 const int16_t *chrFilter, const int16_t **chrUSrc,
420 const int16_t **chrVSrc, int chrFilterSize,
421 const int16_t **alpSrc, uint8_t *dest, int dstW,
422 int y, enum AVPixelFormat target)
424 const uint8_t * const d128 = ff_dither_8x8_220[y&7];
429 for (i = 0; i < dstW; i += 2) {
434 for (j = 0; j < lumFilterSize; j++) {
435 Y1 += lumSrc[j][i] * lumFilter[j];
436 Y2 += lumSrc[j][i+1] * lumFilter[j];
440 if ((Y1 | Y2) & 0x100) {
441 Y1 = av_clip_uint8(Y1);
442 Y2 = av_clip_uint8(Y2);
444 if (c->dither == SWS_DITHER_ED) {
445 Y1 += (7*err + 1*c->dither_error[0][i] + 5*c->dither_error[0][i+1] + 3*c->dither_error[0][i+2] + 8 - 256)>>4;
446 c->dither_error[0][i] = err;
447 acc = 2*acc + (Y1 >= 128);
450 err = Y2 + ((7*Y1 + 1*c->dither_error[0][i+1] + 5*c->dither_error[0][i+2] + 3*c->dither_error[0][i+3] + 8 - 256)>>4);
451 c->dither_error[0][i+1] = Y1;
452 acc = 2*acc + (err >= 128);
455 accumulate_bit(acc, Y1 + d128[(i + 0) & 7]);
456 accumulate_bit(acc, Y2 + d128[(i + 1) & 7]);
459 output_pixel(*dest++, acc);
462 c->dither_error[0][i] = err;
465 output_pixel(*dest, acc);
469 static av_always_inline void
470 yuv2mono_2_c_template(SwsContext *c, const int16_t *buf[2],
471 const int16_t *ubuf[2], const int16_t *vbuf[2],
472 const int16_t *abuf[2], uint8_t *dest, int dstW,
473 int yalpha, int uvalpha, int y,
474 enum AVPixelFormat target)
476 const int16_t *buf0 = buf[0], *buf1 = buf[1];
477 const uint8_t * const d128 = ff_dither_8x8_220[y & 7];
478 int yalpha1 = 4096 - yalpha;
480 av_assert2(yalpha <= 4096U);
482 if (c->dither == SWS_DITHER_ED) {
485 for (i = 0; i < dstW; i +=2) {
488 Y = (buf0[i + 0] * yalpha1 + buf1[i + 0] * yalpha) >> 19;
489 Y += (7*err + 1*c->dither_error[0][i] + 5*c->dither_error[0][i+1] + 3*c->dither_error[0][i+2] + 8 - 256)>>4;
490 c->dither_error[0][i] = err;
491 acc = 2*acc + (Y >= 128);
494 err = (buf0[i + 1] * yalpha1 + buf1[i + 1] * yalpha) >> 19;
495 err += (7*Y + 1*c->dither_error[0][i+1] + 5*c->dither_error[0][i+2] + 3*c->dither_error[0][i+3] + 8 - 256)>>4;
496 c->dither_error[0][i+1] = Y;
497 acc = 2*acc + (err >= 128);
501 output_pixel(*dest++, acc);
503 c->dither_error[0][i] = err;
505 for (i = 0; i < dstW; i += 8) {
508 Y = (buf0[i + 0] * yalpha1 + buf1[i + 0] * yalpha) >> 19;
509 accumulate_bit(acc, Y + d128[0]);
510 Y = (buf0[i + 1] * yalpha1 + buf1[i + 1] * yalpha) >> 19;
511 accumulate_bit(acc, Y + d128[1]);
512 Y = (buf0[i + 2] * yalpha1 + buf1[i + 2] * yalpha) >> 19;
513 accumulate_bit(acc, Y + d128[2]);
514 Y = (buf0[i + 3] * yalpha1 + buf1[i + 3] * yalpha) >> 19;
515 accumulate_bit(acc, Y + d128[3]);
516 Y = (buf0[i + 4] * yalpha1 + buf1[i + 4] * yalpha) >> 19;
517 accumulate_bit(acc, Y + d128[4]);
518 Y = (buf0[i + 5] * yalpha1 + buf1[i + 5] * yalpha) >> 19;
519 accumulate_bit(acc, Y + d128[5]);
520 Y = (buf0[i + 6] * yalpha1 + buf1[i + 6] * yalpha) >> 19;
521 accumulate_bit(acc, Y + d128[6]);
522 Y = (buf0[i + 7] * yalpha1 + buf1[i + 7] * yalpha) >> 19;
523 accumulate_bit(acc, Y + d128[7]);
525 output_pixel(*dest++, acc);
530 static av_always_inline void
531 yuv2mono_1_c_template(SwsContext *c, const int16_t *buf0,
532 const int16_t *ubuf[2], const int16_t *vbuf[2],
533 const int16_t *abuf0, uint8_t *dest, int dstW,
534 int uvalpha, int y, enum AVPixelFormat target)
536 const uint8_t * const d128 = ff_dither_8x8_220[y & 7];
539 if (c->dither == SWS_DITHER_ED) {
542 for (i = 0; i < dstW; i +=2) {
545 Y = ((buf0[i + 0] + 64) >> 7);
546 Y += (7*err + 1*c->dither_error[0][i] + 5*c->dither_error[0][i+1] + 3*c->dither_error[0][i+2] + 8 - 256)>>4;
547 c->dither_error[0][i] = err;
548 acc = 2*acc + (Y >= 128);
551 err = ((buf0[i + 1] + 64) >> 7);
552 err += (7*Y + 1*c->dither_error[0][i+1] + 5*c->dither_error[0][i+2] + 3*c->dither_error[0][i+3] + 8 - 256)>>4;
553 c->dither_error[0][i+1] = Y;
554 acc = 2*acc + (err >= 128);
558 output_pixel(*dest++, acc);
560 c->dither_error[0][i] = err;
562 for (i = 0; i < dstW; i += 8) {
564 accumulate_bit(acc, ((buf0[i + 0] + 64) >> 7) + d128[0]);
565 accumulate_bit(acc, ((buf0[i + 1] + 64) >> 7) + d128[1]);
566 accumulate_bit(acc, ((buf0[i + 2] + 64) >> 7) + d128[2]);
567 accumulate_bit(acc, ((buf0[i + 3] + 64) >> 7) + d128[3]);
568 accumulate_bit(acc, ((buf0[i + 4] + 64) >> 7) + d128[4]);
569 accumulate_bit(acc, ((buf0[i + 5] + 64) >> 7) + d128[5]);
570 accumulate_bit(acc, ((buf0[i + 6] + 64) >> 7) + d128[6]);
571 accumulate_bit(acc, ((buf0[i + 7] + 64) >> 7) + d128[7]);
573 output_pixel(*dest++, acc);
579 #undef accumulate_bit
581 #define YUV2PACKEDWRAPPER(name, base, ext, fmt) \
582 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
583 const int16_t **lumSrc, int lumFilterSize, \
584 const int16_t *chrFilter, const int16_t **chrUSrc, \
585 const int16_t **chrVSrc, int chrFilterSize, \
586 const int16_t **alpSrc, uint8_t *dest, int dstW, \
589 name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
590 chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
591 alpSrc, dest, dstW, y, fmt); \
594 static void name ## ext ## _2_c(SwsContext *c, const int16_t *buf[2], \
595 const int16_t *ubuf[2], const int16_t *vbuf[2], \
596 const int16_t *abuf[2], uint8_t *dest, int dstW, \
597 int yalpha, int uvalpha, int y) \
599 name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
600 dest, dstW, yalpha, uvalpha, y, fmt); \
603 static void name ## ext ## _1_c(SwsContext *c, const int16_t *buf0, \
604 const int16_t *ubuf[2], const int16_t *vbuf[2], \
605 const int16_t *abuf0, uint8_t *dest, int dstW, \
606 int uvalpha, int y) \
608 name ## base ## _1_c_template(c, buf0, ubuf, vbuf, \
609 abuf0, dest, dstW, uvalpha, \
613 YUV2PACKEDWRAPPER(yuv2mono,, white, AV_PIX_FMT_MONOWHITE)
614 YUV2PACKEDWRAPPER(yuv2mono,, black, AV_PIX_FMT_MONOBLACK)
616 #define output_pixels(pos, Y1, U, Y2, V) \
617 if (target == AV_PIX_FMT_YUYV422) { \
618 dest[pos + 0] = Y1; \
620 dest[pos + 2] = Y2; \
622 } else if (target == AV_PIX_FMT_YVYU422) { \
623 dest[pos + 0] = Y1; \
625 dest[pos + 2] = Y2; \
627 } else { /* AV_PIX_FMT_UYVY422 */ \
629 dest[pos + 1] = Y1; \
631 dest[pos + 3] = Y2; \
634 static av_always_inline void
635 yuv2422_X_c_template(SwsContext *c, const int16_t *lumFilter,
636 const int16_t **lumSrc, int lumFilterSize,
637 const int16_t *chrFilter, const int16_t **chrUSrc,
638 const int16_t **chrVSrc, int chrFilterSize,
639 const int16_t **alpSrc, uint8_t *dest, int dstW,
640 int y, enum AVPixelFormat target)
644 for (i = 0; i < ((dstW + 1) >> 1); i++) {
651 for (j = 0; j < lumFilterSize; j++) {
652 Y1 += lumSrc[j][i * 2] * lumFilter[j];
653 Y2 += lumSrc[j][i * 2 + 1] * lumFilter[j];
655 for (j = 0; j < chrFilterSize; j++) {
656 U += chrUSrc[j][i] * chrFilter[j];
657 V += chrVSrc[j][i] * chrFilter[j];
663 if ((Y1 | Y2 | U | V) & 0x100) {
664 Y1 = av_clip_uint8(Y1);
665 Y2 = av_clip_uint8(Y2);
666 U = av_clip_uint8(U);
667 V = av_clip_uint8(V);
669 output_pixels(4*i, Y1, U, Y2, V);
673 static av_always_inline void
674 yuv2422_2_c_template(SwsContext *c, const int16_t *buf[2],
675 const int16_t *ubuf[2], const int16_t *vbuf[2],
676 const int16_t *abuf[2], uint8_t *dest, int dstW,
677 int yalpha, int uvalpha, int y,
678 enum AVPixelFormat target)
680 const int16_t *buf0 = buf[0], *buf1 = buf[1],
681 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
682 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1];
683 int yalpha1 = 4096 - yalpha;
684 int uvalpha1 = 4096 - uvalpha;
686 av_assert2(yalpha <= 4096U);
687 av_assert2(uvalpha <= 4096U);
689 for (i = 0; i < ((dstW + 1) >> 1); i++) {
690 int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 19;
691 int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 19;
692 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha) >> 19;
693 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha) >> 19;
695 if ((Y1 | Y2 | U | V) & 0x100) {
696 Y1 = av_clip_uint8(Y1);
697 Y2 = av_clip_uint8(Y2);
698 U = av_clip_uint8(U);
699 V = av_clip_uint8(V);
702 output_pixels(i * 4, Y1, U, Y2, V);
706 static av_always_inline void
707 yuv2422_1_c_template(SwsContext *c, const int16_t *buf0,
708 const int16_t *ubuf[2], const int16_t *vbuf[2],
709 const int16_t *abuf0, uint8_t *dest, int dstW,
710 int uvalpha, int y, enum AVPixelFormat target)
712 const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
715 if (uvalpha < 2048) {
716 for (i = 0; i < ((dstW + 1) >> 1); i++) {
717 int Y1 = (buf0[i * 2 ]+64) >> 7;
718 int Y2 = (buf0[i * 2 + 1]+64) >> 7;
719 int U = (ubuf0[i] +64) >> 7;
720 int V = (vbuf0[i] +64) >> 7;
722 if ((Y1 | Y2 | U | V) & 0x100) {
723 Y1 = av_clip_uint8(Y1);
724 Y2 = av_clip_uint8(Y2);
725 U = av_clip_uint8(U);
726 V = av_clip_uint8(V);
729 Y1 = av_clip_uint8(Y1);
730 Y2 = av_clip_uint8(Y2);
731 U = av_clip_uint8(U);
732 V = av_clip_uint8(V);
734 output_pixels(i * 4, Y1, U, Y2, V);
737 const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
738 for (i = 0; i < ((dstW + 1) >> 1); i++) {
739 int Y1 = (buf0[i * 2 ] + 64) >> 7;
740 int Y2 = (buf0[i * 2 + 1] + 64) >> 7;
741 int U = (ubuf0[i] + ubuf1[i]+128) >> 8;
742 int V = (vbuf0[i] + vbuf1[i]+128) >> 8;
744 if ((Y1 | Y2 | U | V) & 0x100) {
745 Y1 = av_clip_uint8(Y1);
746 Y2 = av_clip_uint8(Y2);
747 U = av_clip_uint8(U);
748 V = av_clip_uint8(V);
751 Y1 = av_clip_uint8(Y1);
752 Y2 = av_clip_uint8(Y2);
753 U = av_clip_uint8(U);
754 V = av_clip_uint8(V);
756 output_pixels(i * 4, Y1, U, Y2, V);
763 YUV2PACKEDWRAPPER(yuv2, 422, yuyv422, AV_PIX_FMT_YUYV422)
764 YUV2PACKEDWRAPPER(yuv2, 422, yvyu422, AV_PIX_FMT_YVYU422)
765 YUV2PACKEDWRAPPER(yuv2, 422, uyvy422, AV_PIX_FMT_UYVY422)
767 #define R_B ((target == AV_PIX_FMT_RGB48LE || target == AV_PIX_FMT_RGB48BE || target == AV_PIX_FMT_RGBA64LE || target == AV_PIX_FMT_RGBA64BE) ? R : B)
768 #define B_R ((target == AV_PIX_FMT_RGB48LE || target == AV_PIX_FMT_RGB48BE || target == AV_PIX_FMT_RGBA64LE || target == AV_PIX_FMT_RGBA64BE) ? B : R)
769 #define output_pixel(pos, val) \
770 if (isBE(target)) { \
776 static av_always_inline void
777 yuv2rgba64_X_c_template(SwsContext *c, const int16_t *lumFilter,
778 const int32_t **lumSrc, int lumFilterSize,
779 const int16_t *chrFilter, const int32_t **chrUSrc,
780 const int32_t **chrVSrc, int chrFilterSize,
781 const int32_t **alpSrc, uint16_t *dest, int dstW,
782 int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
785 int A1 = 0xffff<<14, A2 = 0xffff<<14;
787 for (i = 0; i < ((dstW + 1) >> 1); i++) {
789 int Y1 = -0x40000000;
790 int Y2 = -0x40000000;
791 int U = -(128 << 23); // 19
792 int V = -(128 << 23);
795 for (j = 0; j < lumFilterSize; j++) {
796 Y1 += lumSrc[j][i * 2] * (unsigned)lumFilter[j];
797 Y2 += lumSrc[j][i * 2 + 1] * (unsigned)lumFilter[j];
799 for (j = 0; j < chrFilterSize; j++) {;
800 U += chrUSrc[j][i] * (unsigned)chrFilter[j];
801 V += chrVSrc[j][i] * (unsigned)chrFilter[j];
807 for (j = 0; j < lumFilterSize; j++) {
808 A1 += alpSrc[j][i * 2] * (unsigned)lumFilter[j];
809 A2 += alpSrc[j][i * 2 + 1] * (unsigned)lumFilter[j];
817 // 8 bits: 12+15=27; 16 bits: 12+19=31
825 // 8 bits: 27 -> 17 bits, 16 bits: 31 - 14 = 17 bits
826 Y1 -= c->yuv2rgb_y_offset;
827 Y2 -= c->yuv2rgb_y_offset;
828 Y1 *= c->yuv2rgb_y_coeff;
829 Y2 *= c->yuv2rgb_y_coeff;
832 // 8 bits: 17 + 13 bits = 30 bits, 16 bits: 17 + 13 bits = 30 bits
834 R = V * c->yuv2rgb_v2r_coeff;
835 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
836 B = U * c->yuv2rgb_u2b_coeff;
838 // 8 bits: 30 - 22 = 8 bits, 16 bits: 30 bits - 14 = 16 bits
839 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
840 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
841 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
843 output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
844 output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
845 output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
846 output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
847 output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
850 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
851 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
852 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
858 static av_always_inline void
859 yuv2rgba64_2_c_template(SwsContext *c, const int32_t *buf[2],
860 const int32_t *ubuf[2], const int32_t *vbuf[2],
861 const int32_t *abuf[2], uint16_t *dest, int dstW,
862 int yalpha, int uvalpha, int y,
863 enum AVPixelFormat target, int hasAlpha, int eightbytes)
865 const int32_t *buf0 = buf[0], *buf1 = buf[1],
866 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
867 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
868 *abuf0 = hasAlpha ? abuf[0] : NULL,
869 *abuf1 = hasAlpha ? abuf[1] : NULL;
870 int yalpha1 = 4096 - yalpha;
871 int uvalpha1 = 4096 - uvalpha;
873 int A1 = 0xffff<<14, A2 = 0xffff<<14;
875 av_assert2(yalpha <= 4096U);
876 av_assert2(uvalpha <= 4096U);
878 for (i = 0; i < ((dstW + 1) >> 1); i++) {
879 int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 14;
880 int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 14;
881 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha - (128 << 23)) >> 14;
882 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha - (128 << 23)) >> 14;
885 Y1 -= c->yuv2rgb_y_offset;
886 Y2 -= c->yuv2rgb_y_offset;
887 Y1 *= c->yuv2rgb_y_coeff;
888 Y2 *= c->yuv2rgb_y_coeff;
892 R = V * c->yuv2rgb_v2r_coeff;
893 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
894 B = U * c->yuv2rgb_u2b_coeff;
897 A1 = (abuf0[i * 2 ] * yalpha1 + abuf1[i * 2 ] * yalpha) >> 1;
898 A2 = (abuf0[i * 2 + 1] * yalpha1 + abuf1[i * 2 + 1] * yalpha) >> 1;
904 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
905 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
906 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
908 output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
909 output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
910 output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
911 output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
912 output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
915 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
916 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
917 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
923 static av_always_inline void
924 yuv2rgba64_1_c_template(SwsContext *c, const int32_t *buf0,
925 const int32_t *ubuf[2], const int32_t *vbuf[2],
926 const int32_t *abuf0, uint16_t *dest, int dstW,
927 int uvalpha, int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
929 const int32_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
931 int A1 = 0xffff<<14, A2= 0xffff<<14;
933 if (uvalpha < 2048) {
934 for (i = 0; i < ((dstW + 1) >> 1); i++) {
935 int Y1 = (buf0[i * 2] ) >> 2;
936 int Y2 = (buf0[i * 2 + 1]) >> 2;
937 int U = (ubuf0[i] - (128 << 11)) >> 2;
938 int V = (vbuf0[i] - (128 << 11)) >> 2;
941 Y1 -= c->yuv2rgb_y_offset;
942 Y2 -= c->yuv2rgb_y_offset;
943 Y1 *= c->yuv2rgb_y_coeff;
944 Y2 *= c->yuv2rgb_y_coeff;
949 A1 = abuf0[i * 2 ] << 11;
950 A2 = abuf0[i * 2 + 1] << 11;
956 R = V * c->yuv2rgb_v2r_coeff;
957 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
958 B = U * c->yuv2rgb_u2b_coeff;
960 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
961 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
962 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
964 output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
965 output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
966 output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
967 output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
968 output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
971 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
972 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
973 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
978 const int32_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
979 int A1 = 0xffff<<14, A2 = 0xffff<<14;
980 for (i = 0; i < ((dstW + 1) >> 1); i++) {
981 int Y1 = (buf0[i * 2] ) >> 2;
982 int Y2 = (buf0[i * 2 + 1]) >> 2;
983 int U = (ubuf0[i] + ubuf1[i] - (128 << 12)) >> 3;
984 int V = (vbuf0[i] + vbuf1[i] - (128 << 12)) >> 3;
987 Y1 -= c->yuv2rgb_y_offset;
988 Y2 -= c->yuv2rgb_y_offset;
989 Y1 *= c->yuv2rgb_y_coeff;
990 Y2 *= c->yuv2rgb_y_coeff;
995 A1 = abuf0[i * 2 ] << 11;
996 A2 = abuf0[i * 2 + 1] << 11;
1002 R = V * c->yuv2rgb_v2r_coeff;
1003 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1004 B = U * c->yuv2rgb_u2b_coeff;
1006 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
1007 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
1008 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
1010 output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
1011 output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
1012 output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
1013 output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
1014 output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
1017 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
1018 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
1019 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
1026 static av_always_inline void
1027 yuv2rgba64_full_X_c_template(SwsContext *c, const int16_t *lumFilter,
1028 const int32_t **lumSrc, int lumFilterSize,
1029 const int16_t *chrFilter, const int32_t **chrUSrc,
1030 const int32_t **chrVSrc, int chrFilterSize,
1031 const int32_t **alpSrc, uint16_t *dest, int dstW,
1032 int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
1037 for (i = 0; i < dstW; i++) {
1039 int Y = -0x40000000;
1040 int U = -(128 << 23); // 19
1041 int V = -(128 << 23);
1044 for (j = 0; j < lumFilterSize; j++) {
1045 Y += lumSrc[j][i] * (unsigned)lumFilter[j];
1047 for (j = 0; j < chrFilterSize; j++) {;
1048 U += chrUSrc[j][i] * (unsigned)chrFilter[j];
1049 V += chrVSrc[j][i] * (unsigned)chrFilter[j];
1054 for (j = 0; j < lumFilterSize; j++) {
1055 A += alpSrc[j][i] * (unsigned)lumFilter[j];
1061 // 8bit: 12+15=27; 16-bit: 12+19=31
1067 // 8bit: 27 -> 17bit, 16bit: 31 - 14 = 17bit
1068 Y -= c->yuv2rgb_y_offset;
1069 Y *= c->yuv2rgb_y_coeff;
1071 // 8bit: 17 + 13bit = 30bit, 16bit: 17 + 13bit = 30bit
1073 R = V * c->yuv2rgb_v2r_coeff;
1074 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1075 B = U * c->yuv2rgb_u2b_coeff;
1077 // 8bit: 30 - 22 = 8bit, 16bit: 30bit - 14 = 16bit
1078 output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1079 output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
1080 output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1082 output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1090 static av_always_inline void
1091 yuv2rgba64_full_2_c_template(SwsContext *c, const int32_t *buf[2],
1092 const int32_t *ubuf[2], const int32_t *vbuf[2],
1093 const int32_t *abuf[2], uint16_t *dest, int dstW,
1094 int yalpha, int uvalpha, int y,
1095 enum AVPixelFormat target, int hasAlpha, int eightbytes)
1097 const int32_t *buf0 = buf[0], *buf1 = buf[1],
1098 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
1099 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
1100 *abuf0 = hasAlpha ? abuf[0] : NULL,
1101 *abuf1 = hasAlpha ? abuf[1] : NULL;
1102 int yalpha1 = 4096 - yalpha;
1103 int uvalpha1 = 4096 - uvalpha;
1107 av_assert2(yalpha <= 4096U);
1108 av_assert2(uvalpha <= 4096U);
1110 for (i = 0; i < dstW; i++) {
1111 int Y = (buf0[i] * yalpha1 + buf1[i] * yalpha) >> 14;
1112 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha - (128 << 23)) >> 14;
1113 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha - (128 << 23)) >> 14;
1116 Y -= c->yuv2rgb_y_offset;
1117 Y *= c->yuv2rgb_y_coeff;
1120 R = V * c->yuv2rgb_v2r_coeff;
1121 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1122 B = U * c->yuv2rgb_u2b_coeff;
1125 A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha) >> 1;
1130 output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1131 output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
1132 output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1134 output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1142 static av_always_inline void
1143 yuv2rgba64_full_1_c_template(SwsContext *c, const int32_t *buf0,
1144 const int32_t *ubuf[2], const int32_t *vbuf[2],
1145 const int32_t *abuf0, uint16_t *dest, int dstW,
1146 int uvalpha, int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
1148 const int32_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1152 if (uvalpha < 2048) {
1153 for (i = 0; i < dstW; i++) {
1154 int Y = (buf0[i]) >> 2;
1155 int U = (ubuf0[i] - (128 << 11)) >> 2;
1156 int V = (vbuf0[i] - (128 << 11)) >> 2;
1159 Y -= c->yuv2rgb_y_offset;
1160 Y *= c->yuv2rgb_y_coeff;
1169 R = V * c->yuv2rgb_v2r_coeff;
1170 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1171 B = U * c->yuv2rgb_u2b_coeff;
1173 output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1174 output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
1175 output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1177 output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1184 const int32_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1186 for (i = 0; i < dstW; i++) {
1187 int Y = (buf0[i] ) >> 2;
1188 int U = (ubuf0[i] + ubuf1[i] - (128 << 12)) >> 3;
1189 int V = (vbuf0[i] + vbuf1[i] - (128 << 12)) >> 3;
1192 Y -= c->yuv2rgb_y_offset;
1193 Y *= c->yuv2rgb_y_coeff;
1202 R = V * c->yuv2rgb_v2r_coeff;
1203 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1204 B = U * c->yuv2rgb_u2b_coeff;
1206 output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1207 output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
1208 output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1210 output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1223 #define YUV2PACKED16WRAPPER(name, base, ext, fmt, hasAlpha, eightbytes) \
1224 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
1225 const int16_t **_lumSrc, int lumFilterSize, \
1226 const int16_t *chrFilter, const int16_t **_chrUSrc, \
1227 const int16_t **_chrVSrc, int chrFilterSize, \
1228 const int16_t **_alpSrc, uint8_t *_dest, int dstW, \
1231 const int32_t **lumSrc = (const int32_t **) _lumSrc, \
1232 **chrUSrc = (const int32_t **) _chrUSrc, \
1233 **chrVSrc = (const int32_t **) _chrVSrc, \
1234 **alpSrc = (const int32_t **) _alpSrc; \
1235 uint16_t *dest = (uint16_t *) _dest; \
1236 name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
1237 chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
1238 alpSrc, dest, dstW, y, fmt, hasAlpha, eightbytes); \
1241 static void name ## ext ## _2_c(SwsContext *c, const int16_t *_buf[2], \
1242 const int16_t *_ubuf[2], const int16_t *_vbuf[2], \
1243 const int16_t *_abuf[2], uint8_t *_dest, int dstW, \
1244 int yalpha, int uvalpha, int y) \
1246 const int32_t **buf = (const int32_t **) _buf, \
1247 **ubuf = (const int32_t **) _ubuf, \
1248 **vbuf = (const int32_t **) _vbuf, \
1249 **abuf = (const int32_t **) _abuf; \
1250 uint16_t *dest = (uint16_t *) _dest; \
1251 name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
1252 dest, dstW, yalpha, uvalpha, y, fmt, hasAlpha, eightbytes); \
1255 static void name ## ext ## _1_c(SwsContext *c, const int16_t *_buf0, \
1256 const int16_t *_ubuf[2], const int16_t *_vbuf[2], \
1257 const int16_t *_abuf0, uint8_t *_dest, int dstW, \
1258 int uvalpha, int y) \
1260 const int32_t *buf0 = (const int32_t *) _buf0, \
1261 **ubuf = (const int32_t **) _ubuf, \
1262 **vbuf = (const int32_t **) _vbuf, \
1263 *abuf0 = (const int32_t *) _abuf0; \
1264 uint16_t *dest = (uint16_t *) _dest; \
1265 name ## base ## _1_c_template(c, buf0, ubuf, vbuf, abuf0, dest, \
1266 dstW, uvalpha, y, fmt, hasAlpha, eightbytes); \
1269 YUV2PACKED16WRAPPER(yuv2, rgba64, rgb48be, AV_PIX_FMT_RGB48BE, 0, 0)
1270 YUV2PACKED16WRAPPER(yuv2, rgba64, rgb48le, AV_PIX_FMT_RGB48LE, 0, 0)
1271 YUV2PACKED16WRAPPER(yuv2, rgba64, bgr48be, AV_PIX_FMT_BGR48BE, 0, 0)
1272 YUV2PACKED16WRAPPER(yuv2, rgba64, bgr48le, AV_PIX_FMT_BGR48LE, 0, 0)
1273 YUV2PACKED16WRAPPER(yuv2, rgba64, rgba64be, AV_PIX_FMT_RGBA64BE, 1, 1)
1274 YUV2PACKED16WRAPPER(yuv2, rgba64, rgba64le, AV_PIX_FMT_RGBA64LE, 1, 1)
1275 YUV2PACKED16WRAPPER(yuv2, rgba64, rgbx64be, AV_PIX_FMT_RGBA64BE, 0, 1)
1276 YUV2PACKED16WRAPPER(yuv2, rgba64, rgbx64le, AV_PIX_FMT_RGBA64LE, 0, 1)
1277 YUV2PACKED16WRAPPER(yuv2, rgba64, bgra64be, AV_PIX_FMT_BGRA64BE, 1, 1)
1278 YUV2PACKED16WRAPPER(yuv2, rgba64, bgra64le, AV_PIX_FMT_BGRA64LE, 1, 1)
1279 YUV2PACKED16WRAPPER(yuv2, rgba64, bgrx64be, AV_PIX_FMT_BGRA64BE, 0, 1)
1280 YUV2PACKED16WRAPPER(yuv2, rgba64, bgrx64le, AV_PIX_FMT_BGRA64LE, 0, 1)
1282 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgb48be_full, AV_PIX_FMT_RGB48BE, 0, 0)
1283 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgb48le_full, AV_PIX_FMT_RGB48LE, 0, 0)
1284 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgr48be_full, AV_PIX_FMT_BGR48BE, 0, 0)
1285 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgr48le_full, AV_PIX_FMT_BGR48LE, 0, 0)
1286 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgba64be_full, AV_PIX_FMT_RGBA64BE, 1, 1)
1287 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgba64le_full, AV_PIX_FMT_RGBA64LE, 1, 1)
1288 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgbx64be_full, AV_PIX_FMT_RGBA64BE, 0, 1)
1289 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgbx64le_full, AV_PIX_FMT_RGBA64LE, 0, 1)
1290 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgra64be_full, AV_PIX_FMT_BGRA64BE, 1, 1)
1291 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgra64le_full, AV_PIX_FMT_BGRA64LE, 1, 1)
1292 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgrx64be_full, AV_PIX_FMT_BGRA64BE, 0, 1)
1293 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgrx64le_full, AV_PIX_FMT_BGRA64LE, 0, 1)
1296 * Write out 2 RGB pixels in the target pixel format. This function takes a
1297 * R/G/B LUT as generated by ff_yuv2rgb_c_init_tables(), which takes care of
1298 * things like endianness conversion and shifting. The caller takes care of
1299 * setting the correct offset in these tables from the chroma (U/V) values.
1300 * This function then uses the luminance (Y1/Y2) values to write out the
1301 * correct RGB values into the destination buffer.
1303 static av_always_inline void
1304 yuv2rgb_write(uint8_t *_dest, int i, int Y1, int Y2,
1305 unsigned A1, unsigned A2,
1306 const void *_r, const void *_g, const void *_b, int y,
1307 enum AVPixelFormat target, int hasAlpha)
1309 if (target == AV_PIX_FMT_ARGB || target == AV_PIX_FMT_RGBA ||
1310 target == AV_PIX_FMT_ABGR || target == AV_PIX_FMT_BGRA) {
1311 uint32_t *dest = (uint32_t *) _dest;
1312 const uint32_t *r = (const uint32_t *) _r;
1313 const uint32_t *g = (const uint32_t *) _g;
1314 const uint32_t *b = (const uint32_t *) _b;
1317 int sh = hasAlpha ? ((target == AV_PIX_FMT_RGB32_1 || target == AV_PIX_FMT_BGR32_1) ? 0 : 24) : 0;
1319 dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1] + (hasAlpha ? A1 << sh : 0);
1320 dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2] + (hasAlpha ? A2 << sh : 0);
1323 int sh = (target == AV_PIX_FMT_RGB32_1 || target == AV_PIX_FMT_BGR32_1) ? 0 : 24;
1325 av_assert2((((r[Y1] + g[Y1] + b[Y1]) >> sh) & 0xFF) == 0);
1326 dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1] + (A1 << sh);
1327 dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2] + (A2 << sh);
1329 #if defined(ASSERT_LEVEL) && ASSERT_LEVEL > 1
1330 int sh = (target == AV_PIX_FMT_RGB32_1 || target == AV_PIX_FMT_BGR32_1) ? 0 : 24;
1332 av_assert2((((r[Y1] + g[Y1] + b[Y1]) >> sh) & 0xFF) == 0xFF);
1334 dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1];
1335 dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2];
1338 } else if (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) {
1339 uint8_t *dest = (uint8_t *) _dest;
1340 const uint8_t *r = (const uint8_t *) _r;
1341 const uint8_t *g = (const uint8_t *) _g;
1342 const uint8_t *b = (const uint8_t *) _b;
1344 #define r_b ((target == AV_PIX_FMT_RGB24) ? r : b)
1345 #define b_r ((target == AV_PIX_FMT_RGB24) ? b : r)
1347 dest[i * 6 + 0] = r_b[Y1];
1348 dest[i * 6 + 1] = g[Y1];
1349 dest[i * 6 + 2] = b_r[Y1];
1350 dest[i * 6 + 3] = r_b[Y2];
1351 dest[i * 6 + 4] = g[Y2];
1352 dest[i * 6 + 5] = b_r[Y2];
1355 } else if (target == AV_PIX_FMT_RGB565 || target == AV_PIX_FMT_BGR565 ||
1356 target == AV_PIX_FMT_RGB555 || target == AV_PIX_FMT_BGR555 ||
1357 target == AV_PIX_FMT_RGB444 || target == AV_PIX_FMT_BGR444) {
1358 uint16_t *dest = (uint16_t *) _dest;
1359 const uint16_t *r = (const uint16_t *) _r;
1360 const uint16_t *g = (const uint16_t *) _g;
1361 const uint16_t *b = (const uint16_t *) _b;
1362 int dr1, dg1, db1, dr2, dg2, db2;
1364 if (target == AV_PIX_FMT_RGB565 || target == AV_PIX_FMT_BGR565) {
1365 dr1 = ff_dither_2x2_8[ y & 1 ][0];
1366 dg1 = ff_dither_2x2_4[ y & 1 ][0];
1367 db1 = ff_dither_2x2_8[(y & 1) ^ 1][0];
1368 dr2 = ff_dither_2x2_8[ y & 1 ][1];
1369 dg2 = ff_dither_2x2_4[ y & 1 ][1];
1370 db2 = ff_dither_2x2_8[(y & 1) ^ 1][1];
1371 } else if (target == AV_PIX_FMT_RGB555 || target == AV_PIX_FMT_BGR555) {
1372 dr1 = ff_dither_2x2_8[ y & 1 ][0];
1373 dg1 = ff_dither_2x2_8[ y & 1 ][1];
1374 db1 = ff_dither_2x2_8[(y & 1) ^ 1][0];
1375 dr2 = ff_dither_2x2_8[ y & 1 ][1];
1376 dg2 = ff_dither_2x2_8[ y & 1 ][0];
1377 db2 = ff_dither_2x2_8[(y & 1) ^ 1][1];
1379 dr1 = ff_dither_4x4_16[ y & 3 ][0];
1380 dg1 = ff_dither_4x4_16[ y & 3 ][1];
1381 db1 = ff_dither_4x4_16[(y & 3) ^ 3][0];
1382 dr2 = ff_dither_4x4_16[ y & 3 ][1];
1383 dg2 = ff_dither_4x4_16[ y & 3 ][0];
1384 db2 = ff_dither_4x4_16[(y & 3) ^ 3][1];
1387 dest[i * 2 + 0] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1];
1388 dest[i * 2 + 1] = r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2];
1389 } else /* 8/4 bits */ {
1390 uint8_t *dest = (uint8_t *) _dest;
1391 const uint8_t *r = (const uint8_t *) _r;
1392 const uint8_t *g = (const uint8_t *) _g;
1393 const uint8_t *b = (const uint8_t *) _b;
1394 int dr1, dg1, db1, dr2, dg2, db2;
1396 if (target == AV_PIX_FMT_RGB8 || target == AV_PIX_FMT_BGR8) {
1397 const uint8_t * const d64 = ff_dither_8x8_73[y & 7];
1398 const uint8_t * const d32 = ff_dither_8x8_32[y & 7];
1399 dr1 = dg1 = d32[(i * 2 + 0) & 7];
1400 db1 = d64[(i * 2 + 0) & 7];
1401 dr2 = dg2 = d32[(i * 2 + 1) & 7];
1402 db2 = d64[(i * 2 + 1) & 7];
1404 const uint8_t * const d64 = ff_dither_8x8_73 [y & 7];
1405 const uint8_t * const d128 = ff_dither_8x8_220[y & 7];
1406 dr1 = db1 = d128[(i * 2 + 0) & 7];
1407 dg1 = d64[(i * 2 + 0) & 7];
1408 dr2 = db2 = d128[(i * 2 + 1) & 7];
1409 dg2 = d64[(i * 2 + 1) & 7];
1412 if (target == AV_PIX_FMT_RGB4 || target == AV_PIX_FMT_BGR4) {
1413 dest[i] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1] +
1414 ((r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2]) << 4);
1416 dest[i * 2 + 0] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1];
1417 dest[i * 2 + 1] = r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2];
1422 static av_always_inline void
1423 yuv2rgb_X_c_template(SwsContext *c, const int16_t *lumFilter,
1424 const int16_t **lumSrc, int lumFilterSize,
1425 const int16_t *chrFilter, const int16_t **chrUSrc,
1426 const int16_t **chrVSrc, int chrFilterSize,
1427 const int16_t **alpSrc, uint8_t *dest, int dstW,
1428 int y, enum AVPixelFormat target, int hasAlpha)
1432 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1438 const void *r, *g, *b;
1440 for (j = 0; j < lumFilterSize; j++) {
1441 Y1 += lumSrc[j][i * 2] * lumFilter[j];
1442 Y2 += lumSrc[j][i * 2 + 1] * lumFilter[j];
1444 for (j = 0; j < chrFilterSize; j++) {
1445 U += chrUSrc[j][i] * chrFilter[j];
1446 V += chrVSrc[j][i] * chrFilter[j];
1455 for (j = 0; j < lumFilterSize; j++) {
1456 A1 += alpSrc[j][i * 2 ] * lumFilter[j];
1457 A2 += alpSrc[j][i * 2 + 1] * lumFilter[j];
1461 if ((A1 | A2) & 0x100) {
1462 A1 = av_clip_uint8(A1);
1463 A2 = av_clip_uint8(A2);
1467 r = c->table_rV[V + YUVRGB_TABLE_HEADROOM];
1468 g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]);
1469 b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1471 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1472 r, g, b, y, target, hasAlpha);
1476 static av_always_inline void
1477 yuv2rgb_2_c_template(SwsContext *c, const int16_t *buf[2],
1478 const int16_t *ubuf[2], const int16_t *vbuf[2],
1479 const int16_t *abuf[2], uint8_t *dest, int dstW,
1480 int yalpha, int uvalpha, int y,
1481 enum AVPixelFormat target, int hasAlpha)
1483 const int16_t *buf0 = buf[0], *buf1 = buf[1],
1484 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
1485 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
1486 *abuf0 = hasAlpha ? abuf[0] : NULL,
1487 *abuf1 = hasAlpha ? abuf[1] : NULL;
1488 int yalpha1 = 4096 - yalpha;
1489 int uvalpha1 = 4096 - uvalpha;
1491 av_assert2(yalpha <= 4096U);
1492 av_assert2(uvalpha <= 4096U);
1494 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1495 int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 19;
1496 int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 19;
1497 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha) >> 19;
1498 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha) >> 19;
1500 const void *r = c->table_rV[V + YUVRGB_TABLE_HEADROOM],
1501 *g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]),
1502 *b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1505 A1 = (abuf0[i * 2 ] * yalpha1 + abuf1[i * 2 ] * yalpha) >> 19;
1506 A2 = (abuf0[i * 2 + 1] * yalpha1 + abuf1[i * 2 + 1] * yalpha) >> 19;
1507 A1 = av_clip_uint8(A1);
1508 A2 = av_clip_uint8(A2);
1511 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1512 r, g, b, y, target, hasAlpha);
1516 static av_always_inline void
1517 yuv2rgb_1_c_template(SwsContext *c, const int16_t *buf0,
1518 const int16_t *ubuf[2], const int16_t *vbuf[2],
1519 const int16_t *abuf0, uint8_t *dest, int dstW,
1520 int uvalpha, int y, enum AVPixelFormat target,
1523 const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1526 if (uvalpha < 2048) {
1527 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1528 int Y1 = (buf0[i * 2 ] + 64) >> 7;
1529 int Y2 = (buf0[i * 2 + 1] + 64) >> 7;
1530 int U = (ubuf0[i] + 64) >> 7;
1531 int V = (vbuf0[i] + 64) >> 7;
1533 const void *r = c->table_rV[V + YUVRGB_TABLE_HEADROOM],
1534 *g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]),
1535 *b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1538 A1 = abuf0[i * 2 ] * 255 + 16384 >> 15;
1539 A2 = abuf0[i * 2 + 1] * 255 + 16384 >> 15;
1540 A1 = av_clip_uint8(A1);
1541 A2 = av_clip_uint8(A2);
1544 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1545 r, g, b, y, target, hasAlpha);
1548 const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1549 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1550 int Y1 = (buf0[i * 2 ] + 64) >> 7;
1551 int Y2 = (buf0[i * 2 + 1] + 64) >> 7;
1552 int U = (ubuf0[i] + ubuf1[i] + 128) >> 8;
1553 int V = (vbuf0[i] + vbuf1[i] + 128) >> 8;
1555 const void *r = c->table_rV[V + YUVRGB_TABLE_HEADROOM],
1556 *g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]),
1557 *b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1560 A1 = (abuf0[i * 2 ] + 64) >> 7;
1561 A2 = (abuf0[i * 2 + 1] + 64) >> 7;
1562 A1 = av_clip_uint8(A1);
1563 A2 = av_clip_uint8(A2);
1566 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1567 r, g, b, y, target, hasAlpha);
1572 #define YUV2RGBWRAPPERX(name, base, ext, fmt, hasAlpha) \
1573 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
1574 const int16_t **lumSrc, int lumFilterSize, \
1575 const int16_t *chrFilter, const int16_t **chrUSrc, \
1576 const int16_t **chrVSrc, int chrFilterSize, \
1577 const int16_t **alpSrc, uint8_t *dest, int dstW, \
1580 name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
1581 chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
1582 alpSrc, dest, dstW, y, fmt, hasAlpha); \
1585 #define YUV2RGBWRAPPERX2(name, base, ext, fmt, hasAlpha) \
1586 YUV2RGBWRAPPERX(name, base, ext, fmt, hasAlpha) \
1587 static void name ## ext ## _2_c(SwsContext *c, const int16_t *buf[2], \
1588 const int16_t *ubuf[2], const int16_t *vbuf[2], \
1589 const int16_t *abuf[2], uint8_t *dest, int dstW, \
1590 int yalpha, int uvalpha, int y) \
1592 name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
1593 dest, dstW, yalpha, uvalpha, y, fmt, hasAlpha); \
1596 #define YUV2RGBWRAPPER(name, base, ext, fmt, hasAlpha) \
1597 YUV2RGBWRAPPERX2(name, base, ext, fmt, hasAlpha) \
1598 static void name ## ext ## _1_c(SwsContext *c, const int16_t *buf0, \
1599 const int16_t *ubuf[2], const int16_t *vbuf[2], \
1600 const int16_t *abuf0, uint8_t *dest, int dstW, \
1601 int uvalpha, int y) \
1603 name ## base ## _1_c_template(c, buf0, ubuf, vbuf, abuf0, dest, \
1604 dstW, uvalpha, y, fmt, hasAlpha); \
1608 YUV2RGBWRAPPER(yuv2rgb,, 32_1, AV_PIX_FMT_RGB32_1, CONFIG_SWSCALE_ALPHA && c->needAlpha)
1609 YUV2RGBWRAPPER(yuv2rgb,, 32, AV_PIX_FMT_RGB32, CONFIG_SWSCALE_ALPHA && c->needAlpha)
1611 #if CONFIG_SWSCALE_ALPHA
1612 YUV2RGBWRAPPER(yuv2rgb,, a32_1, AV_PIX_FMT_RGB32_1, 1)
1613 YUV2RGBWRAPPER(yuv2rgb,, a32, AV_PIX_FMT_RGB32, 1)
1615 YUV2RGBWRAPPER(yuv2rgb,, x32_1, AV_PIX_FMT_RGB32_1, 0)
1616 YUV2RGBWRAPPER(yuv2rgb,, x32, AV_PIX_FMT_RGB32, 0)
1618 YUV2RGBWRAPPER(yuv2, rgb, rgb24, AV_PIX_FMT_RGB24, 0)
1619 YUV2RGBWRAPPER(yuv2, rgb, bgr24, AV_PIX_FMT_BGR24, 0)
1620 YUV2RGBWRAPPER(yuv2rgb,, 16, AV_PIX_FMT_RGB565, 0)
1621 YUV2RGBWRAPPER(yuv2rgb,, 15, AV_PIX_FMT_RGB555, 0)
1622 YUV2RGBWRAPPER(yuv2rgb,, 12, AV_PIX_FMT_RGB444, 0)
1623 YUV2RGBWRAPPER(yuv2rgb,, 8, AV_PIX_FMT_RGB8, 0)
1624 YUV2RGBWRAPPER(yuv2rgb,, 4, AV_PIX_FMT_RGB4, 0)
1625 YUV2RGBWRAPPER(yuv2rgb,, 4b, AV_PIX_FMT_RGB4_BYTE, 0)
1627 static av_always_inline void yuv2rgb_write_full(SwsContext *c,
1628 uint8_t *dest, int i, int Y, int A, int U, int V,
1629 int y, enum AVPixelFormat target, int hasAlpha, int err[4])
1632 int isrgb8 = target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8;
1634 Y -= c->yuv2rgb_y_offset;
1635 Y *= c->yuv2rgb_y_coeff;
1637 R = Y + V*c->yuv2rgb_v2r_coeff;
1638 G = Y + V*c->yuv2rgb_v2g_coeff + U*c->yuv2rgb_u2g_coeff;
1639 B = Y + U*c->yuv2rgb_u2b_coeff;
1640 if ((R | G | B) & 0xC0000000) {
1641 R = av_clip_uintp2(R, 30);
1642 G = av_clip_uintp2(G, 30);
1643 B = av_clip_uintp2(B, 30);
1647 case AV_PIX_FMT_ARGB:
1648 dest[0] = hasAlpha ? A : 255;
1653 case AV_PIX_FMT_RGB24:
1658 case AV_PIX_FMT_RGBA:
1662 dest[3] = hasAlpha ? A : 255;
1664 case AV_PIX_FMT_ABGR:
1665 dest[0] = hasAlpha ? A : 255;
1670 case AV_PIX_FMT_BGR24:
1675 case AV_PIX_FMT_BGRA:
1679 dest[3] = hasAlpha ? A : 255;
1681 case AV_PIX_FMT_BGR4_BYTE:
1682 case AV_PIX_FMT_RGB4_BYTE:
1683 case AV_PIX_FMT_BGR8:
1684 case AV_PIX_FMT_RGB8:
1688 switch (c->dither) {
1690 case SWS_DITHER_AUTO:
1695 R += (7*err[0] + 1*c->dither_error[0][i] + 5*c->dither_error[0][i+1] + 3*c->dither_error[0][i+2])>>4;
1696 G += (7*err[1] + 1*c->dither_error[1][i] + 5*c->dither_error[1][i+1] + 3*c->dither_error[1][i+2])>>4;
1697 B += (7*err[2] + 1*c->dither_error[2][i] + 5*c->dither_error[2][i+1] + 3*c->dither_error[2][i+2])>>4;
1698 c->dither_error[0][i] = err[0];
1699 c->dither_error[1][i] = err[1];
1700 c->dither_error[2][i] = err[2];
1701 r = R >> (isrgb8 ? 5 : 7);
1702 g = G >> (isrgb8 ? 5 : 6);
1703 b = B >> (isrgb8 ? 6 : 7);
1704 r = av_clip(r, 0, isrgb8 ? 7 : 1);
1705 g = av_clip(g, 0, isrgb8 ? 7 : 3);
1706 b = av_clip(b, 0, isrgb8 ? 3 : 1);
1707 err[0] = R - r*(isrgb8 ? 36 : 255);
1708 err[1] = G - g*(isrgb8 ? 36 : 85);
1709 err[2] = B - b*(isrgb8 ? 85 : 255);
1711 case SWS_DITHER_A_DITHER:
1713 /* see http://pippin.gimp.org/a_dither/ for details/origin */
1714 #define A_DITHER(u,v) (((((u)+((v)*236))*119)&0xff))
1715 r = (((R >> 19) + A_DITHER(i,y) -96)>>8);
1716 g = (((G >> 19) + A_DITHER(i + 17,y) - 96)>>8);
1717 b = (((B >> 20) + A_DITHER(i + 17*2,y) -96)>>8);
1718 r = av_clip_uintp2(r, 3);
1719 g = av_clip_uintp2(g, 3);
1720 b = av_clip_uintp2(b, 2);
1722 r = (((R >> 21) + A_DITHER(i,y)-256)>>8);
1723 g = (((G >> 19) + A_DITHER(i + 17,y)-256)>>8);
1724 b = (((B >> 21) + A_DITHER(i + 17*2,y)-256)>>8);
1725 r = av_clip_uintp2(r, 1);
1726 g = av_clip_uintp2(g, 2);
1727 b = av_clip_uintp2(b, 1);
1730 case SWS_DITHER_X_DITHER:
1732 /* see http://pippin.gimp.org/a_dither/ for details/origin */
1733 #define X_DITHER(u,v) (((((u)^((v)*237))*181)&0x1ff)/2)
1734 r = (((R >> 19) + X_DITHER(i,y) - 96)>>8);
1735 g = (((G >> 19) + X_DITHER(i + 17,y) - 96)>>8);
1736 b = (((B >> 20) + X_DITHER(i + 17*2,y) - 96)>>8);
1737 r = av_clip_uintp2(r, 3);
1738 g = av_clip_uintp2(g, 3);
1739 b = av_clip_uintp2(b, 2);
1741 r = (((R >> 21) + X_DITHER(i,y)-256)>>8);
1742 g = (((G >> 19) + X_DITHER(i + 17,y)-256)>>8);
1743 b = (((B >> 21) + X_DITHER(i + 17*2,y)-256)>>8);
1744 r = av_clip_uintp2(r, 1);
1745 g = av_clip_uintp2(g, 2);
1746 b = av_clip_uintp2(b, 1);
1752 if(target == AV_PIX_FMT_BGR4_BYTE) {
1753 dest[0] = r + 2*g + 8*b;
1754 } else if(target == AV_PIX_FMT_RGB4_BYTE) {
1755 dest[0] = b + 2*g + 8*r;
1756 } else if(target == AV_PIX_FMT_BGR8) {
1757 dest[0] = r + 8*g + 64*b;
1758 } else if(target == AV_PIX_FMT_RGB8) {
1759 dest[0] = b + 4*g + 32*r;
1766 static av_always_inline void
1767 yuv2rgb_full_X_c_template(SwsContext *c, const int16_t *lumFilter,
1768 const int16_t **lumSrc, int lumFilterSize,
1769 const int16_t *chrFilter, const int16_t **chrUSrc,
1770 const int16_t **chrVSrc, int chrFilterSize,
1771 const int16_t **alpSrc, uint8_t *dest,
1772 int dstW, int y, enum AVPixelFormat target, int hasAlpha)
1775 int step = (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) ? 3 : 4;
1777 int A = 0; //init to silence warning
1779 if( target == AV_PIX_FMT_BGR4_BYTE || target == AV_PIX_FMT_RGB4_BYTE
1780 || target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8)
1783 for (i = 0; i < dstW; i++) {
1786 int U = (1<<9)-(128 << 19);
1787 int V = (1<<9)-(128 << 19);
1789 for (j = 0; j < lumFilterSize; j++) {
1790 Y += lumSrc[j][i] * lumFilter[j];
1792 for (j = 0; j < chrFilterSize; j++) {
1793 U += chrUSrc[j][i] * chrFilter[j];
1794 V += chrVSrc[j][i] * chrFilter[j];
1801 for (j = 0; j < lumFilterSize; j++) {
1802 A += alpSrc[j][i] * lumFilter[j];
1806 A = av_clip_uint8(A);
1808 yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
1811 c->dither_error[0][i] = err[0];
1812 c->dither_error[1][i] = err[1];
1813 c->dither_error[2][i] = err[2];
1816 static av_always_inline void
1817 yuv2rgb_full_2_c_template(SwsContext *c, const int16_t *buf[2],
1818 const int16_t *ubuf[2], const int16_t *vbuf[2],
1819 const int16_t *abuf[2], uint8_t *dest, int dstW,
1820 int yalpha, int uvalpha, int y,
1821 enum AVPixelFormat target, int hasAlpha)
1823 const int16_t *buf0 = buf[0], *buf1 = buf[1],
1824 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
1825 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
1826 *abuf0 = hasAlpha ? abuf[0] : NULL,
1827 *abuf1 = hasAlpha ? abuf[1] : NULL;
1828 int yalpha1 = 4096 - yalpha;
1829 int uvalpha1 = 4096 - uvalpha;
1831 int step = (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) ? 3 : 4;
1833 int A = 0; // init to silcene warning
1835 av_assert2(yalpha <= 4096U);
1836 av_assert2(uvalpha <= 4096U);
1838 if( target == AV_PIX_FMT_BGR4_BYTE || target == AV_PIX_FMT_RGB4_BYTE
1839 || target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8)
1842 for (i = 0; i < dstW; i++) {
1843 int Y = ( buf0[i] * yalpha1 + buf1[i] * yalpha ) >> 10; //FIXME rounding
1844 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha-(128 << 19)) >> 10;
1845 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha-(128 << 19)) >> 10;
1848 A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha + (1<<18)) >> 19;
1850 A = av_clip_uint8(A);
1853 yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
1856 c->dither_error[0][i] = err[0];
1857 c->dither_error[1][i] = err[1];
1858 c->dither_error[2][i] = err[2];
1861 static av_always_inline void
1862 yuv2rgb_full_1_c_template(SwsContext *c, const int16_t *buf0,
1863 const int16_t *ubuf[2], const int16_t *vbuf[2],
1864 const int16_t *abuf0, uint8_t *dest, int dstW,
1865 int uvalpha, int y, enum AVPixelFormat target,
1868 const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1870 int step = (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) ? 3 : 4;
1873 if( target == AV_PIX_FMT_BGR4_BYTE || target == AV_PIX_FMT_RGB4_BYTE
1874 || target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8)
1877 if (uvalpha < 2048) {
1878 int A = 0; //init to silence warning
1879 for (i = 0; i < dstW; i++) {
1880 int Y = buf0[i] << 2;
1881 int U = (ubuf0[i] - (128<<7)) * 4;
1882 int V = (vbuf0[i] - (128<<7)) * 4;
1885 A = (abuf0[i] + 64) >> 7;
1887 A = av_clip_uint8(A);
1890 yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
1894 const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1895 int A = 0; //init to silence warning
1896 for (i = 0; i < dstW; i++) {
1897 int Y = buf0[i] << 2;
1898 int U = (ubuf0[i] + ubuf1[i] - (128<<8)) << 1;
1899 int V = (vbuf0[i] + vbuf1[i] - (128<<8)) << 1;
1902 A = (abuf0[i] + 64) >> 7;
1904 A = av_clip_uint8(A);
1907 yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
1912 c->dither_error[0][i] = err[0];
1913 c->dither_error[1][i] = err[1];
1914 c->dither_error[2][i] = err[2];
1918 YUV2RGBWRAPPER(yuv2, rgb_full, bgra32_full, AV_PIX_FMT_BGRA, CONFIG_SWSCALE_ALPHA && c->needAlpha)
1919 YUV2RGBWRAPPER(yuv2, rgb_full, abgr32_full, AV_PIX_FMT_ABGR, CONFIG_SWSCALE_ALPHA && c->needAlpha)
1920 YUV2RGBWRAPPER(yuv2, rgb_full, rgba32_full, AV_PIX_FMT_RGBA, CONFIG_SWSCALE_ALPHA && c->needAlpha)
1921 YUV2RGBWRAPPER(yuv2, rgb_full, argb32_full, AV_PIX_FMT_ARGB, CONFIG_SWSCALE_ALPHA && c->needAlpha)
1923 #if CONFIG_SWSCALE_ALPHA
1924 YUV2RGBWRAPPER(yuv2, rgb_full, bgra32_full, AV_PIX_FMT_BGRA, 1)
1925 YUV2RGBWRAPPER(yuv2, rgb_full, abgr32_full, AV_PIX_FMT_ABGR, 1)
1926 YUV2RGBWRAPPER(yuv2, rgb_full, rgba32_full, AV_PIX_FMT_RGBA, 1)
1927 YUV2RGBWRAPPER(yuv2, rgb_full, argb32_full, AV_PIX_FMT_ARGB, 1)
1929 YUV2RGBWRAPPER(yuv2, rgb_full, bgrx32_full, AV_PIX_FMT_BGRA, 0)
1930 YUV2RGBWRAPPER(yuv2, rgb_full, xbgr32_full, AV_PIX_FMT_ABGR, 0)
1931 YUV2RGBWRAPPER(yuv2, rgb_full, rgbx32_full, AV_PIX_FMT_RGBA, 0)
1932 YUV2RGBWRAPPER(yuv2, rgb_full, xrgb32_full, AV_PIX_FMT_ARGB, 0)
1934 YUV2RGBWRAPPER(yuv2, rgb_full, bgr24_full, AV_PIX_FMT_BGR24, 0)
1935 YUV2RGBWRAPPER(yuv2, rgb_full, rgb24_full, AV_PIX_FMT_RGB24, 0)
1937 YUV2RGBWRAPPER(yuv2, rgb_full, bgr4_byte_full, AV_PIX_FMT_BGR4_BYTE, 0)
1938 YUV2RGBWRAPPER(yuv2, rgb_full, rgb4_byte_full, AV_PIX_FMT_RGB4_BYTE, 0)
1939 YUV2RGBWRAPPER(yuv2, rgb_full, bgr8_full, AV_PIX_FMT_BGR8, 0)
1940 YUV2RGBWRAPPER(yuv2, rgb_full, rgb8_full, AV_PIX_FMT_RGB8, 0)
1943 yuv2gbrp_full_X_c(SwsContext *c, const int16_t *lumFilter,
1944 const int16_t **lumSrc, int lumFilterSize,
1945 const int16_t *chrFilter, const int16_t **chrUSrc,
1946 const int16_t **chrVSrc, int chrFilterSize,
1947 const int16_t **alpSrc, uint8_t **dest,
1950 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(c->dstFormat);
1952 int hasAlpha = (desc->flags & AV_PIX_FMT_FLAG_ALPHA) && alpSrc;
1953 uint16_t **dest16 = (uint16_t**)dest;
1954 int SH = 22 + 8 - desc->comp[0].depth;
1955 int A = 0; // init to silence warning
1957 for (i = 0; i < dstW; i++) {
1960 int U = (1 << 9) - (128 << 19);
1961 int V = (1 << 9) - (128 << 19);
1964 for (j = 0; j < lumFilterSize; j++)
1965 Y += lumSrc[j][i] * lumFilter[j];
1967 for (j = 0; j < chrFilterSize; j++) {
1968 U += chrUSrc[j][i] * chrFilter[j];
1969 V += chrVSrc[j][i] * chrFilter[j];
1979 for (j = 0; j < lumFilterSize; j++)
1980 A += alpSrc[j][i] * lumFilter[j];
1983 A = av_clip_uintp2(A, 27);
1986 Y -= c->yuv2rgb_y_offset;
1987 Y *= c->yuv2rgb_y_coeff;
1989 R = Y + V * c->yuv2rgb_v2r_coeff;
1990 G = Y + V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1991 B = Y + U * c->yuv2rgb_u2b_coeff;
1993 if ((R | G | B) & 0xC0000000) {
1994 R = av_clip_uintp2(R, 30);
1995 G = av_clip_uintp2(G, 30);
1996 B = av_clip_uintp2(B, 30);
2000 dest16[0][i] = G >> SH;
2001 dest16[1][i] = B >> SH;
2002 dest16[2][i] = R >> SH;
2004 dest16[3][i] = A >> (SH - 3);
2006 dest[0][i] = G >> 22;
2007 dest[1][i] = B >> 22;
2008 dest[2][i] = R >> 22;
2010 dest[3][i] = A >> 19;
2013 if (SH != 22 && (!isBE(c->dstFormat)) != (!HAVE_BIGENDIAN)) {
2014 for (i = 0; i < dstW; i++) {
2015 dest16[0][i] = av_bswap16(dest16[0][i]);
2016 dest16[1][i] = av_bswap16(dest16[1][i]);
2017 dest16[2][i] = av_bswap16(dest16[2][i]);
2019 dest16[3][i] = av_bswap16(dest16[3][i]);
2025 yuv2gbrp16_full_X_c(SwsContext *c, const int16_t *lumFilter,
2026 const int16_t **lumSrcx, int lumFilterSize,
2027 const int16_t *chrFilter, const int16_t **chrUSrcx,
2028 const int16_t **chrVSrcx, int chrFilterSize,
2029 const int16_t **alpSrc, uint8_t **dest,
2032 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(c->dstFormat);
2034 int hasAlpha = (desc->flags & AV_PIX_FMT_FLAG_ALPHA) && alpSrc;
2035 uint16_t **dest16 = (uint16_t**)dest;
2036 const int32_t **lumSrc = (const int32_t**)lumSrcx;
2037 const int32_t **chrUSrc = (const int32_t**)chrUSrcx;
2038 const int32_t **chrVSrc = (const int32_t**)chrVSrcx;
2039 int A = 0; // init to silence warning
2041 for (i = 0; i < dstW; i++) {
2043 int Y = -0x40000000;
2044 int U = -(128 << 23);
2045 int V = -(128 << 23);
2048 for (j = 0; j < lumFilterSize; j++)
2049 Y += lumSrc[j][i] * (unsigned)lumFilter[j];
2051 for (j = 0; j < chrFilterSize; j++) {
2052 U += chrUSrc[j][i] * (unsigned)chrFilter[j];
2053 V += chrVSrc[j][i] * (unsigned)chrFilter[j];
2064 for (j = 0; j < lumFilterSize; j++)
2065 A += alpSrc[j][i] * lumFilter[j];
2068 A = av_clip_uintp2(A, 27);
2071 Y -= c->yuv2rgb_y_offset;
2072 Y *= c->yuv2rgb_y_coeff;
2074 R = V * c->yuv2rgb_v2r_coeff;
2075 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
2076 B = U * c->yuv2rgb_u2b_coeff;
2078 R = av_clip_uintp2(Y + R, 30);
2079 G = av_clip_uintp2(Y + G, 30);
2080 B = av_clip_uintp2(Y + B, 30);
2082 dest16[0][i] = G >> 14;
2083 dest16[1][i] = B >> 14;
2084 dest16[2][i] = R >> 14;
2086 dest16[3][i] = A >> 11;
2088 if ((!isBE(c->dstFormat)) != (!HAVE_BIGENDIAN)) {
2089 for (i = 0; i < dstW; i++) {
2090 dest16[0][i] = av_bswap16(dest16[0][i]);
2091 dest16[1][i] = av_bswap16(dest16[1][i]);
2092 dest16[2][i] = av_bswap16(dest16[2][i]);
2094 dest16[3][i] = av_bswap16(dest16[3][i]);
2100 yuv2ya8_1_c(SwsContext *c, const int16_t *buf0,
2101 const int16_t *ubuf[2], const int16_t *vbuf[2],
2102 const int16_t *abuf0, uint8_t *dest, int dstW,
2105 int hasAlpha = !!abuf0;
2108 for (i = 0; i < dstW; i++) {
2109 int Y = (buf0[i] + 64) >> 7;
2112 Y = av_clip_uint8(Y);
2115 A = (abuf0[i] + 64) >> 7;
2117 A = av_clip_uint8(A);
2121 dest[i * 2 + 1] = hasAlpha ? A : 255;
2126 yuv2ya8_2_c(SwsContext *c, const int16_t *buf[2],
2127 const int16_t *ubuf[2], const int16_t *vbuf[2],
2128 const int16_t *abuf[2], uint8_t *dest, int dstW,
2129 int yalpha, int uvalpha, int y)
2131 int hasAlpha = abuf && abuf[0] && abuf[1];
2132 const int16_t *buf0 = buf[0], *buf1 = buf[1],
2133 *abuf0 = hasAlpha ? abuf[0] : NULL,
2134 *abuf1 = hasAlpha ? abuf[1] : NULL;
2135 int yalpha1 = 4096 - yalpha;
2138 av_assert2(yalpha <= 4096U);
2140 for (i = 0; i < dstW; i++) {
2141 int Y = (buf0[i] * yalpha1 + buf1[i] * yalpha) >> 19;
2144 Y = av_clip_uint8(Y);
2147 A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha) >> 19;
2148 A = av_clip_uint8(A);
2152 dest[i * 2 + 1] = hasAlpha ? A : 255;
2157 yuv2ya8_X_c(SwsContext *c, const int16_t *lumFilter,
2158 const int16_t **lumSrc, int lumFilterSize,
2159 const int16_t *chrFilter, const int16_t **chrUSrc,
2160 const int16_t **chrVSrc, int chrFilterSize,
2161 const int16_t **alpSrc, uint8_t *dest, int dstW, int y)
2163 int hasAlpha = !!alpSrc;
2166 for (i = 0; i < dstW; i++) {
2168 int Y = 1 << 18, A = 1 << 18;
2170 for (j = 0; j < lumFilterSize; j++)
2171 Y += lumSrc[j][i] * lumFilter[j];
2175 Y = av_clip_uint8(Y);
2178 for (j = 0; j < lumFilterSize; j++)
2179 A += alpSrc[j][i] * lumFilter[j];
2184 A = av_clip_uint8(A);
2188 dest[2 * i + 1] = hasAlpha ? A : 255;
2193 yuv2ayuv64le_X_c(SwsContext *c, const int16_t *lumFilter,
2194 const int16_t **_lumSrc, int lumFilterSize,
2195 const int16_t *chrFilter, const int16_t **_chrUSrc,
2196 const int16_t **_chrVSrc, int chrFilterSize,
2197 const int16_t **_alpSrc, uint8_t *dest, int dstW, int y)
2199 const int32_t **lumSrc = (const int32_t **) _lumSrc,
2200 **chrUSrc = (const int32_t **) _chrUSrc,
2201 **chrVSrc = (const int32_t **) _chrVSrc,
2202 **alpSrc = (const int32_t **) _alpSrc;
2203 int hasAlpha = !!alpSrc;
2206 for (i = 0; i < dstW; i++) {
2207 int Y = 1 << 14, U = 1 << 14;
2208 int V = 1 << 14, A = 1 << 14;
2216 for (j = 0; j < lumFilterSize; j++)
2217 Y += lumSrc[j][i] * (unsigned)lumFilter[j];
2219 for (j = 0; j < chrFilterSize; j++)
2220 U += chrUSrc[j][i] * (unsigned)chrFilter[j];
2222 for (j = 0; j < chrFilterSize; j++)
2223 V += chrVSrc[j][i] * (unsigned)chrFilter[j];
2226 for (j = 0; j < lumFilterSize; j++)
2227 A += alpSrc[j][i] * (unsigned)lumFilter[j];
2229 Y = 0x8000 + av_clip_int16(Y >> 15);
2230 U = 0x8000 + av_clip_int16(U >> 15);
2231 V = 0x8000 + av_clip_int16(V >> 15);
2232 A = 0x8000 + av_clip_int16(A >> 15);
2234 AV_WL16(dest + 8 * i, hasAlpha ? A : 65535);
2235 AV_WL16(dest + 8 * i + 2, Y);
2236 AV_WL16(dest + 8 * i + 4, U);
2237 AV_WL16(dest + 8 * i + 6, V);
2241 av_cold void ff_sws_init_output_funcs(SwsContext *c,
2242 yuv2planar1_fn *yuv2plane1,
2243 yuv2planarX_fn *yuv2planeX,
2244 yuv2interleavedX_fn *yuv2nv12cX,
2245 yuv2packed1_fn *yuv2packed1,
2246 yuv2packed2_fn *yuv2packed2,
2247 yuv2packedX_fn *yuv2packedX,
2248 yuv2anyX_fn *yuv2anyX)
2250 enum AVPixelFormat dstFormat = c->dstFormat;
2251 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(dstFormat);
2253 if (dstFormat == AV_PIX_FMT_P010LE || dstFormat == AV_PIX_FMT_P010BE) {
2254 *yuv2plane1 = isBE(dstFormat) ? yuv2p010l1_BE_c : yuv2p010l1_LE_c;
2255 *yuv2planeX = isBE(dstFormat) ? yuv2p010lX_BE_c : yuv2p010lX_LE_c;
2256 *yuv2nv12cX = yuv2p010cX_c;
2257 } else if (is16BPS(dstFormat)) {
2258 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_16BE_c : yuv2planeX_16LE_c;
2259 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_16BE_c : yuv2plane1_16LE_c;
2260 } else if (isNBPS(dstFormat)) {
2261 if (desc->comp[0].depth == 9) {
2262 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_9BE_c : yuv2planeX_9LE_c;
2263 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_9BE_c : yuv2plane1_9LE_c;
2264 } else if (desc->comp[0].depth == 10) {
2265 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_10BE_c : yuv2planeX_10LE_c;
2266 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_10BE_c : yuv2plane1_10LE_c;
2267 } else if (desc->comp[0].depth == 12) {
2268 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_12BE_c : yuv2planeX_12LE_c;
2269 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_12BE_c : yuv2plane1_12LE_c;
2270 } else if (desc->comp[0].depth == 14) {
2271 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_14BE_c : yuv2planeX_14LE_c;
2272 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_14BE_c : yuv2plane1_14LE_c;
2276 *yuv2plane1 = yuv2plane1_8_c;
2277 *yuv2planeX = yuv2planeX_8_c;
2278 if (dstFormat == AV_PIX_FMT_NV12 || dstFormat == AV_PIX_FMT_NV21)
2279 *yuv2nv12cX = yuv2nv12cX_c;
2282 if(c->flags & SWS_FULL_CHR_H_INT) {
2283 switch (dstFormat) {
2284 case AV_PIX_FMT_RGBA:
2286 *yuv2packedX = yuv2rgba32_full_X_c;
2287 *yuv2packed2 = yuv2rgba32_full_2_c;
2288 *yuv2packed1 = yuv2rgba32_full_1_c;
2290 #if CONFIG_SWSCALE_ALPHA
2292 *yuv2packedX = yuv2rgba32_full_X_c;
2293 *yuv2packed2 = yuv2rgba32_full_2_c;
2294 *yuv2packed1 = yuv2rgba32_full_1_c;
2296 #endif /* CONFIG_SWSCALE_ALPHA */
2298 *yuv2packedX = yuv2rgbx32_full_X_c;
2299 *yuv2packed2 = yuv2rgbx32_full_2_c;
2300 *yuv2packed1 = yuv2rgbx32_full_1_c;
2302 #endif /* !CONFIG_SMALL */
2304 case AV_PIX_FMT_ARGB:
2306 *yuv2packedX = yuv2argb32_full_X_c;
2307 *yuv2packed2 = yuv2argb32_full_2_c;
2308 *yuv2packed1 = yuv2argb32_full_1_c;
2310 #if CONFIG_SWSCALE_ALPHA
2312 *yuv2packedX = yuv2argb32_full_X_c;
2313 *yuv2packed2 = yuv2argb32_full_2_c;
2314 *yuv2packed1 = yuv2argb32_full_1_c;
2316 #endif /* CONFIG_SWSCALE_ALPHA */
2318 *yuv2packedX = yuv2xrgb32_full_X_c;
2319 *yuv2packed2 = yuv2xrgb32_full_2_c;
2320 *yuv2packed1 = yuv2xrgb32_full_1_c;
2322 #endif /* !CONFIG_SMALL */
2324 case AV_PIX_FMT_BGRA:
2326 *yuv2packedX = yuv2bgra32_full_X_c;
2327 *yuv2packed2 = yuv2bgra32_full_2_c;
2328 *yuv2packed1 = yuv2bgra32_full_1_c;
2330 #if CONFIG_SWSCALE_ALPHA
2332 *yuv2packedX = yuv2bgra32_full_X_c;
2333 *yuv2packed2 = yuv2bgra32_full_2_c;
2334 *yuv2packed1 = yuv2bgra32_full_1_c;
2336 #endif /* CONFIG_SWSCALE_ALPHA */
2338 *yuv2packedX = yuv2bgrx32_full_X_c;
2339 *yuv2packed2 = yuv2bgrx32_full_2_c;
2340 *yuv2packed1 = yuv2bgrx32_full_1_c;
2342 #endif /* !CONFIG_SMALL */
2344 case AV_PIX_FMT_ABGR:
2346 *yuv2packedX = yuv2abgr32_full_X_c;
2347 *yuv2packed2 = yuv2abgr32_full_2_c;
2348 *yuv2packed1 = yuv2abgr32_full_1_c;
2350 #if CONFIG_SWSCALE_ALPHA
2352 *yuv2packedX = yuv2abgr32_full_X_c;
2353 *yuv2packed2 = yuv2abgr32_full_2_c;
2354 *yuv2packed1 = yuv2abgr32_full_1_c;
2356 #endif /* CONFIG_SWSCALE_ALPHA */
2358 *yuv2packedX = yuv2xbgr32_full_X_c;
2359 *yuv2packed2 = yuv2xbgr32_full_2_c;
2360 *yuv2packed1 = yuv2xbgr32_full_1_c;
2362 #endif /* !CONFIG_SMALL */
2364 case AV_PIX_FMT_RGBA64LE:
2365 #if CONFIG_SWSCALE_ALPHA
2367 *yuv2packedX = yuv2rgba64le_full_X_c;
2368 *yuv2packed2 = yuv2rgba64le_full_2_c;
2369 *yuv2packed1 = yuv2rgba64le_full_1_c;
2371 #endif /* CONFIG_SWSCALE_ALPHA */
2373 *yuv2packedX = yuv2rgbx64le_full_X_c;
2374 *yuv2packed2 = yuv2rgbx64le_full_2_c;
2375 *yuv2packed1 = yuv2rgbx64le_full_1_c;
2378 case AV_PIX_FMT_RGBA64BE:
2379 #if CONFIG_SWSCALE_ALPHA
2381 *yuv2packedX = yuv2rgba64be_full_X_c;
2382 *yuv2packed2 = yuv2rgba64be_full_2_c;
2383 *yuv2packed1 = yuv2rgba64be_full_1_c;
2385 #endif /* CONFIG_SWSCALE_ALPHA */
2387 *yuv2packedX = yuv2rgbx64be_full_X_c;
2388 *yuv2packed2 = yuv2rgbx64be_full_2_c;
2389 *yuv2packed1 = yuv2rgbx64be_full_1_c;
2392 case AV_PIX_FMT_BGRA64LE:
2393 #if CONFIG_SWSCALE_ALPHA
2395 *yuv2packedX = yuv2bgra64le_full_X_c;
2396 *yuv2packed2 = yuv2bgra64le_full_2_c;
2397 *yuv2packed1 = yuv2bgra64le_full_1_c;
2399 #endif /* CONFIG_SWSCALE_ALPHA */
2401 *yuv2packedX = yuv2bgrx64le_full_X_c;
2402 *yuv2packed2 = yuv2bgrx64le_full_2_c;
2403 *yuv2packed1 = yuv2bgrx64le_full_1_c;
2406 case AV_PIX_FMT_BGRA64BE:
2407 #if CONFIG_SWSCALE_ALPHA
2409 *yuv2packedX = yuv2bgra64be_full_X_c;
2410 *yuv2packed2 = yuv2bgra64be_full_2_c;
2411 *yuv2packed1 = yuv2bgra64be_full_1_c;
2413 #endif /* CONFIG_SWSCALE_ALPHA */
2415 *yuv2packedX = yuv2bgrx64be_full_X_c;
2416 *yuv2packed2 = yuv2bgrx64be_full_2_c;
2417 *yuv2packed1 = yuv2bgrx64be_full_1_c;
2421 case AV_PIX_FMT_RGB24:
2422 *yuv2packedX = yuv2rgb24_full_X_c;
2423 *yuv2packed2 = yuv2rgb24_full_2_c;
2424 *yuv2packed1 = yuv2rgb24_full_1_c;
2426 case AV_PIX_FMT_BGR24:
2427 *yuv2packedX = yuv2bgr24_full_X_c;
2428 *yuv2packed2 = yuv2bgr24_full_2_c;
2429 *yuv2packed1 = yuv2bgr24_full_1_c;
2431 case AV_PIX_FMT_RGB48LE:
2432 *yuv2packedX = yuv2rgb48le_full_X_c;
2433 *yuv2packed2 = yuv2rgb48le_full_2_c;
2434 *yuv2packed1 = yuv2rgb48le_full_1_c;
2436 case AV_PIX_FMT_BGR48LE:
2437 *yuv2packedX = yuv2bgr48le_full_X_c;
2438 *yuv2packed2 = yuv2bgr48le_full_2_c;
2439 *yuv2packed1 = yuv2bgr48le_full_1_c;
2441 case AV_PIX_FMT_RGB48BE:
2442 *yuv2packedX = yuv2rgb48be_full_X_c;
2443 *yuv2packed2 = yuv2rgb48be_full_2_c;
2444 *yuv2packed1 = yuv2rgb48be_full_1_c;
2446 case AV_PIX_FMT_BGR48BE:
2447 *yuv2packedX = yuv2bgr48be_full_X_c;
2448 *yuv2packed2 = yuv2bgr48be_full_2_c;
2449 *yuv2packed1 = yuv2bgr48be_full_1_c;
2451 case AV_PIX_FMT_BGR4_BYTE:
2452 *yuv2packedX = yuv2bgr4_byte_full_X_c;
2453 *yuv2packed2 = yuv2bgr4_byte_full_2_c;
2454 *yuv2packed1 = yuv2bgr4_byte_full_1_c;
2456 case AV_PIX_FMT_RGB4_BYTE:
2457 *yuv2packedX = yuv2rgb4_byte_full_X_c;
2458 *yuv2packed2 = yuv2rgb4_byte_full_2_c;
2459 *yuv2packed1 = yuv2rgb4_byte_full_1_c;
2461 case AV_PIX_FMT_BGR8:
2462 *yuv2packedX = yuv2bgr8_full_X_c;
2463 *yuv2packed2 = yuv2bgr8_full_2_c;
2464 *yuv2packed1 = yuv2bgr8_full_1_c;
2466 case AV_PIX_FMT_RGB8:
2467 *yuv2packedX = yuv2rgb8_full_X_c;
2468 *yuv2packed2 = yuv2rgb8_full_2_c;
2469 *yuv2packed1 = yuv2rgb8_full_1_c;
2471 case AV_PIX_FMT_GBRP:
2472 case AV_PIX_FMT_GBRP9BE:
2473 case AV_PIX_FMT_GBRP9LE:
2474 case AV_PIX_FMT_GBRP10BE:
2475 case AV_PIX_FMT_GBRP10LE:
2476 case AV_PIX_FMT_GBRP12BE:
2477 case AV_PIX_FMT_GBRP12LE:
2478 case AV_PIX_FMT_GBRP14BE:
2479 case AV_PIX_FMT_GBRP14LE:
2480 case AV_PIX_FMT_GBRAP:
2481 case AV_PIX_FMT_GBRAP10BE:
2482 case AV_PIX_FMT_GBRAP10LE:
2483 case AV_PIX_FMT_GBRAP12BE:
2484 case AV_PIX_FMT_GBRAP12LE:
2485 *yuv2anyX = yuv2gbrp_full_X_c;
2487 case AV_PIX_FMT_GBRP16BE:
2488 case AV_PIX_FMT_GBRP16LE:
2489 case AV_PIX_FMT_GBRAP16BE:
2490 case AV_PIX_FMT_GBRAP16LE:
2491 *yuv2anyX = yuv2gbrp16_full_X_c;
2494 if (!*yuv2packedX && !*yuv2anyX)
2498 switch (dstFormat) {
2499 case AV_PIX_FMT_RGBA64LE:
2500 #if CONFIG_SWSCALE_ALPHA
2502 *yuv2packed1 = yuv2rgba64le_1_c;
2503 *yuv2packed2 = yuv2rgba64le_2_c;
2504 *yuv2packedX = yuv2rgba64le_X_c;
2506 #endif /* CONFIG_SWSCALE_ALPHA */
2508 *yuv2packed1 = yuv2rgbx64le_1_c;
2509 *yuv2packed2 = yuv2rgbx64le_2_c;
2510 *yuv2packedX = yuv2rgbx64le_X_c;
2513 case AV_PIX_FMT_RGBA64BE:
2514 #if CONFIG_SWSCALE_ALPHA
2516 *yuv2packed1 = yuv2rgba64be_1_c;
2517 *yuv2packed2 = yuv2rgba64be_2_c;
2518 *yuv2packedX = yuv2rgba64be_X_c;
2520 #endif /* CONFIG_SWSCALE_ALPHA */
2522 *yuv2packed1 = yuv2rgbx64be_1_c;
2523 *yuv2packed2 = yuv2rgbx64be_2_c;
2524 *yuv2packedX = yuv2rgbx64be_X_c;
2527 case AV_PIX_FMT_BGRA64LE:
2528 #if CONFIG_SWSCALE_ALPHA
2530 *yuv2packed1 = yuv2bgra64le_1_c;
2531 *yuv2packed2 = yuv2bgra64le_2_c;
2532 *yuv2packedX = yuv2bgra64le_X_c;
2534 #endif /* CONFIG_SWSCALE_ALPHA */
2536 *yuv2packed1 = yuv2bgrx64le_1_c;
2537 *yuv2packed2 = yuv2bgrx64le_2_c;
2538 *yuv2packedX = yuv2bgrx64le_X_c;
2541 case AV_PIX_FMT_BGRA64BE:
2542 #if CONFIG_SWSCALE_ALPHA
2544 *yuv2packed1 = yuv2bgra64be_1_c;
2545 *yuv2packed2 = yuv2bgra64be_2_c;
2546 *yuv2packedX = yuv2bgra64be_X_c;
2548 #endif /* CONFIG_SWSCALE_ALPHA */
2550 *yuv2packed1 = yuv2bgrx64be_1_c;
2551 *yuv2packed2 = yuv2bgrx64be_2_c;
2552 *yuv2packedX = yuv2bgrx64be_X_c;
2555 case AV_PIX_FMT_RGB48LE:
2556 *yuv2packed1 = yuv2rgb48le_1_c;
2557 *yuv2packed2 = yuv2rgb48le_2_c;
2558 *yuv2packedX = yuv2rgb48le_X_c;
2560 case AV_PIX_FMT_RGB48BE:
2561 *yuv2packed1 = yuv2rgb48be_1_c;
2562 *yuv2packed2 = yuv2rgb48be_2_c;
2563 *yuv2packedX = yuv2rgb48be_X_c;
2565 case AV_PIX_FMT_BGR48LE:
2566 *yuv2packed1 = yuv2bgr48le_1_c;
2567 *yuv2packed2 = yuv2bgr48le_2_c;
2568 *yuv2packedX = yuv2bgr48le_X_c;
2570 case AV_PIX_FMT_BGR48BE:
2571 *yuv2packed1 = yuv2bgr48be_1_c;
2572 *yuv2packed2 = yuv2bgr48be_2_c;
2573 *yuv2packedX = yuv2bgr48be_X_c;
2575 case AV_PIX_FMT_RGB32:
2576 case AV_PIX_FMT_BGR32:
2578 *yuv2packed1 = yuv2rgb32_1_c;
2579 *yuv2packed2 = yuv2rgb32_2_c;
2580 *yuv2packedX = yuv2rgb32_X_c;
2582 #if CONFIG_SWSCALE_ALPHA
2584 *yuv2packed1 = yuv2rgba32_1_c;
2585 *yuv2packed2 = yuv2rgba32_2_c;
2586 *yuv2packedX = yuv2rgba32_X_c;
2588 #endif /* CONFIG_SWSCALE_ALPHA */
2590 *yuv2packed1 = yuv2rgbx32_1_c;
2591 *yuv2packed2 = yuv2rgbx32_2_c;
2592 *yuv2packedX = yuv2rgbx32_X_c;
2594 #endif /* !CONFIG_SMALL */
2596 case AV_PIX_FMT_RGB32_1:
2597 case AV_PIX_FMT_BGR32_1:
2599 *yuv2packed1 = yuv2rgb32_1_1_c;
2600 *yuv2packed2 = yuv2rgb32_1_2_c;
2601 *yuv2packedX = yuv2rgb32_1_X_c;
2603 #if CONFIG_SWSCALE_ALPHA
2605 *yuv2packed1 = yuv2rgba32_1_1_c;
2606 *yuv2packed2 = yuv2rgba32_1_2_c;
2607 *yuv2packedX = yuv2rgba32_1_X_c;
2609 #endif /* CONFIG_SWSCALE_ALPHA */
2611 *yuv2packed1 = yuv2rgbx32_1_1_c;
2612 *yuv2packed2 = yuv2rgbx32_1_2_c;
2613 *yuv2packedX = yuv2rgbx32_1_X_c;
2615 #endif /* !CONFIG_SMALL */
2617 case AV_PIX_FMT_RGB24:
2618 *yuv2packed1 = yuv2rgb24_1_c;
2619 *yuv2packed2 = yuv2rgb24_2_c;
2620 *yuv2packedX = yuv2rgb24_X_c;
2622 case AV_PIX_FMT_BGR24:
2623 *yuv2packed1 = yuv2bgr24_1_c;
2624 *yuv2packed2 = yuv2bgr24_2_c;
2625 *yuv2packedX = yuv2bgr24_X_c;
2627 case AV_PIX_FMT_RGB565LE:
2628 case AV_PIX_FMT_RGB565BE:
2629 case AV_PIX_FMT_BGR565LE:
2630 case AV_PIX_FMT_BGR565BE:
2631 *yuv2packed1 = yuv2rgb16_1_c;
2632 *yuv2packed2 = yuv2rgb16_2_c;
2633 *yuv2packedX = yuv2rgb16_X_c;
2635 case AV_PIX_FMT_RGB555LE:
2636 case AV_PIX_FMT_RGB555BE:
2637 case AV_PIX_FMT_BGR555LE:
2638 case AV_PIX_FMT_BGR555BE:
2639 *yuv2packed1 = yuv2rgb15_1_c;
2640 *yuv2packed2 = yuv2rgb15_2_c;
2641 *yuv2packedX = yuv2rgb15_X_c;
2643 case AV_PIX_FMT_RGB444LE:
2644 case AV_PIX_FMT_RGB444BE:
2645 case AV_PIX_FMT_BGR444LE:
2646 case AV_PIX_FMT_BGR444BE:
2647 *yuv2packed1 = yuv2rgb12_1_c;
2648 *yuv2packed2 = yuv2rgb12_2_c;
2649 *yuv2packedX = yuv2rgb12_X_c;
2651 case AV_PIX_FMT_RGB8:
2652 case AV_PIX_FMT_BGR8:
2653 *yuv2packed1 = yuv2rgb8_1_c;
2654 *yuv2packed2 = yuv2rgb8_2_c;
2655 *yuv2packedX = yuv2rgb8_X_c;
2657 case AV_PIX_FMT_RGB4:
2658 case AV_PIX_FMT_BGR4:
2659 *yuv2packed1 = yuv2rgb4_1_c;
2660 *yuv2packed2 = yuv2rgb4_2_c;
2661 *yuv2packedX = yuv2rgb4_X_c;
2663 case AV_PIX_FMT_RGB4_BYTE:
2664 case AV_PIX_FMT_BGR4_BYTE:
2665 *yuv2packed1 = yuv2rgb4b_1_c;
2666 *yuv2packed2 = yuv2rgb4b_2_c;
2667 *yuv2packedX = yuv2rgb4b_X_c;
2671 switch (dstFormat) {
2672 case AV_PIX_FMT_MONOWHITE:
2673 *yuv2packed1 = yuv2monowhite_1_c;
2674 *yuv2packed2 = yuv2monowhite_2_c;
2675 *yuv2packedX = yuv2monowhite_X_c;
2677 case AV_PIX_FMT_MONOBLACK:
2678 *yuv2packed1 = yuv2monoblack_1_c;
2679 *yuv2packed2 = yuv2monoblack_2_c;
2680 *yuv2packedX = yuv2monoblack_X_c;
2682 case AV_PIX_FMT_YUYV422:
2683 *yuv2packed1 = yuv2yuyv422_1_c;
2684 *yuv2packed2 = yuv2yuyv422_2_c;
2685 *yuv2packedX = yuv2yuyv422_X_c;
2687 case AV_PIX_FMT_YVYU422:
2688 *yuv2packed1 = yuv2yvyu422_1_c;
2689 *yuv2packed2 = yuv2yvyu422_2_c;
2690 *yuv2packedX = yuv2yvyu422_X_c;
2692 case AV_PIX_FMT_UYVY422:
2693 *yuv2packed1 = yuv2uyvy422_1_c;
2694 *yuv2packed2 = yuv2uyvy422_2_c;
2695 *yuv2packedX = yuv2uyvy422_X_c;
2697 case AV_PIX_FMT_YA8:
2698 *yuv2packed1 = yuv2ya8_1_c;
2699 *yuv2packed2 = yuv2ya8_2_c;
2700 *yuv2packedX = yuv2ya8_X_c;
2702 case AV_PIX_FMT_AYUV64LE:
2703 *yuv2packedX = yuv2ayuv64le_X_c;