FFmpeg
output.c
Go to the documentation of this file.
1 /*
2  * Copyright (C) 2001-2012 Michael Niedermayer <michaelni@gmx.at>
3  *
4  * This file is part of FFmpeg.
5  *
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.
10  *
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.
15  *
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
19  */
20 
21 #include <math.h>
22 #include <stdint.h>
23 #include <stdio.h>
24 #include <string.h>
25 
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"
34 #include "config.h"
35 #include "rgb2rgb.h"
36 #include "swscale.h"
37 #include "swscale_internal.h"
38 
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, },
43 };
44 
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, },
49 };
50 
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, },
57 };
58 
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, },
69 };
70 
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, },
81 };
82 
83 #if 1
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, },
94 };
95 #elif 1
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, },
107 };
108 #elif 1
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, },
120 };
121 #else
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, },
133 };
134 #endif
135 
136 #define output_pixel(pos, val, bias, signedness) \
137  if (big_endian) { \
138  AV_WB16(pos, bias + av_clip_ ## signedness ## 16(val >> shift)); \
139  } else { \
140  AV_WL16(pos, bias + av_clip_ ## signedness ## 16(val >> shift)); \
141  }
142 
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)
146 {
147  int i;
148  int shift = 3;
149  av_assert0(output_bits == 16);
150 
151  for (i = 0; i < dstW; i++) {
152  int val = src[i] + (1 << (shift - 1));
153  output_pixel(&dest[i], val, 0, uint);
154  }
155 }
156 
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)
161 {
162  int i;
163  int shift = 15;
164  av_assert0(output_bits == 16);
165 
166  for (i = 0; i < dstW; i++) {
167  int val = 1 << (shift - 1);
168  int j;
169 
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. */
175  val -= 0x40000000;
176  for (j = 0; j < filterSize; j++)
177  val += src[j][i] * (unsigned)filter[j];
178 
179  output_pixel(&dest[i], val, 0x8000, int);
180  }
181 }
182 
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)
186 {
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;
190  int shift = 15;
191  int big_endian = c->dstFormat == AV_PIX_FMT_P016BE;
192  int i, j;
193 
194  for (i = 0; i < chrDstW; i++) {
195  int u = 1 << (shift - 1);
196  int v = 1 << (shift - 1);
197 
198  /* See yuv2planeX_16_c_template for details. */
199  u -= 0x40000000;
200  v -= 0x40000000;
201  for (j = 0; j < chrFilterSize; j++) {
202  u += uSrc[j][i] * (unsigned)chrFilter[j];
203  v += vSrc[j][i] * (unsigned)chrFilter[j];
204  }
205 
206  output_pixel(&dest[2*i] , u, 0x8000, int);
207  output_pixel(&dest[2*i+1], v, 0x8000, int);
208  }
209 }
210 
211 static av_always_inline void
212 yuv2plane1_float_c_template(const int32_t *src, float *dest, int dstW)
213 {
214  static const int big_endian = HAVE_BIGENDIAN;
215  static const int shift = 3;
216  static const float float_mult = 1.0f / 65535.0f;
217  int i, val;
218  uint16_t val_uint;
219 
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;
224  }
225 }
226 
227 static av_always_inline void
228 yuv2plane1_float_bswap_c_template(const int32_t *src, uint32_t *dest, int dstW)
229 {
230  static const int big_endian = HAVE_BIGENDIAN;
231  static const int shift = 3;
232  static const float float_mult = 1.0f / 65535.0f;
233  int i, val;
234  uint16_t val_uint;
235 
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));
240  }
241 }
242 
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)
246 {
247  static const int big_endian = HAVE_BIGENDIAN;
248  static const int shift = 15;
249  static const float float_mult = 1.0f / 65535.0f;
250  int i, j, val;
251  uint16_t val_uint;
252 
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];
257  }
258  output_pixel(&val_uint, val, 0x8000, int);
259  dest[i] = float_mult * (float)val_uint;
260  }
261 }
262 
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)
266 {
267  static const int big_endian = HAVE_BIGENDIAN;
268  static const int shift = 15;
269  static const float float_mult = 1.0f / 65535.0f;
270  int i, j, val;
271  uint16_t val_uint;
272 
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];
277  }
278  output_pixel(&val_uint, val, 0x8000, int);
279  dest[i] = av_bswap32(av_float2int(float_mult * (float)val_uint));
280  }
281 }
282 
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) \
286 { \
287  template((const int32_t *)src, (dest_type *)dest, dstW); \
288 }
289 
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) \
294 { \
295  template(filter, filterSize, (const int32_t **)src, (dest_type *)dest, dstW); \
296 }
297 
298 #if HAVE_BIGENDIAN
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)
303 #else
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)
308 #endif
309 
310 #undef output_pixel
311 
312 #define output_pixel(pos, val) \
313  if (big_endian) { \
314  AV_WB16(pos, av_clip_uintp2(val >> shift, output_bits)); \
315  } else { \
316  AV_WL16(pos, av_clip_uintp2(val >> shift, output_bits)); \
317  }
318 
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)
322 {
323  int i;
324  int shift = 15 - output_bits;
325 
326  for (i = 0; i < dstW; i++) {
327  int val = src[i] + (1 << (shift - 1));
328  output_pixel(&dest[i], val);
329  }
330 }
331 
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)
336 {
337  int i;
338  int shift = 11 + 16 - output_bits;
339 
340  for (i = 0; i < dstW; i++) {
341  int val = 1 << (shift - 1);
342  int j;
343 
344  for (j = 0; j < filterSize; j++)
345  val += src[j][i] * filter[j];
346 
347  output_pixel(&dest[i], val);
348  }
349 }
350 
351 #undef output_pixel
352 
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)\
357 { \
358  yuv2plane1_ ## template_size ## _c_template((const typeX_t *) src, \
359  (uint16_t *) dest, dstW, is_be, bits); \
360 }\
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)\
364 { \
365  yuv2planeX_## template_size ## _c_template(filter, \
366  filterSize, (const typeX_t **) src, \
367  (uint16_t *) dest, dstW, is_be, bits); \
368 }
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)
379 
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)
383 {
384  int i;
385  for (i=0; i<dstW; i++) {
386  int val = dither[(i + offset) & 7] << 12;
387  int j;
388  for (j=0; j<filterSize; j++)
389  val += src[j][i] * filter[j];
390 
391  dest[i]= av_clip_uint8(val>>19);
392  }
393 }
394 
395 static void yuv2plane1_8_c(const int16_t *src, uint8_t *dest, int dstW,
396  const uint8_t *dither, int offset)
397 {
398  int i;
399  for (i=0; i<dstW; i++) {
400  int val = (src[i] + dither[(i + offset) & 7]) >> 7;
401  dest[i]= av_clip_uint8(val);
402  }
403 }
404 
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)
408 {
409  enum AVPixelFormat dstFormat = c->dstFormat;
410  const uint8_t *chrDither = c->chrDither8;
411  int i;
412 
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;
418  int j;
419  for (j=0; j<chrFilterSize; j++) {
420  u += chrUSrc[j][i] * chrFilter[j];
421  v += chrVSrc[j][i] * chrFilter[j];
422  }
423 
424  dest[2*i]= av_clip_uint8(u>>19);
425  dest[2*i+1]= av_clip_uint8(v>>19);
426  }
427  else
428  for (i=0; i<chrDstW; i++) {
429  int u = chrDither[i & 7] << 12;
430  int v = chrDither[(i + 3) & 7] << 12;
431  int j;
432  for (j=0; j<chrFilterSize; j++) {
433  u += chrUSrc[j][i] * chrFilter[j];
434  v += chrVSrc[j][i] * chrFilter[j];
435  }
436 
437  dest[2*i]= av_clip_uint8(v>>19);
438  dest[2*i+1]= av_clip_uint8(u>>19);
439  }
440 }
441 
442 
443 #define output_pixel(pos, val) \
444  if (big_endian) { \
445  AV_WB16(pos, av_clip_uintp2(val >> shift, 10) << 6); \
446  } else { \
447  AV_WL16(pos, av_clip_uintp2(val >> shift, 10) << 6); \
448  }
449 
450 static void yuv2p010l1_c(const int16_t *src,
451  uint16_t *dest, int dstW,
452  int big_endian)
453 {
454  int i;
455  int shift = 5;
456 
457  for (i = 0; i < dstW; i++) {
458  int val = src[i] + (1 << (shift - 1));
459  output_pixel(&dest[i], val);
460  }
461 }
462 
463 static void yuv2p010lX_c(const int16_t *filter, int filterSize,
464  const int16_t **src, uint16_t *dest, int dstW,
465  int big_endian)
466 {
467  int i, j;
468  int shift = 17;
469 
470  for (i = 0; i < dstW; i++) {
471  int val = 1 << (shift - 1);
472 
473  for (j = 0; j < filterSize; j++)
474  val += src[j][i] * filter[j];
475 
476  output_pixel(&dest[i], val);
477  }
478 }
479 
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)
483 {
484  uint16_t *dest = (uint16_t*)dest8;
485  int shift = 17;
486  int big_endian = c->dstFormat == AV_PIX_FMT_P010BE;
487  int i, j;
488 
489  for (i = 0; i < chrDstW; i++) {
490  int u = 1 << (shift - 1);
491  int v = 1 << (shift - 1);
492 
493  for (j = 0; j < chrFilterSize; j++) {
494  u += chrUSrc[j][i] * chrFilter[j];
495  v += chrVSrc[j][i] * chrFilter[j];
496  }
497 
498  output_pixel(&dest[2*i] , u);
499  output_pixel(&dest[2*i+1], v);
500  }
501 }
502 
503 static void yuv2p010l1_LE_c(const int16_t *src,
504  uint8_t *dest, int dstW,
505  const uint8_t *dither, int offset)
506 {
507  yuv2p010l1_c(src, (uint16_t*)dest, dstW, 0);
508 }
509 
510 static void yuv2p010l1_BE_c(const int16_t *src,
511  uint8_t *dest, int dstW,
512  const uint8_t *dither, int offset)
513 {
514  yuv2p010l1_c(src, (uint16_t*)dest, dstW, 1);
515 }
516 
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)
520 {
521  yuv2p010lX_c(filter, filterSize, src, (uint16_t*)dest, dstW, 0);
522 }
523 
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)
527 {
528  yuv2p010lX_c(filter, filterSize, src, (uint16_t*)dest, dstW, 1);
529 }
530 
531 #undef output_pixel
532 
533 
534 #define accumulate_bit(acc, val) \
535  acc <<= 1; \
536  acc |= (val) >= 234
537 #define output_pixel(pos, acc) \
538  if (target == AV_PIX_FMT_MONOBLACK) { \
539  pos = acc; \
540  } else { \
541  pos = ~acc; \
542  }
543 
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)
551 {
552  const uint8_t * const d128 = ff_dither_8x8_220[y&7];
553  int i;
554  unsigned acc = 0;
555  int err = 0;
556 
557  for (i = 0; i < dstW; i += 2) {
558  int j;
559  int Y1 = 1 << 18;
560  int Y2 = 1 << 18;
561 
562  for (j = 0; j < lumFilterSize; j++) {
563  Y1 += lumSrc[j][i] * lumFilter[j];
564  Y2 += lumSrc[j][i+1] * lumFilter[j];
565  }
566  Y1 >>= 19;
567  Y2 >>= 19;
568  if ((Y1 | Y2) & 0x100) {
569  Y1 = av_clip_uint8(Y1);
570  Y2 = av_clip_uint8(Y2);
571  }
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);
576  Y1 -= 220*(acc&1);
577 
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);
581  err -= 220*(acc&1);
582  } else {
583  accumulate_bit(acc, Y1 + d128[(i + 0) & 7]);
584  accumulate_bit(acc, Y2 + d128[(i + 1) & 7]);
585  }
586  if ((i & 7) == 6) {
587  output_pixel(*dest++, acc);
588  }
589  }
590  c->dither_error[0][i] = err;
591 
592  if (i & 6) {
593  output_pixel(*dest, acc);
594  }
595 }
596 
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)
603 {
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;
607  int i;
608  av_assert2(yalpha <= 4096U);
609 
610  if (c->dither == SWS_DITHER_ED) {
611  int err = 0;
612  int acc = 0;
613  for (i = 0; i < dstW; i +=2) {
614  int Y;
615 
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);
620  Y -= 220*(acc&1);
621 
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);
626  err -= 220*(acc&1);
627 
628  if ((i & 7) == 6)
629  output_pixel(*dest++, acc);
630  }
631  c->dither_error[0][i] = err;
632  } else {
633  for (i = 0; i < dstW; i += 8) {
634  int Y, acc = 0;
635 
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]);
652 
653  output_pixel(*dest++, acc);
654  }
655  }
656 }
657 
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)
663 {
664  const uint8_t * const d128 = ff_dither_8x8_220[y & 7];
665  int i;
666 
667  if (c->dither == SWS_DITHER_ED) {
668  int err = 0;
669  int acc = 0;
670  for (i = 0; i < dstW; i +=2) {
671  int Y;
672 
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);
677  Y -= 220*(acc&1);
678 
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);
683  err -= 220*(acc&1);
684 
685  if ((i & 7) == 6)
686  output_pixel(*dest++, acc);
687  }
688  c->dither_error[0][i] = err;
689  } else {
690  for (i = 0; i < dstW; i += 8) {
691  int acc = 0;
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]);
700 
701  output_pixel(*dest++, acc);
702  }
703  }
704 }
705 
706 #undef output_pixel
707 #undef accumulate_bit
708 
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, \
715  int y) \
716 { \
717  name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
718  chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
719  alpSrc, dest, dstW, y, fmt); \
720 } \
721  \
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) \
726 { \
727  name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
728  dest, dstW, yalpha, uvalpha, y, fmt); \
729 } \
730  \
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) \
735 { \
736  name ## base ## _1_c_template(c, buf0, ubuf, vbuf, \
737  abuf0, dest, dstW, uvalpha, \
738  y, fmt); \
739 }
740 
741 YUV2PACKEDWRAPPER(yuv2mono,, white, AV_PIX_FMT_MONOWHITE)
742 YUV2PACKEDWRAPPER(yuv2mono,, black, AV_PIX_FMT_MONOBLACK)
743 
744 #define output_pixels(pos, Y1, U, Y2, V) \
745  if (target == AV_PIX_FMT_YUYV422) { \
746  dest[pos + 0] = Y1; \
747  dest[pos + 1] = U; \
748  dest[pos + 2] = Y2; \
749  dest[pos + 3] = V; \
750  } else if (target == AV_PIX_FMT_YVYU422) { \
751  dest[pos + 0] = Y1; \
752  dest[pos + 1] = V; \
753  dest[pos + 2] = Y2; \
754  dest[pos + 3] = U; \
755  } else { /* AV_PIX_FMT_UYVY422 */ \
756  dest[pos + 0] = U; \
757  dest[pos + 1] = Y1; \
758  dest[pos + 2] = V; \
759  dest[pos + 3] = Y2; \
760  }
761 
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)
769 {
770  int i;
771 
772  for (i = 0; i < ((dstW + 1) >> 1); i++) {
773  int j;
774  int Y1 = 1 << 18;
775  int Y2 = 1 << 18;
776  int U = 1 << 18;
777  int V = 1 << 18;
778 
779  for (j = 0; j < lumFilterSize; j++) {
780  Y1 += lumSrc[j][i * 2] * lumFilter[j];
781  Y2 += lumSrc[j][i * 2 + 1] * lumFilter[j];
782  }
783  for (j = 0; j < chrFilterSize; j++) {
784  U += chrUSrc[j][i] * chrFilter[j];
785  V += chrVSrc[j][i] * chrFilter[j];
786  }
787  Y1 >>= 19;
788  Y2 >>= 19;
789  U >>= 19;
790  V >>= 19;
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);
796  }
797  output_pixels(4*i, Y1, U, Y2, V);
798  }
799 }
800 
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)
807 {
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;
813  int i;
814  av_assert2(yalpha <= 4096U);
815  av_assert2(uvalpha <= 4096U);
816 
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;
822 
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);
828  }
829 
830  output_pixels(i * 4, Y1, U, Y2, V);
831  }
832 }
833 
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)
839 {
840  const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
841  int i;
842 
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;
849 
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);
855  }
856 
857  output_pixels(i * 4, Y1, U, Y2, V);
858  }
859  } else {
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;
866 
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);
872  }
873 
874  output_pixels(i * 4, Y1, U, Y2, V);
875  }
876  }
877 }
878 
879 #undef output_pixels
880 
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)
884 
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)) { \
889  AV_WB16(pos, val); \
890  } else { \
891  AV_WL16(pos, val); \
892  }
893 
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)
901 {
902  int hasAlpha = !!alpSrc;
903  int i;
904 
905  for (i = 0; i < dstW; i++) {
906  int j;
907  int Y = 1 << 18;
908  int64_t A = 0xffff<<14;
909 
910  for (j = 0; j < lumFilterSize; j++)
911  Y += lumSrc[j][i] * lumFilter[j];
912 
913  Y >>= 15;
914  Y = av_clip_uint16(Y);
915 
916  if (hasAlpha) {
917  for (j = 0; j < lumFilterSize; j++)
918  A += alpSrc[j][i] * lumFilter[j];
919 
920  A >>= 15;
921  A = av_clip_uint16(A);
922  }
923 
924  output_pixel(&dest[2 * i ], Y);
925  output_pixel(&dest[2 * i + 1], hasAlpha ? A : 65535);
926  }
927 }
928 
929 static av_always_inline void
930 yuv2ya16_2_c_template(SwsContext *c, const int32_t *buf[2],
931  const int32_t *unused_ubuf[2], const int32_t *unused_vbuf[2],
932  const int32_t *abuf[2], uint16_t *dest, int dstW,
933  int yalpha, int unused_uvalpha, int y,
934  enum AVPixelFormat target, int unused_hasAlpha, int unused_eightbytes)
935 {
936  int hasAlpha = abuf && abuf[0] && abuf[1];
937  const int32_t *buf0 = buf[0], *buf1 = buf[1],
938  *abuf0 = hasAlpha ? abuf[0] : NULL,
939  *abuf1 = hasAlpha ? abuf[1] : NULL;
940  int yalpha1 = 4096 - yalpha;
941  int i;
942 
943  av_assert2(yalpha <= 4096U);
944 
945  for (i = 0; i < dstW; i++) {
946  int Y = (buf0[i] * yalpha1 + buf1[i] * yalpha) >> 15;
947  int A;
948 
949  Y = av_clip_uint16(Y);
950 
951  if (hasAlpha) {
952  A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha) >> 15;
953  A = av_clip_uint16(A);
954  }
955 
956  output_pixel(&dest[2 * i ], Y);
957  output_pixel(&dest[2 * i + 1], hasAlpha ? A : 65535);
958  }
959 }
960 
961 static av_always_inline void
962 yuv2ya16_1_c_template(SwsContext *c, const int32_t *buf0,
963  const int32_t *unused_ubuf[2], const int32_t *unused_vbuf[2],
964  const int32_t *abuf0, uint16_t *dest, int dstW,
965  int unused_uvalpha, int y, enum AVPixelFormat target, int unused_hasAlpha, int unused_eightbytes)
966 {
967  int hasAlpha = !!abuf0;
968  int i;
969 
970  for (i = 0; i < dstW; i++) {
971  int Y = buf0[i] >> 3;/* 19 - 16 */
972  int A;
973 
974  Y = av_clip_uint16(Y);
975 
976  if (hasAlpha) {
977  A = abuf0[i] >> 3;
978  if (A & 0x100)
979  A = av_clip_uint16(A);
980  }
981 
982  output_pixel(&dest[2 * i ], Y);
983  output_pixel(&dest[2 * i + 1], hasAlpha ? A : 65535);
984  }
985 }
986 
987 static av_always_inline void
988 yuv2rgba64_X_c_template(SwsContext *c, const int16_t *lumFilter,
989  const int32_t **lumSrc, int lumFilterSize,
990  const int16_t *chrFilter, const int32_t **chrUSrc,
991  const int32_t **chrVSrc, int chrFilterSize,
992  const int32_t **alpSrc, uint16_t *dest, int dstW,
993  int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
994 {
995  int i;
996  int A1 = 0xffff<<14, A2 = 0xffff<<14;
997 
998  for (i = 0; i < ((dstW + 1) >> 1); i++) {
999  int j;
1000  int Y1 = -0x40000000;
1001  int Y2 = -0x40000000;
1002  int U = -(128 << 23); // 19
1003  int V = -(128 << 23);
1004  int R, G, B;
1005 
1006  for (j = 0; j < lumFilterSize; j++) {
1007  Y1 += lumSrc[j][i * 2] * (unsigned)lumFilter[j];
1008  Y2 += lumSrc[j][i * 2 + 1] * (unsigned)lumFilter[j];
1009  }
1010  for (j = 0; j < chrFilterSize; j++) {;
1011  U += chrUSrc[j][i] * (unsigned)chrFilter[j];
1012  V += chrVSrc[j][i] * (unsigned)chrFilter[j];
1013  }
1014 
1015  if (hasAlpha) {
1016  A1 = -0x40000000;
1017  A2 = -0x40000000;
1018  for (j = 0; j < lumFilterSize; j++) {
1019  A1 += alpSrc[j][i * 2] * (unsigned)lumFilter[j];
1020  A2 += alpSrc[j][i * 2 + 1] * (unsigned)lumFilter[j];
1021  }
1022  A1 >>= 1;
1023  A1 += 0x20002000;
1024  A2 >>= 1;
1025  A2 += 0x20002000;
1026  }
1027 
1028  // 8 bits: 12+15=27; 16 bits: 12+19=31
1029  Y1 >>= 14; // 10
1030  Y1 += 0x10000;
1031  Y2 >>= 14;
1032  Y2 += 0x10000;
1033  U >>= 14;
1034  V >>= 14;
1035 
1036  // 8 bits: 27 -> 17 bits, 16 bits: 31 - 14 = 17 bits
1037  Y1 -= c->yuv2rgb_y_offset;
1038  Y2 -= c->yuv2rgb_y_offset;
1039  Y1 *= c->yuv2rgb_y_coeff;
1040  Y2 *= c->yuv2rgb_y_coeff;
1041  Y1 += 1 << 13; // 21
1042  Y2 += 1 << 13;
1043  // 8 bits: 17 + 13 bits = 30 bits, 16 bits: 17 + 13 bits = 30 bits
1044 
1045  R = V * c->yuv2rgb_v2r_coeff;
1046  G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1047  B = U * c->yuv2rgb_u2b_coeff;
1048 
1049  // 8 bits: 30 - 22 = 8 bits, 16 bits: 30 bits - 14 = 16 bits
1050  output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
1051  output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
1052  output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
1053  if (eightbytes) {
1054  output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
1055  output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
1056  output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
1057  output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
1058  output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
1059  dest += 8;
1060  } else {
1061  output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
1062  output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
1063  output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
1064  dest += 6;
1065  }
1066  }
1067 }
1068 
1069 static av_always_inline void
1070 yuv2rgba64_2_c_template(SwsContext *c, const int32_t *buf[2],
1071  const int32_t *ubuf[2], const int32_t *vbuf[2],
1072  const int32_t *abuf[2], uint16_t *dest, int dstW,
1073  int yalpha, int uvalpha, int y,
1074  enum AVPixelFormat target, int hasAlpha, int eightbytes)
1075 {
1076  const int32_t *buf0 = buf[0], *buf1 = buf[1],
1077  *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
1078  *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
1079  *abuf0 = hasAlpha ? abuf[0] : NULL,
1080  *abuf1 = hasAlpha ? abuf[1] : NULL;
1081  int yalpha1 = 4096 - yalpha;
1082  int uvalpha1 = 4096 - uvalpha;
1083  int i;
1084  int A1 = 0xffff<<14, A2 = 0xffff<<14;
1085 
1086  av_assert2(yalpha <= 4096U);
1087  av_assert2(uvalpha <= 4096U);
1088 
1089  for (i = 0; i < ((dstW + 1) >> 1); i++) {
1090  int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 14;
1091  int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 14;
1092  int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha - (128 << 23)) >> 14;
1093  int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha - (128 << 23)) >> 14;
1094  int R, G, B;
1095 
1096  Y1 -= c->yuv2rgb_y_offset;
1097  Y2 -= c->yuv2rgb_y_offset;
1098  Y1 *= c->yuv2rgb_y_coeff;
1099  Y2 *= c->yuv2rgb_y_coeff;
1100  Y1 += 1 << 13;
1101  Y2 += 1 << 13;
1102 
1103  R = V * c->yuv2rgb_v2r_coeff;
1104  G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1105  B = U * c->yuv2rgb_u2b_coeff;
1106 
1107  if (hasAlpha) {
1108  A1 = (abuf0[i * 2 ] * yalpha1 + abuf1[i * 2 ] * yalpha) >> 1;
1109  A2 = (abuf0[i * 2 + 1] * yalpha1 + abuf1[i * 2 + 1] * yalpha) >> 1;
1110 
1111  A1 += 1 << 13;
1112  A2 += 1 << 13;
1113  }
1114 
1115  output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
1116  output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
1117  output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
1118  if (eightbytes) {
1119  output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
1120  output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
1121  output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
1122  output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
1123  output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
1124  dest += 8;
1125  } else {
1126  output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
1127  output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
1128  output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
1129  dest += 6;
1130  }
1131  }
1132 }
1133 
1134 static av_always_inline void
1135 yuv2rgba64_1_c_template(SwsContext *c, const int32_t *buf0,
1136  const int32_t *ubuf[2], const int32_t *vbuf[2],
1137  const int32_t *abuf0, uint16_t *dest, int dstW,
1138  int uvalpha, int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
1139 {
1140  const int32_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1141  int i;
1142  int A1 = 0xffff<<14, A2= 0xffff<<14;
1143 
1144  if (uvalpha < 2048) {
1145  for (i = 0; i < ((dstW + 1) >> 1); i++) {
1146  int Y1 = (buf0[i * 2] ) >> 2;
1147  int Y2 = (buf0[i * 2 + 1]) >> 2;
1148  int U = (ubuf0[i] - (128 << 11)) >> 2;
1149  int V = (vbuf0[i] - (128 << 11)) >> 2;
1150  int R, G, B;
1151 
1152  Y1 -= c->yuv2rgb_y_offset;
1153  Y2 -= c->yuv2rgb_y_offset;
1154  Y1 *= c->yuv2rgb_y_coeff;
1155  Y2 *= c->yuv2rgb_y_coeff;
1156  Y1 += 1 << 13;
1157  Y2 += 1 << 13;
1158 
1159  if (hasAlpha) {
1160  A1 = abuf0[i * 2 ] << 11;
1161  A2 = abuf0[i * 2 + 1] << 11;
1162 
1163  A1 += 1 << 13;
1164  A2 += 1 << 13;
1165  }
1166 
1167  R = V * c->yuv2rgb_v2r_coeff;
1168  G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1169  B = U * c->yuv2rgb_u2b_coeff;
1170 
1171  output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
1172  output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
1173  output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
1174  if (eightbytes) {
1175  output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
1176  output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
1177  output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
1178  output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
1179  output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
1180  dest += 8;
1181  } else {
1182  output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
1183  output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
1184  output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
1185  dest += 6;
1186  }
1187  }
1188  } else {
1189  const int32_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1190  int A1 = 0xffff<<14, A2 = 0xffff<<14;
1191  for (i = 0; i < ((dstW + 1) >> 1); i++) {
1192  int Y1 = (buf0[i * 2] ) >> 2;
1193  int Y2 = (buf0[i * 2 + 1]) >> 2;
1194  int U = (ubuf0[i] + ubuf1[i] - (128 << 12)) >> 3;
1195  int V = (vbuf0[i] + vbuf1[i] - (128 << 12)) >> 3;
1196  int R, G, B;
1197 
1198  Y1 -= c->yuv2rgb_y_offset;
1199  Y2 -= c->yuv2rgb_y_offset;
1200  Y1 *= c->yuv2rgb_y_coeff;
1201  Y2 *= c->yuv2rgb_y_coeff;
1202  Y1 += 1 << 13;
1203  Y2 += 1 << 13;
1204 
1205  if (hasAlpha) {
1206  A1 = abuf0[i * 2 ] << 11;
1207  A2 = abuf0[i * 2 + 1] << 11;
1208 
1209  A1 += 1 << 13;
1210  A2 += 1 << 13;
1211  }
1212 
1213  R = V * c->yuv2rgb_v2r_coeff;
1214  G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1215  B = U * c->yuv2rgb_u2b_coeff;
1216 
1217  output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
1218  output_pixel(&dest[1], av_clip_uintp2( G + Y1, 30) >> 14);
1219  output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
1220  if (eightbytes) {
1221  output_pixel(&dest[3], av_clip_uintp2(A1 , 30) >> 14);
1222  output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
1223  output_pixel(&dest[5], av_clip_uintp2( G + Y2, 30) >> 14);
1224  output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
1225  output_pixel(&dest[7], av_clip_uintp2(A2 , 30) >> 14);
1226  dest += 8;
1227  } else {
1228  output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
1229  output_pixel(&dest[4], av_clip_uintp2( G + Y2, 30) >> 14);
1230  output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
1231  dest += 6;
1232  }
1233  }
1234  }
1235 }
1236 
1237 static av_always_inline void
1238 yuv2rgba64_full_X_c_template(SwsContext *c, const int16_t *lumFilter,
1239  const int32_t **lumSrc, int lumFilterSize,
1240  const int16_t *chrFilter, const int32_t **chrUSrc,
1241  const int32_t **chrVSrc, int chrFilterSize,
1242  const int32_t **alpSrc, uint16_t *dest, int dstW,
1243  int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
1244 {
1245  int i;
1246  int A = 0xffff<<14;
1247 
1248  for (i = 0; i < dstW; i++) {
1249  int j;
1250  int Y = -0x40000000;
1251  int U = -(128 << 23); // 19
1252  int V = -(128 << 23);
1253  int R, G, B;
1254 
1255  for (j = 0; j < lumFilterSize; j++) {
1256  Y += lumSrc[j][i] * (unsigned)lumFilter[j];
1257  }
1258  for (j = 0; j < chrFilterSize; j++) {;
1259  U += chrUSrc[j][i] * (unsigned)chrFilter[j];
1260  V += chrVSrc[j][i] * (unsigned)chrFilter[j];
1261  }
1262 
1263  if (hasAlpha) {
1264  A = -0x40000000;
1265  for (j = 0; j < lumFilterSize; j++) {
1266  A += alpSrc[j][i] * (unsigned)lumFilter[j];
1267  }
1268  A >>= 1;
1269  A += 0x20002000;
1270  }
1271 
1272  // 8bit: 12+15=27; 16-bit: 12+19=31
1273  Y >>= 14; // 10
1274  Y += 0x10000;
1275  U >>= 14;
1276  V >>= 14;
1277 
1278  // 8bit: 27 -> 17bit, 16bit: 31 - 14 = 17bit
1279  Y -= c->yuv2rgb_y_offset;
1280  Y *= c->yuv2rgb_y_coeff;
1281  Y += 1 << 13; // 21
1282  // 8bit: 17 + 13bit = 30bit, 16bit: 17 + 13bit = 30bit
1283 
1284  R = V * c->yuv2rgb_v2r_coeff;
1285  G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1286  B = U * c->yuv2rgb_u2b_coeff;
1287 
1288  // 8bit: 30 - 22 = 8bit, 16bit: 30bit - 14 = 16bit
1289  output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1290  output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
1291  output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1292  if (eightbytes) {
1293  output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1294  dest += 4;
1295  } else {
1296  dest += 3;
1297  }
1298  }
1299 }
1300 
1301 static av_always_inline void
1302 yuv2rgba64_full_2_c_template(SwsContext *c, const int32_t *buf[2],
1303  const int32_t *ubuf[2], const int32_t *vbuf[2],
1304  const int32_t *abuf[2], uint16_t *dest, int dstW,
1305  int yalpha, int uvalpha, int y,
1306  enum AVPixelFormat target, int hasAlpha, int eightbytes)
1307 {
1308  const int32_t *buf0 = buf[0], *buf1 = buf[1],
1309  *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
1310  *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
1311  *abuf0 = hasAlpha ? abuf[0] : NULL,
1312  *abuf1 = hasAlpha ? abuf[1] : NULL;
1313  int yalpha1 = 4096 - yalpha;
1314  int uvalpha1 = 4096 - uvalpha;
1315  int i;
1316  int A = 0xffff<<14;
1317 
1318  av_assert2(yalpha <= 4096U);
1319  av_assert2(uvalpha <= 4096U);
1320 
1321  for (i = 0; i < dstW; i++) {
1322  int Y = (buf0[i] * yalpha1 + buf1[i] * yalpha) >> 14;
1323  int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha - (128 << 23)) >> 14;
1324  int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha - (128 << 23)) >> 14;
1325  int R, G, B;
1326 
1327  Y -= c->yuv2rgb_y_offset;
1328  Y *= c->yuv2rgb_y_coeff;
1329  Y += 1 << 13;
1330 
1331  R = V * c->yuv2rgb_v2r_coeff;
1332  G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1333  B = U * c->yuv2rgb_u2b_coeff;
1334 
1335  if (hasAlpha) {
1336  A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha) >> 1;
1337 
1338  A += 1 << 13;
1339  }
1340 
1341  output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1342  output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
1343  output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1344  if (eightbytes) {
1345  output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1346  dest += 4;
1347  } else {
1348  dest += 3;
1349  }
1350  }
1351 }
1352 
1353 static av_always_inline void
1354 yuv2rgba64_full_1_c_template(SwsContext *c, const int32_t *buf0,
1355  const int32_t *ubuf[2], const int32_t *vbuf[2],
1356  const int32_t *abuf0, uint16_t *dest, int dstW,
1357  int uvalpha, int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
1358 {
1359  const int32_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1360  int i;
1361  int A = 0xffff<<14;
1362 
1363  if (uvalpha < 2048) {
1364  for (i = 0; i < dstW; i++) {
1365  int Y = (buf0[i]) >> 2;
1366  int U = (ubuf0[i] - (128 << 11)) >> 2;
1367  int V = (vbuf0[i] - (128 << 11)) >> 2;
1368  int R, G, B;
1369 
1370  Y -= c->yuv2rgb_y_offset;
1371  Y *= c->yuv2rgb_y_coeff;
1372  Y += 1 << 13;
1373 
1374  if (hasAlpha) {
1375  A = abuf0[i] << 11;
1376 
1377  A += 1 << 13;
1378  }
1379 
1380  R = V * c->yuv2rgb_v2r_coeff;
1381  G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1382  B = U * c->yuv2rgb_u2b_coeff;
1383 
1384  output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1385  output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
1386  output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1387  if (eightbytes) {
1388  output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1389  dest += 4;
1390  } else {
1391  dest += 3;
1392  }
1393  }
1394  } else {
1395  const int32_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1396  int A = 0xffff<<14;
1397  for (i = 0; i < dstW; i++) {
1398  int Y = (buf0[i] ) >> 2;
1399  int U = (ubuf0[i] + ubuf1[i] - (128 << 12)) >> 3;
1400  int V = (vbuf0[i] + vbuf1[i] - (128 << 12)) >> 3;
1401  int R, G, B;
1402 
1403  Y -= c->yuv2rgb_y_offset;
1404  Y *= c->yuv2rgb_y_coeff;
1405  Y += 1 << 13;
1406 
1407  if (hasAlpha) {
1408  A = abuf0[i] << 11;
1409 
1410  A += 1 << 13;
1411  }
1412 
1413  R = V * c->yuv2rgb_v2r_coeff;
1414  G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1415  B = U * c->yuv2rgb_u2b_coeff;
1416 
1417  output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1418  output_pixel(&dest[1], av_clip_uintp2( G + Y, 30) >> 14);
1419  output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1420  if (eightbytes) {
1421  output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1422  dest += 4;
1423  } else {
1424  dest += 3;
1425  }
1426  }
1427  }
1428 }
1429 
1430 #undef output_pixel
1431 #undef r_b
1432 #undef b_r
1433 
1434 #define YUV2PACKED16WRAPPER(name, base, ext, fmt, hasAlpha, eightbytes) \
1435 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
1436  const int16_t **_lumSrc, int lumFilterSize, \
1437  const int16_t *chrFilter, const int16_t **_chrUSrc, \
1438  const int16_t **_chrVSrc, int chrFilterSize, \
1439  const int16_t **_alpSrc, uint8_t *_dest, int dstW, \
1440  int y) \
1441 { \
1442  const int32_t **lumSrc = (const int32_t **) _lumSrc, \
1443  **chrUSrc = (const int32_t **) _chrUSrc, \
1444  **chrVSrc = (const int32_t **) _chrVSrc, \
1445  **alpSrc = (const int32_t **) _alpSrc; \
1446  uint16_t *dest = (uint16_t *) _dest; \
1447  name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
1448  chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
1449  alpSrc, dest, dstW, y, fmt, hasAlpha, eightbytes); \
1450 } \
1451  \
1452 static void name ## ext ## _2_c(SwsContext *c, const int16_t *_buf[2], \
1453  const int16_t *_ubuf[2], const int16_t *_vbuf[2], \
1454  const int16_t *_abuf[2], uint8_t *_dest, int dstW, \
1455  int yalpha, int uvalpha, int y) \
1456 { \
1457  const int32_t **buf = (const int32_t **) _buf, \
1458  **ubuf = (const int32_t **) _ubuf, \
1459  **vbuf = (const int32_t **) _vbuf, \
1460  **abuf = (const int32_t **) _abuf; \
1461  uint16_t *dest = (uint16_t *) _dest; \
1462  name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
1463  dest, dstW, yalpha, uvalpha, y, fmt, hasAlpha, eightbytes); \
1464 } \
1465  \
1466 static void name ## ext ## _1_c(SwsContext *c, const int16_t *_buf0, \
1467  const int16_t *_ubuf[2], const int16_t *_vbuf[2], \
1468  const int16_t *_abuf0, uint8_t *_dest, int dstW, \
1469  int uvalpha, int y) \
1470 { \
1471  const int32_t *buf0 = (const int32_t *) _buf0, \
1472  **ubuf = (const int32_t **) _ubuf, \
1473  **vbuf = (const int32_t **) _vbuf, \
1474  *abuf0 = (const int32_t *) _abuf0; \
1475  uint16_t *dest = (uint16_t *) _dest; \
1476  name ## base ## _1_c_template(c, buf0, ubuf, vbuf, abuf0, dest, \
1477  dstW, uvalpha, y, fmt, hasAlpha, eightbytes); \
1478 }
1479 
1480 YUV2PACKED16WRAPPER(yuv2, rgba64, rgb48be, AV_PIX_FMT_RGB48BE, 0, 0)
1481 YUV2PACKED16WRAPPER(yuv2, rgba64, rgb48le, AV_PIX_FMT_RGB48LE, 0, 0)
1482 YUV2PACKED16WRAPPER(yuv2, rgba64, bgr48be, AV_PIX_FMT_BGR48BE, 0, 0)
1483 YUV2PACKED16WRAPPER(yuv2, rgba64, bgr48le, AV_PIX_FMT_BGR48LE, 0, 0)
1484 YUV2PACKED16WRAPPER(yuv2, rgba64, rgba64be, AV_PIX_FMT_RGBA64BE, 1, 1)
1485 YUV2PACKED16WRAPPER(yuv2, rgba64, rgba64le, AV_PIX_FMT_RGBA64LE, 1, 1)
1486 YUV2PACKED16WRAPPER(yuv2, rgba64, rgbx64be, AV_PIX_FMT_RGBA64BE, 0, 1)
1487 YUV2PACKED16WRAPPER(yuv2, rgba64, rgbx64le, AV_PIX_FMT_RGBA64LE, 0, 1)
1488 YUV2PACKED16WRAPPER(yuv2, rgba64, bgra64be, AV_PIX_FMT_BGRA64BE, 1, 1)
1489 YUV2PACKED16WRAPPER(yuv2, rgba64, bgra64le, AV_PIX_FMT_BGRA64LE, 1, 1)
1490 YUV2PACKED16WRAPPER(yuv2, rgba64, bgrx64be, AV_PIX_FMT_BGRA64BE, 0, 1)
1491 YUV2PACKED16WRAPPER(yuv2, rgba64, bgrx64le, AV_PIX_FMT_BGRA64LE, 0, 1)
1492 YUV2PACKED16WRAPPER(yuv2, ya16, ya16be, AV_PIX_FMT_YA16BE, 1, 0)
1493 YUV2PACKED16WRAPPER(yuv2, ya16, ya16le, AV_PIX_FMT_YA16LE, 1, 0)
1494 
1495 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgb48be_full, AV_PIX_FMT_RGB48BE, 0, 0)
1496 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgb48le_full, AV_PIX_FMT_RGB48LE, 0, 0)
1497 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgr48be_full, AV_PIX_FMT_BGR48BE, 0, 0)
1498 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgr48le_full, AV_PIX_FMT_BGR48LE, 0, 0)
1499 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgba64be_full, AV_PIX_FMT_RGBA64BE, 1, 1)
1500 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgba64le_full, AV_PIX_FMT_RGBA64LE, 1, 1)
1501 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgbx64be_full, AV_PIX_FMT_RGBA64BE, 0, 1)
1502 YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgbx64le_full, AV_PIX_FMT_RGBA64LE, 0, 1)
1503 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgra64be_full, AV_PIX_FMT_BGRA64BE, 1, 1)
1504 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgra64le_full, AV_PIX_FMT_BGRA64LE, 1, 1)
1505 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgrx64be_full, AV_PIX_FMT_BGRA64BE, 0, 1)
1506 YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgrx64le_full, AV_PIX_FMT_BGRA64LE, 0, 1)
1507 
1508 /*
1509  * Write out 2 RGB pixels in the target pixel format. This function takes a
1510  * R/G/B LUT as generated by ff_yuv2rgb_c_init_tables(), which takes care of
1511  * things like endianness conversion and shifting. The caller takes care of
1512  * setting the correct offset in these tables from the chroma (U/V) values.
1513  * This function then uses the luminance (Y1/Y2) values to write out the
1514  * correct RGB values into the destination buffer.
1515  */
1516 static av_always_inline void
1517 yuv2rgb_write(uint8_t *_dest, int i, int Y1, int Y2,
1518  unsigned A1, unsigned A2,
1519  const void *_r, const void *_g, const void *_b, int y,
1520  enum AVPixelFormat target, int hasAlpha)
1521 {
1522  if (target == AV_PIX_FMT_ARGB || target == AV_PIX_FMT_RGBA ||
1523  target == AV_PIX_FMT_ABGR || target == AV_PIX_FMT_BGRA) {
1524  uint32_t *dest = (uint32_t *) _dest;
1525  const uint32_t *r = (const uint32_t *) _r;
1526  const uint32_t *g = (const uint32_t *) _g;
1527  const uint32_t *b = (const uint32_t *) _b;
1528 
1529 #if CONFIG_SMALL
1530  int sh = hasAlpha ? ((target == AV_PIX_FMT_RGB32_1 || target == AV_PIX_FMT_BGR32_1) ? 0 : 24) : 0;
1531 
1532  dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1] + (hasAlpha ? A1 << sh : 0);
1533  dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2] + (hasAlpha ? A2 << sh : 0);
1534 #else
1535  if (hasAlpha) {
1536  int sh = (target == AV_PIX_FMT_RGB32_1 || target == AV_PIX_FMT_BGR32_1) ? 0 : 24;
1537 
1538  av_assert2((((r[Y1] + g[Y1] + b[Y1]) >> sh) & 0xFF) == 0);
1539  dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1] + (A1 << sh);
1540  dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2] + (A2 << sh);
1541  } else {
1542 #if defined(ASSERT_LEVEL) && ASSERT_LEVEL > 1
1543  int sh = (target == AV_PIX_FMT_RGB32_1 || target == AV_PIX_FMT_BGR32_1) ? 0 : 24;
1544 
1545  av_assert2((((r[Y1] + g[Y1] + b[Y1]) >> sh) & 0xFF) == 0xFF);
1546 #endif
1547  dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1];
1548  dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2];
1549  }
1550 #endif
1551  } else if (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) {
1552  uint8_t *dest = (uint8_t *) _dest;
1553  const uint8_t *r = (const uint8_t *) _r;
1554  const uint8_t *g = (const uint8_t *) _g;
1555  const uint8_t *b = (const uint8_t *) _b;
1556 
1557 #define r_b ((target == AV_PIX_FMT_RGB24) ? r : b)
1558 #define b_r ((target == AV_PIX_FMT_RGB24) ? b : r)
1559 
1560  dest[i * 6 + 0] = r_b[Y1];
1561  dest[i * 6 + 1] = g[Y1];
1562  dest[i * 6 + 2] = b_r[Y1];
1563  dest[i * 6 + 3] = r_b[Y2];
1564  dest[i * 6 + 4] = g[Y2];
1565  dest[i * 6 + 5] = b_r[Y2];
1566 #undef r_b
1567 #undef b_r
1568  } else if (target == AV_PIX_FMT_RGB565 || target == AV_PIX_FMT_BGR565 ||
1569  target == AV_PIX_FMT_RGB555 || target == AV_PIX_FMT_BGR555 ||
1570  target == AV_PIX_FMT_RGB444 || target == AV_PIX_FMT_BGR444) {
1571  uint16_t *dest = (uint16_t *) _dest;
1572  const uint16_t *r = (const uint16_t *) _r;
1573  const uint16_t *g = (const uint16_t *) _g;
1574  const uint16_t *b = (const uint16_t *) _b;
1575  int dr1, dg1, db1, dr2, dg2, db2;
1576 
1577  if (target == AV_PIX_FMT_RGB565 || target == AV_PIX_FMT_BGR565) {
1578  dr1 = ff_dither_2x2_8[ y & 1 ][0];
1579  dg1 = ff_dither_2x2_4[ y & 1 ][0];
1580  db1 = ff_dither_2x2_8[(y & 1) ^ 1][0];
1581  dr2 = ff_dither_2x2_8[ y & 1 ][1];
1582  dg2 = ff_dither_2x2_4[ y & 1 ][1];
1583  db2 = ff_dither_2x2_8[(y & 1) ^ 1][1];
1584  } else if (target == AV_PIX_FMT_RGB555 || target == AV_PIX_FMT_BGR555) {
1585  dr1 = ff_dither_2x2_8[ y & 1 ][0];
1586  dg1 = ff_dither_2x2_8[ y & 1 ][1];
1587  db1 = ff_dither_2x2_8[(y & 1) ^ 1][0];
1588  dr2 = ff_dither_2x2_8[ y & 1 ][1];
1589  dg2 = ff_dither_2x2_8[ y & 1 ][0];
1590  db2 = ff_dither_2x2_8[(y & 1) ^ 1][1];
1591  } else {
1592  dr1 = ff_dither_4x4_16[ y & 3 ][0];
1593  dg1 = ff_dither_4x4_16[ y & 3 ][1];
1594  db1 = ff_dither_4x4_16[(y & 3) ^ 3][0];
1595  dr2 = ff_dither_4x4_16[ y & 3 ][1];
1596  dg2 = ff_dither_4x4_16[ y & 3 ][0];
1597  db2 = ff_dither_4x4_16[(y & 3) ^ 3][1];
1598  }
1599 
1600  dest[i * 2 + 0] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1];
1601  dest[i * 2 + 1] = r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2];
1602  } else /* 8/4 bits */ {
1603  uint8_t *dest = (uint8_t *) _dest;
1604  const uint8_t *r = (const uint8_t *) _r;
1605  const uint8_t *g = (const uint8_t *) _g;
1606  const uint8_t *b = (const uint8_t *) _b;
1607  int dr1, dg1, db1, dr2, dg2, db2;
1608 
1609  if (target == AV_PIX_FMT_RGB8 || target == AV_PIX_FMT_BGR8) {
1610  const uint8_t * const d64 = ff_dither_8x8_73[y & 7];
1611  const uint8_t * const d32 = ff_dither_8x8_32[y & 7];
1612  dr1 = dg1 = d32[(i * 2 + 0) & 7];
1613  db1 = d64[(i * 2 + 0) & 7];
1614  dr2 = dg2 = d32[(i * 2 + 1) & 7];
1615  db2 = d64[(i * 2 + 1) & 7];
1616  } else {
1617  const uint8_t * const d64 = ff_dither_8x8_73 [y & 7];
1618  const uint8_t * const d128 = ff_dither_8x8_220[y & 7];
1619  dr1 = db1 = d128[(i * 2 + 0) & 7];
1620  dg1 = d64[(i * 2 + 0) & 7];
1621  dr2 = db2 = d128[(i * 2 + 1) & 7];
1622  dg2 = d64[(i * 2 + 1) & 7];
1623  }
1624 
1625  if (target == AV_PIX_FMT_RGB4 || target == AV_PIX_FMT_BGR4) {
1626  dest[i] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1] +
1627  ((r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2]) << 4);
1628  } else {
1629  dest[i * 2 + 0] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1];
1630  dest[i * 2 + 1] = r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2];
1631  }
1632  }
1633 }
1634 
1635 static av_always_inline void
1636 yuv2rgb_X_c_template(SwsContext *c, const int16_t *lumFilter,
1637  const int16_t **lumSrc, int lumFilterSize,
1638  const int16_t *chrFilter, const int16_t **chrUSrc,
1639  const int16_t **chrVSrc, int chrFilterSize,
1640  const int16_t **alpSrc, uint8_t *dest, int dstW,
1641  int y, enum AVPixelFormat target, int hasAlpha)
1642 {
1643  int i;
1644 
1645  for (i = 0; i < ((dstW + 1) >> 1); i++) {
1646  int j, A1, A2;
1647  int Y1 = 1 << 18;
1648  int Y2 = 1 << 18;
1649  int U = 1 << 18;
1650  int V = 1 << 18;
1651  const void *r, *g, *b;
1652 
1653  for (j = 0; j < lumFilterSize; j++) {
1654  Y1 += lumSrc[j][i * 2] * lumFilter[j];
1655  Y2 += lumSrc[j][i * 2 + 1] * lumFilter[j];
1656  }
1657  for (j = 0; j < chrFilterSize; j++) {
1658  U += chrUSrc[j][i] * chrFilter[j];
1659  V += chrVSrc[j][i] * chrFilter[j];
1660  }
1661  Y1 >>= 19;
1662  Y2 >>= 19;
1663  U >>= 19;
1664  V >>= 19;
1665  if (hasAlpha) {
1666  A1 = 1 << 18;
1667  A2 = 1 << 18;
1668  for (j = 0; j < lumFilterSize; j++) {
1669  A1 += alpSrc[j][i * 2 ] * lumFilter[j];
1670  A2 += alpSrc[j][i * 2 + 1] * lumFilter[j];
1671  }
1672  A1 >>= 19;
1673  A2 >>= 19;
1674  if ((A1 | A2) & 0x100) {
1675  A1 = av_clip_uint8(A1);
1676  A2 = av_clip_uint8(A2);
1677  }
1678  }
1679 
1680  r = c->table_rV[V + YUVRGB_TABLE_HEADROOM];
1681  g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]);
1682  b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1683 
1684  yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1685  r, g, b, y, target, hasAlpha);
1686  }
1687 }
1688 
1689 static av_always_inline void
1690 yuv2rgb_2_c_template(SwsContext *c, const int16_t *buf[2],
1691  const int16_t *ubuf[2], const int16_t *vbuf[2],
1692  const int16_t *abuf[2], uint8_t *dest, int dstW,
1693  int yalpha, int uvalpha, int y,
1694  enum AVPixelFormat target, int hasAlpha)
1695 {
1696  const int16_t *buf0 = buf[0], *buf1 = buf[1],
1697  *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
1698  *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
1699  *abuf0 = hasAlpha ? abuf[0] : NULL,
1700  *abuf1 = hasAlpha ? abuf[1] : NULL;
1701  int yalpha1 = 4096 - yalpha;
1702  int uvalpha1 = 4096 - uvalpha;
1703  int i;
1704  av_assert2(yalpha <= 4096U);
1705  av_assert2(uvalpha <= 4096U);
1706 
1707  for (i = 0; i < ((dstW + 1) >> 1); i++) {
1708  int Y1 = (buf0[i * 2] * yalpha1 + buf1[i * 2] * yalpha) >> 19;
1709  int Y2 = (buf0[i * 2 + 1] * yalpha1 + buf1[i * 2 + 1] * yalpha) >> 19;
1710  int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha) >> 19;
1711  int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha) >> 19;
1712  int A1, A2;
1713  const void *r = c->table_rV[V + YUVRGB_TABLE_HEADROOM],
1714  *g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]),
1715  *b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1716 
1717  if (hasAlpha) {
1718  A1 = (abuf0[i * 2 ] * yalpha1 + abuf1[i * 2 ] * yalpha) >> 19;
1719  A2 = (abuf0[i * 2 + 1] * yalpha1 + abuf1[i * 2 + 1] * yalpha) >> 19;
1720  A1 = av_clip_uint8(A1);
1721  A2 = av_clip_uint8(A2);
1722  }
1723 
1724  yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1725  r, g, b, y, target, hasAlpha);
1726  }
1727 }
1728 
1729 static av_always_inline void
1730 yuv2rgb_1_c_template(SwsContext *c, const int16_t *buf0,
1731  const int16_t *ubuf[2], const int16_t *vbuf[2],
1732  const int16_t *abuf0, uint8_t *dest, int dstW,
1733  int uvalpha, int y, enum AVPixelFormat target,
1734  int hasAlpha)
1735 {
1736  const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1737  int i;
1738 
1739  if (uvalpha < 2048) {
1740  for (i = 0; i < ((dstW + 1) >> 1); i++) {
1741  int Y1 = (buf0[i * 2 ] + 64) >> 7;
1742  int Y2 = (buf0[i * 2 + 1] + 64) >> 7;
1743  int U = (ubuf0[i] + 64) >> 7;
1744  int V = (vbuf0[i] + 64) >> 7;
1745  int A1, A2;
1746  const void *r = c->table_rV[V + YUVRGB_TABLE_HEADROOM],
1747  *g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]),
1748  *b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1749 
1750  if (hasAlpha) {
1751  A1 = abuf0[i * 2 ] * 255 + 16384 >> 15;
1752  A2 = abuf0[i * 2 + 1] * 255 + 16384 >> 15;
1753  A1 = av_clip_uint8(A1);
1754  A2 = av_clip_uint8(A2);
1755  }
1756 
1757  yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1758  r, g, b, y, target, hasAlpha);
1759  }
1760  } else {
1761  const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1762  for (i = 0; i < ((dstW + 1) >> 1); i++) {
1763  int Y1 = (buf0[i * 2 ] + 64) >> 7;
1764  int Y2 = (buf0[i * 2 + 1] + 64) >> 7;
1765  int U = (ubuf0[i] + ubuf1[i] + 128) >> 8;
1766  int V = (vbuf0[i] + vbuf1[i] + 128) >> 8;
1767  int A1, A2;
1768  const void *r = c->table_rV[V + YUVRGB_TABLE_HEADROOM],
1769  *g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]),
1770  *b = c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1771 
1772  if (hasAlpha) {
1773  A1 = (abuf0[i * 2 ] + 64) >> 7;
1774  A2 = (abuf0[i * 2 + 1] + 64) >> 7;
1775  A1 = av_clip_uint8(A1);
1776  A2 = av_clip_uint8(A2);
1777  }
1778 
1779  yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1780  r, g, b, y, target, hasAlpha);
1781  }
1782  }
1783 }
1784 
1785 #define YUV2RGBWRAPPERX(name, base, ext, fmt, hasAlpha) \
1786 static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
1787  const int16_t **lumSrc, int lumFilterSize, \
1788  const int16_t *chrFilter, const int16_t **chrUSrc, \
1789  const int16_t **chrVSrc, int chrFilterSize, \
1790  const int16_t **alpSrc, uint8_t *dest, int dstW, \
1791  int y) \
1792 { \
1793  name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
1794  chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
1795  alpSrc, dest, dstW, y, fmt, hasAlpha); \
1796 }
1797 
1798 #define YUV2RGBWRAPPERX2(name, base, ext, fmt, hasAlpha) \
1799 YUV2RGBWRAPPERX(name, base, ext, fmt, hasAlpha) \
1800 static void name ## ext ## _2_c(SwsContext *c, const int16_t *buf[2], \
1801  const int16_t *ubuf[2], const int16_t *vbuf[2], \
1802  const int16_t *abuf[2], uint8_t *dest, int dstW, \
1803  int yalpha, int uvalpha, int y) \
1804 { \
1805  name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
1806  dest, dstW, yalpha, uvalpha, y, fmt, hasAlpha); \
1807 }
1808 
1809 #define YUV2RGBWRAPPER(name, base, ext, fmt, hasAlpha) \
1810 YUV2RGBWRAPPERX2(name, base, ext, fmt, hasAlpha) \
1811 static void name ## ext ## _1_c(SwsContext *c, const int16_t *buf0, \
1812  const int16_t *ubuf[2], const int16_t *vbuf[2], \
1813  const int16_t *abuf0, uint8_t *dest, int dstW, \
1814  int uvalpha, int y) \
1815 { \
1816  name ## base ## _1_c_template(c, buf0, ubuf, vbuf, abuf0, dest, \
1817  dstW, uvalpha, y, fmt, hasAlpha); \
1818 }
1819 
1820 #if CONFIG_SMALL
1821 YUV2RGBWRAPPER(yuv2rgb,, 32_1, AV_PIX_FMT_RGB32_1, CONFIG_SWSCALE_ALPHA && c->needAlpha)
1822 YUV2RGBWRAPPER(yuv2rgb,, 32, AV_PIX_FMT_RGB32, CONFIG_SWSCALE_ALPHA && c->needAlpha)
1823 #else
1824 #if CONFIG_SWSCALE_ALPHA
1825 YUV2RGBWRAPPER(yuv2rgb,, a32_1, AV_PIX_FMT_RGB32_1, 1)
1826 YUV2RGBWRAPPER(yuv2rgb,, a32, AV_PIX_FMT_RGB32, 1)
1827 #endif
1828 YUV2RGBWRAPPER(yuv2rgb,, x32_1, AV_PIX_FMT_RGB32_1, 0)
1829 YUV2RGBWRAPPER(yuv2rgb,, x32, AV_PIX_FMT_RGB32, 0)
1830 #endif
1831 YUV2RGBWRAPPER(yuv2, rgb, rgb24, AV_PIX_FMT_RGB24, 0)
1832 YUV2RGBWRAPPER(yuv2, rgb, bgr24, AV_PIX_FMT_BGR24, 0)
1833 YUV2RGBWRAPPER(yuv2rgb,, 16, AV_PIX_FMT_RGB565, 0)
1834 YUV2RGBWRAPPER(yuv2rgb,, 15, AV_PIX_FMT_RGB555, 0)
1835 YUV2RGBWRAPPER(yuv2rgb,, 12, AV_PIX_FMT_RGB444, 0)
1836 YUV2RGBWRAPPER(yuv2rgb,, 8, AV_PIX_FMT_RGB8, 0)
1837 YUV2RGBWRAPPER(yuv2rgb,, 4, AV_PIX_FMT_RGB4, 0)
1838 YUV2RGBWRAPPER(yuv2rgb,, 4b, AV_PIX_FMT_RGB4_BYTE, 0)
1839 
1840 static av_always_inline void yuv2rgb_write_full(SwsContext *c,
1841  uint8_t *dest, int i, int Y, int A, int U, int V,
1842  int y, enum AVPixelFormat target, int hasAlpha, int err[4])
1843 {
1844  int R, G, B;
1845  int isrgb8 = target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8;
1846 
1847  Y -= c->yuv2rgb_y_offset;
1848  Y *= c->yuv2rgb_y_coeff;
1849  Y += 1 << 21;
1850  R = (unsigned)Y + V*c->yuv2rgb_v2r_coeff;
1851  G = (unsigned)Y + V*c->yuv2rgb_v2g_coeff + U*c->yuv2rgb_u2g_coeff;
1852  B = (unsigned)Y + U*c->yuv2rgb_u2b_coeff;
1853  if ((R | G | B) & 0xC0000000) {
1854  R = av_clip_uintp2(R, 30);
1855  G = av_clip_uintp2(G, 30);
1856  B = av_clip_uintp2(B, 30);
1857  }
1858 
1859  switch(target) {
1860  case AV_PIX_FMT_ARGB:
1861  dest[0] = hasAlpha ? A : 255;
1862  dest[1] = R >> 22;
1863  dest[2] = G >> 22;
1864  dest[3] = B >> 22;
1865  break;
1866  case AV_PIX_FMT_RGB24:
1867  dest[0] = R >> 22;
1868  dest[1] = G >> 22;
1869  dest[2] = B >> 22;
1870  break;
1871  case AV_PIX_FMT_RGBA:
1872  dest[0] = R >> 22;
1873  dest[1] = G >> 22;
1874  dest[2] = B >> 22;
1875  dest[3] = hasAlpha ? A : 255;
1876  break;
1877  case AV_PIX_FMT_ABGR:
1878  dest[0] = hasAlpha ? A : 255;
1879  dest[1] = B >> 22;
1880  dest[2] = G >> 22;
1881  dest[3] = R >> 22;
1882  break;
1883  case AV_PIX_FMT_BGR24:
1884  dest[0] = B >> 22;
1885  dest[1] = G >> 22;
1886  dest[2] = R >> 22;
1887  break;
1888  case AV_PIX_FMT_BGRA:
1889  dest[0] = B >> 22;
1890  dest[1] = G >> 22;
1891  dest[2] = R >> 22;
1892  dest[3] = hasAlpha ? A : 255;
1893  break;
1894  case AV_PIX_FMT_BGR4_BYTE:
1895  case AV_PIX_FMT_RGB4_BYTE:
1896  case AV_PIX_FMT_BGR8:
1897  case AV_PIX_FMT_RGB8:
1898  {
1899  int r,g,b;
1900 
1901  switch (c->dither) {
1902  default:
1903  case SWS_DITHER_AUTO:
1904  case SWS_DITHER_ED:
1905  R >>= 22;
1906  G >>= 22;
1907  B >>= 22;
1908  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;
1909  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;
1910  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;
1911  c->dither_error[0][i] = err[0];
1912  c->dither_error[1][i] = err[1];
1913  c->dither_error[2][i] = err[2];
1914  r = R >> (isrgb8 ? 5 : 7);
1915  g = G >> (isrgb8 ? 5 : 6);
1916  b = B >> (isrgb8 ? 6 : 7);
1917  r = av_clip(r, 0, isrgb8 ? 7 : 1);
1918  g = av_clip(g, 0, isrgb8 ? 7 : 3);
1919  b = av_clip(b, 0, isrgb8 ? 3 : 1);
1920  err[0] = R - r*(isrgb8 ? 36 : 255);
1921  err[1] = G - g*(isrgb8 ? 36 : 85);
1922  err[2] = B - b*(isrgb8 ? 85 : 255);
1923  break;
1924  case SWS_DITHER_A_DITHER:
1925  if (isrgb8) {
1926  /* see http://pippin.gimp.org/a_dither/ for details/origin */
1927 #define A_DITHER(u,v) (((((u)+((v)*236))*119)&0xff))
1928  r = (((R >> 19) + A_DITHER(i,y) -96)>>8);
1929  g = (((G >> 19) + A_DITHER(i + 17,y) - 96)>>8);
1930  b = (((B >> 20) + A_DITHER(i + 17*2,y) -96)>>8);
1931  r = av_clip_uintp2(r, 3);
1932  g = av_clip_uintp2(g, 3);
1933  b = av_clip_uintp2(b, 2);
1934  } else {
1935  r = (((R >> 21) + A_DITHER(i,y)-256)>>8);
1936  g = (((G >> 19) + A_DITHER(i + 17,y)-256)>>8);
1937  b = (((B >> 21) + A_DITHER(i + 17*2,y)-256)>>8);
1938  r = av_clip_uintp2(r, 1);
1939  g = av_clip_uintp2(g, 2);
1940  b = av_clip_uintp2(b, 1);
1941  }
1942  break;
1943  case SWS_DITHER_X_DITHER:
1944  if (isrgb8) {
1945  /* see http://pippin.gimp.org/a_dither/ for details/origin */
1946 #define X_DITHER(u,v) (((((u)^((v)*237))*181)&0x1ff)/2)
1947  r = (((R >> 19) + X_DITHER(i,y) - 96)>>8);
1948  g = (((G >> 19) + X_DITHER(i + 17,y) - 96)>>8);
1949  b = (((B >> 20) + X_DITHER(i + 17*2,y) - 96)>>8);
1950  r = av_clip_uintp2(r, 3);
1951  g = av_clip_uintp2(g, 3);
1952  b = av_clip_uintp2(b, 2);
1953  } else {
1954  r = (((R >> 21) + X_DITHER(i,y)-256)>>8);
1955  g = (((G >> 19) + X_DITHER(i + 17,y)-256)>>8);
1956  b = (((B >> 21) + X_DITHER(i + 17*2,y)-256)>>8);
1957  r = av_clip_uintp2(r, 1);
1958  g = av_clip_uintp2(g, 2);
1959  b = av_clip_uintp2(b, 1);
1960  }
1961 
1962  break;
1963  }
1964 
1965  if(target == AV_PIX_FMT_BGR4_BYTE) {
1966  dest[0] = r + 2*g + 8*b;
1967  } else if(target == AV_PIX_FMT_RGB4_BYTE) {
1968  dest[0] = b + 2*g + 8*r;
1969  } else if(target == AV_PIX_FMT_BGR8) {
1970  dest[0] = r + 8*g + 64*b;
1971  } else if(target == AV_PIX_FMT_RGB8) {
1972  dest[0] = b + 4*g + 32*r;
1973  } else
1974  av_assert2(0);
1975  break;}
1976  }
1977 }
1978 
1979 static av_always_inline void
1980 yuv2rgb_full_X_c_template(SwsContext *c, const int16_t *lumFilter,
1981  const int16_t **lumSrc, int lumFilterSize,
1982  const int16_t *chrFilter, const int16_t **chrUSrc,
1983  const int16_t **chrVSrc, int chrFilterSize,
1984  const int16_t **alpSrc, uint8_t *dest,
1985  int dstW, int y, enum AVPixelFormat target, int hasAlpha)
1986 {
1987  int i;
1988  int step = (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) ? 3 : 4;
1989  int err[4] = {0};
1990  int A = 0; //init to silence warning
1991 
1992  if( target == AV_PIX_FMT_BGR4_BYTE || target == AV_PIX_FMT_RGB4_BYTE
1993  || target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8)
1994  step = 1;
1995 
1996  for (i = 0; i < dstW; i++) {
1997  int j;
1998  int Y = 1<<9;
1999  int U = (1<<9)-(128 << 19);
2000  int V = (1<<9)-(128 << 19);
2001 
2002  for (j = 0; j < lumFilterSize; j++) {
2003  Y += lumSrc[j][i] * lumFilter[j];
2004  }
2005  for (j = 0; j < chrFilterSize; j++) {
2006  U += chrUSrc[j][i] * chrFilter[j];
2007  V += chrVSrc[j][i] * chrFilter[j];
2008  }
2009  Y >>= 10;
2010  U >>= 10;
2011  V >>= 10;
2012  if (hasAlpha) {
2013  A = 1 << 18;
2014  for (j = 0; j < lumFilterSize; j++) {
2015  A += alpSrc[j][i] * lumFilter[j];
2016  }
2017  A >>= 19;
2018  if (A & 0x100)
2019  A = av_clip_uint8(A);
2020  }
2021  yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
2022  dest += step;
2023  }
2024  c->dither_error[0][i] = err[0];
2025  c->dither_error[1][i] = err[1];
2026  c->dither_error[2][i] = err[2];
2027 }
2028 
2029 static av_always_inline void
2030 yuv2rgb_full_2_c_template(SwsContext *c, const int16_t *buf[2],
2031  const int16_t *ubuf[2], const int16_t *vbuf[2],
2032  const int16_t *abuf[2], uint8_t *dest, int dstW,
2033  int yalpha, int uvalpha, int y,
2034  enum AVPixelFormat target, int hasAlpha)
2035 {
2036  const int16_t *buf0 = buf[0], *buf1 = buf[1],
2037  *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
2038  *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
2039  *abuf0 = hasAlpha ? abuf[0] : NULL,
2040  *abuf1 = hasAlpha ? abuf[1] : NULL;
2041  int yalpha1 = 4096 - yalpha;
2042  int uvalpha1 = 4096 - uvalpha;
2043  int i;
2044  int step = (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) ? 3 : 4;
2045  int err[4] = {0};
2046  int A = 0; // init to silcene warning
2047 
2048  av_assert2(yalpha <= 4096U);
2049  av_assert2(uvalpha <= 4096U);
2050 
2051  if( target == AV_PIX_FMT_BGR4_BYTE || target == AV_PIX_FMT_RGB4_BYTE
2052  || target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8)
2053  step = 1;
2054 
2055  for (i = 0; i < dstW; i++) {
2056  int Y = ( buf0[i] * yalpha1 + buf1[i] * yalpha ) >> 10; //FIXME rounding
2057  int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha-(128 << 19)) >> 10;
2058  int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha-(128 << 19)) >> 10;
2059 
2060  if (hasAlpha) {
2061  A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha + (1<<18)) >> 19;
2062  if (A & 0x100)
2063  A = av_clip_uint8(A);
2064  }
2065 
2066  yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
2067  dest += step;
2068  }
2069  c->dither_error[0][i] = err[0];
2070  c->dither_error[1][i] = err[1];
2071  c->dither_error[2][i] = err[2];
2072 }
2073 
2074 static av_always_inline void
2075 yuv2rgb_full_1_c_template(SwsContext *c, const int16_t *buf0,
2076  const int16_t *ubuf[2], const int16_t *vbuf[2],
2077  const int16_t *abuf0, uint8_t *dest, int dstW,
2078  int uvalpha, int y, enum AVPixelFormat target,
2079  int hasAlpha)
2080 {
2081  const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
2082  int i;
2083  int step = (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) ? 3 : 4;
2084  int err[4] = {0};
2085 
2086  if( target == AV_PIX_FMT_BGR4_BYTE || target == AV_PIX_FMT_RGB4_BYTE
2087  || target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8)
2088  step = 1;
2089 
2090  if (uvalpha < 2048) {
2091  int A = 0; //init to silence warning
2092  for (i = 0; i < dstW; i++) {
2093  int Y = buf0[i] * 4;
2094  int U = (ubuf0[i] - (128<<7)) * 4;
2095  int V = (vbuf0[i] - (128<<7)) * 4;
2096 
2097  if (hasAlpha) {
2098  A = (abuf0[i] + 64) >> 7;
2099  if (A & 0x100)
2100  A = av_clip_uint8(A);
2101  }
2102 
2103  yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
2104  dest += step;
2105  }
2106  } else {
2107  const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
2108  int A = 0; //init to silence warning
2109  for (i = 0; i < dstW; i++) {
2110  int Y = buf0[i] * 4;
2111  int U = (ubuf0[i] + ubuf1[i] - (128<<8)) * 2;
2112  int V = (vbuf0[i] + vbuf1[i] - (128<<8)) * 2;
2113 
2114  if (hasAlpha) {
2115  A = (abuf0[i] + 64) >> 7;
2116  if (A & 0x100)
2117  A = av_clip_uint8(A);
2118  }
2119 
2120  yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
2121  dest += step;
2122  }
2123  }
2124 
2125  c->dither_error[0][i] = err[0];
2126  c->dither_error[1][i] = err[1];
2127  c->dither_error[2][i] = err[2];
2128 }
2129 
2130 #if CONFIG_SMALL
2131 YUV2RGBWRAPPER(yuv2, rgb_full, bgra32_full, AV_PIX_FMT_BGRA, CONFIG_SWSCALE_ALPHA && c->needAlpha)
2132 YUV2RGBWRAPPER(yuv2, rgb_full, abgr32_full, AV_PIX_FMT_ABGR, CONFIG_SWSCALE_ALPHA && c->needAlpha)
2133 YUV2RGBWRAPPER(yuv2, rgb_full, rgba32_full, AV_PIX_FMT_RGBA, CONFIG_SWSCALE_ALPHA && c->needAlpha)
2134 YUV2RGBWRAPPER(yuv2, rgb_full, argb32_full, AV_PIX_FMT_ARGB, CONFIG_SWSCALE_ALPHA && c->needAlpha)
2135 #else
2136 #if CONFIG_SWSCALE_ALPHA
2137 YUV2RGBWRAPPER(yuv2, rgb_full, bgra32_full, AV_PIX_FMT_BGRA, 1)
2138 YUV2RGBWRAPPER(yuv2, rgb_full, abgr32_full, AV_PIX_FMT_ABGR, 1)
2139 YUV2RGBWRAPPER(yuv2, rgb_full, rgba32_full, AV_PIX_FMT_RGBA, 1)
2140 YUV2RGBWRAPPER(yuv2, rgb_full, argb32_full, AV_PIX_FMT_ARGB, 1)
2141 #endif
2142 YUV2RGBWRAPPER(yuv2, rgb_full, bgrx32_full, AV_PIX_FMT_BGRA, 0)
2143 YUV2RGBWRAPPER(yuv2, rgb_full, xbgr32_full, AV_PIX_FMT_ABGR, 0)
2144 YUV2RGBWRAPPER(yuv2, rgb_full, rgbx32_full, AV_PIX_FMT_RGBA, 0)
2145 YUV2RGBWRAPPER(yuv2, rgb_full, xrgb32_full, AV_PIX_FMT_ARGB, 0)
2146 #endif
2147 YUV2RGBWRAPPER(yuv2, rgb_full, bgr24_full, AV_PIX_FMT_BGR24, 0)
2148 YUV2RGBWRAPPER(yuv2, rgb_full, rgb24_full, AV_PIX_FMT_RGB24, 0)
2149 
2150 YUV2RGBWRAPPER(yuv2, rgb_full, bgr4_byte_full, AV_PIX_FMT_BGR4_BYTE, 0)
2151 YUV2RGBWRAPPER(yuv2, rgb_full, rgb4_byte_full, AV_PIX_FMT_RGB4_BYTE, 0)
2152 YUV2RGBWRAPPER(yuv2, rgb_full, bgr8_full, AV_PIX_FMT_BGR8, 0)
2153 YUV2RGBWRAPPER(yuv2, rgb_full, rgb8_full, AV_PIX_FMT_RGB8, 0)
2154 
2155 static void
2156 yuv2gbrp_full_X_c(SwsContext *c, const int16_t *lumFilter,
2157  const int16_t **lumSrc, int lumFilterSize,
2158  const int16_t *chrFilter, const int16_t **chrUSrc,
2159  const int16_t **chrVSrc, int chrFilterSize,
2160  const int16_t **alpSrc, uint8_t **dest,
2161  int dstW, int y)
2162 {
2163  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(c->dstFormat);
2164  int i;
2165  int hasAlpha = (desc->flags & AV_PIX_FMT_FLAG_ALPHA) && alpSrc;
2166  uint16_t **dest16 = (uint16_t**)dest;
2167  int SH = 22 + 8 - desc->comp[0].depth;
2168  int A = 0; // init to silence warning
2169 
2170  for (i = 0; i < dstW; i++) {
2171  int j;
2172  int Y = 1 << 9;
2173  int U = (1 << 9) - (128 << 19);
2174  int V = (1 << 9) - (128 << 19);
2175  int R, G, B;
2176 
2177  for (j = 0; j < lumFilterSize; j++)
2178  Y += lumSrc[j][i] * lumFilter[j];
2179 
2180  for (j = 0; j < chrFilterSize; j++) {
2181  U += chrUSrc[j][i] * chrFilter[j];
2182  V += chrVSrc[j][i] * chrFilter[j];
2183  }
2184 
2185  Y >>= 10;
2186  U >>= 10;
2187  V >>= 10;
2188 
2189  if (hasAlpha) {
2190  A = 1 << 18;
2191 
2192  for (j = 0; j < lumFilterSize; j++)
2193  A += alpSrc[j][i] * lumFilter[j];
2194 
2195  if (A & 0xF8000000)
2196  A = av_clip_uintp2(A, 27);
2197  }
2198 
2199  Y -= c->yuv2rgb_y_offset;
2200  Y *= c->yuv2rgb_y_coeff;
2201  Y += 1 << (SH-1);
2202  R = Y + V * c->yuv2rgb_v2r_coeff;
2203  G = Y + V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
2204  B = Y + U * c->yuv2rgb_u2b_coeff;
2205 
2206  if ((R | G | B) & 0xC0000000) {
2207  R = av_clip_uintp2(R, 30);
2208  G = av_clip_uintp2(G, 30);
2209  B = av_clip_uintp2(B, 30);
2210  }
2211 
2212  if (SH != 22) {
2213  dest16[0][i] = G >> SH;
2214  dest16[1][i] = B >> SH;
2215  dest16[2][i] = R >> SH;
2216  if (hasAlpha)
2217  dest16[3][i] = A >> (SH - 3);
2218  } else {
2219  dest[0][i] = G >> 22;
2220  dest[1][i] = B >> 22;
2221  dest[2][i] = R >> 22;
2222  if (hasAlpha)
2223  dest[3][i] = A >> 19;
2224  }
2225  }
2226  if (SH != 22 && (!isBE(c->dstFormat)) != (!HAVE_BIGENDIAN)) {
2227  for (i = 0; i < dstW; i++) {
2228  dest16[0][i] = av_bswap16(dest16[0][i]);
2229  dest16[1][i] = av_bswap16(dest16[1][i]);
2230  dest16[2][i] = av_bswap16(dest16[2][i]);
2231  if (hasAlpha)
2232  dest16[3][i] = av_bswap16(dest16[3][i]);
2233  }
2234  }
2235 }
2236 
2237 static void
2238 yuv2gbrp16_full_X_c(SwsContext *c, const int16_t *lumFilter,
2239  const int16_t **lumSrcx, int lumFilterSize,
2240  const int16_t *chrFilter, const int16_t **chrUSrcx,
2241  const int16_t **chrVSrcx, int chrFilterSize,
2242  const int16_t **alpSrcx, uint8_t **dest,
2243  int dstW, int y)
2244 {
2245  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(c->dstFormat);
2246  int i;
2247  int hasAlpha = (desc->flags & AV_PIX_FMT_FLAG_ALPHA) && alpSrcx;
2248  uint16_t **dest16 = (uint16_t**)dest;
2249  const int32_t **lumSrc = (const int32_t**)lumSrcx;
2250  const int32_t **chrUSrc = (const int32_t**)chrUSrcx;
2251  const int32_t **chrVSrc = (const int32_t**)chrVSrcx;
2252  const int32_t **alpSrc = (const int32_t**)alpSrcx;
2253 
2254  for (i = 0; i < dstW; i++) {
2255  int j;
2256  int Y = -0x40000000;
2257  int U = -(128 << 23);
2258  int V = -(128 << 23);
2259  int R, G, B, A;
2260 
2261  for (j = 0; j < lumFilterSize; j++)
2262  Y += lumSrc[j][i] * (unsigned)lumFilter[j];
2263 
2264  for (j = 0; j < chrFilterSize; j++) {
2265  U += chrUSrc[j][i] * (unsigned)chrFilter[j];
2266  V += chrVSrc[j][i] * (unsigned)chrFilter[j];
2267  }
2268 
2269  Y >>= 14;
2270  Y += 0x10000;
2271  U >>= 14;
2272  V >>= 14;
2273 
2274  if (hasAlpha) {
2275  A = -0x40000000;
2276 
2277  for (j = 0; j < lumFilterSize; j++)
2278  A += alpSrc[j][i] * (unsigned)lumFilter[j];
2279 
2280  A >>= 1;
2281  A += 0x20002000;
2282  }
2283 
2284  Y -= c->yuv2rgb_y_offset;
2285  Y *= c->yuv2rgb_y_coeff;
2286  Y += 1 << 13;
2287  R = V * c->yuv2rgb_v2r_coeff;
2288  G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
2289  B = U * c->yuv2rgb_u2b_coeff;
2290 
2291  R = av_clip_uintp2(Y + R, 30);
2292  G = av_clip_uintp2(Y + G, 30);
2293  B = av_clip_uintp2(Y + B, 30);
2294 
2295  dest16[0][i] = G >> 14;
2296  dest16[1][i] = B >> 14;
2297  dest16[2][i] = R >> 14;
2298  if (hasAlpha)
2299  dest16[3][i] = av_clip_uintp2(A, 30) >> 14;
2300  }
2301  if ((!isBE(c->dstFormat)) != (!HAVE_BIGENDIAN)) {
2302  for (i = 0; i < dstW; i++) {
2303  dest16[0][i] = av_bswap16(dest16[0][i]);
2304  dest16[1][i] = av_bswap16(dest16[1][i]);
2305  dest16[2][i] = av_bswap16(dest16[2][i]);
2306  if (hasAlpha)
2307  dest16[3][i] = av_bswap16(dest16[3][i]);
2308  }
2309  }
2310 }
2311 
2312 static void
2313 yuv2ya8_1_c(SwsContext *c, const int16_t *buf0,
2314  const int16_t *ubuf[2], const int16_t *vbuf[2],
2315  const int16_t *abuf0, uint8_t *dest, int dstW,
2316  int uvalpha, int y)
2317 {
2318  int hasAlpha = !!abuf0;
2319  int i;
2320 
2321  for (i = 0; i < dstW; i++) {
2322  int Y = (buf0[i] + 64) >> 7;
2323  int A;
2324 
2325  Y = av_clip_uint8(Y);
2326 
2327  if (hasAlpha) {
2328  A = (abuf0[i] + 64) >> 7;
2329  if (A & 0x100)
2330  A = av_clip_uint8(A);
2331  }
2332 
2333  dest[i * 2 ] = Y;
2334  dest[i * 2 + 1] = hasAlpha ? A : 255;
2335  }
2336 }
2337 
2338 static void
2339 yuv2ya8_2_c(SwsContext *c, const int16_t *buf[2],
2340  const int16_t *ubuf[2], const int16_t *vbuf[2],
2341  const int16_t *abuf[2], uint8_t *dest, int dstW,
2342  int yalpha, int uvalpha, int y)
2343 {
2344  int hasAlpha = abuf && abuf[0] && abuf[1];
2345  const int16_t *buf0 = buf[0], *buf1 = buf[1],
2346  *abuf0 = hasAlpha ? abuf[0] : NULL,
2347  *abuf1 = hasAlpha ? abuf[1] : NULL;
2348  int yalpha1 = 4096 - yalpha;
2349  int i;
2350 
2351  av_assert2(yalpha <= 4096U);
2352 
2353  for (i = 0; i < dstW; i++) {
2354  int Y = (buf0[i] * yalpha1 + buf1[i] * yalpha) >> 19;
2355  int A;
2356 
2357  Y = av_clip_uint8(Y);
2358 
2359  if (hasAlpha) {
2360  A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha) >> 19;
2361  A = av_clip_uint8(A);
2362  }
2363 
2364  dest[i * 2 ] = Y;
2365  dest[i * 2 + 1] = hasAlpha ? A : 255;
2366  }
2367 }
2368 
2369 static void
2370 yuv2ya8_X_c(SwsContext *c, const int16_t *lumFilter,
2371  const int16_t **lumSrc, int lumFilterSize,
2372  const int16_t *chrFilter, const int16_t **chrUSrc,
2373  const int16_t **chrVSrc, int chrFilterSize,
2374  const int16_t **alpSrc, uint8_t *dest, int dstW, int y)
2375 {
2376  int hasAlpha = !!alpSrc;
2377  int i;
2378 
2379  for (i = 0; i < dstW; i++) {
2380  int j;
2381  int Y = 1 << 18, A = 1 << 18;
2382 
2383  for (j = 0; j < lumFilterSize; j++)
2384  Y += lumSrc[j][i] * lumFilter[j];
2385 
2386  Y >>= 19;
2387  if (Y & 0x100)
2388  Y = av_clip_uint8(Y);
2389 
2390  if (hasAlpha) {
2391  for (j = 0; j < lumFilterSize; j++)
2392  A += alpSrc[j][i] * lumFilter[j];
2393 
2394  A >>= 19;
2395 
2396  if (A & 0x100)
2397  A = av_clip_uint8(A);
2398  }
2399 
2400  dest[2 * i ] = Y;
2401  dest[2 * i + 1] = hasAlpha ? A : 255;
2402  }
2403 }
2404 
2405 static void
2406 yuv2ayuv64le_X_c(SwsContext *c, const int16_t *lumFilter,
2407  const int16_t **_lumSrc, int lumFilterSize,
2408  const int16_t *chrFilter, const int16_t **_chrUSrc,
2409  const int16_t **_chrVSrc, int chrFilterSize,
2410  const int16_t **_alpSrc, uint8_t *dest, int dstW, int y)
2411 {
2412  const int32_t **lumSrc = (const int32_t **) _lumSrc,
2413  **chrUSrc = (const int32_t **) _chrUSrc,
2414  **chrVSrc = (const int32_t **) _chrVSrc,
2415  **alpSrc = (const int32_t **) _alpSrc;
2416  int hasAlpha = !!alpSrc;
2417  int i;
2418 
2419  for (i = 0; i < dstW; i++) {
2420  int Y = 1 << 14, U = 1 << 14;
2421  int V = 1 << 14, A = 1 << 14;
2422  int j;
2423 
2424  Y -= 0x40000000;
2425  U -= 0x40000000;
2426  V -= 0x40000000;
2427  A -= 0x40000000;
2428 
2429  for (j = 0; j < lumFilterSize; j++)
2430  Y += lumSrc[j][i] * (unsigned)lumFilter[j];
2431 
2432  for (j = 0; j < chrFilterSize; j++)
2433  U += chrUSrc[j][i] * (unsigned)chrFilter[j];
2434 
2435  for (j = 0; j < chrFilterSize; j++)
2436  V += chrVSrc[j][i] * (unsigned)chrFilter[j];
2437 
2438  if (hasAlpha)
2439  for (j = 0; j < lumFilterSize; j++)
2440  A += alpSrc[j][i] * (unsigned)lumFilter[j];
2441 
2442  Y = 0x8000 + av_clip_int16(Y >> 15);
2443  U = 0x8000 + av_clip_int16(U >> 15);
2444  V = 0x8000 + av_clip_int16(V >> 15);
2445  A = 0x8000 + av_clip_int16(A >> 15);
2446 
2447  AV_WL16(dest + 8 * i, hasAlpha ? A : 65535);
2448  AV_WL16(dest + 8 * i + 2, Y);
2449  AV_WL16(dest + 8 * i + 4, U);
2450  AV_WL16(dest + 8 * i + 6, V);
2451  }
2452 }
2453 
2454 av_cold void ff_sws_init_output_funcs(SwsContext *c,
2455  yuv2planar1_fn *yuv2plane1,
2456  yuv2planarX_fn *yuv2planeX,
2457  yuv2interleavedX_fn *yuv2nv12cX,
2458  yuv2packed1_fn *yuv2packed1,
2459  yuv2packed2_fn *yuv2packed2,
2460  yuv2packedX_fn *yuv2packedX,
2461  yuv2anyX_fn *yuv2anyX)
2462 {
2463  enum AVPixelFormat dstFormat = c->dstFormat;
2464  const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(dstFormat);
2465 
2466  if (dstFormat == AV_PIX_FMT_P010LE || dstFormat == AV_PIX_FMT_P010BE) {
2467  *yuv2plane1 = isBE(dstFormat) ? yuv2p010l1_BE_c : yuv2p010l1_LE_c;
2468  *yuv2planeX = isBE(dstFormat) ? yuv2p010lX_BE_c : yuv2p010lX_LE_c;
2469  *yuv2nv12cX = yuv2p010cX_c;
2470  } else if (is16BPS(dstFormat)) {
2471  *yuv2planeX = isBE(dstFormat) ? yuv2planeX_16BE_c : yuv2planeX_16LE_c;
2472  *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_16BE_c : yuv2plane1_16LE_c;
2473  if (dstFormat == AV_PIX_FMT_P016LE || dstFormat == AV_PIX_FMT_P016BE) {
2474  *yuv2nv12cX = yuv2p016cX_c;
2475  }
2476  } else if (isNBPS(dstFormat)) {
2477  if (desc->comp[0].depth == 9) {
2478  *yuv2planeX = isBE(dstFormat) ? yuv2planeX_9BE_c : yuv2planeX_9LE_c;
2479  *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_9BE_c : yuv2plane1_9LE_c;
2480  } else if (desc->comp[0].depth == 10) {
2481  *yuv2planeX = isBE(dstFormat) ? yuv2planeX_10BE_c : yuv2planeX_10LE_c;
2482  *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_10BE_c : yuv2plane1_10LE_c;
2483  } else if (desc->comp[0].depth == 12) {
2484  *yuv2planeX = isBE(dstFormat) ? yuv2planeX_12BE_c : yuv2planeX_12LE_c;
2485  *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_12BE_c : yuv2plane1_12LE_c;
2486  } else if (desc->comp[0].depth == 14) {
2487  *yuv2planeX = isBE(dstFormat) ? yuv2planeX_14BE_c : yuv2planeX_14LE_c;
2488  *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_14BE_c : yuv2plane1_14LE_c;
2489  } else
2490  av_assert0(0);
2491  } else if (dstFormat == AV_PIX_FMT_GRAYF32BE) {
2492  *yuv2planeX = yuv2planeX_floatBE_c;
2493  *yuv2plane1 = yuv2plane1_floatBE_c;
2494  } else if (dstFormat == AV_PIX_FMT_GRAYF32LE) {
2495  *yuv2planeX = yuv2planeX_floatLE_c;
2496  *yuv2plane1 = yuv2plane1_floatLE_c;
2497  } else {
2498  *yuv2plane1 = yuv2plane1_8_c;
2499  *yuv2planeX = yuv2planeX_8_c;
2500  if (dstFormat == AV_PIX_FMT_NV12 || dstFormat == AV_PIX_FMT_NV21 ||
2501  dstFormat == AV_PIX_FMT_NV24 || dstFormat == AV_PIX_FMT_NV42)
2502  *yuv2nv12cX = yuv2nv12cX_c;
2503  }
2504 
2505  if(c->flags & SWS_FULL_CHR_H_INT) {
2506  switch (dstFormat) {
2507  case AV_PIX_FMT_RGBA:
2508 #if CONFIG_SMALL
2509  *yuv2packedX = yuv2rgba32_full_X_c;
2510  *yuv2packed2 = yuv2rgba32_full_2_c;
2511  *yuv2packed1 = yuv2rgba32_full_1_c;
2512 #else
2513 #if CONFIG_SWSCALE_ALPHA
2514  if (c->needAlpha) {
2515  *yuv2packedX = yuv2rgba32_full_X_c;
2516  *yuv2packed2 = yuv2rgba32_full_2_c;
2517  *yuv2packed1 = yuv2rgba32_full_1_c;
2518  } else
2519 #endif /* CONFIG_SWSCALE_ALPHA */
2520  {
2521  *yuv2packedX = yuv2rgbx32_full_X_c;
2522  *yuv2packed2 = yuv2rgbx32_full_2_c;
2523  *yuv2packed1 = yuv2rgbx32_full_1_c;
2524  }
2525 #endif /* !CONFIG_SMALL */
2526  break;
2527  case AV_PIX_FMT_ARGB:
2528 #if CONFIG_SMALL
2529  *yuv2packedX = yuv2argb32_full_X_c;
2530  *yuv2packed2 = yuv2argb32_full_2_c;
2531  *yuv2packed1 = yuv2argb32_full_1_c;
2532 #else
2533 #if CONFIG_SWSCALE_ALPHA
2534  if (c->needAlpha) {
2535  *yuv2packedX = yuv2argb32_full_X_c;
2536  *yuv2packed2 = yuv2argb32_full_2_c;
2537  *yuv2packed1 = yuv2argb32_full_1_c;
2538  } else
2539 #endif /* CONFIG_SWSCALE_ALPHA */
2540  {
2541  *yuv2packedX = yuv2xrgb32_full_X_c;
2542  *yuv2packed2 = yuv2xrgb32_full_2_c;
2543  *yuv2packed1 = yuv2xrgb32_full_1_c;
2544  }
2545 #endif /* !CONFIG_SMALL */
2546  break;
2547  case AV_PIX_FMT_BGRA:
2548 #if CONFIG_SMALL
2549  *yuv2packedX = yuv2bgra32_full_X_c;
2550  *yuv2packed2 = yuv2bgra32_full_2_c;
2551  *yuv2packed1 = yuv2bgra32_full_1_c;
2552 #else
2553 #if CONFIG_SWSCALE_ALPHA
2554  if (c->needAlpha) {
2555  *yuv2packedX = yuv2bgra32_full_X_c;
2556  *yuv2packed2 = yuv2bgra32_full_2_c;
2557  *yuv2packed1 = yuv2bgra32_full_1_c;
2558  } else
2559 #endif /* CONFIG_SWSCALE_ALPHA */
2560  {
2561  *yuv2packedX = yuv2bgrx32_full_X_c;
2562  *yuv2packed2 = yuv2bgrx32_full_2_c;
2563  *yuv2packed1 = yuv2bgrx32_full_1_c;
2564  }
2565 #endif /* !CONFIG_SMALL */
2566  break;
2567  case AV_PIX_FMT_ABGR:
2568 #if CONFIG_SMALL
2569  *yuv2packedX = yuv2abgr32_full_X_c;
2570  *yuv2packed2 = yuv2abgr32_full_2_c;
2571  *yuv2packed1 = yuv2abgr32_full_1_c;
2572 #else
2573 #if CONFIG_SWSCALE_ALPHA
2574  if (c->needAlpha) {
2575  *yuv2packedX = yuv2abgr32_full_X_c;
2576  *yuv2packed2 = yuv2abgr32_full_2_c;
2577  *yuv2packed1 = yuv2abgr32_full_1_c;
2578  } else
2579 #endif /* CONFIG_SWSCALE_ALPHA */
2580  {
2581  *yuv2packedX = yuv2xbgr32_full_X_c;
2582  *yuv2packed2 = yuv2xbgr32_full_2_c;
2583  *yuv2packed1 = yuv2xbgr32_full_1_c;
2584  }
2585 #endif /* !CONFIG_SMALL */
2586  break;
2587  case AV_PIX_FMT_RGBA64LE:
2588 #if CONFIG_SWSCALE_ALPHA
2589  if (c->needAlpha) {
2590  *yuv2packedX = yuv2rgba64le_full_X_c;
2591  *yuv2packed2 = yuv2rgba64le_full_2_c;
2592  *yuv2packed1 = yuv2rgba64le_full_1_c;
2593  } else
2594 #endif /* CONFIG_SWSCALE_ALPHA */
2595  {
2596  *yuv2packedX = yuv2rgbx64le_full_X_c;
2597  *yuv2packed2 = yuv2rgbx64le_full_2_c;
2598  *yuv2packed1 = yuv2rgbx64le_full_1_c;
2599  }
2600  break;
2601  case AV_PIX_FMT_RGBA64BE:
2602 #if CONFIG_SWSCALE_ALPHA
2603  if (c->needAlpha) {
2604  *yuv2packedX = yuv2rgba64be_full_X_c;
2605  *yuv2packed2 = yuv2rgba64be_full_2_c;
2606  *yuv2packed1 = yuv2rgba64be_full_1_c;
2607  } else
2608 #endif /* CONFIG_SWSCALE_ALPHA */
2609  {
2610  *yuv2packedX = yuv2rgbx64be_full_X_c;
2611  *yuv2packed2 = yuv2rgbx64be_full_2_c;
2612  *yuv2packed1 = yuv2rgbx64be_full_1_c;
2613  }
2614  break;
2615  case AV_PIX_FMT_BGRA64LE:
2616 #if CONFIG_SWSCALE_ALPHA
2617  if (c->needAlpha) {
2618  *yuv2packedX = yuv2bgra64le_full_X_c;
2619  *yuv2packed2 = yuv2bgra64le_full_2_c;
2620  *yuv2packed1 = yuv2bgra64le_full_1_c;
2621  } else
2622 #endif /* CONFIG_SWSCALE_ALPHA */
2623  {
2624  *yuv2packedX = yuv2bgrx64le_full_X_c;
2625  *yuv2packed2 = yuv2bgrx64le_full_2_c;
2626  *yuv2packed1 = yuv2bgrx64le_full_1_c;
2627  }
2628  break;
2629  case AV_PIX_FMT_BGRA64BE:
2630 #if CONFIG_SWSCALE_ALPHA
2631  if (c->needAlpha) {
2632  *yuv2packedX = yuv2bgra64be_full_X_c;
2633  *yuv2packed2 = yuv2bgra64be_full_2_c;
2634  *yuv2packed1 = yuv2bgra64be_full_1_c;
2635  } else
2636 #endif /* CONFIG_SWSCALE_ALPHA */
2637  {
2638  *yuv2packedX = yuv2bgrx64be_full_X_c;
2639  *yuv2packed2 = yuv2bgrx64be_full_2_c;
2640  *yuv2packed1 = yuv2bgrx64be_full_1_c;
2641  }
2642  break;
2643 
2644  case AV_PIX_FMT_RGB24:
2645  *yuv2packedX = yuv2rgb24_full_X_c;
2646  *yuv2packed2 = yuv2rgb24_full_2_c;
2647  *yuv2packed1 = yuv2rgb24_full_1_c;
2648  break;
2649  case AV_PIX_FMT_BGR24:
2650  *yuv2packedX = yuv2bgr24_full_X_c;
2651  *yuv2packed2 = yuv2bgr24_full_2_c;
2652  *yuv2packed1 = yuv2bgr24_full_1_c;
2653  break;
2654  case AV_PIX_FMT_RGB48LE:
2655  *yuv2packedX = yuv2rgb48le_full_X_c;
2656  *yuv2packed2 = yuv2rgb48le_full_2_c;
2657  *yuv2packed1 = yuv2rgb48le_full_1_c;
2658  break;
2659  case AV_PIX_FMT_BGR48LE:
2660  *yuv2packedX = yuv2bgr48le_full_X_c;
2661  *yuv2packed2 = yuv2bgr48le_full_2_c;
2662  *yuv2packed1 = yuv2bgr48le_full_1_c;
2663  break;
2664  case AV_PIX_FMT_RGB48BE:
2665  *yuv2packedX = yuv2rgb48be_full_X_c;
2666  *yuv2packed2 = yuv2rgb48be_full_2_c;
2667  *yuv2packed1 = yuv2rgb48be_full_1_c;
2668  break;
2669  case AV_PIX_FMT_BGR48BE:
2670  *yuv2packedX = yuv2bgr48be_full_X_c;
2671  *yuv2packed2 = yuv2bgr48be_full_2_c;
2672  *yuv2packed1 = yuv2bgr48be_full_1_c;
2673  break;
2674  case AV_PIX_FMT_BGR4_BYTE:
2675  *yuv2packedX = yuv2bgr4_byte_full_X_c;
2676  *yuv2packed2 = yuv2bgr4_byte_full_2_c;
2677  *yuv2packed1 = yuv2bgr4_byte_full_1_c;
2678  break;
2679  case AV_PIX_FMT_RGB4_BYTE:
2680  *yuv2packedX = yuv2rgb4_byte_full_X_c;
2681  *yuv2packed2 = yuv2rgb4_byte_full_2_c;
2682  *yuv2packed1 = yuv2rgb4_byte_full_1_c;
2683  break;
2684  case AV_PIX_FMT_BGR8:
2685  *yuv2packedX = yuv2bgr8_full_X_c;
2686  *yuv2packed2 = yuv2bgr8_full_2_c;
2687  *yuv2packed1 = yuv2bgr8_full_1_c;
2688  break;
2689  case AV_PIX_FMT_RGB8:
2690  *yuv2packedX = yuv2rgb8_full_X_c;
2691  *yuv2packed2 = yuv2rgb8_full_2_c;
2692  *yuv2packed1 = yuv2rgb8_full_1_c;
2693  break;
2694  case AV_PIX_FMT_GBRP:
2695  case AV_PIX_FMT_GBRP9BE:
2696  case AV_PIX_FMT_GBRP9LE:
2697  case AV_PIX_FMT_GBRP10BE:
2698  case AV_PIX_FMT_GBRP10LE:
2699  case AV_PIX_FMT_GBRP12BE:
2700  case AV_PIX_FMT_GBRP12LE:
2701  case AV_PIX_FMT_GBRP14BE:
2702  case AV_PIX_FMT_GBRP14LE:
2703  case AV_PIX_FMT_GBRAP:
2704  case AV_PIX_FMT_GBRAP10BE:
2705  case AV_PIX_FMT_GBRAP10LE:
2706  case AV_PIX_FMT_GBRAP12BE:
2707  case AV_PIX_FMT_GBRAP12LE:
2708  *yuv2anyX = yuv2gbrp_full_X_c;
2709  break;
2710  case AV_PIX_FMT_GBRP16BE:
2711  case AV_PIX_FMT_GBRP16LE:
2712  case AV_PIX_FMT_GBRAP16BE:
2713  case AV_PIX_FMT_GBRAP16LE:
2714  *yuv2anyX = yuv2gbrp16_full_X_c;
2715  break;
2716  }
2717  if (!*yuv2packedX && !*yuv2anyX)
2718  goto YUV_PACKED;
2719  } else {
2720  YUV_PACKED:
2721  switch (dstFormat) {
2722  case AV_PIX_FMT_RGBA64LE:
2723 #if CONFIG_SWSCALE_ALPHA
2724  if (c->needAlpha) {
2725  *yuv2packed1 = yuv2rgba64le_1_c;
2726  *yuv2packed2 = yuv2rgba64le_2_c;
2727  *yuv2packedX = yuv2rgba64le_X_c;
2728  } else
2729 #endif /* CONFIG_SWSCALE_ALPHA */
2730  {
2731  *yuv2packed1 = yuv2rgbx64le_1_c;
2732  *yuv2packed2 = yuv2rgbx64le_2_c;
2733  *yuv2packedX = yuv2rgbx64le_X_c;
2734  }
2735  break;
2736  case AV_PIX_FMT_RGBA64BE:
2737 #if CONFIG_SWSCALE_ALPHA
2738  if (c->needAlpha) {
2739  *yuv2packed1 = yuv2rgba64be_1_c;
2740  *yuv2packed2 = yuv2rgba64be_2_c;
2741  *yuv2packedX = yuv2rgba64be_X_c;
2742  } else
2743 #endif /* CONFIG_SWSCALE_ALPHA */
2744  {
2745  *yuv2packed1 = yuv2rgbx64be_1_c;
2746  *yuv2packed2 = yuv2rgbx64be_2_c;
2747  *yuv2packedX = yuv2rgbx64be_X_c;
2748  }
2749  break;
2750  case AV_PIX_FMT_BGRA64LE:
2751 #if CONFIG_SWSCALE_ALPHA
2752  if (c->needAlpha) {
2753  *yuv2packed1 = yuv2bgra64le_1_c;
2754  *yuv2packed2 = yuv2bgra64le_2_c;
2755  *yuv2packedX = yuv2bgra64le_X_c;
2756  } else
2757 #endif /* CONFIG_SWSCALE_ALPHA */
2758  {
2759  *yuv2packed1 = yuv2bgrx64le_1_c;
2760  *yuv2packed2 = yuv2bgrx64le_2_c;
2761  *yuv2packedX = yuv2bgrx64le_X_c;
2762  }
2763  break;
2764  case AV_PIX_FMT_BGRA64BE:
2765 #if CONFIG_SWSCALE_ALPHA
2766  if (c->needAlpha) {
2767  *yuv2packed1 = yuv2bgra64be_1_c;
2768  *yuv2packed2 = yuv2bgra64be_2_c;
2769  *yuv2packedX = yuv2bgra64be_X_c;
2770  } else
2771 #endif /* CONFIG_SWSCALE_ALPHA */
2772  {
2773  *yuv2packed1 = yuv2bgrx64be_1_c;
2774  *yuv2packed2 = yuv2bgrx64be_2_c;
2775  *yuv2packedX = yuv2bgrx64be_X_c;
2776  }
2777  break;
2778  case AV_PIX_FMT_RGB48LE:
2779  *yuv2packed1 = yuv2rgb48le_1_c;
2780  *yuv2packed2 = yuv2rgb48le_2_c;
2781  *yuv2packedX = yuv2rgb48le_X_c;
2782  break;
2783  case AV_PIX_FMT_RGB48BE:
2784  *yuv2packed1 = yuv2rgb48be_1_c;
2785  *yuv2packed2 = yuv2rgb48be_2_c;
2786  *yuv2packedX = yuv2rgb48be_X_c;
2787  break;
2788  case AV_PIX_FMT_BGR48LE:
2789  *yuv2packed1 = yuv2bgr48le_1_c;
2790  *yuv2packed2 = yuv2bgr48le_2_c;
2791  *yuv2packedX = yuv2bgr48le_X_c;
2792  break;
2793  case AV_PIX_FMT_BGR48BE:
2794  *yuv2packed1 = yuv2bgr48be_1_c;
2795  *yuv2packed2 = yuv2bgr48be_2_c;
2796  *yuv2packedX = yuv2bgr48be_X_c;
2797  break;
2798  case AV_PIX_FMT_RGB32:
2799  case AV_PIX_FMT_BGR32:
2800 #if CONFIG_SMALL
2801  *yuv2packed1 = yuv2rgb32_1_c;
2802  *yuv2packed2 = yuv2rgb32_2_c;
2803  *yuv2packedX = yuv2rgb32_X_c;
2804 #else
2805 #if CONFIG_SWSCALE_ALPHA
2806  if (c->needAlpha) {
2807  *yuv2packed1 = yuv2rgba32_1_c;
2808  *yuv2packed2 = yuv2rgba32_2_c;
2809  *yuv2packedX = yuv2rgba32_X_c;
2810  } else
2811 #endif /* CONFIG_SWSCALE_ALPHA */
2812  {
2813  *yuv2packed1 = yuv2rgbx32_1_c;
2814  *yuv2packed2 = yuv2rgbx32_2_c;
2815  *yuv2packedX = yuv2rgbx32_X_c;
2816  }
2817 #endif /* !CONFIG_SMALL */
2818  break;
2819  case AV_PIX_FMT_RGB32_1:
2820  case AV_PIX_FMT_BGR32_1:
2821 #if CONFIG_SMALL
2822  *yuv2packed1 = yuv2rgb32_1_1_c;
2823  *yuv2packed2 = yuv2rgb32_1_2_c;
2824  *yuv2packedX = yuv2rgb32_1_X_c;
2825 #else
2826 #if CONFIG_SWSCALE_ALPHA
2827  if (c->needAlpha) {
2828  *yuv2packed1 = yuv2rgba32_1_1_c;
2829  *yuv2packed2 = yuv2rgba32_1_2_c;
2830  *yuv2packedX = yuv2rgba32_1_X_c;
2831  } else
2832 #endif /* CONFIG_SWSCALE_ALPHA */
2833  {
2834  *yuv2packed1 = yuv2rgbx32_1_1_c;
2835  *yuv2packed2 = yuv2rgbx32_1_2_c;
2836  *yuv2packedX = yuv2rgbx32_1_X_c;
2837  }
2838 #endif /* !CONFIG_SMALL */
2839  break;
2840  case AV_PIX_FMT_RGB24:
2841  *yuv2packed1 = yuv2rgb24_1_c;
2842  *yuv2packed2 = yuv2rgb24_2_c;
2843  *yuv2packedX = yuv2rgb24_X_c;
2844  break;
2845  case AV_PIX_FMT_BGR24:
2846  *yuv2packed1 = yuv2bgr24_1_c;
2847  *yuv2packed2 = yuv2bgr24_2_c;
2848  *yuv2packedX = yuv2bgr24_X_c;
2849  break;
2850  case AV_PIX_FMT_RGB565LE:
2851  case AV_PIX_FMT_RGB565BE:
2852  case AV_PIX_FMT_BGR565LE:
2853  case AV_PIX_FMT_BGR565BE:
2854  *yuv2packed1 = yuv2rgb16_1_c;
2855  *yuv2packed2 = yuv2rgb16_2_c;
2856  *yuv2packedX = yuv2rgb16_X_c;
2857  break;
2858  case AV_PIX_FMT_RGB555LE:
2859  case AV_PIX_FMT_RGB555BE:
2860  case AV_PIX_FMT_BGR555LE:
2861  case AV_PIX_FMT_BGR555BE:
2862  *yuv2packed1 = yuv2rgb15_1_c;
2863  *yuv2packed2 = yuv2rgb15_2_c;
2864  *yuv2packedX = yuv2rgb15_X_c;
2865  break;
2866  case AV_PIX_FMT_RGB444LE:
2867  case AV_PIX_FMT_RGB444BE:
2868  case AV_PIX_FMT_BGR444LE:
2869  case AV_PIX_FMT_BGR444BE:
2870  *yuv2packed1 = yuv2rgb12_1_c;
2871  *yuv2packed2 = yuv2rgb12_2_c;
2872  *yuv2packedX = yuv2rgb12_X_c;
2873  break;
2874  case AV_PIX_FMT_RGB8:
2875  case AV_PIX_FMT_BGR8:
2876  *yuv2packed1 = yuv2rgb8_1_c;
2877  *yuv2packed2 = yuv2rgb8_2_c;
2878  *yuv2packedX = yuv2rgb8_X_c;
2879  break;
2880  case AV_PIX_FMT_RGB4:
2881  case AV_PIX_FMT_BGR4:
2882  *yuv2packed1 = yuv2rgb4_1_c;
2883  *yuv2packed2 = yuv2rgb4_2_c;
2884  *yuv2packedX = yuv2rgb4_X_c;
2885  break;
2886  case AV_PIX_FMT_RGB4_BYTE:
2887  case AV_PIX_FMT_BGR4_BYTE:
2888  *yuv2packed1 = yuv2rgb4b_1_c;
2889  *yuv2packed2 = yuv2rgb4b_2_c;
2890  *yuv2packedX = yuv2rgb4b_X_c;
2891  break;
2892  }
2893  }
2894  switch (dstFormat) {
2895  case AV_PIX_FMT_MONOWHITE:
2896  *yuv2packed1 = yuv2monowhite_1_c;
2897  *yuv2packed2 = yuv2monowhite_2_c;
2898  *yuv2packedX = yuv2monowhite_X_c;
2899  break;
2900  case AV_PIX_FMT_MONOBLACK:
2901  *yuv2packed1 = yuv2monoblack_1_c;
2902  *yuv2packed2 = yuv2monoblack_2_c;
2903  *yuv2packedX = yuv2monoblack_X_c;
2904  break;
2905  case AV_PIX_FMT_YUYV422:
2906  *yuv2packed1 = yuv2yuyv422_1_c;
2907  *yuv2packed2 = yuv2yuyv422_2_c;
2908  *yuv2packedX = yuv2yuyv422_X_c;
2909  break;
2910  case AV_PIX_FMT_YVYU422:
2911  *yuv2packed1 = yuv2yvyu422_1_c;
2912  *yuv2packed2 = yuv2yvyu422_2_c;
2913  *yuv2packedX = yuv2yvyu422_X_c;
2914  break;
2915  case AV_PIX_FMT_UYVY422:
2916  *yuv2packed1 = yuv2uyvy422_1_c;
2917  *yuv2packed2 = yuv2uyvy422_2_c;
2918  *yuv2packedX = yuv2uyvy422_X_c;
2919  break;
2920  case AV_PIX_FMT_YA8:
2921  *yuv2packed1 = yuv2ya8_1_c;
2922  *yuv2packed2 = yuv2ya8_2_c;
2923  *yuv2packedX = yuv2ya8_X_c;
2924  break;
2925  case AV_PIX_FMT_YA16LE:
2926  *yuv2packed1 = yuv2ya16le_1_c;
2927  *yuv2packed2 = yuv2ya16le_2_c;
2928  *yuv2packedX = yuv2ya16le_X_c;
2929  break;
2930  case AV_PIX_FMT_YA16BE:
2931  *yuv2packed1 = yuv2ya16be_1_c;
2932  *yuv2packed2 = yuv2ya16be_2_c;
2933  *yuv2packedX = yuv2ya16be_X_c;
2934  break;
2935  case AV_PIX_FMT_AYUV64LE:
2936  *yuv2packedX = yuv2ayuv64le_X_c;
2937  break;
2938  }
2939 }
yuv2packed2_fn
void(* yuv2packed2_fn)(struct SwsContext *c, const int16_t *lumSrc[2], const int16_t *chrUSrc[2], const int16_t *chrVSrc[2], const int16_t *alpSrc[2], uint8_t *dest, int dstW, int yalpha, int uvalpha, int y)
Write one line of horizontally scaled Y/U/V/A to packed-pixel YUV/RGB output by doing bilinear scalin...
Definition: swscale_internal.h:202
yuv2planar1_fn
void(* yuv2planar1_fn)(const int16_t *src, uint8_t *dest, int dstW, const uint8_t *dither, int offset)
Write one line of horizontally scaled data to planar output without any additional vertical scaling (...
Definition: swscale_internal.h:98
yuv2packed1_fn
void(* yuv2packed1_fn)(struct SwsContext *c, const int16_t *lumSrc, const int16_t *chrUSrc[2], const int16_t *chrVSrc[2], const int16_t *alpSrc, uint8_t *dest, int dstW, int uvalpha, int y)
Write one line of horizontally scaled Y/U/V/A to packed-pixel YUV/RGB output without any additional v...
Definition: swscale_internal.h:169
YUV2PACKEDWRAPPER
#define YUV2PACKEDWRAPPER(name, base, ext, fmt)
Definition: output.c:709
AVPixelFormat
AVPixelFormat
Pixel format.
Definition: pixfmt.h:64
AV_PIX_FMT_BGR48LE
@ AV_PIX_FMT_BGR48LE
packed RGB 16:16:16, 48bpp, 16B, 16G, 16R, the 2-byte value for each R/G/B component is stored as lit...
Definition: pixfmt.h:149
ff_dither_4x4_16
const uint8_t ff_dither_4x4_16[][8]
Definition: output.c:51
X_DITHER
#define X_DITHER(u, v)
r
const char * r
Definition: vf_curves.c:114
acc
int acc
Definition: yuv2rgb.c:555
AV_PIX_FMT_YA8
@ AV_PIX_FMT_YA8
8 bits gray, 8 bits alpha
Definition: pixfmt.h:143
AV_PIX_FMT_BGRA64BE
@ AV_PIX_FMT_BGRA64BE
packed RGBA 16:16:16:16, 64bpp, 16B, 16G, 16R, 16A, the 2-byte value for each R/G/B/A component is st...
Definition: pixfmt.h:207
yuv2rgb_X_c_template
static av_always_inline void yuv2rgb_X_c_template(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize, const int16_t *chrFilter, const int16_t **chrUSrc, const int16_t **chrVSrc, int chrFilterSize, const int16_t **alpSrc, uint8_t *dest, int dstW, int y, enum AVPixelFormat target, int hasAlpha)
Definition: output.c:1636
AV_PIX_FMT_BGR32
#define AV_PIX_FMT_BGR32
Definition: pixfmt.h:362
AV_PIX_FMT_RGB444LE
@ AV_PIX_FMT_RGB444LE
packed RGB 4:4:4, 16bpp, (msb)4X 4R 4G 4B(lsb), little-endian, X=unused/undefined
Definition: pixfmt.h:139
u
#define u(width, name, range_min, range_max)
Definition: cbs_h2645.c:252
AV_PIX_FMT_GBRP16BE
@ AV_PIX_FMT_GBRP16BE
planar GBR 4:4:4 48bpp, big-endian
Definition: pixfmt.h:174
AV_PIX_FMT_GBRP10BE
@ AV_PIX_FMT_GBRP10BE
planar GBR 4:4:4 30bpp, big-endian
Definition: pixfmt.h:172
av_pix_fmt_desc_get
const AVPixFmtDescriptor * av_pix_fmt_desc_get(enum AVPixelFormat pix_fmt)
Definition: pixdesc.c:2522
ff_dither_8x8_32
const uint8_t ff_dither_8x8_32[][8]
Definition: output.c:59
yuv2rgb_full_1_c_template
static av_always_inline void yuv2rgb_full_1_c_template(SwsContext *c, const int16_t *buf0, const int16_t *ubuf[2], const int16_t *vbuf[2], const int16_t *abuf0, uint8_t *dest, int dstW, int uvalpha, int y, enum AVPixelFormat target, int hasAlpha)
Definition: output.c:2075
yuv2rgba64_full_X_c_template
static av_always_inline void yuv2rgba64_full_X_c_template(SwsContext *c, const int16_t *lumFilter, const int32_t **lumSrc, int lumFilterSize, const int16_t *chrFilter, const int32_t **chrUSrc, const int32_t **chrVSrc, int chrFilterSize, const int32_t **alpSrc, uint16_t *dest, int dstW, int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
Definition: output.c:1238
SWS_DITHER_A_DITHER
@ SWS_DITHER_A_DITHER
Definition: swscale_internal.h:70
accumulate_bit
#define accumulate_bit(acc, val)
Definition: output.c:534
pixdesc.h
AV_PIX_FMT_RGBA64BE
@ AV_PIX_FMT_RGBA64BE
packed RGBA 16:16:16:16, 64bpp, 16R, 16G, 16B, 16A, the 2-byte value for each R/G/B/A component is st...
Definition: pixfmt.h:205
step
trying all byte sequences megabyte in length and selecting the best looking sequence will yield cases to try But a word about which is also called distortion Distortion can be quantified by almost any quality measurement one chooses the sum of squared differences is used but more complex methods that consider psychovisual effects can be used as well It makes no difference in this discussion First step
Definition: rate_distortion.txt:58
R
#define R
Definition: huffyuvdsp.h:34
b
#define b
Definition: input.c:41
yuv2planeX
static void FUNC() yuv2planeX(const int16_t *filter, int filterSize, const int16_t **src, uint8_t *dest, int dstW, const uint8_t *dither, int offset)
Definition: swscale_ppc_template.c:81
AV_PIX_FMT_MONOWHITE
@ AV_PIX_FMT_MONOWHITE
Y , 1bpp, 0 is white, 1 is black, in each byte pixels are ordered from the msb to the lsb.
Definition: pixfmt.h:75
AV_PIX_FMT_P010BE
@ AV_PIX_FMT_P010BE
like NV12, with 10bpp per component, data in the high bits, zeros in the low bits,...
Definition: pixfmt.h:285
AV_PIX_FMT_RGB32_1
#define AV_PIX_FMT_RGB32_1
Definition: pixfmt.h:361
yuv2rgba64_X_c_template
static av_always_inline void yuv2rgba64_X_c_template(SwsContext *c, const int16_t *lumFilter, const int32_t **lumSrc, int lumFilterSize, const int16_t *chrFilter, const int32_t **chrUSrc, const int32_t **chrVSrc, int chrFilterSize, const int32_t **alpSrc, uint16_t *dest, int dstW, int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
Definition: output.c:988
b_r
#define b_r
AV_PIX_FMT_GBRP14BE
@ AV_PIX_FMT_GBRP14BE
planar GBR 4:4:4 42bpp, big-endian
Definition: pixfmt.h:256
AV_PIX_FMT_BGR24
@ AV_PIX_FMT_BGR24
packed RGB 8:8:8, 24bpp, BGRBGR...
Definition: pixfmt.h:69
AV_PIX_FMT_BGRA
@ AV_PIX_FMT_BGRA
packed BGRA 8:8:8:8, 32bpp, BGRABGRA...
Definition: pixfmt.h:95
mathematics.h
yuv2rgb_full_X_c_template
static av_always_inline void yuv2rgb_full_X_c_template(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize, const int16_t *chrFilter, const int16_t **chrUSrc, const int16_t **chrVSrc, int chrFilterSize, const int16_t **alpSrc, uint8_t *dest, int dstW, int y, enum AVPixelFormat target, int hasAlpha)
Definition: output.c:1980
filter
filter_frame For filters that do not use the this method is called when a frame is pushed to the filter s input It can be called at any time except in a reentrant way If the input frame is enough to produce then the filter should push the output frames on the output link immediately As an exception to the previous rule if the input frame is enough to produce several output frames then the filter needs output only at least one per link The additional frames can be left buffered in the filter
Definition: filter_design.txt:228
yuv2422_2_c_template
static av_always_inline void yuv2422_2_c_template(SwsContext *c, const int16_t *buf[2], const int16_t *ubuf[2], const int16_t *vbuf[2], const int16_t *abuf[2], uint8_t *dest, int dstW, int yalpha, int uvalpha, int y, enum AVPixelFormat target)
Definition: output.c:802
av_float2int
static av_always_inline uint32_t av_float2int(float f)
Reinterpret a float as a 32-bit integer.
Definition: intfloat.h:50
yuv2plane1_8_c
static void yuv2plane1_8_c(const int16_t *src, uint8_t *dest, int dstW, const uint8_t *dither, int offset)
Definition: output.c:395
AV_PIX_FMT_GRAYF32LE
@ AV_PIX_FMT_GRAYF32LE
IEEE-754 single precision Y, 32bpp, little-endian.
Definition: pixfmt.h:341
yuv2planeX_10_c_template
static av_always_inline void yuv2planeX_10_c_template(const int16_t *filter, int filterSize, const int16_t **src, uint16_t *dest, int dstW, int big_endian, int output_bits)
Definition: output.c:333
AV_PIX_FMT_RGB555BE
@ AV_PIX_FMT_RGB555BE
packed RGB 5:5:5, 16bpp, (msb)1X 5R 5G 5B(lsb), big-endian , X=unused/undefined
Definition: pixfmt.h:107
yuv2p016cX_c
static void yuv2p016cX_c(SwsContext *c, const int16_t *chrFilter, int chrFilterSize, const int16_t **chrUSrc, const int16_t **chrVSrc, uint8_t *dest8, int chrDstW)
Definition: output.c:183
AV_PIX_FMT_AYUV64LE
@ AV_PIX_FMT_AYUV64LE
packed AYUV 4:4:4,64bpp (1 Cr & Cb sample per 1x1 Y & A samples), little-endian
Definition: pixfmt.h:279
SH
#define SH(val, pdst)
Definition: generic_macros_msa.h:156
AV_PIX_FMT_GBRAP12LE
@ AV_PIX_FMT_GBRAP12LE
planar GBR 4:4:4:4 48bpp, little-endian
Definition: pixfmt.h:288
A
#define A(x)
Definition: vp56_arith.h:28
is16BPS
static av_always_inline int is16BPS(enum AVPixelFormat pix_fmt)
Definition: swscale_internal.h:646
AV_PIX_FMT_GBRAP
@ AV_PIX_FMT_GBRAP
planar GBRA 4:4:4:4 32bpp
Definition: pixfmt.h:215
yuv2anyX_fn
void(* yuv2anyX_fn)(struct SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize, const int16_t *chrFilter, const int16_t **chrUSrc, const int16_t **chrVSrc, int chrFilterSize, const int16_t **alpSrc, uint8_t **dest, int dstW, int y)
Write one line of horizontally scaled Y/U/V/A to YUV/RGB output by doing multi-point vertical scaling...
Definition: swscale_internal.h:268
yuv2422_X_c_template
static av_always_inline void yuv2422_X_c_template(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize, const int16_t *chrFilter, const int16_t **chrUSrc, const int16_t **chrVSrc, int chrFilterSize, const int16_t **alpSrc, uint8_t *dest, int dstW, int y, enum AVPixelFormat target)
Definition: output.c:763
U
#define U(x)
Definition: vp56_arith.h:37
yuv2mono_1_c_template
static av_always_inline void yuv2mono_1_c_template(SwsContext *c, const int16_t *buf0, const int16_t *ubuf[2], const int16_t *vbuf[2], const int16_t *abuf0, uint8_t *dest, int dstW, int uvalpha, int y, enum AVPixelFormat target)
Definition: output.c:659
yuv2plane1_16_c_template
static av_always_inline void yuv2plane1_16_c_template(const int32_t *src, uint16_t *dest, int dstW, int big_endian, int output_bits)
Definition: output.c:144
yuv2422_1_c_template
static av_always_inline void yuv2422_1_c_template(SwsContext *c, const int16_t *buf0, const int16_t *ubuf[2], const int16_t *vbuf[2], const int16_t *abuf0, uint8_t *dest, int dstW, int uvalpha, int y, enum AVPixelFormat target)
Definition: output.c:835
yuv2gbrp_full_X_c
static void yuv2gbrp_full_X_c(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize, const int16_t *chrFilter, const int16_t **chrUSrc, const int16_t **chrVSrc, int chrFilterSize, const int16_t **alpSrc, uint8_t **dest, int dstW, int y)
Definition: output.c:2156
isNBPS
static av_always_inline int isNBPS(enum AVPixelFormat pix_fmt)
Definition: swscale_internal.h:653
R_B
#define R_B
Definition: output.c:885
src
#define src
Definition: vp8dsp.c:254
yuv2plane1_float
#define yuv2plane1_float(template, dest_type, BE_LE)
Definition: output.c:283
av_bswap32
#define av_bswap32
Definition: bswap.h:33
yuv2planeX_16_c_template
static av_always_inline void yuv2planeX_16_c_template(const int16_t *filter, int filterSize, const int32_t **src, uint16_t *dest, int dstW, int big_endian, int output_bits)
Definition: output.c:158
r_b
#define r_b
AV_PIX_FMT_BGR8
@ AV_PIX_FMT_BGR8
packed RGB 3:3:2, 8bpp, (msb)2B 3G 3R(lsb)
Definition: pixfmt.h:83
avassert.h
buf
void * buf
Definition: avisynth_c.h:766
av_cold
#define av_cold
Definition: attributes.h:84
yuv2mono_2_c_template
static av_always_inline void yuv2mono_2_c_template(SwsContext *c, const int16_t *buf[2], const int16_t *ubuf[2], const int16_t *vbuf[2], const int16_t *abuf[2], uint8_t *dest, int dstW, int yalpha, int uvalpha, int y, enum AVPixelFormat target)
Definition: output.c:598
yuv2planeX_float
#define yuv2planeX_float(template, dest_type, BE_LE)
Definition: output.c:290
YUVRGB_TABLE_HEADROOM
#define YUVRGB_TABLE_HEADROOM
Definition: swscale_internal.h:38
SWS_DITHER_ED
@ SWS_DITHER_ED
Definition: swscale_internal.h:69
AV_PIX_FMT_GBRAP16BE
@ AV_PIX_FMT_GBRAP16BE
planar GBRA 4:4:4:4 64bpp, big-endian
Definition: pixfmt.h:216
yuv2rgb_full_2_c_template
static av_always_inline void yuv2rgb_full_2_c_template(SwsContext *c, const int16_t *buf[2], const int16_t *ubuf[2], const int16_t *vbuf[2], const int16_t *abuf[2], uint8_t *dest, int dstW, int yalpha, int uvalpha, int y, enum AVPixelFormat target, int hasAlpha)
Definition: output.c:2030
intreadwrite.h
AV_PIX_FMT_GBRP16LE
@ AV_PIX_FMT_GBRP16LE
planar GBR 4:4:4 48bpp, little-endian
Definition: pixfmt.h:175
output_pixels
#define output_pixels(pos, Y1, U, Y2, V)
Definition: output.c:744
g
const char * g
Definition: vf_curves.c:115
yuv2p010l1_LE_c
static void yuv2p010l1_LE_c(const int16_t *src, uint8_t *dest, int dstW, const uint8_t *dither, int offset)
Definition: output.c:503
AV_PIX_FMT_P016BE
@ AV_PIX_FMT_P016BE
like NV12, with 16bpp per component, big-endian
Definition: pixfmt.h:301
AV_PIX_FMT_GBRP12LE
@ AV_PIX_FMT_GBRP12LE
planar GBR 4:4:4 36bpp, little-endian
Definition: pixfmt.h:255
av_assert0
#define av_assert0(cond)
assert() equivalent, that is always enabled.
Definition: avassert.h:37
yuv2plane1_float_bswap_c_template
static av_always_inline void yuv2plane1_float_bswap_c_template(const int32_t *src, uint32_t *dest, int dstW)
Definition: output.c:228
ff_dither_2x2_4
const uint8_t ff_dither_2x2_4[][8]
Definition: output.c:39
ff_dither_8x8_220
const uint8_t ff_dither_8x8_220[][8]
Definition: output.c:84
AV_PIX_FMT_FLAG_ALPHA
#define AV_PIX_FMT_FLAG_ALPHA
The pixel format has an alpha channel.
Definition: pixdesc.h:177
yuv2p010cX_c
static void yuv2p010cX_c(SwsContext *c, const int16_t *chrFilter, int chrFilterSize, const int16_t **chrUSrc, const int16_t **chrVSrc, uint8_t *dest8, int chrDstW)
Definition: output.c:480
yuv2ya8_X_c
static void yuv2ya8_X_c(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize, const int16_t *chrFilter, const int16_t **chrUSrc, const int16_t **chrVSrc, int chrFilterSize, const int16_t **alpSrc, uint8_t *dest, int dstW, int y)
Definition: output.c:2370
YUV2PACKED16WRAPPER
#define YUV2PACKED16WRAPPER(name, base, ext, fmt, hasAlpha, eightbytes)
Definition: output.c:1434
yuv2mono_X_c_template
static av_always_inline void yuv2mono_X_c_template(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize, const int16_t *chrFilter, const int16_t **chrUSrc, const int16_t **chrVSrc, int chrFilterSize, const int16_t **alpSrc, uint8_t *dest, int dstW, int y, enum AVPixelFormat target)
Definition: output.c:545
AV_PIX_FMT_RGB4
@ AV_PIX_FMT_RGB4
packed RGB 1:2:1 bitstream, 4bpp, (msb)1R 2G 1B(lsb), a byte contains two pixels, the first pixel in ...
Definition: pixfmt.h:87
AV_PIX_FMT_GBRP10LE
@ AV_PIX_FMT_GBRP10LE
planar GBR 4:4:4 30bpp, little-endian
Definition: pixfmt.h:173
AV_PIX_FMT_BGR32_1
#define AV_PIX_FMT_BGR32_1
Definition: pixfmt.h:363
yuv2rgba64_full_2_c_template
static av_always_inline void yuv2rgba64_full_2_c_template(SwsContext *c, const int32_t *buf[2], const int32_t *ubuf[2], const int32_t *vbuf[2], const int32_t *abuf[2], uint16_t *dest, int dstW, int yalpha, int uvalpha, int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
Definition: output.c:1302
AV_PIX_FMT_RGBA
@ AV_PIX_FMT_RGBA
packed RGBA 8:8:8:8, 32bpp, RGBARGBA...
Definition: pixfmt.h:93
int32_t
int32_t
Definition: audio_convert.c:194
if
if(ret)
Definition: filter_design.txt:179
yuv2NBPS
#define yuv2NBPS(bits, BE_LE, is_be, template_size, typeX_t)
Definition: output.c:353
AV_PIX_FMT_RGB565LE
@ AV_PIX_FMT_RGB565LE
packed RGB 5:6:5, 16bpp, (msb) 5R 6G 5B(lsb), little-endian
Definition: pixfmt.h:106
AV_PIX_FMT_GBRAP12BE
@ AV_PIX_FMT_GBRAP12BE
planar GBR 4:4:4:4 48bpp, big-endian
Definition: pixfmt.h:287
NULL
#define NULL
Definition: coverity.c:32
yuv2rgba64_full_1_c_template
static av_always_inline void yuv2rgba64_full_1_c_template(SwsContext *c, const int32_t *buf0, const int32_t *ubuf[2], const int32_t *vbuf[2], const int32_t *abuf0, uint16_t *dest, int dstW, int uvalpha, int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
Definition: output.c:1354
AV_PIX_FMT_YUYV422
@ AV_PIX_FMT_YUYV422
packed YUV 4:2:2, 16bpp, Y0 Cb Y1 Cr
Definition: pixfmt.h:67
AV_PIX_FMT_RGB48LE
@ AV_PIX_FMT_RGB48LE
packed RGB 16:16:16, 48bpp, 16R, 16G, 16B, the 2-byte value for each R/G/B component is stored as lit...
Definition: pixfmt.h:103
AV_PIX_FMT_YA16LE
@ AV_PIX_FMT_YA16LE
16 bits gray, 16 bits alpha (little-endian)
Definition: pixfmt.h:213
yuv2gbrp16_full_X_c
static void yuv2gbrp16_full_X_c(SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrcx, int lumFilterSize, const int16_t *chrFilter, const int16_t **chrUSrcx, const int16_t **chrVSrcx, int chrFilterSize, const int16_t **alpSrcx, uint8_t **dest, int dstW, int y)
Definition: output.c:2238
AV_PIX_FMT_MONOBLACK
@ AV_PIX_FMT_MONOBLACK
Y , 1bpp, 0 is black, 1 is white, in each byte pixels are ordered from the msb to the lsb.
Definition: pixfmt.h:76
V
#define V
Definition: avdct.c:30
AV_PIX_FMT_BGR565LE
@ AV_PIX_FMT_BGR565LE
packed BGR 5:6:5, 16bpp, (msb) 5B 6G 5R(lsb), little-endian
Definition: pixfmt.h:111
AV_PIX_FMT_RGBA64LE
@ AV_PIX_FMT_RGBA64LE
packed RGBA 16:16:16:16, 64bpp, 16R, 16G, 16B, 16A, the 2-byte value for each R/G/B/A component is st...
Definition: pixfmt.h:206
AV_PIX_FMT_RGB8
@ AV_PIX_FMT_RGB8
packed RGB 3:3:2, 8bpp, (msb)2R 3G 3B(lsb)
Definition: pixfmt.h:86
yuv2ya8_2_c
static void yuv2ya8_2_c(SwsContext *c, const int16_t *buf[2], const int16_t *ubuf[2], const int16_t *vbuf[2], const int16_t *abuf[2], uint8_t *dest, int dstW, int yalpha, int uvalpha, int y)
Definition: output.c:2339
AV_PIX_FMT_BGR4
@ AV_PIX_FMT_BGR4
packed RGB 1:2:1 bitstream, 4bpp, (msb)1B 2G 1R(lsb), a byte contains two pixels, the first pixel in ...
Definition: pixfmt.h:84
AV_PIX_FMT_BGR555BE
@ AV_PIX_FMT_BGR555BE
packed BGR 5:5:5, 16bpp, (msb)1X 5B 5G 5R(lsb), big-endian , X=unused/undefined
Definition: pixfmt.h:112
yuv2plane1_float_c_template
static av_always_inline void yuv2plane1_float_c_template(const int32_t *src, float *dest, int dstW)
Definition: output.c:212
A2
#define A2
Definition: binkdsp.c:32
AV_PIX_FMT_ABGR
@ AV_PIX_FMT_ABGR
packed ABGR 8:8:8:8, 32bpp, ABGRABGR...
Definition: pixfmt.h:94
yuv2p010lX_c
static void yuv2p010lX_c(const int16_t *filter, int filterSize, const int16_t **src, uint16_t *dest, int dstW, int big_endian)
Definition: output.c:463
c
Undefined Behavior In the C some operations are like signed integer dereferencing freed accessing outside allocated Undefined Behavior must not occur in a C it is not safe even if the output of undefined operations is unused The unsafety may seem nit picking but Optimizing compilers have in fact optimized code on the assumption that no undefined Behavior occurs Optimizing code based on wrong assumptions can and has in some cases lead to effects beyond the output of computations The signed integer overflow problem in speed critical code Code which is highly optimized and works with signed integers sometimes has the problem that often the output of the computation does not c
Definition: undefined.txt:32
AV_PIX_FMT_BGR4_BYTE
@ AV_PIX_FMT_BGR4_BYTE
packed RGB 1:2:1, 8bpp, (msb)1B 2G 1R(lsb)
Definition: pixfmt.h:85
A_DITHER
#define A_DITHER(u, v)
d64
const uint8_t * d64
Definition: yuv2rgb.c:503
isBE
static av_always_inline int isBE(enum AVPixelFormat pix_fmt)
Definition: swscale_internal.h:660
desc
const char * desc
Definition: nvenc.c:68
AV_PIX_FMT_RGB24
@ AV_PIX_FMT_RGB24
packed RGB 8:8:8, 24bpp, RGBRGB...
Definition: pixfmt.h:68
cpu.h
AV_PIX_FMT_RGB444BE
@ AV_PIX_FMT_RGB444BE
packed RGB 4:4:4, 16bpp, (msb)4X 4R 4G 4B(lsb), big-endian, X=unused/undefined
Definition: pixfmt.h:140
SWS_FULL_CHR_H_INT
#define SWS_FULL_CHR_H_INT
Definition: swscale.h:79
yuv2planeX_float_bswap_c_template
static av_always_inline void yuv2planeX_float_bswap_c_template(const int16_t *filter, int filterSize, const int32_t **src, uint32_t *dest, int dstW)
Definition: output.c:264
AV_PIX_FMT_YA16BE
@ AV_PIX_FMT_YA16BE
16 bits gray, 16 bits alpha (big-endian)
Definition: pixfmt.h:212
AV_PIX_FMT_BGR555
#define AV_PIX_FMT_BGR555
Definition: pixfmt.h:380
AV_PIX_FMT_GBRP9BE
@ AV_PIX_FMT_GBRP9BE
planar GBR 4:4:4 27bpp, big-endian
Definition: pixfmt.h:170
AV_PIX_FMT_BGR444BE
@ AV_PIX_FMT_BGR444BE
packed BGR 4:4:4, 16bpp, (msb)4X 4B 4G 4R(lsb), big-endian, X=unused/undefined
Definition: pixfmt.h:142
val
const char const char void * val
Definition: avisynth_c.h:863
output_pixel
#define output_pixel(pos, val, bias, signedness)
Definition: output.c:887
AV_PIX_FMT_GBRP9LE
@ AV_PIX_FMT_GBRP9LE
planar GBR 4:4:4 27bpp, little-endian
Definition: pixfmt.h:171
yuv2ya16_2_c_template
static av_always_inline void yuv2ya16_2_c_template(SwsContext *c, const int32_t *buf[2], const int32_t *unused_ubuf[2], const int32_t *unused_vbuf[2], const int32_t *abuf[2], uint16_t *dest, int dstW, int yalpha, int unused_uvalpha, int y, enum AVPixelFormat target, int unused_hasAlpha, int unused_eightbytes)
Definition: output.c:930
AV_WL16
#define AV_WL16(p, v)
Definition: intreadwrite.h:412
AV_PIX_FMT_RGB32
#define AV_PIX_FMT_RGB32
Definition: pixfmt.h:360
AV_PIX_FMT_GBRAP10LE
@ AV_PIX_FMT_GBRAP10LE
planar GBR 4:4:4:4 40bpp, little-endian
Definition: pixfmt.h:291
AV_PIX_FMT_BGR565BE
@ AV_PIX_FMT_BGR565BE
packed BGR 5:6:5, 16bpp, (msb) 5B 6G 5R(lsb), big-endian
Definition: pixfmt.h:110
av_bswap16
#define av_bswap16
Definition: bswap.h:31
offset
it s the only field you need to keep assuming you have a context There is some magic you don t need to care about around this just let it vf offset
Definition: writing_filters.txt:86
attributes.h
yuv2rgb_write_full
static av_always_inline void yuv2rgb_write_full(SwsContext *c, uint8_t *dest, int i, int Y, int A, int U, int V, int y, enum AVPixelFormat target, int hasAlpha, int err[4])
Definition: output.c:1840
ff_dither_8x8_73
const uint8_t ff_dither_8x8_73[][8]
Definition: output.c:71
Y
#define Y
Definition: boxblur.h:38
yuv2rgb_write
static av_always_inline void yuv2rgb_write(uint8_t *_dest, int i, int Y1, int Y2, unsigned A1, unsigned A2, const void *_r, const void *_g, const void *_b, int y, enum AVPixelFormat target, int hasAlpha)
Definition: output.c:1517
AV_PIX_FMT_ARGB
@ AV_PIX_FMT_ARGB
packed ARGB 8:8:8:8, 32bpp, ARGBARGB...
Definition: pixfmt.h:92
AV_PIX_FMT_BGRA64LE
@ AV_PIX_FMT_BGRA64LE
packed RGBA 16:16:16:16, 64bpp, 16B, 16G, 16R, 16A, the 2-byte value for each R/G/B/A component is st...
Definition: pixfmt.h:208
DECLARE_ALIGNED
#define DECLARE_ALIGNED(n, t, v)
Definition: mem.h:112
av_assert2
#define av_assert2(cond)
assert() equivalent, that does lie in speed critical code.
Definition: avassert.h:64
AV_PIX_FMT_RGB555LE
@ AV_PIX_FMT_RGB555LE
packed RGB 5:5:5, 16bpp, (msb)1X 5R 5G 5B(lsb), little-endian, X=unused/undefined
Definition: pixfmt.h:108
AV_PIX_FMT_RGB48BE
@ AV_PIX_FMT_RGB48BE
packed RGB 16:16:16, 48bpp, 16R, 16G, 16B, the 2-byte value for each R/G/B component is stored as big...
Definition: pixfmt.h:102
i
#define i(width, name, range_min, range_max)
Definition: cbs_h2645.c:259
yuv2p010l1_c
static void yuv2p010l1_c(const int16_t *src, uint16_t *dest, int dstW, int big_endian)
Definition: output.c:450
yuv2ya16_1_c_template
static av_always_inline void yuv2ya16_1_c_template(SwsContext *c, const int32_t *buf0, const int32_t *unused_ubuf[2], const int32_t *unused_vbuf[2], const int32_t *abuf0, uint16_t *dest, int dstW, int unused_uvalpha, int y, enum AVPixelFormat target, int unused_hasAlpha, int unused_eightbytes)
Definition: output.c:962
AV_PIX_FMT_NV24
@ AV_PIX_FMT_NV24
planar YUV 4:4:4, 24bpp, 1 plane for Y and 1 plane for the UV components, which are interleaved (firs...
Definition: pixfmt.h:348
AV_PIX_FMT_BGR444
#define AV_PIX_FMT_BGR444
Definition: pixfmt.h:381
AV_PIX_FMT_RGB555
#define AV_PIX_FMT_RGB555
Definition: pixfmt.h:375
yuv2planeX_float_c_template
static av_always_inline void yuv2planeX_float_c_template(const int16_t *filter, int filterSize, const int32_t **src, float *dest, int dstW)
Definition: output.c:244
av_always_inline
#define av_always_inline
Definition: attributes.h:43
swscale_internal.h
uint8_t
uint8_t
Definition: audio_convert.c:194
yuv2nv12cX_c
static void yuv2nv12cX_c(SwsContext *c, const int16_t *chrFilter, int chrFilterSize, const int16_t **chrUSrc, const int16_t **chrVSrc, uint8_t *dest, int chrDstW)
Definition: output.c:405
AV_PIX_FMT_NV21
@ AV_PIX_FMT_NV21
as above, but U and V bytes are swapped
Definition: pixfmt.h:90
AV_PIX_FMT_BGR565
#define AV_PIX_FMT_BGR565
Definition: pixfmt.h:379
AV_PIX_FMT_RGB4_BYTE
@ AV_PIX_FMT_RGB4_BYTE
packed RGB 1:2:1, 8bpp, (msb)1R 2G 1B(lsb)
Definition: pixfmt.h:88
AV_PIX_FMT_NV42
@ AV_PIX_FMT_NV42
as above, but U and V bytes are swapped
Definition: pixfmt.h:349
AV_PIX_FMT_RGB565
#define AV_PIX_FMT_RGB565
Definition: pixfmt.h:374
AV_PIX_FMT_GBRAP16LE
@ AV_PIX_FMT_GBRAP16LE
planar GBRA 4:4:4:4 64bpp, little-endian
Definition: pixfmt.h:217
AV_PIX_FMT_YVYU422
@ AV_PIX_FMT_YVYU422
packed YUV 4:2:2, 16bpp, Y0 Cr Y1 Cb
Definition: pixfmt.h:210
ff_sws_init_output_funcs
av_cold void ff_sws_init_output_funcs(SwsContext *c, yuv2planar1_fn *yuv2plane1, yuv2planarX_fn *yuv2planeX, yuv2interleavedX_fn *yuv2nv12cX, yuv2packed1_fn *yuv2packed1, yuv2packed2_fn *yuv2packed2, yuv2packedX_fn *yuv2packedX, yuv2anyX_fn *yuv2anyX)
Definition: output.c:2454
G
#define G
Definition: huffyuvdsp.h:33
bswap.h
AV_PIX_FMT_NV12
@ AV_PIX_FMT_NV12
planar YUV 4:2:0, 12bpp, 1 plane for Y and 1 plane for the UV components, which are interleaved (firs...
Definition: pixfmt.h:89
AV_PIX_FMT_P016LE
@ AV_PIX_FMT_P016LE
like NV12, with 16bpp per component, little-endian
Definition: pixfmt.h:300
yuv2p010lX_LE_c
static void yuv2p010lX_LE_c(const int16_t *filter, int filterSize, const int16_t **src, uint8_t *dest, int dstW, const uint8_t *dither, int offset)
Definition: output.c:517
AV_PIX_FMT_GRAYF32BE
@ AV_PIX_FMT_GRAYF32BE
IEEE-754 single precision Y, 32bpp, big-endian.
Definition: pixfmt.h:340
yuv2ya8_1_c
static void yuv2ya8_1_c(SwsContext *c, const int16_t *buf0, const int16_t *ubuf[2], const int16_t *vbuf[2], const int16_t *abuf0, uint8_t *dest, int dstW, int uvalpha, int y)
Definition: output.c:2313
YUV2RGBWRAPPER
#define YUV2RGBWRAPPER(name, base, ext, fmt, hasAlpha)
Definition: output.c:1809
d128
const uint8_t * d128
Definition: yuv2rgb.c:554
AV_PIX_FMT_GBRP12BE
@ AV_PIX_FMT_GBRP12BE
planar GBR 4:4:4 36bpp, big-endian
Definition: pixfmt.h:254
AV_PIX_FMT_UYVY422
@ AV_PIX_FMT_UYVY422
packed YUV 4:2:2, 16bpp, Cb Y0 Cr Y1
Definition: pixfmt.h:81
yuv2planarX_fn
void(* yuv2planarX_fn)(const int16_t *filter, int filterSize, const int16_t **src, uint8_t *dest, int dstW, const uint8_t *dither, int offset)
Write one line of horizontally scaled data to planar output with multi-point vertical scaling between...
Definition: swscale_internal.h:114
B
#define B
Definition: huffyuvdsp.h:32
yuv2interleavedX_fn
void(* yuv2interleavedX_fn)(struct SwsContext *c, const int16_t *chrFilter, int chrFilterSize, const int16_t **chrUSrc, const int16_t **chrVSrc, uint8_t *dest, int dstW)
Write one line of horizontally scaled chroma to interleaved output with multi-point vertical scaling ...
Definition: swscale_internal.h:133
config.h
yuv2packedX_fn
void(* yuv2packedX_fn)(struct SwsContext *c, const int16_t *lumFilter, const int16_t **lumSrc, int lumFilterSize, const int16_t *chrFilter, const int16_t **chrUSrc, const int16_t **chrVSrc, int chrFilterSize, const int16_t **alpSrc, uint8_t *dest, int dstW, int y)
Write one line of horizontally scaled Y/U/V/A to packed-pixel YUV/RGB output by doing multi-point ver...
Definition: swscale_internal.h:234
yuv2p010lX_BE_c
static void yuv2p010lX_BE_c(const int16_t *filter, int filterSize, const int16_t **src, uint8_t *dest, int dstW, const uint8_t *dither, int offset)
Definition: output.c:524
yuv2ya16_X_c_template
static av_always_inline void yuv2ya16_X_c_template(SwsContext *c, const int16_t *lumFilter, const int32_t **lumSrc, int lumFilterSize, const int16_t *chrFilter, const int32_t **unused_chrUSrc, const int32_t **unused_chrVSrc, int unused_chrFilterSize, const int32_t **alpSrc, uint16_t *dest, int dstW, int y, enum AVPixelFormat target, int unused_hasAlpha, int unused_eightbytes)
Definition: output.c:895
AV_PIX_FMT_RGB565BE
@ AV_PIX_FMT_RGB565BE
packed RGB 5:6:5, 16bpp, (msb) 5R 6G 5B(lsb), big-endian
Definition: pixfmt.h:105
shift
static int shift(int a, int b)
Definition: sonic.c:82
AV_PIX_FMT_GBRP
@ AV_PIX_FMT_GBRP
planar GBR 4:4:4 24bpp
Definition: pixfmt.h:168
d32
const uint8_t * d32
Definition: yuv2rgb.c:502
avutil.h
AV_PIX_FMT_P010LE
@ AV_PIX_FMT_P010LE
like NV12, with 10bpp per component, data in the high bits, zeros in the low bits,...
Definition: pixfmt.h:284
AVPixFmtDescriptor
Descriptor that unambiguously describes how the bits of a pixel are stored in the up to 4 data planes...
Definition: pixdesc.h:81
AV_PIX_FMT_BGR555LE
@ AV_PIX_FMT_BGR555LE
packed BGR 5:5:5, 16bpp, (msb)1X 5B 5G 5R(lsb), little-endian, X=unused/undefined
Definition: pixfmt.h:113
yuv2rgb_1_c_template
static av_always_inline void yuv2rgb_1_c_template(SwsContext *c, const int16_t *buf0, const int16_t *ubuf[2], const int16_t *vbuf[2], const int16_t *abuf0, uint8_t *dest, int dstW, int uvalpha, int y, enum AVPixelFormat target, int hasAlpha)
Definition: output.c:1730
SWS_DITHER_AUTO
@ SWS_DITHER_AUTO
Definition: swscale_internal.h:67
B_R
#define B_R
Definition: output.c:886
AV_PIX_FMT_GBRP14LE
@ AV_PIX_FMT_GBRP14LE
planar GBR 4:4:4 42bpp, little-endian
Definition: pixfmt.h:257
yuv2rgb_2_c_template
static av_always_inline void yuv2rgb_2_c_template(SwsContext *c, const int16_t *buf[2], const int16_t *ubuf[2], const int16_t *vbuf[2], const int16_t *abuf[2], uint8_t *dest, int dstW, int yalpha, int uvalpha, int y, enum AVPixelFormat target, int hasAlpha)
Definition: output.c:1690
A1
#define A1
Definition: binkdsp.c:31
AV_PIX_FMT_GBRAP10BE
@ AV_PIX_FMT_GBRAP10BE
planar GBR 4:4:4:4 40bpp, big-endian
Definition: pixfmt.h:290
yuv2p010l1_BE_c
static void yuv2p010l1_BE_c(const int16_t *src, uint8_t *dest, int dstW, const uint8_t *dither, int offset)
Definition: output.c:510
yuv2rgba64_1_c_template
static av_always_inline void yuv2rgba64_1_c_template(SwsContext *c, const int32_t *buf0, const int32_t *ubuf[2], const int32_t *vbuf[2], const int32_t *abuf0, uint16_t *dest, int dstW, int uvalpha, int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
Definition: output.c:1135
yuv2rgba64_2_c_template
static av_always_inline void yuv2rgba64_2_c_template(SwsContext *c, const int32_t *buf[2], const int32_t *ubuf[2], const int32_t *vbuf[2], const int32_t *abuf[2], uint16_t *dest, int dstW, int yalpha, int uvalpha, int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
Definition: output.c:1070
SWS_DITHER_X_DITHER
@ SWS_DITHER_X_DITHER
Definition: swscale_internal.h:71
SwsContext
Definition: swscale_internal.h:280
AV_PIX_FMT_BGR444LE
@ AV_PIX_FMT_BGR444LE
packed BGR 4:4:4, 16bpp, (msb)4X 4B 4G 4R(lsb), little-endian, X=unused/undefined
Definition: pixfmt.h:141
yuv2rgb
static void yuv2rgb(uint8_t *out, int ridx, int Y, int U, int V)
Definition: g2meet.c:280
rgb2rgb.h
swscale.h
yuv2ayuv64le_X_c
static void yuv2ayuv64le_X_c(SwsContext *c, const int16_t *lumFilter, const int16_t **_lumSrc, int lumFilterSize, const int16_t *chrFilter, const int16_t **_chrUSrc, const int16_t **_chrVSrc, int chrFilterSize, const int16_t **_alpSrc, uint8_t *dest, int dstW, int y)
Definition: output.c:2406
ff_dither_2x2_8
const uint8_t ff_dither_2x2_8[][8]
Definition: output.c:45
AV_PIX_FMT_BGR48BE
@ AV_PIX_FMT_BGR48BE
packed RGB 16:16:16, 48bpp, 16B, 16G, 16R, the 2-byte value for each R/G/B component is stored as big...
Definition: pixfmt.h:148
AV_PIX_FMT_RGB444
#define AV_PIX_FMT_RGB444
Definition: pixfmt.h:376
dither
static const uint8_t dither[8][8]
Definition: vf_fspp.c:57
Generated on Wed Aug 24 2022 21:27:30 for FFmpeg by   doxygen 1.8.17