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);
183 static void yuv2p016cX_c(SwsContext *c, const int16_t *chrFilter, int chrFilterSize,
184 const int16_t **chrUSrc, const int16_t **chrVSrc,
185 uint8_t *dest8, int chrDstW)
187 uint16_t *dest = (uint16_t*)dest8;
188 const int32_t **uSrc = (const int32_t **)chrUSrc;
189 const int32_t **vSrc = (const int32_t **)chrVSrc;
191 int big_endian = c->dstFormat == AV_PIX_FMT_P016BE;
194 for (i = 0; i < chrDstW; i++) {
195 int u = 1 << (shift - 1);
196 int v = 1 << (shift - 1);
198 /* See yuv2planeX_16_c_template for details. */
201 for (j = 0; j < chrFilterSize; j++) {
202 u += uSrc[j][i] * (unsigned)chrFilter[j];
203 v += vSrc[j][i] * (unsigned)chrFilter[j];
206 output_pixel(&dest[2*i] , u, 0x8000, int);
207 output_pixel(&dest[2*i+1], v, 0x8000, int);
211 static av_always_inline void
212 yuv2plane1_float_c_template(const int32_t *src, float *dest, int dstW)
214 static const int big_endian = HAVE_BIGENDIAN;
215 static const int shift = 3;
216 static const float float_mult = 1.0f / 65535.0f;
220 for (i = 0; i < dstW; ++i){
221 val = src[i] + (1 << (shift - 1));
222 output_pixel(&val_uint, val, 0, uint);
223 dest[i] = float_mult * (float)val_uint;
227 static av_always_inline void
228 yuv2plane1_float_bswap_c_template(const int32_t *src, uint32_t *dest, int dstW)
230 static const int big_endian = HAVE_BIGENDIAN;
231 static const int shift = 3;
232 static const float float_mult = 1.0f / 65535.0f;
236 for (i = 0; i < dstW; ++i){
237 val = src[i] + (1 << (shift - 1));
238 output_pixel(&val_uint, val, 0, uint);
239 dest[i] = av_bswap32(av_float2int(float_mult * (float)val_uint));
243 static av_always_inline void
244 yuv2planeX_float_c_template(const int16_t *filter, int filterSize, const int32_t **src,
245 float *dest, int dstW)
247 static const int big_endian = HAVE_BIGENDIAN;
248 static const int shift = 15;
249 static const float float_mult = 1.0f / 65535.0f;
253 for (i = 0; i < dstW; ++i){
254 val = (1 << (shift - 1)) - 0x40000000;
255 for (j = 0; j < filterSize; ++j){
256 val += src[j][i] * (unsigned)filter[j];
258 output_pixel(&val_uint, val, 0x8000, int);
259 dest[i] = float_mult * (float)val_uint;
263 static av_always_inline void
264 yuv2planeX_float_bswap_c_template(const int16_t *filter, int filterSize, const int32_t **src,
265 uint32_t *dest, int dstW)
267 static const int big_endian = HAVE_BIGENDIAN;
268 static const int shift = 15;
269 static const float float_mult = 1.0f / 65535.0f;
273 for (i = 0; i < dstW; ++i){
274 val = (1 << (shift - 1)) - 0x40000000;
275 for (j = 0; j < filterSize; ++j){
276 val += src[j][i] * (unsigned)filter[j];
278 output_pixel(&val_uint, val, 0x8000, int);
279 dest[i] = av_bswap32(av_float2int(float_mult * (float)val_uint));
283 #define yuv2plane1_float(template, dest_type, BE_LE) \
284 static void yuv2plane1_float ## BE_LE ## _c(const int16_t *src, uint8_t *dest, int dstW, \
285 const uint8_t *dither, int offset) \
287 template((const int32_t *)src, (dest_type *)dest, dstW); \
290 #define yuv2planeX_float(template, dest_type, BE_LE) \
291 static void yuv2planeX_float ## BE_LE ## _c(const int16_t *filter, int filterSize, \
292 const int16_t **src, uint8_t *dest, int dstW, \
293 const uint8_t *dither, int offset) \
295 template(filter, filterSize, (const int32_t **)src, (dest_type *)dest, dstW); \
299 yuv2plane1_float(yuv2plane1_float_c_template, float, BE)
300 yuv2plane1_float(yuv2plane1_float_bswap_c_template, uint32_t, LE)
301 yuv2planeX_float(yuv2planeX_float_c_template, float, BE)
302 yuv2planeX_float(yuv2planeX_float_bswap_c_template, uint32_t, LE)
304 yuv2plane1_float(yuv2plane1_float_c_template, float, LE)
305 yuv2plane1_float(yuv2plane1_float_bswap_c_template, uint32_t, BE)
306 yuv2planeX_float(yuv2planeX_float_c_template, float, LE)
307 yuv2planeX_float(yuv2planeX_float_bswap_c_template, uint32_t, BE)
312 #define output_pixel(pos, val) \
314 AV_WB16(pos, av_clip_uintp2(val >> shift, output_bits)); \
316 AV_WL16(pos, av_clip_uintp2(val >> shift, output_bits)); \
319 static av_always_inline void
320 yuv2plane1_10_c_template(const int16_t *src, uint16_t *dest, int dstW,
321 int big_endian, int output_bits)
324 int shift = 15 - output_bits;
326 for (i = 0; i < dstW; i++) {
327 int val = src[i] + (1 << (shift - 1));
328 output_pixel(&dest[i], val);
332 static av_always_inline void
333 yuv2planeX_10_c_template(const int16_t *filter, int filterSize,
334 const int16_t **src, uint16_t *dest, int dstW,
335 int big_endian, int output_bits)
338 int shift = 11 + 16 - output_bits;
340 for (i = 0; i < dstW; i++) {
341 int val = 1 << (shift - 1);
344 for (j = 0; j < filterSize; j++)
345 val += src[j][i] * filter[j];
347 output_pixel(&dest[i], val);
353 #define yuv2NBPS(bits, BE_LE, is_be, template_size, typeX_t) \
354 static void yuv2plane1_ ## bits ## BE_LE ## _c(const int16_t *src, \
355 uint8_t *dest, int dstW, \
356 const uint8_t *dither, int offset)\
358 yuv2plane1_ ## template_size ## _c_template((const typeX_t *) src, \
359 (uint16_t *) dest, dstW, is_be, bits); \
361 static void yuv2planeX_ ## bits ## BE_LE ## _c(const int16_t *filter, int filterSize, \
362 const int16_t **src, uint8_t *dest, int dstW, \
363 const uint8_t *dither, int offset)\
365 yuv2planeX_## template_size ## _c_template(filter, \
366 filterSize, (const typeX_t **) src, \
367 (uint16_t *) dest, dstW, is_be, bits); \
369 yuv2NBPS( 9, BE, 1, 10, int16_t)
370 yuv2NBPS( 9, LE, 0, 10, int16_t)
371 yuv2NBPS(10, BE, 1, 10, int16_t)
372 yuv2NBPS(10, LE, 0, 10, int16_t)
373 yuv2NBPS(12, BE, 1, 10, int16_t)
374 yuv2NBPS(12, LE, 0, 10, int16_t)
375 yuv2NBPS(14, BE, 1, 10, int16_t)
376 yuv2NBPS(14, LE, 0, 10, int16_t)
377 yuv2NBPS(16, BE, 1, 16, int32_t)
378 yuv2NBPS(16, LE, 0, 16, int32_t)
380 static void yuv2planeX_8_c(const int16_t *filter, int filterSize,
381 const int16_t **src, uint8_t *dest, int dstW,
382 const uint8_t *dither, int offset)
385 for (i=0; i<dstW; i++) {
386 int val = dither[(i + offset) & 7] << 12;
388 for (j=0; j<filterSize; j++)
389 val += src[j][i] * filter[j];
391 dest[i]= av_clip_uint8(val>>19);
395 static void yuv2plane1_8_c(const int16_t *src, uint8_t *dest, int dstW,
396 const uint8_t *dither, int offset)
399 for (i=0; i<dstW; i++) {
400 int val = (src[i] + dither[(i + offset) & 7]) >> 7;
401 dest[i]= av_clip_uint8(val);
405 static void yuv2nv12cX_c(SwsContext *c, const int16_t *chrFilter, int chrFilterSize,
406 const int16_t **chrUSrc, const int16_t **chrVSrc,
407 uint8_t *dest, int chrDstW)
409 enum AVPixelFormat dstFormat = c->dstFormat;
410 const uint8_t *chrDither = c->chrDither8;
413 if (dstFormat == AV_PIX_FMT_NV12 ||
414 dstFormat == AV_PIX_FMT_NV24)
415 for (i=0; i<chrDstW; i++) {
416 int u = chrDither[i & 7] << 12;
417 int v = chrDither[(i + 3) & 7] << 12;
419 for (j=0; j<chrFilterSize; j++) {
420 u += chrUSrc[j][i] * chrFilter[j];
421 v += chrVSrc[j][i] * chrFilter[j];
424 dest[2*i]= av_clip_uint8(u>>19);
425 dest[2*i+1]= av_clip_uint8(v>>19);
428 for (i=0; i<chrDstW; i++) {
429 int u = chrDither[i & 7] << 12;
430 int v = chrDither[(i + 3) & 7] << 12;
432 for (j=0; j<chrFilterSize; j++) {
433 u += chrUSrc[j][i] * chrFilter[j];
434 v += chrVSrc[j][i] * chrFilter[j];
437 dest[2*i]= av_clip_uint8(v>>19);
438 dest[2*i+1]= av_clip_uint8(u>>19);
443 #define output_pixel(pos, val) \
445 AV_WB16(pos, av_clip_uintp2(val >> shift, 10) << 6); \
447 AV_WL16(pos, av_clip_uintp2(val >> shift, 10) << 6); \
450 static void yuv2p010l1_c(const int16_t *src,
451 uint16_t *dest, int dstW,
457 for (i = 0; i < dstW; i++) {
458 int val = src[i] + (1 << (shift - 1));
459 output_pixel(&dest[i], val);
463 static void yuv2p010lX_c(const int16_t *filter, int filterSize,
464 const int16_t **src, uint16_t *dest, int dstW,
470 for (i = 0; i < dstW; i++) {
471 int val = 1 << (shift - 1);
473 for (j = 0; j < filterSize; j++)
474 val += src[j][i] * filter[j];
476 output_pixel(&dest[i], val);
480 static void yuv2p010cX_c(SwsContext *c, const int16_t *chrFilter, int chrFilterSize,
481 const int16_t **chrUSrc, const int16_t **chrVSrc,
482 uint8_t *dest8, int chrDstW)
484 uint16_t *dest = (uint16_t*)dest8;
486 int big_endian = c->dstFormat == AV_PIX_FMT_P010BE;
489 for (i = 0; i < chrDstW; i++) {
490 int u = 1 << (shift - 1);
491 int v = 1 << (shift - 1);
493 for (j = 0; j < chrFilterSize; j++) {
494 u += chrUSrc[j][i] * chrFilter[j];
495 v += chrVSrc[j][i] * chrFilter[j];
498 output_pixel(&dest[2*i] , u);
499 output_pixel(&dest[2*i+1], v);
503 static void yuv2p010l1_LE_c(const int16_t *src,
504 uint8_t *dest, int dstW,
505 const uint8_t *dither, int offset)
507 yuv2p010l1_c(src, (uint16_t*)dest, dstW, 0);
510 static void yuv2p010l1_BE_c(const int16_t *src,
511 uint8_t *dest, int dstW,
512 const uint8_t *dither, int offset)
514 yuv2p010l1_c(src, (uint16_t*)dest, dstW, 1);
517 static void yuv2p010lX_LE_c(const int16_t *filter, int filterSize,
518 const int16_t **src, uint8_t *dest, int dstW,
519 const uint8_t *dither, int offset)
521 yuv2p010lX_c(filter, filterSize, src, (uint16_t*)dest, dstW, 0);
524 static void yuv2p010lX_BE_c(const int16_t *filter, int filterSize,
525 const int16_t **src, uint8_t *dest, int dstW,
526 const uint8_t *dither, int offset)
528 yuv2p010lX_c(filter, filterSize, src, (uint16_t*)dest, dstW, 1);
534 #define accumulate_bit(acc, val) \
537 #define output_pixel(pos, acc) \
538 if (target == AV_PIX_FMT_MONOBLACK) { \
544 static av_always_inline void
545 yuv2mono_X_c_template(SwsContext *c, const int16_t *lumFilter,
546 const int16_t **lumSrc, int lumFilterSize,
547 const int16_t *chrFilter, const int16_t **chrUSrc,
548 const int16_t **chrVSrc, int chrFilterSize,
549 const int16_t **alpSrc, uint8_t *dest, int dstW,
550 int y, enum AVPixelFormat target)
552 const uint8_t * const d128 = ff_dither_8x8_220[y&7];
557 for (i = 0; i < dstW; i += 2) {
562 for (j = 0; j < lumFilterSize; j++) {
563 Y1 += lumSrc[j][i] * lumFilter[j];
564 Y2 += lumSrc[j][i+1] * lumFilter[j];
568 if ((Y1 | Y2) & 0x100) {
569 Y1 = av_clip_uint8(Y1);
570 Y2 = av_clip_uint8(Y2);
572 if (c->dither == SWS_DITHER_ED) {
573 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;
574 c->dither_error[0][i] = err;
575 acc = 2*acc + (Y1 >= 128);
578 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);
579 c->dither_error[0][i+1] = Y1;
580 acc = 2*acc + (err >= 128);
583 accumulate_bit(acc, Y1 + d128[(i + 0) & 7]);
584 accumulate_bit(acc, Y2 + d128[(i + 1) & 7]);
587 output_pixel(*dest++, acc);
590 c->dither_error[0][i] = err;
593 output_pixel(*dest, acc);
597 static av_always_inline void
598 yuv2mono_2_c_template(SwsContext *c, const int16_t *buf[2],
599 const int16_t *ubuf[2], const int16_t *vbuf[2],
600 const int16_t *abuf[2], uint8_t *dest, int dstW,
601 int yalpha, int uvalpha, int y,
602 enum AVPixelFormat target)
604 const int16_t *buf0 = buf[0], *buf1 = buf[1];
605 const uint8_t * const d128 = ff_dither_8x8_220[y & 7];
606 int yalpha1 = 4096 - yalpha;
608 av_assert2(yalpha <= 4096U);
610 if (c->dither == SWS_DITHER_ED) {
613 for (i = 0; i < dstW; i +=2) {
616 Y = (buf0[i + 0] * yalpha1 + buf1[i + 0] * yalpha) >> 19;
617 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;
618 c->dither_error[0][i] = err;
619 acc = 2*acc + (Y >= 128);
622 err = (buf0[i + 1] * yalpha1 + buf1[i + 1] * yalpha) >> 19;
623 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;
624 c->dither_error[0][i+1] = Y;
625 acc = 2*acc + (err >= 128);
629 output_pixel(*dest++, acc);
631 c->dither_error[0][i] = err;
633 for (i = 0; i < dstW; i += 8) {
636 Y = (buf0[i + 0] * yalpha1 + buf1[i + 0] * yalpha) >> 19;
637 accumulate_bit(acc, Y + d128[0]);
638 Y = (buf0[i + 1] * yalpha1 + buf1[i + 1] * yalpha) >> 19;
639 accumulate_bit(acc, Y + d128[1]);
640 Y = (buf0[i + 2] * yalpha1 + buf1[i + 2] * yalpha) >> 19;
641 accumulate_bit(acc, Y + d128[2]);
642 Y = (buf0[i + 3] * yalpha1 + buf1[i + 3] * yalpha) >> 19;
643 accumulate_bit(acc, Y + d128[3]);
644 Y = (buf0[i + 4] * yalpha1 + buf1[i + 4] * yalpha) >> 19;
645 accumulate_bit(acc, Y + d128[4]);
646 Y = (buf0[i + 5] * yalpha1 + buf1[i + 5] * yalpha) >> 19;
647 accumulate_bit(acc, Y + d128[5]);
648 Y = (buf0[i + 6] * yalpha1 + buf1[i + 6] * yalpha) >> 19;
649 accumulate_bit(acc, Y + d128[6]);
650 Y = (buf0[i + 7] * yalpha1 + buf1[i + 7] * yalpha) >> 19;
651 accumulate_bit(acc, Y + d128[7]);
653 output_pixel(*dest++, acc);
658 static av_always_inline void
659 yuv2mono_1_c_template(SwsContext *c, const int16_t *buf0,
660 const int16_t *ubuf[2], const int16_t *vbuf[2],
661 const int16_t *abuf0, uint8_t *dest, int dstW,
662 int uvalpha, int y, enum AVPixelFormat target)
664 const uint8_t * const d128 = ff_dither_8x8_220[y & 7];
667 if (c->dither == SWS_DITHER_ED) {
670 for (i = 0; i < dstW; i +=2) {
673 Y = ((buf0[i + 0] + 64) >> 7);
674 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;
675 c->dither_error[0][i] = err;
676 acc = 2*acc + (Y >= 128);
679 err = ((buf0[i + 1] + 64) >> 7);
680 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;
681 c->dither_error[0][i+1] = Y;
682 acc = 2*acc + (err >= 128);
686 output_pixel(*dest++, acc);
688 c->dither_error[0][i] = err;
690 for (i = 0; i < dstW; i += 8) {
692 accumulate_bit(acc, ((buf0[i + 0] + 64) >> 7) + d128[0]);
693 accumulate_bit(acc, ((buf0[i + 1] + 64) >> 7) + d128[1]);
694 accumulate_bit(acc, ((buf0[i + 2] + 64) >> 7) + d128[2]);
695 accumulate_bit(acc, ((buf0[i + 3] + 64) >> 7) + d128[3]);
696 accumulate_bit(acc, ((buf0[i + 4] + 64) >> 7) + d128[4]);
697 accumulate_bit(acc, ((buf0[i + 5] + 64) >> 7) + d128[5]);
698 accumulate_bit(acc, ((buf0[i + 6] + 64) >> 7) + d128[6]);
699 accumulate_bit(acc, ((buf0[i + 7] + 64) >> 7) + d128[7]);
701 output_pixel(*dest++, acc);
707 #undef accumulate_bit
709 #define YUV2PACKEDWRAPPER(name, base, ext, fmt) \
710 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
711 const int16_t **lumSrc, int lumFilterSize, \
712 const int16_t *chrFilter, const int16_t **chrUSrc, \
713 const int16_t **chrVSrc, int chrFilterSize, \
714 const int16_t **alpSrc, uint8_t *dest, int dstW, \
717 name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
718 chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
719 alpSrc, dest, dstW, y, fmt); \
722 static void name ## ext ## _2_c(SwsContext *c, const int16_t *buf[2], \
723 const int16_t *ubuf[2], const int16_t *vbuf[2], \
724 const int16_t *abuf[2], uint8_t *dest, int dstW, \
725 int yalpha, int uvalpha, int y) \
727 name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
728 dest, dstW, yalpha, uvalpha, y, fmt); \
731 static void name ## ext ## _1_c(SwsContext *c, const int16_t *buf0, \
732 const int16_t *ubuf[2], const int16_t *vbuf[2], \
733 const int16_t *abuf0, uint8_t *dest, int dstW, \
734 int uvalpha, int y) \
736 name ## base ## _1_c_template(c, buf0, ubuf, vbuf, \
737 abuf0, dest, dstW, uvalpha, \
741 YUV2PACKEDWRAPPER(yuv2mono,, white, AV_PIX_FMT_MONOWHITE)
742 YUV2PACKEDWRAPPER(yuv2mono,, black, AV_PIX_FMT_MONOBLACK)
744 #define output_pixels(pos, Y1, U, Y2, V) \
745 if (target == AV_PIX_FMT_YUYV422) { \
746 dest[pos + 0] = Y1; \
748 dest[pos + 2] = Y2; \
750 } else if (target == AV_PIX_FMT_YVYU422) { \
751 dest[pos + 0] = Y1; \
753 dest[pos + 2] = Y2; \
755 } else { /* AV_PIX_FMT_UYVY422 */ \
757 dest[pos + 1] = Y1; \
759 dest[pos + 3] = Y2; \
762 static av_always_inline void
763 yuv2422_X_c_template(SwsContext *c, const int16_t *lumFilter,
764 const int16_t **lumSrc, int lumFilterSize,
765 const int16_t *chrFilter, const int16_t **chrUSrc,
766 const int16_t **chrVSrc, int chrFilterSize,
767 const int16_t **alpSrc, uint8_t *dest, int dstW,
768 int y, enum AVPixelFormat target)
772 for (i = 0; i < ((dstW + 1) >> 1); i++) {
779 for (j = 0; j < lumFilterSize; j++) {
780 Y1 += lumSrc[j][i * 2] * lumFilter[j];
781 Y2 += lumSrc[j][i * 2 + 1] * lumFilter[j];
783 for (j = 0; j < chrFilterSize; j++) {
784 U += chrUSrc[j][i] * chrFilter[j];
785 V += chrVSrc[j][i] * chrFilter[j];
791 if ((Y1 | Y2 | U | V) & 0x100) {
792 Y1 = av_clip_uint8(Y1);
793 Y2 = av_clip_uint8(Y2);
794 U = av_clip_uint8(U);
795 V = av_clip_uint8(V);
797 output_pixels(4*i, Y1, U, Y2, V);
801 static av_always_inline void
802 yuv2422_2_c_template(SwsContext *c, const int16_t *buf[2],
803 const int16_t *ubuf[2], const int16_t *vbuf[2],
804 const int16_t *abuf[2], uint8_t *dest, int dstW,
805 int yalpha, int uvalpha, int y,
806 enum AVPixelFormat target)
808 const int16_t *buf0 = buf[0], *buf1 = buf[1],
809 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
810 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1];
811 int yalpha1 = 4096 - yalpha;
812 int uvalpha1 = 4096 - uvalpha;
814 av_assert2(yalpha <= 4096U);
815 av_assert2(uvalpha <= 4096U);
817 for (i = 0; i < ((dstW + 1) >> 1); i++) {
818 int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 19;
819 int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 19;
820 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha) >> 19;
821 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha) >> 19;
823 if ((Y1 | Y2 | U | V) & 0x100) {
824 Y1 = av_clip_uint8(Y1);
825 Y2 = av_clip_uint8(Y2);
826 U = av_clip_uint8(U);
827 V = av_clip_uint8(V);
830 output_pixels(i * 4, Y1, U, Y2, V);
834 static av_always_inline void
835 yuv2422_1_c_template(SwsContext *c, const int16_t *buf0,
836 const int16_t *ubuf[2], const int16_t *vbuf[2],
837 const int16_t *abuf0, uint8_t *dest, int dstW,
838 int uvalpha, int y, enum AVPixelFormat target)
840 const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
843 if (uvalpha < 2048) {
844 for (i = 0; i < ((dstW + 1) >> 1); i++) {
845 int Y1 = (buf0[i * 2 ]+64) >> 7;
846 int Y2 = (buf0[i * 2 + 1]+64) >> 7;
847 int U = (ubuf0[i] +64) >> 7;
848 int V = (vbuf0[i] +64) >> 7;
850 if ((Y1 | Y2 | U | V) & 0x100) {
851 Y1 = av_clip_uint8(Y1);
852 Y2 = av_clip_uint8(Y2);
853 U = av_clip_uint8(U);
854 V = av_clip_uint8(V);
857 output_pixels(i * 4, Y1, U, Y2, V);
860 const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
861 for (i = 0; i < ((dstW + 1) >> 1); i++) {
862 int Y1 = (buf0[i * 2 ] + 64) >> 7;
863 int Y2 = (buf0[i * 2 + 1] + 64) >> 7;
864 int U = (ubuf0[i] + ubuf1[i]+128) >> 8;
865 int V = (vbuf0[i] + vbuf1[i]+128) >> 8;
867 if ((Y1 | Y2 | U | V) & 0x100) {
868 Y1 = av_clip_uint8(Y1);
869 Y2 = av_clip_uint8(Y2);
870 U = av_clip_uint8(U);
871 V = av_clip_uint8(V);
874 output_pixels(i * 4, Y1, U, Y2, V);
881 YUV2PACKEDWRAPPER(yuv2, 422, yuyv422, AV_PIX_FMT_YUYV422)
882 YUV2PACKEDWRAPPER(yuv2, 422, yvyu422, AV_PIX_FMT_YVYU422)
883 YUV2PACKEDWRAPPER(yuv2, 422, uyvy422, AV_PIX_FMT_UYVY422)
885 #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)
886 #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)
887 #define output_pixel(pos, val) \
888 if (isBE(target)) { \
894 static av_always_inline void
895 yuv2ya16_X_c_template(SwsContext *c, const int16_t *lumFilter,
896 const int32_t **lumSrc, int lumFilterSize,
897 const int16_t *chrFilter, const int32_t **unused_chrUSrc,
898 const int32_t **unused_chrVSrc, int unused_chrFilterSize,
899 const int32_t **alpSrc, uint16_t *dest, int dstW,
900 int y, enum AVPixelFormat target, int unused_hasAlpha, int unused_eightbytes)
902 int hasAlpha = !!alpSrc;
905 for (i = 0; i < dstW; i++) {
910 for (j = 0; j < lumFilterSize; j++)
911 Y += lumSrc[j][i] * lumFilter[j];
914 Y += (1<<3) + 0x8000;
915 Y = av_clip_uint16(Y);
918 A = -0x40000000 + (1<<14);
919 for (j = 0; j < lumFilterSize; j++)
920 A += alpSrc[j][i] * lumFilter[j];
924 A = av_clip_uint16(A);
927 output_pixel(&dest[2 * i ], Y);
928 output_pixel(&dest[2 * i + 1], A);
932 static av_always_inline void
933 yuv2ya16_2_c_template(SwsContext *c, const int32_t *buf[2],
934 const int32_t *unused_ubuf[2], const int32_t *unused_vbuf[2],
935 const int32_t *abuf[2], uint16_t *dest, int dstW,
936 int yalpha, int unused_uvalpha, int y,
937 enum AVPixelFormat target, int unused_hasAlpha, int unused_eightbytes)
939 int hasAlpha = abuf && abuf[0] && abuf[1];
940 const int32_t *buf0 = buf[0], *buf1 = buf[1],
941 *abuf0 = hasAlpha ? abuf[0] : NULL,
942 *abuf1 = hasAlpha ? abuf[1] : NULL;
943 int yalpha1 = 4096 - yalpha;
946 av_assert2(yalpha <= 4096U);
948 for (i = 0; i < dstW; i++) {
949 int Y = (buf0[i] * yalpha1 + buf1[i] * yalpha) >> 15;
952 Y = av_clip_uint16(Y);
955 A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha) >> 15;
956 A = av_clip_uint16(A);
959 output_pixel(&dest[2 * i ], Y);
960 output_pixel(&dest[2 * i + 1], hasAlpha ? A : 65535);
964 static av_always_inline void
965 yuv2ya16_1_c_template(SwsContext *c, const int32_t *buf0,
966 const int32_t *unused_ubuf[2], const int32_t *unused_vbuf[2],
967 const int32_t *abuf0, uint16_t *dest, int dstW,
968 int unused_uvalpha, int y, enum AVPixelFormat target, int unused_hasAlpha, int unused_eightbytes)
970 int hasAlpha = !!abuf0;
973 for (i = 0; i < dstW; i++) {
974 int Y = buf0[i] >> 3;/* 19 - 16 */
977 Y = av_clip_uint16(Y);
982 A = av_clip_uint16(A);
985 output_pixel(&dest[2 * i ], Y);
986 output_pixel(&dest[2 * i + 1], hasAlpha ? A : 65535);
990 static av_always_inline void
991 yuv2rgba64_X_c_template(SwsContext *c, const int16_t *lumFilter,
992 const int32_t **lumSrc, int lumFilterSize,
993 const int16_t *chrFilter, const int32_t **chrUSrc,
994 const int32_t **chrVSrc, int chrFilterSize,
995 const int32_t **alpSrc, uint16_t *dest, int dstW,
996 int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
999 int A1 = 0xffff<<14, A2 = 0xffff<<14;
1001 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1003 int Y1 = -0x40000000;
1004 int Y2 = -0x40000000;
1005 int U = -(128 << 23); // 19
1006 int V = -(128 << 23);
1009 for (j = 0; j < lumFilterSize; j++) {
1010 Y1 += lumSrc[j][i * 2] * (unsigned)lumFilter[j];
1011 Y2 += lumSrc[j][i * 2 + 1] * (unsigned)lumFilter[j];
1013 for (j = 0; j < chrFilterSize; j++) {;
1014 U += chrUSrc[j][i] * (unsigned)chrFilter[j];
1015 V += chrVSrc[j][i] * (unsigned)chrFilter[j];
1021 for (j = 0; j < lumFilterSize; j++) {
1022 A1 += alpSrc[j][i * 2] * (unsigned)lumFilter[j];
1023 A2 += alpSrc[j][i * 2 + 1] * (unsigned)lumFilter[j];
1031 // 8 bits: 12+15=27; 16 bits: 12+19=31
1039 // 8 bits: 27 -> 17 bits, 16 bits: 31 - 14 = 17 bits
1040 Y1 -= c->yuv2rgb_y_offset;
1041 Y2 -= c->yuv2rgb_y_offset;
1042 Y1 *= c->yuv2rgb_y_coeff;
1043 Y2 *= c->yuv2rgb_y_coeff;
1044 Y1 += 1 << 13; // 21
1046 // 8 bits: 17 + 13 bits = 30 bits, 16 bits: 17 + 13 bits = 30 bits
1048 R = V * c->yuv2rgb_v2r_coeff;
1049 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1050 B = U * c->yuv2rgb_u2b_coeff;
1052 // 8 bits: 30 - 22 = 8 bits, 16 bits: 30 bits - 14 = 16 bits
1053 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
1054 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
1055 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
1057 output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
1058 output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
1059 output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
1060 output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
1061 output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
1064 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
1065 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
1066 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
1072 static av_always_inline void
1073 yuv2rgba64_2_c_template(SwsContext *c, const int32_t *buf[2],
1074 const int32_t *ubuf[2], const int32_t *vbuf[2],
1075 const int32_t *abuf[2], uint16_t *dest, int dstW,
1076 int yalpha, int uvalpha, int y,
1077 enum AVPixelFormat target, int hasAlpha, int eightbytes)
1079 const int32_t *buf0 = buf[0], *buf1 = buf[1],
1080 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
1081 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
1082 *abuf0 = hasAlpha ? abuf[0] : NULL,
1083 *abuf1 = hasAlpha ? abuf[1] : NULL;
1084 int yalpha1 = 4096 - yalpha;
1085 int uvalpha1 = 4096 - uvalpha;
1087 int A1 = 0xffff<<14, A2 = 0xffff<<14;
1089 av_assert2(yalpha <= 4096U);
1090 av_assert2(uvalpha <= 4096U);
1092 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1093 int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 14;
1094 int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 14;
1095 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha - (128 << 23)) >> 14;
1096 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha - (128 << 23)) >> 14;
1099 Y1 -= c->yuv2rgb_y_offset;
1100 Y2 -= c->yuv2rgb_y_offset;
1101 Y1 *= c->yuv2rgb_y_coeff;
1102 Y2 *= c->yuv2rgb_y_coeff;
1106 R = V * c->yuv2rgb_v2r_coeff;
1107 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1108 B = U * c->yuv2rgb_u2b_coeff;
1111 A1 = (abuf0[i * 2 ] * yalpha1 + abuf1[i * 2 ] * yalpha) >> 1;
1112 A2 = (abuf0[i * 2 + 1] * yalpha1 + abuf1[i * 2 + 1] * yalpha) >> 1;
1118 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
1119 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
1120 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
1122 output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
1123 output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
1124 output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
1125 output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
1126 output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
1129 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
1130 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
1131 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
1137 static av_always_inline void
1138 yuv2rgba64_1_c_template(SwsContext *c, const int32_t *buf0,
1139 const int32_t *ubuf[2], const int32_t *vbuf[2],
1140 const int32_t *abuf0, uint16_t *dest, int dstW,
1141 int uvalpha, int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
1143 const int32_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1145 int A1 = 0xffff<<14, A2= 0xffff<<14;
1147 if (uvalpha < 2048) {
1148 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1149 int Y1 = (buf0[i * 2] ) >> 2;
1150 int Y2 = (buf0[i * 2 + 1]) >> 2;
1151 int U = (ubuf0[i] - (128 << 11)) >> 2;
1152 int V = (vbuf0[i] - (128 << 11)) >> 2;
1155 Y1 -= c->yuv2rgb_y_offset;
1156 Y2 -= c->yuv2rgb_y_offset;
1157 Y1 *= c->yuv2rgb_y_coeff;
1158 Y2 *= c->yuv2rgb_y_coeff;
1163 A1 = abuf0[i * 2 ] << 11;
1164 A2 = abuf0[i * 2 + 1] << 11;
1170 R = V * c->yuv2rgb_v2r_coeff;
1171 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1172 B = U * c->yuv2rgb_u2b_coeff;
1174 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
1175 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
1176 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
1178 output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
1179 output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
1180 output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
1181 output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
1182 output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
1185 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
1186 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
1187 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
1192 const int32_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1193 int A1 = 0xffff<<14, A2 = 0xffff<<14;
1194 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1195 int Y1 = (buf0[i * 2] ) >> 2;
1196 int Y2 = (buf0[i * 2 + 1]) >> 2;
1197 int U = (ubuf0[i] + ubuf1[i] - (128 << 12)) >> 3;
1198 int V = (vbuf0[i] + vbuf1[i] - (128 << 12)) >> 3;
1201 Y1 -= c->yuv2rgb_y_offset;
1202 Y2 -= c->yuv2rgb_y_offset;
1203 Y1 *= c->yuv2rgb_y_coeff;
1204 Y2 *= c->yuv2rgb_y_coeff;
1209 A1 = abuf0[i * 2 ] << 11;
1210 A2 = abuf0[i * 2 + 1] << 11;
1216 R = V * c->yuv2rgb_v2r_coeff;
1217 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1218 B = U * c->yuv2rgb_u2b_coeff;
1220 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
1221 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
1222 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
1224 output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
1225 output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
1226 output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
1227 output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
1228 output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
1231 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
1232 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
1233 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
1240 static av_always_inline void
1241 yuv2rgba64_full_X_c_template(SwsContext *c, const int16_t *lumFilter,
1242 const int32_t **lumSrc, int lumFilterSize,
1243 const int16_t *chrFilter, const int32_t **chrUSrc,
1244 const int32_t **chrVSrc, int chrFilterSize,
1245 const int32_t **alpSrc, uint16_t *dest, int dstW,
1246 int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
1251 for (i = 0; i < dstW; i++) {
1253 int Y = -0x40000000;
1254 int U = -(128 << 23); // 19
1255 int V = -(128 << 23);
1258 for (j = 0; j < lumFilterSize; j++) {
1259 Y += lumSrc[j][i] * (unsigned)lumFilter[j];
1261 for (j = 0; j < chrFilterSize; j++) {;
1262 U += chrUSrc[j][i] * (unsigned)chrFilter[j];
1263 V += chrVSrc[j][i] * (unsigned)chrFilter[j];
1268 for (j = 0; j < lumFilterSize; j++) {
1269 A += alpSrc[j][i] * (unsigned)lumFilter[j];
1275 // 8bit: 12+15=27; 16-bit: 12+19=31
1281 // 8bit: 27 -> 17bit, 16bit: 31 - 14 = 17bit
1282 Y -= c->yuv2rgb_y_offset;
1283 Y *= c->yuv2rgb_y_coeff;
1285 // 8bit: 17 + 13bit = 30bit, 16bit: 17 + 13bit = 30bit
1287 R = V * c->yuv2rgb_v2r_coeff;
1288 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1289 B = U * c->yuv2rgb_u2b_coeff;
1291 // 8bit: 30 - 22 = 8bit, 16bit: 30bit - 14 = 16bit
1292 output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1293 output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
1294 output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1296 output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1304 static av_always_inline void
1305 yuv2rgba64_full_2_c_template(SwsContext *c, const int32_t *buf[2],
1306 const int32_t *ubuf[2], const int32_t *vbuf[2],
1307 const int32_t *abuf[2], uint16_t *dest, int dstW,
1308 int yalpha, int uvalpha, int y,
1309 enum AVPixelFormat target, int hasAlpha, int eightbytes)
1311 const int32_t *buf0 = buf[0], *buf1 = buf[1],
1312 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
1313 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
1314 *abuf0 = hasAlpha ? abuf[0] : NULL,
1315 *abuf1 = hasAlpha ? abuf[1] : NULL;
1316 int yalpha1 = 4096 - yalpha;
1317 int uvalpha1 = 4096 - uvalpha;
1321 av_assert2(yalpha <= 4096U);
1322 av_assert2(uvalpha <= 4096U);
1324 for (i = 0; i < dstW; i++) {
1325 int Y = (buf0[i] * yalpha1 + buf1[i] * yalpha) >> 14;
1326 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha - (128 << 23)) >> 14;
1327 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha - (128 << 23)) >> 14;
1330 Y -= c->yuv2rgb_y_offset;
1331 Y *= c->yuv2rgb_y_coeff;
1334 R = V * c->yuv2rgb_v2r_coeff;
1335 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1336 B = U * c->yuv2rgb_u2b_coeff;
1339 A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha) >> 1;
1344 output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1345 output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
1346 output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1348 output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1356 static av_always_inline void
1357 yuv2rgba64_full_1_c_template(SwsContext *c, const int32_t *buf0,
1358 const int32_t *ubuf[2], const int32_t *vbuf[2],
1359 const int32_t *abuf0, uint16_t *dest, int dstW,
1360 int uvalpha, int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
1362 const int32_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1366 if (uvalpha < 2048) {
1367 for (i = 0; i < dstW; i++) {
1368 int Y = (buf0[i]) >> 2;
1369 int U = (ubuf0[i] - (128 << 11)) >> 2;
1370 int V = (vbuf0[i] - (128 << 11)) >> 2;
1373 Y -= c->yuv2rgb_y_offset;
1374 Y *= c->yuv2rgb_y_coeff;
1383 R = V * c->yuv2rgb_v2r_coeff;
1384 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1385 B = U * c->yuv2rgb_u2b_coeff;
1387 output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1388 output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
1389 output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1391 output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1398 const int32_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1400 for (i = 0; i < dstW; i++) {
1401 int Y = (buf0[i] ) >> 2;
1402 int U = (ubuf0[i] + ubuf1[i] - (128 << 12)) >> 3;
1403 int V = (vbuf0[i] + vbuf1[i] - (128 << 12)) >> 3;
1406 Y -= c->yuv2rgb_y_offset;
1407 Y *= c->yuv2rgb_y_coeff;
1416 R = V * c->yuv2rgb_v2r_coeff;
1417 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1418 B = U * c->yuv2rgb_u2b_coeff;
1420 output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1421 output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
1422 output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1424 output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1437 #define YUV2PACKED16WRAPPER(name, base, ext, fmt, hasAlpha, eightbytes) \
1438 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
1439 const int16_t **_lumSrc, int lumFilterSize, \
1440 const int16_t *chrFilter, const int16_t **_chrUSrc, \
1441 const int16_t **_chrVSrc, int chrFilterSize, \
1442 const int16_t **_alpSrc, uint8_t *_dest, int dstW, \
1445 const int32_t **lumSrc = (const int32_t **) _lumSrc, \
1446 **chrUSrc = (const int32_t **) _chrUSrc, \
1447 **chrVSrc = (const int32_t **) _chrVSrc, \
1448 **alpSrc = (const int32_t **) _alpSrc; \
1449 uint16_t *dest = (uint16_t *) _dest; \
1450 name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
1451 chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
1452 alpSrc, dest, dstW, y, fmt, hasAlpha, eightbytes); \
1455 static void name ## ext ## _2_c(SwsContext *c, const int16_t *_buf[2], \
1456 const int16_t *_ubuf[2], const int16_t *_vbuf[2], \
1457 const int16_t *_abuf[2], uint8_t *_dest, int dstW, \
1458 int yalpha, int uvalpha, int y) \
1460 const int32_t **buf = (const int32_t **) _buf, \
1461 **ubuf = (const int32_t **) _ubuf, \
1462 **vbuf = (const int32_t **) _vbuf, \
1463 **abuf = (const int32_t **) _abuf; \
1464 uint16_t *dest = (uint16_t *) _dest; \
1465 name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
1466 dest, dstW, yalpha, uvalpha, y, fmt, hasAlpha, eightbytes); \
1469 static void name ## ext ## _1_c(SwsContext *c, const int16_t *_buf0, \
1470 const int16_t *_ubuf[2], const int16_t *_vbuf[2], \
1471 const int16_t *_abuf0, uint8_t *_dest, int dstW, \
1472 int uvalpha, int y) \
1474 const int32_t *buf0 = (const int32_t *) _buf0, \
1475 **ubuf = (const int32_t **) _ubuf, \
1476 **vbuf = (const int32_t **) _vbuf, \
1477 *abuf0 = (const int32_t *) _abuf0; \
1478 uint16_t *dest = (uint16_t *) _dest; \
1479 name ## base ## _1_c_template(c, buf0, ubuf, vbuf, abuf0, dest, \
1480 dstW, uvalpha, y, fmt, hasAlpha, eightbytes); \
1483 YUV2PACKED16WRAPPER(yuv2, rgba64, rgb48be, AV_PIX_FMT_RGB48BE, 0, 0)
1484 YUV2PACKED16WRAPPER(yuv2, rgba64, rgb48le, AV_PIX_FMT_RGB48LE, 0, 0)
1485 YUV2PACKED16WRAPPER(yuv2, rgba64, bgr48be, AV_PIX_FMT_BGR48BE, 0, 0)
1486 YUV2PACKED16WRAPPER(yuv2, rgba64, bgr48le, AV_PIX_FMT_BGR48LE, 0, 0)
1487 YUV2PACKED16WRAPPER(yuv2, rgba64, rgba64be, AV_PIX_FMT_RGBA64BE, 1, 1)
1488 YUV2PACKED16WRAPPER(yuv2, rgba64, rgba64le, AV_PIX_FMT_RGBA64LE, 1, 1)
1489 YUV2PACKED16WRAPPER(yuv2, rgba64, rgbx64be, AV_PIX_FMT_RGBA64BE, 0, 1)
1490 YUV2PACKED16WRAPPER(yuv2, rgba64, rgbx64le, AV_PIX_FMT_RGBA64LE, 0, 1)
1491 YUV2PACKED16WRAPPER(yuv2, rgba64, bgra64be, AV_PIX_FMT_BGRA64BE, 1, 1)
1492 YUV2PACKED16WRAPPER(yuv2, rgba64, bgra64le, AV_PIX_FMT_BGRA64LE, 1, 1)
1493 YUV2PACKED16WRAPPER(yuv2, rgba64, bgrx64be, AV_PIX_FMT_BGRA64BE, 0, 1)
1494 YUV2PACKED16WRAPPER(yuv2, rgba64, bgrx64le, AV_PIX_FMT_BGRA64LE, 0, 1)
1495 YUV2PACKED16WRAPPER(yuv2, ya16, ya16be, AV_PIX_FMT_YA16BE, 1, 0)
1496 YUV2PACKED16WRAPPER(yuv2, ya16, ya16le, AV_PIX_FMT_YA16LE, 1, 0)
1498 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgb48be_full, AV_PIX_FMT_RGB48BE, 0, 0)
1499 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgb48le_full, AV_PIX_FMT_RGB48LE, 0, 0)
1500 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgr48be_full, AV_PIX_FMT_BGR48BE, 0, 0)
1501 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgr48le_full, AV_PIX_FMT_BGR48LE, 0, 0)
1502 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgba64be_full, AV_PIX_FMT_RGBA64BE, 1, 1)
1503 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgba64le_full, AV_PIX_FMT_RGBA64LE, 1, 1)
1504 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgbx64be_full, AV_PIX_FMT_RGBA64BE, 0, 1)
1505 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgbx64le_full, AV_PIX_FMT_RGBA64LE, 0, 1)
1506 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgra64be_full, AV_PIX_FMT_BGRA64BE, 1, 1)
1507 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgra64le_full, AV_PIX_FMT_BGRA64LE, 1, 1)
1508 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgrx64be_full, AV_PIX_FMT_BGRA64BE, 0, 1)
1509 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgrx64le_full, AV_PIX_FMT_BGRA64LE, 0, 1)
1512 * Write out 2 RGB pixels in the target pixel format. This function takes a
1513 * R/G/B LUT as generated by ff_yuv2rgb_c_init_tables(), which takes care of
1514 * things like endianness conversion and shifting. The caller takes care of
1515 * setting the correct offset in these tables from the chroma (U/V) values.
1516 * This function then uses the luminance (Y1/Y2) values to write out the
1517 * correct RGB values into the destination buffer.
1519 static av_always_inline void
1520 yuv2rgb_write(uint8_t *_dest, int i, int Y1, int Y2,
1521 unsigned A1, unsigned A2,
1522 const void *_r, const void *_g, const void *_b, int y,
1523 enum AVPixelFormat target, int hasAlpha)
1525 if (target == AV_PIX_FMT_ARGB || target == AV_PIX_FMT_RGBA ||
1526 target == AV_PIX_FMT_ABGR || target == AV_PIX_FMT_BGRA) {
1527 uint32_t *dest = (uint32_t *) _dest;
1528 const uint32_t *r = (const uint32_t *) _r;
1529 const uint32_t *g = (const uint32_t *) _g;
1530 const uint32_t *b = (const uint32_t *) _b;
1533 int sh = hasAlpha ? ((target == AV_PIX_FMT_RGB32_1 || target == AV_PIX_FMT_BGR32_1) ? 0 : 24) : 0;
1535 dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1] + (hasAlpha ? A1 << sh : 0);
1536 dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2] + (hasAlpha ? A2 << sh : 0);
1539 int sh = (target == AV_PIX_FMT_RGB32_1 || target == AV_PIX_FMT_BGR32_1) ? 0 : 24;
1541 av_assert2((((r[Y1] + g[Y1] + b[Y1]) >> sh) & 0xFF) == 0);
1542 dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1] + (A1 << sh);
1543 dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2] + (A2 << sh);
1545 #if defined(ASSERT_LEVEL) && ASSERT_LEVEL > 1
1546 int sh = (target == AV_PIX_FMT_RGB32_1 || target == AV_PIX_FMT_BGR32_1) ? 0 : 24;
1548 av_assert2((((r[Y1] + g[Y1] + b[Y1]) >> sh) & 0xFF) == 0xFF);
1550 dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1];
1551 dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2];
1554 } else if (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) {
1555 uint8_t *dest = (uint8_t *) _dest;
1556 const uint8_t *r = (const uint8_t *) _r;
1557 const uint8_t *g = (const uint8_t *) _g;
1558 const uint8_t *b = (const uint8_t *) _b;
1560 #define r_b ((target == AV_PIX_FMT_RGB24) ? r : b)
1561 #define b_r ((target == AV_PIX_FMT_RGB24) ? b : r)
1563 dest[i * 6 + 0] = r_b[Y1];
1564 dest[i * 6 + 1] = g[Y1];
1565 dest[i * 6 + 2] = b_r[Y1];
1566 dest[i * 6 + 3] = r_b[Y2];
1567 dest[i * 6 + 4] = g[Y2];
1568 dest[i * 6 + 5] = b_r[Y2];
1571 } else if (target == AV_PIX_FMT_RGB565 || target == AV_PIX_FMT_BGR565 ||
1572 target == AV_PIX_FMT_RGB555 || target == AV_PIX_FMT_BGR555 ||
1573 target == AV_PIX_FMT_RGB444 || target == AV_PIX_FMT_BGR444) {
1574 uint16_t *dest = (uint16_t *) _dest;
1575 const uint16_t *r = (const uint16_t *) _r;
1576 const uint16_t *g = (const uint16_t *) _g;
1577 const uint16_t *b = (const uint16_t *) _b;
1578 int dr1, dg1, db1, dr2, dg2, db2;
1580 if (target == AV_PIX_FMT_RGB565 || target == AV_PIX_FMT_BGR565) {
1581 dr1 = ff_dither_2x2_8[ y & 1 ][0];
1582 dg1 = ff_dither_2x2_4[ y & 1 ][0];
1583 db1 = ff_dither_2x2_8[(y & 1) ^ 1][0];
1584 dr2 = ff_dither_2x2_8[ y & 1 ][1];
1585 dg2 = ff_dither_2x2_4[ y & 1 ][1];
1586 db2 = ff_dither_2x2_8[(y & 1) ^ 1][1];
1587 } else if (target == AV_PIX_FMT_RGB555 || target == AV_PIX_FMT_BGR555) {
1588 dr1 = ff_dither_2x2_8[ y & 1 ][0];
1589 dg1 = ff_dither_2x2_8[ y & 1 ][1];
1590 db1 = ff_dither_2x2_8[(y & 1) ^ 1][0];
1591 dr2 = ff_dither_2x2_8[ y & 1 ][1];
1592 dg2 = ff_dither_2x2_8[ y & 1 ][0];
1593 db2 = ff_dither_2x2_8[(y & 1) ^ 1][1];
1595 dr1 = ff_dither_4x4_16[ y & 3 ][0];
1596 dg1 = ff_dither_4x4_16[ y & 3 ][1];
1597 db1 = ff_dither_4x4_16[(y & 3) ^ 3][0];
1598 dr2 = ff_dither_4x4_16[ y & 3 ][1];
1599 dg2 = ff_dither_4x4_16[ y & 3 ][0];
1600 db2 = ff_dither_4x4_16[(y & 3) ^ 3][1];
1603 dest[i * 2 + 0] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1];
1604 dest[i * 2 + 1] = r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2];
1605 } else if (target == AV_PIX_FMT_X2RGB10) {
1606 uint32_t *dest = (uint32_t *) _dest;
1607 const uint32_t *r = (const uint32_t *) _r;
1608 const uint32_t *g = (const uint32_t *) _g;
1609 const uint32_t *b = (const uint32_t *) _b;
1610 dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1];
1611 dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2];
1612 } else /* 8/4 bits */ {
1613 uint8_t *dest = (uint8_t *) _dest;
1614 const uint8_t *r = (const uint8_t *) _r;
1615 const uint8_t *g = (const uint8_t *) _g;
1616 const uint8_t *b = (const uint8_t *) _b;
1617 int dr1, dg1, db1, dr2, dg2, db2;
1619 if (target == AV_PIX_FMT_RGB8 || target == AV_PIX_FMT_BGR8) {
1620 const uint8_t * const d64 = ff_dither_8x8_73[y & 7];
1621 const uint8_t * const d32 = ff_dither_8x8_32[y & 7];
1622 dr1 = dg1 = d32[(i * 2 + 0) & 7];
1623 db1 = d64[(i * 2 + 0) & 7];
1624 dr2 = dg2 = d32[(i * 2 + 1) & 7];
1625 db2 = d64[(i * 2 + 1) & 7];
1627 const uint8_t * const d64 = ff_dither_8x8_73 [y & 7];
1628 const uint8_t * const d128 = ff_dither_8x8_220[y & 7];
1629 dr1 = db1 = d128[(i * 2 + 0) & 7];
1630 dg1 = d64[(i * 2 + 0) & 7];
1631 dr2 = db2 = d128[(i * 2 + 1) & 7];
1632 dg2 = d64[(i * 2 + 1) & 7];
1635 if (target == AV_PIX_FMT_RGB4 || target == AV_PIX_FMT_BGR4) {
1636 dest[i] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1] +
1637 ((r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2]) << 4);
1639 dest[i * 2 + 0] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1];
1640 dest[i * 2 + 1] = r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2];
1645 static av_always_inline void
1646 yuv2rgb_X_c_template(SwsContext *c, const int16_t *lumFilter,
1647 const int16_t **lumSrc, int lumFilterSize,
1648 const int16_t *chrFilter, const int16_t **chrUSrc,
1649 const int16_t **chrVSrc, int chrFilterSize,
1650 const int16_t **alpSrc, uint8_t *dest, int dstW,
1651 int y, enum AVPixelFormat target, int hasAlpha)
1655 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1661 const void *r, *g, *b;
1663 for (j = 0; j < lumFilterSize; j++) {
1664 Y1 += lumSrc[j][i * 2] * lumFilter[j];
1665 Y2 += lumSrc[j][i * 2 + 1] * lumFilter[j];
1667 for (j = 0; j < chrFilterSize; j++) {
1668 U += chrUSrc[j][i] * chrFilter[j];
1669 V += chrVSrc[j][i] * chrFilter[j];
1678 for (j = 0; j < lumFilterSize; j++) {
1679 A1 += alpSrc[j][i * 2 ] * lumFilter[j];
1680 A2 += alpSrc[j][i * 2 + 1] * lumFilter[j];
1684 if ((A1 | A2) & 0x100) {
1685 A1 = av_clip_uint8(A1);
1686 A2 = av_clip_uint8(A2);
1690 r = c->table_rV[V + YUVRGB_TABLE_HEADROOM];
1691 g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]);
1692 b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1694 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1695 r, g, b, y, target, hasAlpha);
1699 static av_always_inline void
1700 yuv2rgb_2_c_template(SwsContext *c, const int16_t *buf[2],
1701 const int16_t *ubuf[2], const int16_t *vbuf[2],
1702 const int16_t *abuf[2], uint8_t *dest, int dstW,
1703 int yalpha, int uvalpha, int y,
1704 enum AVPixelFormat target, int hasAlpha)
1706 const int16_t *buf0 = buf[0], *buf1 = buf[1],
1707 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
1708 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
1709 *abuf0 = hasAlpha ? abuf[0] : NULL,
1710 *abuf1 = hasAlpha ? abuf[1] : NULL;
1711 int yalpha1 = 4096 - yalpha;
1712 int uvalpha1 = 4096 - uvalpha;
1714 av_assert2(yalpha <= 4096U);
1715 av_assert2(uvalpha <= 4096U);
1717 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1718 int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 19;
1719 int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 19;
1720 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha) >> 19;
1721 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha) >> 19;
1723 const void *r = c->table_rV[V + YUVRGB_TABLE_HEADROOM],
1724 *g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]),
1725 *b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1728 A1 = (abuf0[i * 2 ] * yalpha1 + abuf1[i * 2 ] * yalpha) >> 19;
1729 A2 = (abuf0[i * 2 + 1] * yalpha1 + abuf1[i * 2 + 1] * yalpha) >> 19;
1730 A1 = av_clip_uint8(A1);
1731 A2 = av_clip_uint8(A2);
1734 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1735 r, g, b, y, target, hasAlpha);
1739 static av_always_inline void
1740 yuv2rgb_1_c_template(SwsContext *c, const int16_t *buf0,
1741 const int16_t *ubuf[2], const int16_t *vbuf[2],
1742 const int16_t *abuf0, uint8_t *dest, int dstW,
1743 int uvalpha, int y, enum AVPixelFormat target,
1746 const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1749 if (uvalpha < 2048) {
1750 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1751 int Y1 = (buf0[i * 2 ] + 64) >> 7;
1752 int Y2 = (buf0[i * 2 + 1] + 64) >> 7;
1753 int U = (ubuf0[i] + 64) >> 7;
1754 int V = (vbuf0[i] + 64) >> 7;
1756 const void *r = c->table_rV[V + YUVRGB_TABLE_HEADROOM],
1757 *g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]),
1758 *b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1761 A1 = abuf0[i * 2 ] * 255 + 16384 >> 15;
1762 A2 = abuf0[i * 2 + 1] * 255 + 16384 >> 15;
1763 A1 = av_clip_uint8(A1);
1764 A2 = av_clip_uint8(A2);
1767 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1768 r, g, b, y, target, hasAlpha);
1771 const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1772 for (i = 0; i < ((dstW + 1) >> 1); i++) {
1773 int Y1 = (buf0[i * 2 ] + 64) >> 7;
1774 int Y2 = (buf0[i * 2 + 1] + 64) >> 7;
1775 int U = (ubuf0[i] + ubuf1[i] + 128) >> 8;
1776 int V = (vbuf0[i] + vbuf1[i] + 128) >> 8;
1778 const void *r = c->table_rV[V + YUVRGB_TABLE_HEADROOM],
1779 *g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]),
1780 *b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1783 A1 = (abuf0[i * 2 ] + 64) >> 7;
1784 A2 = (abuf0[i * 2 + 1] + 64) >> 7;
1785 A1 = av_clip_uint8(A1);
1786 A2 = av_clip_uint8(A2);
1789 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1790 r, g, b, y, target, hasAlpha);
1795 #define YUV2RGBWRAPPERX(name, base, ext, fmt, hasAlpha) \
1796 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
1797 const int16_t **lumSrc, int lumFilterSize, \
1798 const int16_t *chrFilter, const int16_t **chrUSrc, \
1799 const int16_t **chrVSrc, int chrFilterSize, \
1800 const int16_t **alpSrc, uint8_t *dest, int dstW, \
1803 name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
1804 chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
1805 alpSrc, dest, dstW, y, fmt, hasAlpha); \
1808 #define YUV2RGBWRAPPERX2(name, base, ext, fmt, hasAlpha) \
1809 YUV2RGBWRAPPERX(name, base, ext, fmt, hasAlpha) \
1810 static void name ## ext ## _2_c(SwsContext *c, const int16_t *buf[2], \
1811 const int16_t *ubuf[2], const int16_t *vbuf[2], \
1812 const int16_t *abuf[2], uint8_t *dest, int dstW, \
1813 int yalpha, int uvalpha, int y) \
1815 name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
1816 dest, dstW, yalpha, uvalpha, y, fmt, hasAlpha); \
1819 #define YUV2RGBWRAPPER(name, base, ext, fmt, hasAlpha) \
1820 YUV2RGBWRAPPERX2(name, base, ext, fmt, hasAlpha) \
1821 static void name ## ext ## _1_c(SwsContext *c, const int16_t *buf0, \
1822 const int16_t *ubuf[2], const int16_t *vbuf[2], \
1823 const int16_t *abuf0, uint8_t *dest, int dstW, \
1824 int uvalpha, int y) \
1826 name ## base ## _1_c_template(c, buf0, ubuf, vbuf, abuf0, dest, \
1827 dstW, uvalpha, y, fmt, hasAlpha); \
1831 YUV2RGBWRAPPER(yuv2rgb,, 32_1, AV_PIX_FMT_RGB32_1, CONFIG_SWSCALE_ALPHA && c->needAlpha)
1832 YUV2RGBWRAPPER(yuv2rgb,, 32, AV_PIX_FMT_RGB32, CONFIG_SWSCALE_ALPHA && c->needAlpha)
1834 #if CONFIG_SWSCALE_ALPHA
1835 YUV2RGBWRAPPER(yuv2rgb,, a32_1, AV_PIX_FMT_RGB32_1, 1)
1836 YUV2RGBWRAPPER(yuv2rgb,, a32, AV_PIX_FMT_RGB32, 1)
1838 YUV2RGBWRAPPER(yuv2rgb,, x32_1, AV_PIX_FMT_RGB32_1, 0)
1839 YUV2RGBWRAPPER(yuv2rgb,, x32, AV_PIX_FMT_RGB32, 0)
1841 YUV2RGBWRAPPER(yuv2, rgb, rgb24, AV_PIX_FMT_RGB24, 0)
1842 YUV2RGBWRAPPER(yuv2, rgb, bgr24, AV_PIX_FMT_BGR24, 0)
1843 YUV2RGBWRAPPER(yuv2rgb,, 16, AV_PIX_FMT_RGB565, 0)
1844 YUV2RGBWRAPPER(yuv2rgb,, 15, AV_PIX_FMT_RGB555, 0)
1845 YUV2RGBWRAPPER(yuv2rgb,, 12, AV_PIX_FMT_RGB444, 0)
1846 YUV2RGBWRAPPER(yuv2rgb,, 8, AV_PIX_FMT_RGB8, 0)
1847 YUV2RGBWRAPPER(yuv2rgb,, 4, AV_PIX_FMT_RGB4, 0)
1848 YUV2RGBWRAPPER(yuv2rgb,, 4b, AV_PIX_FMT_RGB4_BYTE, 0)
1849 YUV2RGBWRAPPER(yuv2, rgb, x2rgb10, AV_PIX_FMT_X2RGB10, 0)
1851 static av_always_inline void yuv2rgb_write_full(SwsContext *c,
1852 uint8_t *dest, int i, int Y, int A, int U, int V,
1853 int y, enum AVPixelFormat target, int hasAlpha, int err[4])
1856 int isrgb8 = target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8;
1858 Y -= c->yuv2rgb_y_offset;
1859 Y *= c->yuv2rgb_y_coeff;
1861 R = (unsigned)Y + V*c->yuv2rgb_v2r_coeff;
1862 G = (unsigned)Y + V*c->yuv2rgb_v2g_coeff + U*c->yuv2rgb_u2g_coeff;
1863 B = (unsigned)Y + U*c->yuv2rgb_u2b_coeff;
1864 if ((R | G | B) & 0xC0000000) {
1865 R = av_clip_uintp2(R, 30);
1866 G = av_clip_uintp2(G, 30);
1867 B = av_clip_uintp2(B, 30);
1871 case AV_PIX_FMT_ARGB:
1872 dest[0] = hasAlpha ? A : 255;
1877 case AV_PIX_FMT_RGB24:
1882 case AV_PIX_FMT_RGBA:
1886 dest[3] = hasAlpha ? A : 255;
1888 case AV_PIX_FMT_ABGR:
1889 dest[0] = hasAlpha ? A : 255;
1894 case AV_PIX_FMT_BGR24:
1899 case AV_PIX_FMT_BGRA:
1903 dest[3] = hasAlpha ? A : 255;
1905 case AV_PIX_FMT_BGR4_BYTE:
1906 case AV_PIX_FMT_RGB4_BYTE:
1907 case AV_PIX_FMT_BGR8:
1908 case AV_PIX_FMT_RGB8:
1912 switch (c->dither) {
1914 case SWS_DITHER_AUTO:
1919 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;
1920 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;
1921 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;
1922 c->dither_error[0][i] = err[0];
1923 c->dither_error[1][i] = err[1];
1924 c->dither_error[2][i] = err[2];
1925 r = R >> (isrgb8 ? 5 : 7);
1926 g = G >> (isrgb8 ? 5 : 6);
1927 b = B >> (isrgb8 ? 6 : 7);
1928 r = av_clip(r, 0, isrgb8 ? 7 : 1);
1929 g = av_clip(g, 0, isrgb8 ? 7 : 3);
1930 b = av_clip(b, 0, isrgb8 ? 3 : 1);
1931 err[0] = R - r*(isrgb8 ? 36 : 255);
1932 err[1] = G - g*(isrgb8 ? 36 : 85);
1933 err[2] = B - b*(isrgb8 ? 85 : 255);
1935 case SWS_DITHER_A_DITHER:
1937 /* see http://pippin.gimp.org/a_dither/ for details/origin */
1938 #define A_DITHER(u,v) (((((u)+((v)*236))*119)&0xff))
1939 r = (((R >> 19) + A_DITHER(i,y) -96)>>8);
1940 g = (((G >> 19) + A_DITHER(i + 17,y) - 96)>>8);
1941 b = (((B >> 20) + A_DITHER(i + 17*2,y) -96)>>8);
1942 r = av_clip_uintp2(r, 3);
1943 g = av_clip_uintp2(g, 3);
1944 b = av_clip_uintp2(b, 2);
1946 r = (((R >> 21) + A_DITHER(i,y)-256)>>8);
1947 g = (((G >> 19) + A_DITHER(i + 17,y)-256)>>8);
1948 b = (((B >> 21) + A_DITHER(i + 17*2,y)-256)>>8);
1949 r = av_clip_uintp2(r, 1);
1950 g = av_clip_uintp2(g, 2);
1951 b = av_clip_uintp2(b, 1);
1954 case SWS_DITHER_X_DITHER:
1956 /* see http://pippin.gimp.org/a_dither/ for details/origin */
1957 #define X_DITHER(u,v) (((((u)^((v)*237))*181)&0x1ff)/2)
1958 r = (((R >> 19) + X_DITHER(i,y) - 96)>>8);
1959 g = (((G >> 19) + X_DITHER(i + 17,y) - 96)>>8);
1960 b = (((B >> 20) + X_DITHER(i + 17*2,y) - 96)>>8);
1961 r = av_clip_uintp2(r, 3);
1962 g = av_clip_uintp2(g, 3);
1963 b = av_clip_uintp2(b, 2);
1965 r = (((R >> 21) + X_DITHER(i,y)-256)>>8);
1966 g = (((G >> 19) + X_DITHER(i + 17,y)-256)>>8);
1967 b = (((B >> 21) + X_DITHER(i + 17*2,y)-256)>>8);
1968 r = av_clip_uintp2(r, 1);
1969 g = av_clip_uintp2(g, 2);
1970 b = av_clip_uintp2(b, 1);
1976 if(target == AV_PIX_FMT_BGR4_BYTE) {
1977 dest[0] = r + 2*g + 8*b;
1978 } else if(target == AV_PIX_FMT_RGB4_BYTE) {
1979 dest[0] = b + 2*g + 8*r;
1980 } else if(target == AV_PIX_FMT_BGR8) {
1981 dest[0] = r + 8*g + 64*b;
1982 } else if(target == AV_PIX_FMT_RGB8) {
1983 dest[0] = b + 4*g + 32*r;
1990 static av_always_inline void
1991 yuv2rgb_full_X_c_template(SwsContext *c, const int16_t *lumFilter,
1992 const int16_t **lumSrc, int lumFilterSize,
1993 const int16_t *chrFilter, const int16_t **chrUSrc,
1994 const int16_t **chrVSrc, int chrFilterSize,
1995 const int16_t **alpSrc, uint8_t *dest,
1996 int dstW, int y, enum AVPixelFormat target, int hasAlpha)
1999 int step = (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) ? 3 : 4;
2001 int A = 0; //init to silence warning
2003 if( target == AV_PIX_FMT_BGR4_BYTE || target == AV_PIX_FMT_RGB4_BYTE
2004 || target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8)
2007 for (i = 0; i < dstW; i++) {
2010 int U = (1<<9)-(128 << 19);
2011 int V = (1<<9)-(128 << 19);
2013 for (j = 0; j < lumFilterSize; j++) {
2014 Y += lumSrc[j][i] * lumFilter[j];
2016 for (j = 0; j < chrFilterSize; j++) {
2017 U += chrUSrc[j][i] * chrFilter[j];
2018 V += chrVSrc[j][i] * chrFilter[j];
2025 for (j = 0; j < lumFilterSize; j++) {
2026 A += alpSrc[j][i] * lumFilter[j];
2030 A = av_clip_uint8(A);
2032 yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
2035 c->dither_error[0][i] = err[0];
2036 c->dither_error[1][i] = err[1];
2037 c->dither_error[2][i] = err[2];
2040 static av_always_inline void
2041 yuv2rgb_full_2_c_template(SwsContext *c, const int16_t *buf[2],
2042 const int16_t *ubuf[2], const int16_t *vbuf[2],
2043 const int16_t *abuf[2], uint8_t *dest, int dstW,
2044 int yalpha, int uvalpha, int y,
2045 enum AVPixelFormat target, int hasAlpha)
2047 const int16_t *buf0 = buf[0], *buf1 = buf[1],
2048 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
2049 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
2050 *abuf0 = hasAlpha ? abuf[0] : NULL,
2051 *abuf1 = hasAlpha ? abuf[1] : NULL;
2052 int yalpha1 = 4096 - yalpha;
2053 int uvalpha1 = 4096 - uvalpha;
2055 int step = (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) ? 3 : 4;
2057 int A = 0; // init to silcene warning
2059 av_assert2(yalpha <= 4096U);
2060 av_assert2(uvalpha <= 4096U);
2062 if( target == AV_PIX_FMT_BGR4_BYTE || target == AV_PIX_FMT_RGB4_BYTE
2063 || target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8)
2066 for (i = 0; i < dstW; i++) {
2067 int Y = ( buf0[i] * yalpha1 + buf1[i] * yalpha ) >> 10; //FIXME rounding
2068 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha-(128 << 19)) >> 10;
2069 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha-(128 << 19)) >> 10;
2072 A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha + (1<<18)) >> 19;
2074 A = av_clip_uint8(A);
2077 yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
2080 c->dither_error[0][i] = err[0];
2081 c->dither_error[1][i] = err[1];
2082 c->dither_error[2][i] = err[2];
2085 static av_always_inline void
2086 yuv2rgb_full_1_c_template(SwsContext *c, const int16_t *buf0,
2087 const int16_t *ubuf[2], const int16_t *vbuf[2],
2088 const int16_t *abuf0, uint8_t *dest, int dstW,
2089 int uvalpha, int y, enum AVPixelFormat target,
2092 const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
2094 int step = (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) ? 3 : 4;
2097 if( target == AV_PIX_FMT_BGR4_BYTE || target == AV_PIX_FMT_RGB4_BYTE
2098 || target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8)
2101 if (uvalpha < 2048) {
2102 int A = 0; //init to silence warning
2103 for (i = 0; i < dstW; i++) {
2104 int Y = buf0[i] * 4;
2105 int U = (ubuf0[i] - (128<<7)) * 4;
2106 int V = (vbuf0[i] - (128<<7)) * 4;
2109 A = (abuf0[i] + 64) >> 7;
2111 A = av_clip_uint8(A);
2114 yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
2118 const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
2119 int A = 0; //init to silence warning
2120 for (i = 0; i < dstW; i++) {
2121 int Y = buf0[i] * 4;
2122 int U = (ubuf0[i] + ubuf1[i] - (128<<8)) * 2;
2123 int V = (vbuf0[i] + vbuf1[i] - (128<<8)) * 2;
2126 A = (abuf0[i] + 64) >> 7;
2128 A = av_clip_uint8(A);
2131 yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
2136 c->dither_error[0][i] = err[0];
2137 c->dither_error[1][i] = err[1];
2138 c->dither_error[2][i] = err[2];
2142 YUV2RGBWRAPPER(yuv2, rgb_full, bgra32_full, AV_PIX_FMT_BGRA, CONFIG_SWSCALE_ALPHA && c->needAlpha)
2143 YUV2RGBWRAPPER(yuv2, rgb_full, abgr32_full, AV_PIX_FMT_ABGR, CONFIG_SWSCALE_ALPHA && c->needAlpha)
2144 YUV2RGBWRAPPER(yuv2, rgb_full, rgba32_full, AV_PIX_FMT_RGBA, CONFIG_SWSCALE_ALPHA && c->needAlpha)
2145 YUV2RGBWRAPPER(yuv2, rgb_full, argb32_full, AV_PIX_FMT_ARGB, CONFIG_SWSCALE_ALPHA && c->needAlpha)
2147 #if CONFIG_SWSCALE_ALPHA
2148 YUV2RGBWRAPPER(yuv2, rgb_full, bgra32_full, AV_PIX_FMT_BGRA, 1)
2149 YUV2RGBWRAPPER(yuv2, rgb_full, abgr32_full, AV_PIX_FMT_ABGR, 1)
2150 YUV2RGBWRAPPER(yuv2, rgb_full, rgba32_full, AV_PIX_FMT_RGBA, 1)
2151 YUV2RGBWRAPPER(yuv2, rgb_full, argb32_full, AV_PIX_FMT_ARGB, 1)
2153 YUV2RGBWRAPPER(yuv2, rgb_full, bgrx32_full, AV_PIX_FMT_BGRA, 0)
2154 YUV2RGBWRAPPER(yuv2, rgb_full, xbgr32_full, AV_PIX_FMT_ABGR, 0)
2155 YUV2RGBWRAPPER(yuv2, rgb_full, rgbx32_full, AV_PIX_FMT_RGBA, 0)
2156 YUV2RGBWRAPPER(yuv2, rgb_full, xrgb32_full, AV_PIX_FMT_ARGB, 0)
2158 YUV2RGBWRAPPER(yuv2, rgb_full, bgr24_full, AV_PIX_FMT_BGR24, 0)
2159 YUV2RGBWRAPPER(yuv2, rgb_full, rgb24_full, AV_PIX_FMT_RGB24, 0)
2161 YUV2RGBWRAPPER(yuv2, rgb_full, bgr4_byte_full, AV_PIX_FMT_BGR4_BYTE, 0)
2162 YUV2RGBWRAPPER(yuv2, rgb_full, rgb4_byte_full, AV_PIX_FMT_RGB4_BYTE, 0)
2163 YUV2RGBWRAPPER(yuv2, rgb_full, bgr8_full, AV_PIX_FMT_BGR8, 0)
2164 YUV2RGBWRAPPER(yuv2, rgb_full, rgb8_full, AV_PIX_FMT_RGB8, 0)
2167 yuv2gbrp_full_X_c(SwsContext *c, const int16_t *lumFilter,
2168 const int16_t **lumSrc, int lumFilterSize,
2169 const int16_t *chrFilter, const int16_t **chrUSrc,
2170 const int16_t **chrVSrc, int chrFilterSize,
2171 const int16_t **alpSrc, uint8_t **dest,
2174 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(c->dstFormat);
2176 int hasAlpha = (desc->flags & AV_PIX_FMT_FLAG_ALPHA) && alpSrc;
2177 uint16_t **dest16 = (uint16_t**)dest;
2178 int SH = 22 + 8 - desc->comp[0].depth;
2179 int A = 0; // init to silence warning
2181 for (i = 0; i < dstW; i++) {
2184 int U = (1 << 9) - (128 << 19);
2185 int V = (1 << 9) - (128 << 19);
2188 for (j = 0; j < lumFilterSize; j++)
2189 Y += lumSrc[j][i] * lumFilter[j];
2191 for (j = 0; j < chrFilterSize; j++) {
2192 U += chrUSrc[j][i] * chrFilter[j];
2193 V += chrVSrc[j][i] * chrFilter[j];
2203 for (j = 0; j < lumFilterSize; j++)
2204 A += alpSrc[j][i] * lumFilter[j];
2207 A = av_clip_uintp2(A, 27);
2210 Y -= c->yuv2rgb_y_offset;
2211 Y *= c->yuv2rgb_y_coeff;
2213 R = Y + V * c->yuv2rgb_v2r_coeff;
2214 G = Y + V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
2215 B = Y + U * c->yuv2rgb_u2b_coeff;
2217 if ((R | G | B) & 0xC0000000) {
2218 R = av_clip_uintp2(R, 30);
2219 G = av_clip_uintp2(G, 30);
2220 B = av_clip_uintp2(B, 30);
2224 dest16[0][i] = G >> SH;
2225 dest16[1][i] = B >> SH;
2226 dest16[2][i] = R >> SH;
2228 dest16[3][i] = A >> (SH - 3);
2230 dest[0][i] = G >> 22;
2231 dest[1][i] = B >> 22;
2232 dest[2][i] = R >> 22;
2234 dest[3][i] = A >> 19;
2237 if (SH != 22 && (!isBE(c->dstFormat)) != (!HAVE_BIGENDIAN)) {
2238 for (i = 0; i < dstW; i++) {
2239 dest16[0][i] = av_bswap16(dest16[0][i]);
2240 dest16[1][i] = av_bswap16(dest16[1][i]);
2241 dest16[2][i] = av_bswap16(dest16[2][i]);
2243 dest16[3][i] = av_bswap16(dest16[3][i]);
2249 yuv2gbrp16_full_X_c(SwsContext *c, const int16_t *lumFilter,
2250 const int16_t **lumSrcx, int lumFilterSize,
2251 const int16_t *chrFilter, const int16_t **chrUSrcx,
2252 const int16_t **chrVSrcx, int chrFilterSize,
2253 const int16_t **alpSrcx, uint8_t **dest,
2256 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(c->dstFormat);
2258 int hasAlpha = (desc->flags & AV_PIX_FMT_FLAG_ALPHA) && alpSrcx;
2259 uint16_t **dest16 = (uint16_t**)dest;
2260 const int32_t **lumSrc = (const int32_t**)lumSrcx;
2261 const int32_t **chrUSrc = (const int32_t**)chrUSrcx;
2262 const int32_t **chrVSrc = (const int32_t**)chrVSrcx;
2263 const int32_t **alpSrc = (const int32_t**)alpSrcx;
2265 for (i = 0; i < dstW; i++) {
2267 int Y = -0x40000000;
2268 int U = -(128 << 23);
2269 int V = -(128 << 23);
2272 for (j = 0; j < lumFilterSize; j++)
2273 Y += lumSrc[j][i] * (unsigned)lumFilter[j];
2275 for (j = 0; j < chrFilterSize; j++) {
2276 U += chrUSrc[j][i] * (unsigned)chrFilter[j];
2277 V += chrVSrc[j][i] * (unsigned)chrFilter[j];
2288 for (j = 0; j < lumFilterSize; j++)
2289 A += alpSrc[j][i] * (unsigned)lumFilter[j];
2295 Y -= c->yuv2rgb_y_offset;
2296 Y *= c->yuv2rgb_y_coeff;
2298 R = V * c->yuv2rgb_v2r_coeff;
2299 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
2300 B = U * c->yuv2rgb_u2b_coeff;
2302 R = av_clip_uintp2(Y + R, 30);
2303 G = av_clip_uintp2(Y + G, 30);
2304 B = av_clip_uintp2(Y + B, 30);
2306 dest16[0][i] = G >> 14;
2307 dest16[1][i] = B >> 14;
2308 dest16[2][i] = R >> 14;
2310 dest16[3][i] = av_clip_uintp2(A, 30) >> 14;
2312 if ((!isBE(c->dstFormat)) != (!HAVE_BIGENDIAN)) {
2313 for (i = 0; i < dstW; i++) {
2314 dest16[0][i] = av_bswap16(dest16[0][i]);
2315 dest16[1][i] = av_bswap16(dest16[1][i]);
2316 dest16[2][i] = av_bswap16(dest16[2][i]);
2318 dest16[3][i] = av_bswap16(dest16[3][i]);
2324 yuv2gbrpf32_full_X_c(SwsContext *c, const int16_t *lumFilter,
2325 const int16_t **lumSrcx, int lumFilterSize,
2326 const int16_t *chrFilter, const int16_t **chrUSrcx,
2327 const int16_t **chrVSrcx, int chrFilterSize,
2328 const int16_t **alpSrcx, uint8_t **dest,
2331 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(c->dstFormat);
2333 int hasAlpha = (desc->flags & AV_PIX_FMT_FLAG_ALPHA) && alpSrcx;
2334 uint32_t **dest32 = (uint32_t**)dest;
2335 const int32_t **lumSrc = (const int32_t**)lumSrcx;
2336 const int32_t **chrUSrc = (const int32_t**)chrUSrcx;
2337 const int32_t **chrVSrc = (const int32_t**)chrVSrcx;
2338 const int32_t **alpSrc = (const int32_t**)alpSrcx;
2339 static const float float_mult = 1.0f / 65535.0f;
2341 for (i = 0; i < dstW; i++) {
2343 int Y = -0x40000000;
2344 int U = -(128 << 23);
2345 int V = -(128 << 23);
2348 for (j = 0; j < lumFilterSize; j++)
2349 Y += lumSrc[j][i] * (unsigned)lumFilter[j];
2351 for (j = 0; j < chrFilterSize; j++) {
2352 U += chrUSrc[j][i] * (unsigned)chrFilter[j];
2353 V += chrVSrc[j][i] * (unsigned)chrFilter[j];
2364 for (j = 0; j < lumFilterSize; j++)
2365 A += alpSrc[j][i] * (unsigned)lumFilter[j];
2371 Y -= c->yuv2rgb_y_offset;
2372 Y *= c->yuv2rgb_y_coeff;
2374 R = V * c->yuv2rgb_v2r_coeff;
2375 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
2376 B = U * c->yuv2rgb_u2b_coeff;
2378 R = av_clip_uintp2(Y + R, 30);
2379 G = av_clip_uintp2(Y + G, 30);
2380 B = av_clip_uintp2(Y + B, 30);
2382 dest32[0][i] = av_float2int(float_mult * (float)(G >> 14));
2383 dest32[1][i] = av_float2int(float_mult * (float)(B >> 14));
2384 dest32[2][i] = av_float2int(float_mult * (float)(R >> 14));
2386 dest32[3][i] = av_float2int(float_mult * (float)(av_clip_uintp2(A, 30) >> 14));
2388 if ((!isBE(c->dstFormat)) != (!HAVE_BIGENDIAN)) {
2389 for (i = 0; i < dstW; i++) {
2390 dest32[0][i] = av_bswap32(dest32[0][i]);
2391 dest32[1][i] = av_bswap32(dest32[1][i]);
2392 dest32[2][i] = av_bswap32(dest32[2][i]);
2394 dest32[3][i] = av_bswap32(dest32[3][i]);
2400 yuv2ya8_1_c(SwsContext *c, const int16_t *buf0,
2401 const int16_t *ubuf[2], const int16_t *vbuf[2],
2402 const int16_t *abuf0, uint8_t *dest, int dstW,
2405 int hasAlpha = !!abuf0;
2408 for (i = 0; i < dstW; i++) {
2409 int Y = (buf0[i] + 64) >> 7;
2412 Y = av_clip_uint8(Y);
2415 A = (abuf0[i] + 64) >> 7;
2417 A = av_clip_uint8(A);
2421 dest[i * 2 + 1] = hasAlpha ? A : 255;
2426 yuv2ya8_2_c(SwsContext *c, const int16_t *buf[2],
2427 const int16_t *ubuf[2], const int16_t *vbuf[2],
2428 const int16_t *abuf[2], uint8_t *dest, int dstW,
2429 int yalpha, int uvalpha, int y)
2431 int hasAlpha = abuf && abuf[0] && abuf[1];
2432 const int16_t *buf0 = buf[0], *buf1 = buf[1],
2433 *abuf0 = hasAlpha ? abuf[0] : NULL,
2434 *abuf1 = hasAlpha ? abuf[1] : NULL;
2435 int yalpha1 = 4096 - yalpha;
2438 av_assert2(yalpha <= 4096U);
2440 for (i = 0; i < dstW; i++) {
2441 int Y = (buf0[i] * yalpha1 + buf1[i] * yalpha) >> 19;
2444 Y = av_clip_uint8(Y);
2447 A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha) >> 19;
2448 A = av_clip_uint8(A);
2452 dest[i * 2 + 1] = hasAlpha ? A : 255;
2457 yuv2ya8_X_c(SwsContext *c, const int16_t *lumFilter,
2458 const int16_t **lumSrc, int lumFilterSize,
2459 const int16_t *chrFilter, const int16_t **chrUSrc,
2460 const int16_t **chrVSrc, int chrFilterSize,
2461 const int16_t **alpSrc, uint8_t *dest, int dstW, int y)
2463 int hasAlpha = !!alpSrc;
2466 for (i = 0; i < dstW; i++) {
2468 int Y = 1 << 18, A = 1 << 18;
2470 for (j = 0; j < lumFilterSize; j++)
2471 Y += lumSrc[j][i] * lumFilter[j];
2475 Y = av_clip_uint8(Y);
2478 for (j = 0; j < lumFilterSize; j++)
2479 A += alpSrc[j][i] * lumFilter[j];
2484 A = av_clip_uint8(A);
2488 dest[2 * i + 1] = hasAlpha ? A : 255;
2493 yuv2ayuv64le_X_c(SwsContext *c, const int16_t *lumFilter,
2494 const int16_t **_lumSrc, int lumFilterSize,
2495 const int16_t *chrFilter, const int16_t **_chrUSrc,
2496 const int16_t **_chrVSrc, int chrFilterSize,
2497 const int16_t **_alpSrc, uint8_t *dest, int dstW, int y)
2499 const int32_t **lumSrc = (const int32_t **) _lumSrc,
2500 **chrUSrc = (const int32_t **) _chrUSrc,
2501 **chrVSrc = (const int32_t **) _chrVSrc,
2502 **alpSrc = (const int32_t **) _alpSrc;
2503 int hasAlpha = !!alpSrc;
2506 for (i = 0; i < dstW; i++) {
2507 int Y = 1 << 14, U = 1 << 14;
2508 int V = 1 << 14, A = 1 << 14;
2516 for (j = 0; j < lumFilterSize; j++)
2517 Y += lumSrc[j][i] * (unsigned)lumFilter[j];
2519 for (j = 0; j < chrFilterSize; j++)
2520 U += chrUSrc[j][i] * (unsigned)chrFilter[j];
2522 for (j = 0; j < chrFilterSize; j++)
2523 V += chrVSrc[j][i] * (unsigned)chrFilter[j];
2526 for (j = 0; j < lumFilterSize; j++)
2527 A += alpSrc[j][i] * (unsigned)lumFilter[j];
2529 Y = 0x8000 + av_clip_int16(Y >> 15);
2530 U = 0x8000 + av_clip_int16(U >> 15);
2531 V = 0x8000 + av_clip_int16(V >> 15);
2532 A = 0x8000 + av_clip_int16(A >> 15);
2534 AV_WL16(dest + 8 * i, hasAlpha ? A : 65535);
2535 AV_WL16(dest + 8 * i + 2, Y);
2536 AV_WL16(dest + 8 * i + 4, U);
2537 AV_WL16(dest + 8 * i + 6, V);
2541 av_cold void ff_sws_init_output_funcs(SwsContext *c,
2542 yuv2planar1_fn *yuv2plane1,
2543 yuv2planarX_fn *yuv2planeX,
2544 yuv2interleavedX_fn *yuv2nv12cX,
2545 yuv2packed1_fn *yuv2packed1,
2546 yuv2packed2_fn *yuv2packed2,
2547 yuv2packedX_fn *yuv2packedX,
2548 yuv2anyX_fn *yuv2anyX)
2550 enum AVPixelFormat dstFormat = c->dstFormat;
2551 const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(dstFormat);
2553 if (dstFormat == AV_PIX_FMT_P010LE || dstFormat == AV_PIX_FMT_P010BE) {
2554 *yuv2plane1 = isBE(dstFormat) ? yuv2p010l1_BE_c : yuv2p010l1_LE_c;
2555 *yuv2planeX = isBE(dstFormat) ? yuv2p010lX_BE_c : yuv2p010lX_LE_c;
2556 *yuv2nv12cX = yuv2p010cX_c;
2557 } else if (is16BPS(dstFormat)) {
2558 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_16BE_c : yuv2planeX_16LE_c;
2559 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_16BE_c : yuv2plane1_16LE_c;
2560 if (dstFormat == AV_PIX_FMT_P016LE || dstFormat == AV_PIX_FMT_P016BE) {
2561 *yuv2nv12cX = yuv2p016cX_c;
2563 } else if (isNBPS(dstFormat)) {
2564 if (desc->comp[0].depth == 9) {
2565 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_9BE_c : yuv2planeX_9LE_c;
2566 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_9BE_c : yuv2plane1_9LE_c;
2567 } else if (desc->comp[0].depth == 10) {
2568 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_10BE_c : yuv2planeX_10LE_c;
2569 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_10BE_c : yuv2plane1_10LE_c;
2570 } else if (desc->comp[0].depth == 12) {
2571 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_12BE_c : yuv2planeX_12LE_c;
2572 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_12BE_c : yuv2plane1_12LE_c;
2573 } else if (desc->comp[0].depth == 14) {
2574 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_14BE_c : yuv2planeX_14LE_c;
2575 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_14BE_c : yuv2plane1_14LE_c;
2578 } else if (dstFormat == AV_PIX_FMT_GRAYF32BE) {
2579 *yuv2planeX = yuv2planeX_floatBE_c;
2580 *yuv2plane1 = yuv2plane1_floatBE_c;
2581 } else if (dstFormat == AV_PIX_FMT_GRAYF32LE) {
2582 *yuv2planeX = yuv2planeX_floatLE_c;
2583 *yuv2plane1 = yuv2plane1_floatLE_c;
2585 *yuv2plane1 = yuv2plane1_8_c;
2586 *yuv2planeX = yuv2planeX_8_c;
2587 if (dstFormat == AV_PIX_FMT_NV12 || dstFormat == AV_PIX_FMT_NV21 ||
2588 dstFormat == AV_PIX_FMT_NV24 || dstFormat == AV_PIX_FMT_NV42)
2589 *yuv2nv12cX = yuv2nv12cX_c;
2592 if(c->flags & SWS_FULL_CHR_H_INT) {
2593 switch (dstFormat) {
2594 case AV_PIX_FMT_RGBA:
2596 *yuv2packedX = yuv2rgba32_full_X_c;
2597 *yuv2packed2 = yuv2rgba32_full_2_c;
2598 *yuv2packed1 = yuv2rgba32_full_1_c;
2600 #if CONFIG_SWSCALE_ALPHA
2602 *yuv2packedX = yuv2rgba32_full_X_c;
2603 *yuv2packed2 = yuv2rgba32_full_2_c;
2604 *yuv2packed1 = yuv2rgba32_full_1_c;
2606 #endif /* CONFIG_SWSCALE_ALPHA */
2608 *yuv2packedX = yuv2rgbx32_full_X_c;
2609 *yuv2packed2 = yuv2rgbx32_full_2_c;
2610 *yuv2packed1 = yuv2rgbx32_full_1_c;
2612 #endif /* !CONFIG_SMALL */
2614 case AV_PIX_FMT_ARGB:
2616 *yuv2packedX = yuv2argb32_full_X_c;
2617 *yuv2packed2 = yuv2argb32_full_2_c;
2618 *yuv2packed1 = yuv2argb32_full_1_c;
2620 #if CONFIG_SWSCALE_ALPHA
2622 *yuv2packedX = yuv2argb32_full_X_c;
2623 *yuv2packed2 = yuv2argb32_full_2_c;
2624 *yuv2packed1 = yuv2argb32_full_1_c;
2626 #endif /* CONFIG_SWSCALE_ALPHA */
2628 *yuv2packedX = yuv2xrgb32_full_X_c;
2629 *yuv2packed2 = yuv2xrgb32_full_2_c;
2630 *yuv2packed1 = yuv2xrgb32_full_1_c;
2632 #endif /* !CONFIG_SMALL */
2634 case AV_PIX_FMT_BGRA:
2636 *yuv2packedX = yuv2bgra32_full_X_c;
2637 *yuv2packed2 = yuv2bgra32_full_2_c;
2638 *yuv2packed1 = yuv2bgra32_full_1_c;
2640 #if CONFIG_SWSCALE_ALPHA
2642 *yuv2packedX = yuv2bgra32_full_X_c;
2643 *yuv2packed2 = yuv2bgra32_full_2_c;
2644 *yuv2packed1 = yuv2bgra32_full_1_c;
2646 #endif /* CONFIG_SWSCALE_ALPHA */
2648 *yuv2packedX = yuv2bgrx32_full_X_c;
2649 *yuv2packed2 = yuv2bgrx32_full_2_c;
2650 *yuv2packed1 = yuv2bgrx32_full_1_c;
2652 #endif /* !CONFIG_SMALL */
2654 case AV_PIX_FMT_ABGR:
2656 *yuv2packedX = yuv2abgr32_full_X_c;
2657 *yuv2packed2 = yuv2abgr32_full_2_c;
2658 *yuv2packed1 = yuv2abgr32_full_1_c;
2660 #if CONFIG_SWSCALE_ALPHA
2662 *yuv2packedX = yuv2abgr32_full_X_c;
2663 *yuv2packed2 = yuv2abgr32_full_2_c;
2664 *yuv2packed1 = yuv2abgr32_full_1_c;
2666 #endif /* CONFIG_SWSCALE_ALPHA */
2668 *yuv2packedX = yuv2xbgr32_full_X_c;
2669 *yuv2packed2 = yuv2xbgr32_full_2_c;
2670 *yuv2packed1 = yuv2xbgr32_full_1_c;
2672 #endif /* !CONFIG_SMALL */
2674 case AV_PIX_FMT_RGBA64LE:
2675 #if CONFIG_SWSCALE_ALPHA
2677 *yuv2packedX = yuv2rgba64le_full_X_c;
2678 *yuv2packed2 = yuv2rgba64le_full_2_c;
2679 *yuv2packed1 = yuv2rgba64le_full_1_c;
2681 #endif /* CONFIG_SWSCALE_ALPHA */
2683 *yuv2packedX = yuv2rgbx64le_full_X_c;
2684 *yuv2packed2 = yuv2rgbx64le_full_2_c;
2685 *yuv2packed1 = yuv2rgbx64le_full_1_c;
2688 case AV_PIX_FMT_RGBA64BE:
2689 #if CONFIG_SWSCALE_ALPHA
2691 *yuv2packedX = yuv2rgba64be_full_X_c;
2692 *yuv2packed2 = yuv2rgba64be_full_2_c;
2693 *yuv2packed1 = yuv2rgba64be_full_1_c;
2695 #endif /* CONFIG_SWSCALE_ALPHA */
2697 *yuv2packedX = yuv2rgbx64be_full_X_c;
2698 *yuv2packed2 = yuv2rgbx64be_full_2_c;
2699 *yuv2packed1 = yuv2rgbx64be_full_1_c;
2702 case AV_PIX_FMT_BGRA64LE:
2703 #if CONFIG_SWSCALE_ALPHA
2705 *yuv2packedX = yuv2bgra64le_full_X_c;
2706 *yuv2packed2 = yuv2bgra64le_full_2_c;
2707 *yuv2packed1 = yuv2bgra64le_full_1_c;
2709 #endif /* CONFIG_SWSCALE_ALPHA */
2711 *yuv2packedX = yuv2bgrx64le_full_X_c;
2712 *yuv2packed2 = yuv2bgrx64le_full_2_c;
2713 *yuv2packed1 = yuv2bgrx64le_full_1_c;
2716 case AV_PIX_FMT_BGRA64BE:
2717 #if CONFIG_SWSCALE_ALPHA
2719 *yuv2packedX = yuv2bgra64be_full_X_c;
2720 *yuv2packed2 = yuv2bgra64be_full_2_c;
2721 *yuv2packed1 = yuv2bgra64be_full_1_c;
2723 #endif /* CONFIG_SWSCALE_ALPHA */
2725 *yuv2packedX = yuv2bgrx64be_full_X_c;
2726 *yuv2packed2 = yuv2bgrx64be_full_2_c;
2727 *yuv2packed1 = yuv2bgrx64be_full_1_c;
2731 case AV_PIX_FMT_RGB24:
2732 *yuv2packedX = yuv2rgb24_full_X_c;
2733 *yuv2packed2 = yuv2rgb24_full_2_c;
2734 *yuv2packed1 = yuv2rgb24_full_1_c;
2736 case AV_PIX_FMT_BGR24:
2737 *yuv2packedX = yuv2bgr24_full_X_c;
2738 *yuv2packed2 = yuv2bgr24_full_2_c;
2739 *yuv2packed1 = yuv2bgr24_full_1_c;
2741 case AV_PIX_FMT_RGB48LE:
2742 *yuv2packedX = yuv2rgb48le_full_X_c;
2743 *yuv2packed2 = yuv2rgb48le_full_2_c;
2744 *yuv2packed1 = yuv2rgb48le_full_1_c;
2746 case AV_PIX_FMT_BGR48LE:
2747 *yuv2packedX = yuv2bgr48le_full_X_c;
2748 *yuv2packed2 = yuv2bgr48le_full_2_c;
2749 *yuv2packed1 = yuv2bgr48le_full_1_c;
2751 case AV_PIX_FMT_RGB48BE:
2752 *yuv2packedX = yuv2rgb48be_full_X_c;
2753 *yuv2packed2 = yuv2rgb48be_full_2_c;
2754 *yuv2packed1 = yuv2rgb48be_full_1_c;
2756 case AV_PIX_FMT_BGR48BE:
2757 *yuv2packedX = yuv2bgr48be_full_X_c;
2758 *yuv2packed2 = yuv2bgr48be_full_2_c;
2759 *yuv2packed1 = yuv2bgr48be_full_1_c;
2761 case AV_PIX_FMT_BGR4_BYTE:
2762 *yuv2packedX = yuv2bgr4_byte_full_X_c;
2763 *yuv2packed2 = yuv2bgr4_byte_full_2_c;
2764 *yuv2packed1 = yuv2bgr4_byte_full_1_c;
2766 case AV_PIX_FMT_RGB4_BYTE:
2767 *yuv2packedX = yuv2rgb4_byte_full_X_c;
2768 *yuv2packed2 = yuv2rgb4_byte_full_2_c;
2769 *yuv2packed1 = yuv2rgb4_byte_full_1_c;
2771 case AV_PIX_FMT_BGR8:
2772 *yuv2packedX = yuv2bgr8_full_X_c;
2773 *yuv2packed2 = yuv2bgr8_full_2_c;
2774 *yuv2packed1 = yuv2bgr8_full_1_c;
2776 case AV_PIX_FMT_RGB8:
2777 *yuv2packedX = yuv2rgb8_full_X_c;
2778 *yuv2packed2 = yuv2rgb8_full_2_c;
2779 *yuv2packed1 = yuv2rgb8_full_1_c;
2781 case AV_PIX_FMT_GBRP:
2782 case AV_PIX_FMT_GBRP9BE:
2783 case AV_PIX_FMT_GBRP9LE:
2784 case AV_PIX_FMT_GBRP10BE:
2785 case AV_PIX_FMT_GBRP10LE:
2786 case AV_PIX_FMT_GBRP12BE:
2787 case AV_PIX_FMT_GBRP12LE:
2788 case AV_PIX_FMT_GBRP14BE:
2789 case AV_PIX_FMT_GBRP14LE:
2790 case AV_PIX_FMT_GBRAP:
2791 case AV_PIX_FMT_GBRAP10BE:
2792 case AV_PIX_FMT_GBRAP10LE:
2793 case AV_PIX_FMT_GBRAP12BE:
2794 case AV_PIX_FMT_GBRAP12LE:
2795 *yuv2anyX = yuv2gbrp_full_X_c;
2797 case AV_PIX_FMT_GBRP16BE:
2798 case AV_PIX_FMT_GBRP16LE:
2799 case AV_PIX_FMT_GBRAP16BE:
2800 case AV_PIX_FMT_GBRAP16LE:
2801 *yuv2anyX = yuv2gbrp16_full_X_c;
2803 case AV_PIX_FMT_GBRPF32BE:
2804 case AV_PIX_FMT_GBRPF32LE:
2805 case AV_PIX_FMT_GBRAPF32BE:
2806 case AV_PIX_FMT_GBRAPF32LE:
2807 *yuv2anyX = yuv2gbrpf32_full_X_c;
2810 if (!*yuv2packedX && !*yuv2anyX)
2814 switch (dstFormat) {
2815 case AV_PIX_FMT_RGBA64LE:
2816 #if CONFIG_SWSCALE_ALPHA
2818 *yuv2packed1 = yuv2rgba64le_1_c;
2819 *yuv2packed2 = yuv2rgba64le_2_c;
2820 *yuv2packedX = yuv2rgba64le_X_c;
2822 #endif /* CONFIG_SWSCALE_ALPHA */
2824 *yuv2packed1 = yuv2rgbx64le_1_c;
2825 *yuv2packed2 = yuv2rgbx64le_2_c;
2826 *yuv2packedX = yuv2rgbx64le_X_c;
2829 case AV_PIX_FMT_RGBA64BE:
2830 #if CONFIG_SWSCALE_ALPHA
2832 *yuv2packed1 = yuv2rgba64be_1_c;
2833 *yuv2packed2 = yuv2rgba64be_2_c;
2834 *yuv2packedX = yuv2rgba64be_X_c;
2836 #endif /* CONFIG_SWSCALE_ALPHA */
2838 *yuv2packed1 = yuv2rgbx64be_1_c;
2839 *yuv2packed2 = yuv2rgbx64be_2_c;
2840 *yuv2packedX = yuv2rgbx64be_X_c;
2843 case AV_PIX_FMT_BGRA64LE:
2844 #if CONFIG_SWSCALE_ALPHA
2846 *yuv2packed1 = yuv2bgra64le_1_c;
2847 *yuv2packed2 = yuv2bgra64le_2_c;
2848 *yuv2packedX = yuv2bgra64le_X_c;
2850 #endif /* CONFIG_SWSCALE_ALPHA */
2852 *yuv2packed1 = yuv2bgrx64le_1_c;
2853 *yuv2packed2 = yuv2bgrx64le_2_c;
2854 *yuv2packedX = yuv2bgrx64le_X_c;
2857 case AV_PIX_FMT_BGRA64BE:
2858 #if CONFIG_SWSCALE_ALPHA
2860 *yuv2packed1 = yuv2bgra64be_1_c;
2861 *yuv2packed2 = yuv2bgra64be_2_c;
2862 *yuv2packedX = yuv2bgra64be_X_c;
2864 #endif /* CONFIG_SWSCALE_ALPHA */
2866 *yuv2packed1 = yuv2bgrx64be_1_c;
2867 *yuv2packed2 = yuv2bgrx64be_2_c;
2868 *yuv2packedX = yuv2bgrx64be_X_c;
2871 case AV_PIX_FMT_RGB48LE:
2872 *yuv2packed1 = yuv2rgb48le_1_c;
2873 *yuv2packed2 = yuv2rgb48le_2_c;
2874 *yuv2packedX = yuv2rgb48le_X_c;
2876 case AV_PIX_FMT_RGB48BE:
2877 *yuv2packed1 = yuv2rgb48be_1_c;
2878 *yuv2packed2 = yuv2rgb48be_2_c;
2879 *yuv2packedX = yuv2rgb48be_X_c;
2881 case AV_PIX_FMT_BGR48LE:
2882 *yuv2packed1 = yuv2bgr48le_1_c;
2883 *yuv2packed2 = yuv2bgr48le_2_c;
2884 *yuv2packedX = yuv2bgr48le_X_c;
2886 case AV_PIX_FMT_BGR48BE:
2887 *yuv2packed1 = yuv2bgr48be_1_c;
2888 *yuv2packed2 = yuv2bgr48be_2_c;
2889 *yuv2packedX = yuv2bgr48be_X_c;
2891 case AV_PIX_FMT_RGB32:
2892 case AV_PIX_FMT_BGR32:
2894 *yuv2packed1 = yuv2rgb32_1_c;
2895 *yuv2packed2 = yuv2rgb32_2_c;
2896 *yuv2packedX = yuv2rgb32_X_c;
2898 #if CONFIG_SWSCALE_ALPHA
2900 *yuv2packed1 = yuv2rgba32_1_c;
2901 *yuv2packed2 = yuv2rgba32_2_c;
2902 *yuv2packedX = yuv2rgba32_X_c;
2904 #endif /* CONFIG_SWSCALE_ALPHA */
2906 *yuv2packed1 = yuv2rgbx32_1_c;
2907 *yuv2packed2 = yuv2rgbx32_2_c;
2908 *yuv2packedX = yuv2rgbx32_X_c;
2910 #endif /* !CONFIG_SMALL */
2912 case AV_PIX_FMT_RGB32_1:
2913 case AV_PIX_FMT_BGR32_1:
2915 *yuv2packed1 = yuv2rgb32_1_1_c;
2916 *yuv2packed2 = yuv2rgb32_1_2_c;
2917 *yuv2packedX = yuv2rgb32_1_X_c;
2919 #if CONFIG_SWSCALE_ALPHA
2921 *yuv2packed1 = yuv2rgba32_1_1_c;
2922 *yuv2packed2 = yuv2rgba32_1_2_c;
2923 *yuv2packedX = yuv2rgba32_1_X_c;
2925 #endif /* CONFIG_SWSCALE_ALPHA */
2927 *yuv2packed1 = yuv2rgbx32_1_1_c;
2928 *yuv2packed2 = yuv2rgbx32_1_2_c;
2929 *yuv2packedX = yuv2rgbx32_1_X_c;
2931 #endif /* !CONFIG_SMALL */
2933 case AV_PIX_FMT_RGB24:
2934 *yuv2packed1 = yuv2rgb24_1_c;
2935 *yuv2packed2 = yuv2rgb24_2_c;
2936 *yuv2packedX = yuv2rgb24_X_c;
2938 case AV_PIX_FMT_BGR24:
2939 *yuv2packed1 = yuv2bgr24_1_c;
2940 *yuv2packed2 = yuv2bgr24_2_c;
2941 *yuv2packedX = yuv2bgr24_X_c;
2943 case AV_PIX_FMT_RGB565LE:
2944 case AV_PIX_FMT_RGB565BE:
2945 case AV_PIX_FMT_BGR565LE:
2946 case AV_PIX_FMT_BGR565BE:
2947 *yuv2packed1 = yuv2rgb16_1_c;
2948 *yuv2packed2 = yuv2rgb16_2_c;
2949 *yuv2packedX = yuv2rgb16_X_c;
2951 case AV_PIX_FMT_RGB555LE:
2952 case AV_PIX_FMT_RGB555BE:
2953 case AV_PIX_FMT_BGR555LE:
2954 case AV_PIX_FMT_BGR555BE:
2955 *yuv2packed1 = yuv2rgb15_1_c;
2956 *yuv2packed2 = yuv2rgb15_2_c;
2957 *yuv2packedX = yuv2rgb15_X_c;
2959 case AV_PIX_FMT_RGB444LE:
2960 case AV_PIX_FMT_RGB444BE:
2961 case AV_PIX_FMT_BGR444LE:
2962 case AV_PIX_FMT_BGR444BE:
2963 *yuv2packed1 = yuv2rgb12_1_c;
2964 *yuv2packed2 = yuv2rgb12_2_c;
2965 *yuv2packedX = yuv2rgb12_X_c;
2967 case AV_PIX_FMT_RGB8:
2968 case AV_PIX_FMT_BGR8:
2969 *yuv2packed1 = yuv2rgb8_1_c;
2970 *yuv2packed2 = yuv2rgb8_2_c;
2971 *yuv2packedX = yuv2rgb8_X_c;
2973 case AV_PIX_FMT_RGB4:
2974 case AV_PIX_FMT_BGR4:
2975 *yuv2packed1 = yuv2rgb4_1_c;
2976 *yuv2packed2 = yuv2rgb4_2_c;
2977 *yuv2packedX = yuv2rgb4_X_c;
2979 case AV_PIX_FMT_RGB4_BYTE:
2980 case AV_PIX_FMT_BGR4_BYTE:
2981 *yuv2packed1 = yuv2rgb4b_1_c;
2982 *yuv2packed2 = yuv2rgb4b_2_c;
2983 *yuv2packedX = yuv2rgb4b_X_c;
2985 case AV_PIX_FMT_X2RGB10LE:
2986 case AV_PIX_FMT_X2RGB10BE:
2987 *yuv2packed1 = yuv2x2rgb10_1_c;
2988 *yuv2packed2 = yuv2x2rgb10_2_c;
2989 *yuv2packedX = yuv2x2rgb10_X_c;
2993 switch (dstFormat) {
2994 case AV_PIX_FMT_MONOWHITE:
2995 *yuv2packed1 = yuv2monowhite_1_c;
2996 *yuv2packed2 = yuv2monowhite_2_c;
2997 *yuv2packedX = yuv2monowhite_X_c;
2999 case AV_PIX_FMT_MONOBLACK:
3000 *yuv2packed1 = yuv2monoblack_1_c;
3001 *yuv2packed2 = yuv2monoblack_2_c;
3002 *yuv2packedX = yuv2monoblack_X_c;
3004 case AV_PIX_FMT_YUYV422:
3005 *yuv2packed1 = yuv2yuyv422_1_c;
3006 *yuv2packed2 = yuv2yuyv422_2_c;
3007 *yuv2packedX = yuv2yuyv422_X_c;
3009 case AV_PIX_FMT_YVYU422:
3010 *yuv2packed1 = yuv2yvyu422_1_c;
3011 *yuv2packed2 = yuv2yvyu422_2_c;
3012 *yuv2packedX = yuv2yvyu422_X_c;
3014 case AV_PIX_FMT_UYVY422:
3015 *yuv2packed1 = yuv2uyvy422_1_c;
3016 *yuv2packed2 = yuv2uyvy422_2_c;
3017 *yuv2packedX = yuv2uyvy422_X_c;
3019 case AV_PIX_FMT_YA8:
3020 *yuv2packed1 = yuv2ya8_1_c;
3021 *yuv2packed2 = yuv2ya8_2_c;
3022 *yuv2packedX = yuv2ya8_X_c;
3024 case AV_PIX_FMT_YA16LE:
3025 *yuv2packed1 = yuv2ya16le_1_c;
3026 *yuv2packed2 = yuv2ya16le_2_c;
3027 *yuv2packedX = yuv2ya16le_X_c;
3029 case AV_PIX_FMT_YA16BE:
3030 *yuv2packed1 = yuv2ya16be_1_c;
3031 *yuv2packed2 = yuv2ya16be_2_c;
3032 *yuv2packedX = yuv2ya16be_X_c;
3034 case AV_PIX_FMT_AYUV64LE:
3035 *yuv2packedX = yuv2ayuv64le_X_c;