2 * Copyright (C) 2001-2003 Michael Niedermayer <michaelni@gmx.at>
4 * This file is part of Libav.
6 * Libav 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 * Libav 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 Libav; if not, write to the Free Software
18 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
27 #include "libavutil/avutil.h"
28 #include "libavutil/bswap.h"
29 #include "libavutil/cpu.h"
30 #include "libavutil/intreadwrite.h"
31 #include "libavutil/mathematics.h"
32 #include "libavutil/pixdesc.h"
36 #include "swscale_internal.h"
38 DECLARE_ALIGNED(8, static const uint8_t, dither_2x2_4)[2][8]={
39 { 1, 3, 1, 3, 1, 3, 1, 3, },
40 { 2, 0, 2, 0, 2, 0, 2, 0, },
43 DECLARE_ALIGNED(8, static const uint8_t, dither_2x2_8)[2][8]={
44 { 6, 2, 6, 2, 6, 2, 6, 2, },
45 { 0, 4, 0, 4, 0, 4, 0, 4, },
48 DECLARE_ALIGNED(8, const uint8_t, dither_4x4_16)[4][8]={
49 { 8, 4, 11, 7, 8, 4, 11, 7, },
50 { 2, 14, 1, 13, 2, 14, 1, 13, },
51 { 10, 6, 9, 5, 10, 6, 9, 5, },
52 { 0, 12, 3, 15, 0, 12, 3, 15, },
55 DECLARE_ALIGNED(8, const uint8_t, dither_8x8_32)[8][8]={
56 { 17, 9, 23, 15, 16, 8, 22, 14, },
57 { 5, 29, 3, 27, 4, 28, 2, 26, },
58 { 21, 13, 19, 11, 20, 12, 18, 10, },
59 { 0, 24, 6, 30, 1, 25, 7, 31, },
60 { 16, 8, 22, 14, 17, 9, 23, 15, },
61 { 4, 28, 2, 26, 5, 29, 3, 27, },
62 { 20, 12, 18, 10, 21, 13, 19, 11, },
63 { 1, 25, 7, 31, 0, 24, 6, 30, },
66 DECLARE_ALIGNED(8, const uint8_t, dither_8x8_73)[8][8]={
67 { 0, 55, 14, 68, 3, 58, 17, 72, },
68 { 37, 18, 50, 32, 40, 22, 54, 35, },
69 { 9, 64, 5, 59, 13, 67, 8, 63, },
70 { 46, 27, 41, 23, 49, 31, 44, 26, },
71 { 2, 57, 16, 71, 1, 56, 15, 70, },
72 { 39, 21, 52, 34, 38, 19, 51, 33, },
73 { 11, 66, 7, 62, 10, 65, 6, 60, },
74 { 48, 30, 43, 25, 47, 29, 42, 24, },
78 DECLARE_ALIGNED(8, const uint8_t, dither_8x8_220)[8][8]={
79 {117, 62, 158, 103, 113, 58, 155, 100, },
80 { 34, 199, 21, 186, 31, 196, 17, 182, },
81 {144, 89, 131, 76, 141, 86, 127, 72, },
82 { 0, 165, 41, 206, 10, 175, 52, 217, },
83 {110, 55, 151, 96, 120, 65, 162, 107, },
84 { 28, 193, 14, 179, 38, 203, 24, 189, },
85 {138, 83, 124, 69, 148, 93, 134, 79, },
86 { 7, 172, 48, 213, 3, 168, 45, 210, },
89 // tries to correct a gamma of 1.5
90 DECLARE_ALIGNED(8, const uint8_t, dither_8x8_220)[8][8]={
91 { 0, 143, 18, 200, 2, 156, 25, 215, },
92 { 78, 28, 125, 64, 89, 36, 138, 74, },
93 { 10, 180, 3, 161, 16, 195, 8, 175, },
94 {109, 51, 93, 38, 121, 60, 105, 47, },
95 { 1, 152, 23, 210, 0, 147, 20, 205, },
96 { 85, 33, 134, 71, 81, 30, 130, 67, },
97 { 14, 190, 6, 171, 12, 185, 5, 166, },
98 {117, 57, 101, 44, 113, 54, 97, 41, },
101 // tries to correct a gamma of 2.0
102 DECLARE_ALIGNED(8, const uint8_t, dither_8x8_220)[8][8]={
103 { 0, 124, 8, 193, 0, 140, 12, 213, },
104 { 55, 14, 104, 42, 66, 19, 119, 52, },
105 { 3, 168, 1, 145, 6, 187, 3, 162, },
106 { 86, 31, 70, 21, 99, 39, 82, 28, },
107 { 0, 134, 11, 206, 0, 129, 9, 200, },
108 { 62, 17, 114, 48, 58, 16, 109, 45, },
109 { 5, 181, 2, 157, 4, 175, 1, 151, },
110 { 95, 36, 78, 26, 90, 34, 74, 24, },
113 // tries to correct a gamma of 2.5
114 DECLARE_ALIGNED(8, const uint8_t, dither_8x8_220)[8][8]={
115 { 0, 107, 3, 187, 0, 125, 6, 212, },
116 { 39, 7, 86, 28, 49, 11, 102, 36, },
117 { 1, 158, 0, 131, 3, 180, 1, 151, },
118 { 68, 19, 52, 12, 81, 25, 64, 17, },
119 { 0, 119, 5, 203, 0, 113, 4, 195, },
120 { 45, 9, 96, 33, 42, 8, 91, 30, },
121 { 2, 172, 1, 144, 2, 165, 0, 137, },
122 { 77, 23, 60, 15, 72, 21, 56, 14, },
126 #define output_pixel(pos, val, bias, signedness) \
128 AV_WB16(pos, bias + av_clip_ ## signedness ## 16(val >> shift)); \
130 AV_WL16(pos, bias + av_clip_ ## signedness ## 16(val >> shift)); \
133 static av_always_inline void
134 yuv2plane1_16_c_template(const int32_t *src, uint16_t *dest, int dstW,
135 int big_endian, int output_bits)
138 int shift = 19 - output_bits;
140 for (i = 0; i < dstW; i++) {
141 int val = src[i] + (1 << (shift - 1));
142 output_pixel(&dest[i], val, 0, uint);
146 static av_always_inline void
147 yuv2planeX_16_c_template(const int16_t *filter, int filterSize,
148 const int32_t **src, uint16_t *dest, int dstW,
149 int big_endian, int output_bits)
152 int shift = 15 + 16 - output_bits;
154 for (i = 0; i < dstW; i++) {
155 int val = 1 << (30-output_bits);
158 /* range of val is [0,0x7FFFFFFF], so 31 bits, but with lanczos/spline
159 * filters (or anything with negative coeffs, the range can be slightly
160 * wider in both directions. To account for this overflow, we subtract
161 * a constant so it always fits in the signed range (assuming a
162 * reasonable filterSize), and re-add that at the end. */
164 for (j = 0; j < filterSize; j++)
165 val += src[j][i] * filter[j];
167 output_pixel(&dest[i], val, 0x8000, int);
173 #define output_pixel(pos, val) \
175 AV_WB16(pos, av_clip_uintp2(val >> shift, output_bits)); \
177 AV_WL16(pos, av_clip_uintp2(val >> shift, output_bits)); \
180 static av_always_inline void
181 yuv2plane1_10_c_template(const int16_t *src, uint16_t *dest, int dstW,
182 int big_endian, int output_bits)
185 int shift = 15 - output_bits;
187 for (i = 0; i < dstW; i++) {
188 int val = src[i] + (1 << (shift - 1));
189 output_pixel(&dest[i], val);
193 static av_always_inline void
194 yuv2planeX_10_c_template(const int16_t *filter, int filterSize,
195 const int16_t **src, uint16_t *dest, int dstW,
196 int big_endian, int output_bits)
199 int shift = 11 + 16 - output_bits;
201 for (i = 0; i < dstW; i++) {
202 int val = 1 << (26-output_bits);
205 for (j = 0; j < filterSize; j++)
206 val += src[j][i] * filter[j];
208 output_pixel(&dest[i], val);
214 #define yuv2NBPS(bits, BE_LE, is_be, template_size, typeX_t) \
215 static void yuv2plane1_ ## bits ## BE_LE ## _c(const int16_t *src, \
216 uint8_t *dest, int dstW, \
217 const uint8_t *dither, int offset)\
219 yuv2plane1_ ## template_size ## _c_template((const typeX_t *) src, \
220 (uint16_t *) dest, dstW, is_be, bits); \
222 static void yuv2planeX_ ## bits ## BE_LE ## _c(const int16_t *filter, int filterSize, \
223 const int16_t **src, uint8_t *dest, int dstW, \
224 const uint8_t *dither, int offset)\
226 yuv2planeX_## template_size ## _c_template(filter, \
227 filterSize, (const typeX_t **) src, \
228 (uint16_t *) dest, dstW, is_be, bits); \
230 yuv2NBPS( 9, BE, 1, 10, int16_t)
231 yuv2NBPS( 9, LE, 0, 10, int16_t)
232 yuv2NBPS(10, BE, 1, 10, int16_t)
233 yuv2NBPS(10, LE, 0, 10, int16_t)
234 yuv2NBPS(16, BE, 1, 16, int32_t)
235 yuv2NBPS(16, LE, 0, 16, int32_t)
237 static void yuv2planeX_8_c(const int16_t *filter, int filterSize,
238 const int16_t **src, uint8_t *dest, int dstW,
239 const uint8_t *dither, int offset)
242 for (i=0; i<dstW; i++) {
243 int val = dither[(i + offset) & 7] << 12;
245 for (j=0; j<filterSize; j++)
246 val += src[j][i] * filter[j];
248 dest[i]= av_clip_uint8(val>>19);
252 static void yuv2plane1_8_c(const int16_t *src, uint8_t *dest, int dstW,
253 const uint8_t *dither, int offset)
256 for (i=0; i<dstW; i++) {
257 int val = (src[i] + dither[(i + offset) & 7]) >> 7;
258 dest[i]= av_clip_uint8(val);
262 static void yuv2nv12cX_c(SwsContext *c, const int16_t *chrFilter, int chrFilterSize,
263 const int16_t **chrUSrc, const int16_t **chrVSrc,
264 uint8_t *dest, int chrDstW)
266 enum PixelFormat dstFormat = c->dstFormat;
267 const uint8_t *chrDither = c->chrDither8;
270 if (dstFormat == PIX_FMT_NV12)
271 for (i=0; i<chrDstW; i++) {
272 int u = chrDither[i & 7] << 12;
273 int v = chrDither[(i + 3) & 7] << 12;
275 for (j=0; j<chrFilterSize; j++) {
276 u += chrUSrc[j][i] * chrFilter[j];
277 v += chrVSrc[j][i] * chrFilter[j];
280 dest[2*i]= av_clip_uint8(u>>19);
281 dest[2*i+1]= av_clip_uint8(v>>19);
284 for (i=0; i<chrDstW; i++) {
285 int u = chrDither[i & 7] << 12;
286 int v = chrDither[(i + 3) & 7] << 12;
288 for (j=0; j<chrFilterSize; j++) {
289 u += chrUSrc[j][i] * chrFilter[j];
290 v += chrVSrc[j][i] * chrFilter[j];
293 dest[2*i]= av_clip_uint8(v>>19);
294 dest[2*i+1]= av_clip_uint8(u>>19);
298 #define accumulate_bit(acc, val) \
300 acc |= (val) >= (128 + 110)
301 #define output_pixel(pos, acc) \
302 if (target == PIX_FMT_MONOBLACK) { \
308 static av_always_inline void
309 yuv2mono_X_c_template(SwsContext *c, const int16_t *lumFilter,
310 const int16_t **lumSrc, int lumFilterSize,
311 const int16_t *chrFilter, const int16_t **chrUSrc,
312 const int16_t **chrVSrc, int chrFilterSize,
313 const int16_t **alpSrc, uint8_t *dest, int dstW,
314 int y, enum PixelFormat target)
316 const uint8_t * const d128=dither_8x8_220[y&7];
320 for (i = 0; i < dstW - 1; i += 2) {
325 for (j = 0; j < lumFilterSize; j++) {
326 Y1 += lumSrc[j][i] * lumFilter[j];
327 Y2 += lumSrc[j][i+1] * lumFilter[j];
331 if ((Y1 | Y2) & 0x100) {
332 Y1 = av_clip_uint8(Y1);
333 Y2 = av_clip_uint8(Y2);
335 accumulate_bit(acc, Y1 + d128[(i + 0) & 7]);
336 accumulate_bit(acc, Y2 + d128[(i + 1) & 7]);
338 output_pixel(*dest++, acc);
343 static av_always_inline void
344 yuv2mono_2_c_template(SwsContext *c, const int16_t *buf[2],
345 const int16_t *ubuf[2], const int16_t *vbuf[2],
346 const int16_t *abuf[2], uint8_t *dest, int dstW,
347 int yalpha, int uvalpha, int y,
348 enum PixelFormat target)
350 const int16_t *buf0 = buf[0], *buf1 = buf[1];
351 const uint8_t * const d128 = dither_8x8_220[y & 7];
352 int yalpha1 = 4095 - yalpha;
355 for (i = 0; i < dstW - 7; i += 8) {
358 Y = (buf0[i + 0] * yalpha1 + buf1[i + 0] * yalpha) >> 19;
359 accumulate_bit(acc, Y + d128[0]);
360 Y = (buf0[i + 1] * yalpha1 + buf1[i + 1] * yalpha) >> 19;
361 accumulate_bit(acc, Y + d128[1]);
362 Y = (buf0[i + 2] * yalpha1 + buf1[i + 2] * yalpha) >> 19;
363 accumulate_bit(acc, Y + d128[2]);
364 Y = (buf0[i + 3] * yalpha1 + buf1[i + 3] * yalpha) >> 19;
365 accumulate_bit(acc, Y + d128[3]);
366 Y = (buf0[i + 4] * yalpha1 + buf1[i + 4] * yalpha) >> 19;
367 accumulate_bit(acc, Y + d128[4]);
368 Y = (buf0[i + 5] * yalpha1 + buf1[i + 5] * yalpha) >> 19;
369 accumulate_bit(acc, Y + d128[5]);
370 Y = (buf0[i + 6] * yalpha1 + buf1[i + 6] * yalpha) >> 19;
371 accumulate_bit(acc, Y + d128[6]);
372 Y = (buf0[i + 7] * yalpha1 + buf1[i + 7] * yalpha) >> 19;
373 accumulate_bit(acc, Y + d128[7]);
375 output_pixel(*dest++, acc);
379 static av_always_inline void
380 yuv2mono_1_c_template(SwsContext *c, const int16_t *buf0,
381 const int16_t *ubuf[2], const int16_t *vbuf[2],
382 const int16_t *abuf0, uint8_t *dest, int dstW,
383 int uvalpha, int y, enum PixelFormat target)
385 const uint8_t * const d128 = dither_8x8_220[y & 7];
388 for (i = 0; i < dstW - 7; i += 8) {
391 accumulate_bit(acc, (buf0[i + 0] >> 7) + d128[0]);
392 accumulate_bit(acc, (buf0[i + 1] >> 7) + d128[1]);
393 accumulate_bit(acc, (buf0[i + 2] >> 7) + d128[2]);
394 accumulate_bit(acc, (buf0[i + 3] >> 7) + d128[3]);
395 accumulate_bit(acc, (buf0[i + 4] >> 7) + d128[4]);
396 accumulate_bit(acc, (buf0[i + 5] >> 7) + d128[5]);
397 accumulate_bit(acc, (buf0[i + 6] >> 7) + d128[6]);
398 accumulate_bit(acc, (buf0[i + 7] >> 7) + d128[7]);
400 output_pixel(*dest++, acc);
405 #undef accumulate_bit
407 #define YUV2PACKEDWRAPPER(name, base, ext, fmt) \
408 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
409 const int16_t **lumSrc, int lumFilterSize, \
410 const int16_t *chrFilter, const int16_t **chrUSrc, \
411 const int16_t **chrVSrc, int chrFilterSize, \
412 const int16_t **alpSrc, uint8_t *dest, int dstW, \
415 name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
416 chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
417 alpSrc, dest, dstW, y, fmt); \
420 static void name ## ext ## _2_c(SwsContext *c, const int16_t *buf[2], \
421 const int16_t *ubuf[2], const int16_t *vbuf[2], \
422 const int16_t *abuf[2], uint8_t *dest, int dstW, \
423 int yalpha, int uvalpha, int y) \
425 name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
426 dest, dstW, yalpha, uvalpha, y, fmt); \
429 static void name ## ext ## _1_c(SwsContext *c, const int16_t *buf0, \
430 const int16_t *ubuf[2], const int16_t *vbuf[2], \
431 const int16_t *abuf0, uint8_t *dest, int dstW, \
432 int uvalpha, int y) \
434 name ## base ## _1_c_template(c, buf0, ubuf, vbuf, \
435 abuf0, dest, dstW, uvalpha, \
439 YUV2PACKEDWRAPPER(yuv2mono,, white, PIX_FMT_MONOWHITE)
440 YUV2PACKEDWRAPPER(yuv2mono,, black, PIX_FMT_MONOBLACK)
442 #define output_pixels(pos, Y1, U, Y2, V) \
443 if (target == PIX_FMT_YUYV422) { \
444 dest[pos + 0] = Y1; \
446 dest[pos + 2] = Y2; \
450 dest[pos + 1] = Y1; \
452 dest[pos + 3] = Y2; \
455 static av_always_inline void
456 yuv2422_X_c_template(SwsContext *c, const int16_t *lumFilter,
457 const int16_t **lumSrc, int lumFilterSize,
458 const int16_t *chrFilter, const int16_t **chrUSrc,
459 const int16_t **chrVSrc, int chrFilterSize,
460 const int16_t **alpSrc, uint8_t *dest, int dstW,
461 int y, enum PixelFormat target)
465 for (i = 0; i < (dstW >> 1); i++) {
472 for (j = 0; j < lumFilterSize; j++) {
473 Y1 += lumSrc[j][i * 2] * lumFilter[j];
474 Y2 += lumSrc[j][i * 2 + 1] * lumFilter[j];
476 for (j = 0; j < chrFilterSize; j++) {
477 U += chrUSrc[j][i] * chrFilter[j];
478 V += chrVSrc[j][i] * chrFilter[j];
484 if ((Y1 | Y2 | U | V) & 0x100) {
485 Y1 = av_clip_uint8(Y1);
486 Y2 = av_clip_uint8(Y2);
487 U = av_clip_uint8(U);
488 V = av_clip_uint8(V);
490 output_pixels(4*i, Y1, U, Y2, V);
494 static av_always_inline void
495 yuv2422_2_c_template(SwsContext *c, const int16_t *buf[2],
496 const int16_t *ubuf[2], const int16_t *vbuf[2],
497 const int16_t *abuf[2], uint8_t *dest, int dstW,
498 int yalpha, int uvalpha, int y,
499 enum PixelFormat target)
501 const int16_t *buf0 = buf[0], *buf1 = buf[1],
502 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
503 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1];
504 int yalpha1 = 4095 - yalpha;
505 int uvalpha1 = 4095 - uvalpha;
508 for (i = 0; i < (dstW >> 1); i++) {
509 int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 19;
510 int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 19;
511 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha) >> 19;
512 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha) >> 19;
514 Y1 = av_clip_uint8(Y1);
515 Y2 = av_clip_uint8(Y2);
516 U = av_clip_uint8(U);
517 V = av_clip_uint8(V);
519 output_pixels(i * 4, Y1, U, Y2, V);
523 static av_always_inline void
524 yuv2422_1_c_template(SwsContext *c, const int16_t *buf0,
525 const int16_t *ubuf[2], const int16_t *vbuf[2],
526 const int16_t *abuf0, uint8_t *dest, int dstW,
527 int uvalpha, int y, enum PixelFormat target)
529 const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
532 if (uvalpha < 2048) {
533 for (i = 0; i < (dstW >> 1); i++) {
534 int Y1 = buf0[i * 2] >> 7;
535 int Y2 = buf0[i * 2 + 1] >> 7;
536 int U = ubuf0[i] >> 7;
537 int V = vbuf0[i] >> 7;
539 Y1 = av_clip_uint8(Y1);
540 Y2 = av_clip_uint8(Y2);
541 U = av_clip_uint8(U);
542 V = av_clip_uint8(V);
544 output_pixels(i * 4, Y1, U, Y2, V);
547 const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
548 for (i = 0; i < (dstW >> 1); i++) {
549 int Y1 = buf0[i * 2] >> 7;
550 int Y2 = buf0[i * 2 + 1] >> 7;
551 int U = (ubuf0[i] + ubuf1[i]) >> 8;
552 int V = (vbuf0[i] + vbuf1[i]) >> 8;
554 Y1 = av_clip_uint8(Y1);
555 Y2 = av_clip_uint8(Y2);
556 U = av_clip_uint8(U);
557 V = av_clip_uint8(V);
559 output_pixels(i * 4, Y1, U, Y2, V);
566 YUV2PACKEDWRAPPER(yuv2, 422, yuyv422, PIX_FMT_YUYV422)
567 YUV2PACKEDWRAPPER(yuv2, 422, uyvy422, PIX_FMT_UYVY422)
569 #define R_B ((target == PIX_FMT_RGB48LE || target == PIX_FMT_RGB48BE) ? R : B)
570 #define B_R ((target == PIX_FMT_RGB48LE || target == PIX_FMT_RGB48BE) ? B : R)
571 #define output_pixel(pos, val) \
572 if (isBE(target)) { \
578 static av_always_inline void
579 yuv2rgb48_X_c_template(SwsContext *c, const int16_t *lumFilter,
580 const int32_t **lumSrc, int lumFilterSize,
581 const int16_t *chrFilter, const int32_t **chrUSrc,
582 const int32_t **chrVSrc, int chrFilterSize,
583 const int32_t **alpSrc, uint16_t *dest, int dstW,
584 int y, enum PixelFormat target)
588 for (i = 0; i < (dstW >> 1); i++) {
590 int Y1 = -0x40000000;
591 int Y2 = -0x40000000;
592 int U = -128 << 23; // 19
596 for (j = 0; j < lumFilterSize; j++) {
597 Y1 += lumSrc[j][i * 2] * lumFilter[j];
598 Y2 += lumSrc[j][i * 2 + 1] * lumFilter[j];
600 for (j = 0; j < chrFilterSize; j++) {
601 U += chrUSrc[j][i] * chrFilter[j];
602 V += chrVSrc[j][i] * chrFilter[j];
605 // 8bit: 12+15=27; 16-bit: 12+19=31
613 // 8bit: 27 -> 17bit, 16bit: 31 - 14 = 17bit
614 Y1 -= c->yuv2rgb_y_offset;
615 Y2 -= c->yuv2rgb_y_offset;
616 Y1 *= c->yuv2rgb_y_coeff;
617 Y2 *= c->yuv2rgb_y_coeff;
620 // 8bit: 17 + 13bit = 30bit, 16bit: 17 + 13bit = 30bit
622 R = V * c->yuv2rgb_v2r_coeff;
623 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
624 B = U * c->yuv2rgb_u2b_coeff;
626 // 8bit: 30 - 22 = 8bit, 16bit: 30bit - 14 = 16bit
627 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
628 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
629 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
630 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
631 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
632 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
637 static av_always_inline void
638 yuv2rgb48_2_c_template(SwsContext *c, const int32_t *buf[2],
639 const int32_t *ubuf[2], const int32_t *vbuf[2],
640 const int32_t *abuf[2], uint16_t *dest, int dstW,
641 int yalpha, int uvalpha, int y,
642 enum PixelFormat target)
644 const int32_t *buf0 = buf[0], *buf1 = buf[1],
645 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
646 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1];
647 int yalpha1 = 4095 - yalpha;
648 int uvalpha1 = 4095 - uvalpha;
651 for (i = 0; i < (dstW >> 1); i++) {
652 int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 14;
653 int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 14;
654 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha + (-128 << 23)) >> 14;
655 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha + (-128 << 23)) >> 14;
658 Y1 -= c->yuv2rgb_y_offset;
659 Y2 -= c->yuv2rgb_y_offset;
660 Y1 *= c->yuv2rgb_y_coeff;
661 Y2 *= c->yuv2rgb_y_coeff;
665 R = V * c->yuv2rgb_v2r_coeff;
666 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
667 B = U * c->yuv2rgb_u2b_coeff;
669 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
670 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
671 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
672 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
673 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
674 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
679 static av_always_inline void
680 yuv2rgb48_1_c_template(SwsContext *c, const int32_t *buf0,
681 const int32_t *ubuf[2], const int32_t *vbuf[2],
682 const int32_t *abuf0, uint16_t *dest, int dstW,
683 int uvalpha, int y, enum PixelFormat target)
685 const int32_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
688 if (uvalpha < 2048) {
689 for (i = 0; i < (dstW >> 1); i++) {
690 int Y1 = (buf0[i * 2] ) >> 2;
691 int Y2 = (buf0[i * 2 + 1]) >> 2;
692 int U = (ubuf0[i] + (-128 << 11)) >> 2;
693 int V = (vbuf0[i] + (-128 << 11)) >> 2;
696 Y1 -= c->yuv2rgb_y_offset;
697 Y2 -= c->yuv2rgb_y_offset;
698 Y1 *= c->yuv2rgb_y_coeff;
699 Y2 *= c->yuv2rgb_y_coeff;
703 R = V * c->yuv2rgb_v2r_coeff;
704 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
705 B = U * c->yuv2rgb_u2b_coeff;
707 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
708 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
709 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
710 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
711 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
712 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
716 const int32_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
717 for (i = 0; i < (dstW >> 1); i++) {
718 int Y1 = (buf0[i * 2] ) >> 2;
719 int Y2 = (buf0[i * 2 + 1]) >> 2;
720 int U = (ubuf0[i] + ubuf1[i] + (-128 << 12)) >> 3;
721 int V = (vbuf0[i] + vbuf1[i] + (-128 << 12)) >> 3;
724 Y1 -= c->yuv2rgb_y_offset;
725 Y2 -= c->yuv2rgb_y_offset;
726 Y1 *= c->yuv2rgb_y_coeff;
727 Y2 *= c->yuv2rgb_y_coeff;
731 R = V * c->yuv2rgb_v2r_coeff;
732 G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
733 B = U * c->yuv2rgb_u2b_coeff;
735 output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
736 output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
737 output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
738 output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
739 output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
740 output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
750 #define YUV2PACKED16WRAPPER(name, base, ext, fmt) \
751 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
752 const int16_t **_lumSrc, int lumFilterSize, \
753 const int16_t *chrFilter, const int16_t **_chrUSrc, \
754 const int16_t **_chrVSrc, int chrFilterSize, \
755 const int16_t **_alpSrc, uint8_t *_dest, int dstW, \
758 const int32_t **lumSrc = (const int32_t **) _lumSrc, \
759 **chrUSrc = (const int32_t **) _chrUSrc, \
760 **chrVSrc = (const int32_t **) _chrVSrc, \
761 **alpSrc = (const int32_t **) _alpSrc; \
762 uint16_t *dest = (uint16_t *) _dest; \
763 name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
764 chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
765 alpSrc, dest, dstW, y, fmt); \
768 static void name ## ext ## _2_c(SwsContext *c, const int16_t *_buf[2], \
769 const int16_t *_ubuf[2], const int16_t *_vbuf[2], \
770 const int16_t *_abuf[2], uint8_t *_dest, int dstW, \
771 int yalpha, int uvalpha, int y) \
773 const int32_t **buf = (const int32_t **) _buf, \
774 **ubuf = (const int32_t **) _ubuf, \
775 **vbuf = (const int32_t **) _vbuf, \
776 **abuf = (const int32_t **) _abuf; \
777 uint16_t *dest = (uint16_t *) _dest; \
778 name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
779 dest, dstW, yalpha, uvalpha, y, fmt); \
782 static void name ## ext ## _1_c(SwsContext *c, const int16_t *_buf0, \
783 const int16_t *_ubuf[2], const int16_t *_vbuf[2], \
784 const int16_t *_abuf0, uint8_t *_dest, int dstW, \
785 int uvalpha, int y) \
787 const int32_t *buf0 = (const int32_t *) _buf0, \
788 **ubuf = (const int32_t **) _ubuf, \
789 **vbuf = (const int32_t **) _vbuf, \
790 *abuf0 = (const int32_t *) _abuf0; \
791 uint16_t *dest = (uint16_t *) _dest; \
792 name ## base ## _1_c_template(c, buf0, ubuf, vbuf, abuf0, dest, \
793 dstW, uvalpha, y, fmt); \
796 YUV2PACKED16WRAPPER(yuv2, rgb48, rgb48be, PIX_FMT_RGB48BE)
797 YUV2PACKED16WRAPPER(yuv2, rgb48, rgb48le, PIX_FMT_RGB48LE)
798 YUV2PACKED16WRAPPER(yuv2, rgb48, bgr48be, PIX_FMT_BGR48BE)
799 YUV2PACKED16WRAPPER(yuv2, rgb48, bgr48le, PIX_FMT_BGR48LE)
802 * Write out 2 RGB pixels in the target pixel format. This function takes a
803 * R/G/B LUT as generated by ff_yuv2rgb_c_init_tables(), which takes care of
804 * things like endianness conversion and shifting. The caller takes care of
805 * setting the correct offset in these tables from the chroma (U/V) values.
806 * This function then uses the luminance (Y1/Y2) values to write out the
807 * correct RGB values into the destination buffer.
809 static av_always_inline void
810 yuv2rgb_write(uint8_t *_dest, int i, unsigned Y1, unsigned Y2,
811 unsigned A1, unsigned A2,
812 const void *_r, const void *_g, const void *_b, int y,
813 enum PixelFormat target, int hasAlpha)
815 if (target == PIX_FMT_ARGB || target == PIX_FMT_RGBA ||
816 target == PIX_FMT_ABGR || target == PIX_FMT_BGRA) {
817 uint32_t *dest = (uint32_t *) _dest;
818 const uint32_t *r = (const uint32_t *) _r;
819 const uint32_t *g = (const uint32_t *) _g;
820 const uint32_t *b = (const uint32_t *) _b;
823 int sh = hasAlpha ? ((target == PIX_FMT_RGB32_1 || target == PIX_FMT_BGR32_1) ? 0 : 24) : 0;
825 dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1] + (hasAlpha ? A1 << sh : 0);
826 dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2] + (hasAlpha ? A2 << sh : 0);
829 int sh = (target == PIX_FMT_RGB32_1 || target == PIX_FMT_BGR32_1) ? 0 : 24;
831 dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1] + (A1 << sh);
832 dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2] + (A2 << sh);
834 dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1];
835 dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2];
838 } else if (target == PIX_FMT_RGB24 || target == PIX_FMT_BGR24) {
839 uint8_t *dest = (uint8_t *) _dest;
840 const uint8_t *r = (const uint8_t *) _r;
841 const uint8_t *g = (const uint8_t *) _g;
842 const uint8_t *b = (const uint8_t *) _b;
844 #define r_b ((target == PIX_FMT_RGB24) ? r : b)
845 #define b_r ((target == PIX_FMT_RGB24) ? b : r)
846 dest[i * 6 + 0] = r_b[Y1];
847 dest[i * 6 + 1] = g[Y1];
848 dest[i * 6 + 2] = b_r[Y1];
849 dest[i * 6 + 3] = r_b[Y2];
850 dest[i * 6 + 4] = g[Y2];
851 dest[i * 6 + 5] = b_r[Y2];
854 } else if (target == PIX_FMT_RGB565 || target == PIX_FMT_BGR565 ||
855 target == PIX_FMT_RGB555 || target == PIX_FMT_BGR555 ||
856 target == PIX_FMT_RGB444 || target == PIX_FMT_BGR444) {
857 uint16_t *dest = (uint16_t *) _dest;
858 const uint16_t *r = (const uint16_t *) _r;
859 const uint16_t *g = (const uint16_t *) _g;
860 const uint16_t *b = (const uint16_t *) _b;
861 int dr1, dg1, db1, dr2, dg2, db2;
863 if (target == PIX_FMT_RGB565 || target == PIX_FMT_BGR565) {
864 dr1 = dither_2x2_8[ y & 1 ][0];
865 dg1 = dither_2x2_4[ y & 1 ][0];
866 db1 = dither_2x2_8[(y & 1) ^ 1][0];
867 dr2 = dither_2x2_8[ y & 1 ][1];
868 dg2 = dither_2x2_4[ y & 1 ][1];
869 db2 = dither_2x2_8[(y & 1) ^ 1][1];
870 } else if (target == PIX_FMT_RGB555 || target == PIX_FMT_BGR555) {
871 dr1 = dither_2x2_8[ y & 1 ][0];
872 dg1 = dither_2x2_8[ y & 1 ][1];
873 db1 = dither_2x2_8[(y & 1) ^ 1][0];
874 dr2 = dither_2x2_8[ y & 1 ][1];
875 dg2 = dither_2x2_8[ y & 1 ][0];
876 db2 = dither_2x2_8[(y & 1) ^ 1][1];
878 dr1 = dither_4x4_16[ y & 3 ][0];
879 dg1 = dither_4x4_16[ y & 3 ][1];
880 db1 = dither_4x4_16[(y & 3) ^ 3][0];
881 dr2 = dither_4x4_16[ y & 3 ][1];
882 dg2 = dither_4x4_16[ y & 3 ][0];
883 db2 = dither_4x4_16[(y & 3) ^ 3][1];
886 dest[i * 2 + 0] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1];
887 dest[i * 2 + 1] = r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2];
888 } else /* 8/4-bit */ {
889 uint8_t *dest = (uint8_t *) _dest;
890 const uint8_t *r = (const uint8_t *) _r;
891 const uint8_t *g = (const uint8_t *) _g;
892 const uint8_t *b = (const uint8_t *) _b;
893 int dr1, dg1, db1, dr2, dg2, db2;
895 if (target == PIX_FMT_RGB8 || target == PIX_FMT_BGR8) {
896 const uint8_t * const d64 = dither_8x8_73[y & 7];
897 const uint8_t * const d32 = dither_8x8_32[y & 7];
898 dr1 = dg1 = d32[(i * 2 + 0) & 7];
899 db1 = d64[(i * 2 + 0) & 7];
900 dr2 = dg2 = d32[(i * 2 + 1) & 7];
901 db2 = d64[(i * 2 + 1) & 7];
903 const uint8_t * const d64 = dither_8x8_73 [y & 7];
904 const uint8_t * const d128 = dither_8x8_220[y & 7];
905 dr1 = db1 = d128[(i * 2 + 0) & 7];
906 dg1 = d64[(i * 2 + 0) & 7];
907 dr2 = db2 = d128[(i * 2 + 1) & 7];
908 dg2 = d64[(i * 2 + 1) & 7];
911 if (target == PIX_FMT_RGB4 || target == PIX_FMT_BGR4) {
912 dest[i] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1] +
913 ((r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2]) << 4);
915 dest[i * 2 + 0] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1];
916 dest[i * 2 + 1] = r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2];
921 static av_always_inline void
922 yuv2rgb_X_c_template(SwsContext *c, const int16_t *lumFilter,
923 const int16_t **lumSrc, int lumFilterSize,
924 const int16_t *chrFilter, const int16_t **chrUSrc,
925 const int16_t **chrVSrc, int chrFilterSize,
926 const int16_t **alpSrc, uint8_t *dest, int dstW,
927 int y, enum PixelFormat target, int hasAlpha)
931 for (i = 0; i < (dstW >> 1); i++) {
937 const void *r, *g, *b;
939 for (j = 0; j < lumFilterSize; j++) {
940 Y1 += lumSrc[j][i * 2] * lumFilter[j];
941 Y2 += lumSrc[j][i * 2 + 1] * lumFilter[j];
943 for (j = 0; j < chrFilterSize; j++) {
944 U += chrUSrc[j][i] * chrFilter[j];
945 V += chrVSrc[j][i] * chrFilter[j];
951 if ((Y1 | Y2 | U | V) & 0x100) {
952 Y1 = av_clip_uint8(Y1);
953 Y2 = av_clip_uint8(Y2);
954 U = av_clip_uint8(U);
955 V = av_clip_uint8(V);
960 for (j = 0; j < lumFilterSize; j++) {
961 A1 += alpSrc[j][i * 2 ] * lumFilter[j];
962 A2 += alpSrc[j][i * 2 + 1] * lumFilter[j];
966 if ((A1 | A2) & 0x100) {
967 A1 = av_clip_uint8(A1);
968 A2 = av_clip_uint8(A2);
972 /* FIXME fix tables so that clipping is not needed and then use _NOCLIP*/
974 g = (c->table_gU[U] + c->table_gV[V]);
977 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
978 r, g, b, y, target, hasAlpha);
982 static av_always_inline void
983 yuv2rgb_2_c_template(SwsContext *c, const int16_t *buf[2],
984 const int16_t *ubuf[2], const int16_t *vbuf[2],
985 const int16_t *abuf[2], uint8_t *dest, int dstW,
986 int yalpha, int uvalpha, int y,
987 enum PixelFormat target, int hasAlpha)
989 const int16_t *buf0 = buf[0], *buf1 = buf[1],
990 *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
991 *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
992 *abuf0 = hasAlpha ? abuf[0] : NULL,
993 *abuf1 = hasAlpha ? abuf[1] : NULL;
994 int yalpha1 = 4095 - yalpha;
995 int uvalpha1 = 4095 - uvalpha;
998 for (i = 0; i < (dstW >> 1); i++) {
999 int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 19;
1000 int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 19;
1001 int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha) >> 19;
1002 int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha) >> 19;
1004 const void *r = c->table_rV[V],
1005 *g = (c->table_gU[U] + c->table_gV[V]),
1006 *b = c->table_bU[U];
1008 Y1 = av_clip_uint8(Y1);
1009 Y2 = av_clip_uint8(Y2);
1010 U = av_clip_uint8(U);
1011 V = av_clip_uint8(V);
1014 A1 = (abuf0[i * 2 ] * yalpha1 + abuf1[i * 2 ] * yalpha) >> 19;
1015 A2 = (abuf0[i * 2 + 1] * yalpha1 + abuf1[i * 2 + 1] * yalpha) >> 19;
1016 A1 = av_clip_uint8(A1);
1017 A2 = av_clip_uint8(A2);
1020 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1021 r, g, b, y, target, hasAlpha);
1025 static av_always_inline void
1026 yuv2rgb_1_c_template(SwsContext *c, const int16_t *buf0,
1027 const int16_t *ubuf[2], const int16_t *vbuf[2],
1028 const int16_t *abuf0, uint8_t *dest, int dstW,
1029 int uvalpha, int y, enum PixelFormat target,
1032 const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1035 if (uvalpha < 2048) {
1036 for (i = 0; i < (dstW >> 1); i++) {
1037 int Y1 = buf0[i * 2] >> 7;
1038 int Y2 = buf0[i * 2 + 1] >> 7;
1039 int U = ubuf0[i] >> 7;
1040 int V = vbuf0[i] >> 7;
1042 const void *r = c->table_rV[V],
1043 *g = (c->table_gU[U] + c->table_gV[V]),
1044 *b = c->table_bU[U];
1046 Y1 = av_clip_uint8(Y1);
1047 Y2 = av_clip_uint8(Y2);
1048 U = av_clip_uint8(U);
1049 V = av_clip_uint8(V);
1052 A1 = abuf0[i * 2 ] >> 7;
1053 A2 = abuf0[i * 2 + 1] >> 7;
1054 A1 = av_clip_uint8(A1);
1055 A2 = av_clip_uint8(A2);
1058 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1059 r, g, b, y, target, hasAlpha);
1062 const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1063 for (i = 0; i < (dstW >> 1); i++) {
1064 int Y1 = buf0[i * 2] >> 7;
1065 int Y2 = buf0[i * 2 + 1] >> 7;
1066 int U = (ubuf0[i] + ubuf1[i]) >> 8;
1067 int V = (vbuf0[i] + vbuf1[i]) >> 8;
1069 const void *r = c->table_rV[V],
1070 *g = (c->table_gU[U] + c->table_gV[V]),
1071 *b = c->table_bU[U];
1073 Y1 = av_clip_uint8(Y1);
1074 Y2 = av_clip_uint8(Y2);
1075 U = av_clip_uint8(U);
1076 V = av_clip_uint8(V);
1079 A1 = abuf0[i * 2 ] >> 7;
1080 A2 = abuf0[i * 2 + 1] >> 7;
1081 A1 = av_clip_uint8(A1);
1082 A2 = av_clip_uint8(A2);
1085 yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1086 r, g, b, y, target, hasAlpha);
1091 #define YUV2RGBWRAPPERX(name, base, ext, fmt, hasAlpha) \
1092 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
1093 const int16_t **lumSrc, int lumFilterSize, \
1094 const int16_t *chrFilter, const int16_t **chrUSrc, \
1095 const int16_t **chrVSrc, int chrFilterSize, \
1096 const int16_t **alpSrc, uint8_t *dest, int dstW, \
1099 name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
1100 chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
1101 alpSrc, dest, dstW, y, fmt, hasAlpha); \
1103 #define YUV2RGBWRAPPER(name, base, ext, fmt, hasAlpha) \
1104 YUV2RGBWRAPPERX(name, base, ext, fmt, hasAlpha) \
1105 static void name ## ext ## _2_c(SwsContext *c, const int16_t *buf[2], \
1106 const int16_t *ubuf[2], const int16_t *vbuf[2], \
1107 const int16_t *abuf[2], uint8_t *dest, int dstW, \
1108 int yalpha, int uvalpha, int y) \
1110 name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
1111 dest, dstW, yalpha, uvalpha, y, fmt, hasAlpha); \
1114 static void name ## ext ## _1_c(SwsContext *c, const int16_t *buf0, \
1115 const int16_t *ubuf[2], const int16_t *vbuf[2], \
1116 const int16_t *abuf0, uint8_t *dest, int dstW, \
1117 int uvalpha, int y) \
1119 name ## base ## _1_c_template(c, buf0, ubuf, vbuf, abuf0, dest, \
1120 dstW, uvalpha, y, fmt, hasAlpha); \
1124 YUV2RGBWRAPPER(yuv2rgb,, 32_1, PIX_FMT_RGB32_1, CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1125 YUV2RGBWRAPPER(yuv2rgb,, 32, PIX_FMT_RGB32, CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1127 #if CONFIG_SWSCALE_ALPHA
1128 YUV2RGBWRAPPER(yuv2rgb,, a32_1, PIX_FMT_RGB32_1, 1)
1129 YUV2RGBWRAPPER(yuv2rgb,, a32, PIX_FMT_RGB32, 1)
1131 YUV2RGBWRAPPER(yuv2rgb,, x32_1, PIX_FMT_RGB32_1, 0)
1132 YUV2RGBWRAPPER(yuv2rgb,, x32, PIX_FMT_RGB32, 0)
1134 YUV2RGBWRAPPER(yuv2, rgb, rgb24, PIX_FMT_RGB24, 0)
1135 YUV2RGBWRAPPER(yuv2, rgb, bgr24, PIX_FMT_BGR24, 0)
1136 YUV2RGBWRAPPER(yuv2rgb,, 16, PIX_FMT_RGB565, 0)
1137 YUV2RGBWRAPPER(yuv2rgb,, 15, PIX_FMT_RGB555, 0)
1138 YUV2RGBWRAPPER(yuv2rgb,, 12, PIX_FMT_RGB444, 0)
1139 YUV2RGBWRAPPER(yuv2rgb,, 8, PIX_FMT_RGB8, 0)
1140 YUV2RGBWRAPPER(yuv2rgb,, 4, PIX_FMT_RGB4, 0)
1141 YUV2RGBWRAPPER(yuv2rgb,, 4b, PIX_FMT_RGB4_BYTE, 0)
1143 static av_always_inline void
1144 yuv2rgb_full_X_c_template(SwsContext *c, const int16_t *lumFilter,
1145 const int16_t **lumSrc, int lumFilterSize,
1146 const int16_t *chrFilter, const int16_t **chrUSrc,
1147 const int16_t **chrVSrc, int chrFilterSize,
1148 const int16_t **alpSrc, uint8_t *dest,
1149 int dstW, int y, enum PixelFormat target, int hasAlpha)
1152 int step = (target == PIX_FMT_RGB24 || target == PIX_FMT_BGR24) ? 3 : 4;
1154 for (i = 0; i < dstW; i++) {
1161 for (j = 0; j < lumFilterSize; j++) {
1162 Y += lumSrc[j][i] * lumFilter[j];
1164 for (j = 0; j < chrFilterSize; j++) {
1165 U += chrUSrc[j][i] * chrFilter[j];
1166 V += chrVSrc[j][i] * chrFilter[j];
1173 for (j = 0; j < lumFilterSize; j++) {
1174 A += alpSrc[j][i] * lumFilter[j];
1178 A = av_clip_uint8(A);
1180 Y -= c->yuv2rgb_y_offset;
1181 Y *= c->yuv2rgb_y_coeff;
1183 R = Y + V*c->yuv2rgb_v2r_coeff;
1184 G = Y + V*c->yuv2rgb_v2g_coeff + U*c->yuv2rgb_u2g_coeff;
1185 B = Y + U*c->yuv2rgb_u2b_coeff;
1186 if ((R | G | B) & 0xC0000000) {
1187 R = av_clip_uintp2(R, 30);
1188 G = av_clip_uintp2(G, 30);
1189 B = av_clip_uintp2(B, 30);
1194 dest[0] = hasAlpha ? A : 255;
1208 dest[3] = hasAlpha ? A : 255;
1211 dest[0] = hasAlpha ? A : 255;
1226 dest[3] = hasAlpha ? A : 255;
1234 YUV2RGBWRAPPERX(yuv2, rgb_full, bgra32_full, PIX_FMT_BGRA, CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1235 YUV2RGBWRAPPERX(yuv2, rgb_full, abgr32_full, PIX_FMT_ABGR, CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1236 YUV2RGBWRAPPERX(yuv2, rgb_full, rgba32_full, PIX_FMT_RGBA, CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1237 YUV2RGBWRAPPERX(yuv2, rgb_full, argb32_full, PIX_FMT_ARGB, CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1239 #if CONFIG_SWSCALE_ALPHA
1240 YUV2RGBWRAPPERX(yuv2, rgb_full, bgra32_full, PIX_FMT_BGRA, 1)
1241 YUV2RGBWRAPPERX(yuv2, rgb_full, abgr32_full, PIX_FMT_ABGR, 1)
1242 YUV2RGBWRAPPERX(yuv2, rgb_full, rgba32_full, PIX_FMT_RGBA, 1)
1243 YUV2RGBWRAPPERX(yuv2, rgb_full, argb32_full, PIX_FMT_ARGB, 1)
1245 YUV2RGBWRAPPERX(yuv2, rgb_full, bgrx32_full, PIX_FMT_BGRA, 0)
1246 YUV2RGBWRAPPERX(yuv2, rgb_full, xbgr32_full, PIX_FMT_ABGR, 0)
1247 YUV2RGBWRAPPERX(yuv2, rgb_full, rgbx32_full, PIX_FMT_RGBA, 0)
1248 YUV2RGBWRAPPERX(yuv2, rgb_full, xrgb32_full, PIX_FMT_ARGB, 0)
1250 YUV2RGBWRAPPERX(yuv2, rgb_full, bgr24_full, PIX_FMT_BGR24, 0)
1251 YUV2RGBWRAPPERX(yuv2, rgb_full, rgb24_full, PIX_FMT_RGB24, 0)
1253 void ff_sws_init_output_funcs(SwsContext *c,
1254 yuv2planar1_fn *yuv2plane1,
1255 yuv2planarX_fn *yuv2planeX,
1256 yuv2interleavedX_fn *yuv2nv12cX,
1257 yuv2packed1_fn *yuv2packed1,
1258 yuv2packed2_fn *yuv2packed2,
1259 yuv2packedX_fn *yuv2packedX)
1261 enum PixelFormat dstFormat = c->dstFormat;
1263 if (is16BPS(dstFormat)) {
1264 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_16BE_c : yuv2planeX_16LE_c;
1265 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_16BE_c : yuv2plane1_16LE_c;
1266 } else if (is9_OR_10BPS(dstFormat)) {
1267 if (av_pix_fmt_descriptors[dstFormat].comp[0].depth_minus1 == 8) {
1268 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_9BE_c : yuv2planeX_9LE_c;
1269 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_9BE_c : yuv2plane1_9LE_c;
1271 *yuv2planeX = isBE(dstFormat) ? yuv2planeX_10BE_c : yuv2planeX_10LE_c;
1272 *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_10BE_c : yuv2plane1_10LE_c;
1275 *yuv2plane1 = yuv2plane1_8_c;
1276 *yuv2planeX = yuv2planeX_8_c;
1277 if (dstFormat == PIX_FMT_NV12 || dstFormat == PIX_FMT_NV21)
1278 *yuv2nv12cX = yuv2nv12cX_c;
1281 if(c->flags & SWS_FULL_CHR_H_INT) {
1282 switch (dstFormat) {
1285 *yuv2packedX = yuv2rgba32_full_X_c;
1287 #if CONFIG_SWSCALE_ALPHA
1289 *yuv2packedX = yuv2rgba32_full_X_c;
1291 #endif /* CONFIG_SWSCALE_ALPHA */
1293 *yuv2packedX = yuv2rgbx32_full_X_c;
1295 #endif /* !CONFIG_SMALL */
1299 *yuv2packedX = yuv2argb32_full_X_c;
1301 #if CONFIG_SWSCALE_ALPHA
1303 *yuv2packedX = yuv2argb32_full_X_c;
1305 #endif /* CONFIG_SWSCALE_ALPHA */
1307 *yuv2packedX = yuv2xrgb32_full_X_c;
1309 #endif /* !CONFIG_SMALL */
1313 *yuv2packedX = yuv2bgra32_full_X_c;
1315 #if CONFIG_SWSCALE_ALPHA
1317 *yuv2packedX = yuv2bgra32_full_X_c;
1319 #endif /* CONFIG_SWSCALE_ALPHA */
1321 *yuv2packedX = yuv2bgrx32_full_X_c;
1323 #endif /* !CONFIG_SMALL */
1327 *yuv2packedX = yuv2abgr32_full_X_c;
1329 #if CONFIG_SWSCALE_ALPHA
1331 *yuv2packedX = yuv2abgr32_full_X_c;
1333 #endif /* CONFIG_SWSCALE_ALPHA */
1335 *yuv2packedX = yuv2xbgr32_full_X_c;
1337 #endif /* !CONFIG_SMALL */
1340 *yuv2packedX = yuv2rgb24_full_X_c;
1343 *yuv2packedX = yuv2bgr24_full_X_c;
1347 switch (dstFormat) {
1348 case PIX_FMT_RGB48LE:
1349 *yuv2packed1 = yuv2rgb48le_1_c;
1350 *yuv2packed2 = yuv2rgb48le_2_c;
1351 *yuv2packedX = yuv2rgb48le_X_c;
1353 case PIX_FMT_RGB48BE:
1354 *yuv2packed1 = yuv2rgb48be_1_c;
1355 *yuv2packed2 = yuv2rgb48be_2_c;
1356 *yuv2packedX = yuv2rgb48be_X_c;
1358 case PIX_FMT_BGR48LE:
1359 *yuv2packed1 = yuv2bgr48le_1_c;
1360 *yuv2packed2 = yuv2bgr48le_2_c;
1361 *yuv2packedX = yuv2bgr48le_X_c;
1363 case PIX_FMT_BGR48BE:
1364 *yuv2packed1 = yuv2bgr48be_1_c;
1365 *yuv2packed2 = yuv2bgr48be_2_c;
1366 *yuv2packedX = yuv2bgr48be_X_c;
1371 *yuv2packed1 = yuv2rgb32_1_c;
1372 *yuv2packed2 = yuv2rgb32_2_c;
1373 *yuv2packedX = yuv2rgb32_X_c;
1375 #if CONFIG_SWSCALE_ALPHA
1377 *yuv2packed1 = yuv2rgba32_1_c;
1378 *yuv2packed2 = yuv2rgba32_2_c;
1379 *yuv2packedX = yuv2rgba32_X_c;
1381 #endif /* CONFIG_SWSCALE_ALPHA */
1383 *yuv2packed1 = yuv2rgbx32_1_c;
1384 *yuv2packed2 = yuv2rgbx32_2_c;
1385 *yuv2packedX = yuv2rgbx32_X_c;
1387 #endif /* !CONFIG_SMALL */
1389 case PIX_FMT_RGB32_1:
1390 case PIX_FMT_BGR32_1:
1392 *yuv2packed1 = yuv2rgb32_1_1_c;
1393 *yuv2packed2 = yuv2rgb32_1_2_c;
1394 *yuv2packedX = yuv2rgb32_1_X_c;
1396 #if CONFIG_SWSCALE_ALPHA
1398 *yuv2packed1 = yuv2rgba32_1_1_c;
1399 *yuv2packed2 = yuv2rgba32_1_2_c;
1400 *yuv2packedX = yuv2rgba32_1_X_c;
1402 #endif /* CONFIG_SWSCALE_ALPHA */
1404 *yuv2packed1 = yuv2rgbx32_1_1_c;
1405 *yuv2packed2 = yuv2rgbx32_1_2_c;
1406 *yuv2packedX = yuv2rgbx32_1_X_c;
1408 #endif /* !CONFIG_SMALL */
1411 *yuv2packed1 = yuv2rgb24_1_c;
1412 *yuv2packed2 = yuv2rgb24_2_c;
1413 *yuv2packedX = yuv2rgb24_X_c;
1416 *yuv2packed1 = yuv2bgr24_1_c;
1417 *yuv2packed2 = yuv2bgr24_2_c;
1418 *yuv2packedX = yuv2bgr24_X_c;
1420 case PIX_FMT_RGB565LE:
1421 case PIX_FMT_RGB565BE:
1422 case PIX_FMT_BGR565LE:
1423 case PIX_FMT_BGR565BE:
1424 *yuv2packed1 = yuv2rgb16_1_c;
1425 *yuv2packed2 = yuv2rgb16_2_c;
1426 *yuv2packedX = yuv2rgb16_X_c;
1428 case PIX_FMT_RGB555LE:
1429 case PIX_FMT_RGB555BE:
1430 case PIX_FMT_BGR555LE:
1431 case PIX_FMT_BGR555BE:
1432 *yuv2packed1 = yuv2rgb15_1_c;
1433 *yuv2packed2 = yuv2rgb15_2_c;
1434 *yuv2packedX = yuv2rgb15_X_c;
1436 case PIX_FMT_RGB444LE:
1437 case PIX_FMT_RGB444BE:
1438 case PIX_FMT_BGR444LE:
1439 case PIX_FMT_BGR444BE:
1440 *yuv2packed1 = yuv2rgb12_1_c;
1441 *yuv2packed2 = yuv2rgb12_2_c;
1442 *yuv2packedX = yuv2rgb12_X_c;
1446 *yuv2packed1 = yuv2rgb8_1_c;
1447 *yuv2packed2 = yuv2rgb8_2_c;
1448 *yuv2packedX = yuv2rgb8_X_c;
1452 *yuv2packed1 = yuv2rgb4_1_c;
1453 *yuv2packed2 = yuv2rgb4_2_c;
1454 *yuv2packedX = yuv2rgb4_X_c;
1456 case PIX_FMT_RGB4_BYTE:
1457 case PIX_FMT_BGR4_BYTE:
1458 *yuv2packed1 = yuv2rgb4b_1_c;
1459 *yuv2packed2 = yuv2rgb4b_2_c;
1460 *yuv2packedX = yuv2rgb4b_X_c;
1464 switch (dstFormat) {
1465 case PIX_FMT_MONOWHITE:
1466 *yuv2packed1 = yuv2monowhite_1_c;
1467 *yuv2packed2 = yuv2monowhite_2_c;
1468 *yuv2packedX = yuv2monowhite_X_c;
1470 case PIX_FMT_MONOBLACK:
1471 *yuv2packed1 = yuv2monoblack_1_c;
1472 *yuv2packed2 = yuv2monoblack_2_c;
1473 *yuv2packedX = yuv2monoblack_X_c;
1475 case PIX_FMT_YUYV422:
1476 *yuv2packed1 = yuv2yuyv422_1_c;
1477 *yuv2packed2 = yuv2yuyv422_2_c;
1478 *yuv2packedX = yuv2yuyv422_X_c;
1480 case PIX_FMT_UYVY422:
1481 *yuv2packed1 = yuv2uyvy422_1_c;
1482 *yuv2packed2 = yuv2uyvy422_2_c;
1483 *yuv2packedX = yuv2uyvy422_X_c;