Statistics
| Branch: | Revision:

ffmpeg / libswscale / yuv2rgb.c @ fc69cc61

History | View | Annotate | Download (24.8 KB)

1
/*
2
 * yuv2rgb.c, Software YUV to RGB converter
3
 *
4
 *  Copyright (C) 1999, Aaron Holtzman <aholtzma@ess.engr.uvic.ca>
5
 *
6
 *  Functions broken out from display_x11.c and several new modes
7
 *  added by HÃ¥kan Hjort <d95hjort@dtek.chalmers.se>
8
 *
9
 *  15 & 16 bpp support by Franck Sicard <Franck.Sicard@solsoft.fr>
10
 *
11
 *  MMX/MMX2 template stuff (needed for fast movntq support),
12
 *  1,4,8bpp support and context / deglobalize stuff
13
 *  by Michael Niedermayer (michaelni@gmx.at)
14
 *
15
 *  This file is part of mpeg2dec, a free MPEG-2 video decoder
16
 *
17
 *  mpeg2dec is free software; you can redistribute it and/or modify
18
 *  it under the terms of the GNU General Public License as published by
19
 *  the Free Software Foundation; either version 2, or (at your option)
20
 *  any later version.
21
 *
22
 *  mpeg2dec is distributed in the hope that it will be useful,
23
 *  but WITHOUT ANY WARRANTY; without even the implied warranty of
24
 *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
25
 *  GNU General Public License for more details.
26
 *
27
 *  You should have received a copy of the GNU General Public License
28
 *  along with mpeg2dec; if not, write to the Free Software
29
 *  Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
30
 */
31

    
32
#include <stdio.h>
33
#include <stdlib.h>
34
#include <inttypes.h>
35
#include <assert.h>
36

    
37
#include "config.h"
38
#include "rgb2rgb.h"
39
#include "swscale.h"
40
#include "swscale_internal.h"
41

    
42
#ifdef HAVE_MLIB
43
#include "yuv2rgb_mlib.c"
44
#endif
45

    
46
#define DITHER1XBPP // only for mmx
47

    
48
const uint8_t  __attribute__((aligned(8))) dither_2x2_4[2][8]={
49
{  1,   3,   1,   3,   1,   3,   1,   3, },
50
{  2,   0,   2,   0,   2,   0,   2,   0, },
51
};
52

    
53
const uint8_t  __attribute__((aligned(8))) dither_2x2_8[2][8]={
54
{  6,   2,   6,   2,   6,   2,   6,   2, },
55
{  0,   4,   0,   4,   0,   4,   0,   4, },
56
};
57

    
58
const uint8_t  __attribute__((aligned(8))) dither_8x8_32[8][8]={
59
{ 17,   9,  23,  15,  16,   8,  22,  14, },
60
{  5,  29,   3,  27,   4,  28,   2,  26, },
61
{ 21,  13,  19,  11,  20,  12,  18,  10, },
62
{  0,  24,   6,  30,   1,  25,   7,  31, },
63
{ 16,   8,  22,  14,  17,   9,  23,  15, },
64
{  4,  28,   2,  26,   5,  29,   3,  27, },
65
{ 20,  12,  18,  10,  21,  13,  19,  11, },
66
{  1,  25,   7,  31,   0,  24,   6,  30, },
67
};
68

    
69
#if 0
70
const uint8_t  __attribute__((aligned(8))) dither_8x8_64[8][8]={
71
{  0,  48,  12,  60,   3,  51,  15,  63, },
72
{ 32,  16,  44,  28,  35,  19,  47,  31, },
73
{  8,  56,   4,  52,  11,  59,   7,  55, },
74
{ 40,  24,  36,  20,  43,  27,  39,  23, },
75
{  2,  50,  14,  62,   1,  49,  13,  61, },
76
{ 34,  18,  46,  30,  33,  17,  45,  29, },
77
{ 10,  58,   6,  54,   9,  57,   5,  53, },
78
{ 42,  26,  38,  22,  41,  25,  37,  21, },
79
};
80
#endif
81

    
82
const uint8_t  __attribute__((aligned(8))) dither_8x8_73[8][8]={
83
{  0,  55,  14,  68,   3,  58,  17,  72, },
84
{ 37,  18,  50,  32,  40,  22,  54,  35, },
85
{  9,  64,   5,  59,  13,  67,   8,  63, },
86
{ 46,  27,  41,  23,  49,  31,  44,  26, },
87
{  2,  57,  16,  71,   1,  56,  15,  70, },
88
{ 39,  21,  52,  34,  38,  19,  51,  33, },
89
{ 11,  66,   7,  62,  10,  65,   6,  60, },
90
{ 48,  30,  43,  25,  47,  29,  42,  24, },
91
};
92

    
93
#if 0
94
const uint8_t  __attribute__((aligned(8))) dither_8x8_128[8][8]={
95
{ 68,  36,  92,  60,  66,  34,  90,  58, },
96
{ 20, 116,  12, 108,  18, 114,  10, 106, },
97
{ 84,  52,  76,  44,  82,  50,  74,  42, },
98
{  0,  96,  24, 120,   6, 102,  30, 126, },
99
{ 64,  32,  88,  56,  70,  38,  94,  62, },
100
{ 16, 112,   8, 104,  22, 118,  14, 110, },
101
{ 80,  48,  72,  40,  86,  54,  78,  46, },
102
{  4, 100,  28, 124,   2,  98,  26, 122, },
103
};
104
#endif
105

    
106
#if 1
107
const uint8_t  __attribute__((aligned(8))) dither_8x8_220[8][8]={
108
{117,  62, 158, 103, 113,  58, 155, 100, },
109
{ 34, 199,  21, 186,  31, 196,  17, 182, },
110
{144,  89, 131,  76, 141,  86, 127,  72, },
111
{  0, 165,  41, 206,  10, 175,  52, 217, },
112
{110,  55, 151,  96, 120,  65, 162, 107, },
113
{ 28, 193,  14, 179,  38, 203,  24, 189, },
114
{138,  83, 124,  69, 148,  93, 134,  79, },
115
{  7, 172,  48, 213,   3, 168,  45, 210, },
116
};
117
#elif 1
118
// tries to correct a gamma of 1.5
119
const uint8_t  __attribute__((aligned(8))) dither_8x8_220[8][8]={
120
{  0, 143,  18, 200,   2, 156,  25, 215, },
121
{ 78,  28, 125,  64,  89,  36, 138,  74, },
122
{ 10, 180,   3, 161,  16, 195,   8, 175, },
123
{109,  51,  93,  38, 121,  60, 105,  47, },
124
{  1, 152,  23, 210,   0, 147,  20, 205, },
125
{ 85,  33, 134,  71,  81,  30, 130,  67, },
126
{ 14, 190,   6, 171,  12, 185,   5, 166, },
127
{117,  57, 101,  44, 113,  54,  97,  41, },
128
};
129
#elif 1
130
// tries to correct a gamma of 2.0
131
const uint8_t  __attribute__((aligned(8))) dither_8x8_220[8][8]={
132
{  0, 124,   8, 193,   0, 140,  12, 213, },
133
{ 55,  14, 104,  42,  66,  19, 119,  52, },
134
{  3, 168,   1, 145,   6, 187,   3, 162, },
135
{ 86,  31,  70,  21,  99,  39,  82,  28, },
136
{  0, 134,  11, 206,   0, 129,   9, 200, },
137
{ 62,  17, 114,  48,  58,  16, 109,  45, },
138
{  5, 181,   2, 157,   4, 175,   1, 151, },
139
{ 95,  36,  78,  26,  90,  34,  74,  24, },
140
};
141
#else
142
// tries to correct a gamma of 2.5
143
const uint8_t  __attribute__((aligned(8))) dither_8x8_220[8][8]={
144
{  0, 107,   3, 187,   0, 125,   6, 212, },
145
{ 39,   7,  86,  28,  49,  11, 102,  36, },
146
{  1, 158,   0, 131,   3, 180,   1, 151, },
147
{ 68,  19,  52,  12,  81,  25,  64,  17, },
148
{  0, 119,   5, 203,   0, 113,   4, 195, },
149
{ 45,   9,  96,  33,  42,   8,  91,  30, },
150
{  2, 172,   1, 144,   2, 165,   0, 137, },
151
{ 77,  23,  60,  15,  72,  21,  56,  14, },
152
};
153
#endif
154

    
155
#ifdef HAVE_MMX
156

    
157
/* hope these constant values are cache line aligned */
158
static uint64_t attribute_used __attribute__((aligned(8))) mmx_00ffw   = 0x00ff00ff00ff00ffULL;
159
static uint64_t attribute_used __attribute__((aligned(8))) mmx_redmask = 0xf8f8f8f8f8f8f8f8ULL;
160
static uint64_t attribute_used __attribute__((aligned(8))) mmx_grnmask = 0xfcfcfcfcfcfcfcfcULL;
161

    
162
static uint64_t attribute_used __attribute__((aligned(8))) M24A=   0x00FF0000FF0000FFULL;
163
static uint64_t attribute_used __attribute__((aligned(8))) M24B=   0xFF0000FF0000FF00ULL;
164
static uint64_t attribute_used __attribute__((aligned(8))) M24C=   0x0000FF0000FF0000ULL;
165

    
166
// the volatile is required because gcc otherwise optimizes some writes away not knowing that these
167
// are read in the asm block
168
static volatile uint64_t attribute_used __attribute__((aligned(8))) b5Dither;
169
static volatile uint64_t attribute_used __attribute__((aligned(8))) g5Dither;
170
static volatile uint64_t attribute_used __attribute__((aligned(8))) g6Dither;
171
static volatile uint64_t attribute_used __attribute__((aligned(8))) r5Dither;
172

    
173
static uint64_t __attribute__((aligned(8))) dither4[2]={
174
    0x0103010301030103LL,
175
    0x0200020002000200LL,};
176

    
177
static uint64_t __attribute__((aligned(8))) dither8[2]={
178
    0x0602060206020602LL,
179
    0x0004000400040004LL,};
180

    
181
#undef HAVE_MMX
182

    
183
//MMX versions
184
#undef RENAME
185
#define HAVE_MMX
186
#undef HAVE_MMX2
187
#undef HAVE_3DNOW
188
#define RENAME(a) a ## _MMX
189
#include "yuv2rgb_template.c"
190

    
191
//MMX2 versions
192
#undef RENAME
193
#define HAVE_MMX
194
#define HAVE_MMX2
195
#undef HAVE_3DNOW
196
#define RENAME(a) a ## _MMX2
197
#include "yuv2rgb_template.c"
198

    
199
#endif /* defined(ARCH_X86) */
200

    
201
const int32_t Inverse_Table_6_9[8][4] = {
202
    {117504, 138453, 13954, 34903}, /* no sequence_display_extension */
203
    {117504, 138453, 13954, 34903}, /* ITU-R Rec. 709 (1990) */
204
    {104597, 132201, 25675, 53279}, /* unspecified */
205
    {104597, 132201, 25675, 53279}, /* reserved */
206
    {104448, 132798, 24759, 53109}, /* FCC */
207
    {104597, 132201, 25675, 53279}, /* ITU-R Rec. 624-4 System B, G */
208
    {104597, 132201, 25675, 53279}, /* SMPTE 170M */
209
    {117579, 136230, 16907, 35559}  /* SMPTE 240M (1987) */
210
};
211

    
212
#define RGB(i)                                      \
213
    U = pu[i];                                      \
214
    V = pv[i];                                      \
215
    r = (void *)c->table_rV[V];                     \
216
    g = (void *)(c->table_gU[U] + c->table_gV[V]);  \
217
    b = (void *)c->table_bU[U];
218

    
219
#define DST1(i)                         \
220
    Y = py_1[2*i];                      \
221
    dst_1[2*i] = r[Y] + g[Y] + b[Y];    \
222
    Y = py_1[2*i+1];                    \
223
    dst_1[2*i+1] = r[Y] + g[Y] + b[Y];
224

    
225
#define DST2(i)                         \
226
    Y = py_2[2*i];                      \
227
    dst_2[2*i] = r[Y] + g[Y] + b[Y];    \
228
    Y = py_2[2*i+1];                    \
229
    dst_2[2*i+1] = r[Y] + g[Y] + b[Y];
230

    
231
#define DST1RGB(i)                                                \
232
    Y = py_1[2*i];                                                \
233
    dst_1[6*i] = r[Y]; dst_1[6*i+1] = g[Y]; dst_1[6*i+2] = b[Y];  \
234
    Y = py_1[2*i+1];                                              \
235
    dst_1[6*i+3] = r[Y]; dst_1[6*i+4] = g[Y]; dst_1[6*i+5] = b[Y];
236

    
237
#define DST2RGB(i)                                                \
238
    Y = py_2[2*i];                                                \
239
    dst_2[6*i] = r[Y]; dst_2[6*i+1] = g[Y]; dst_2[6*i+2] = b[Y];  \
240
    Y = py_2[2*i+1];                                              \
241
    dst_2[6*i+3] = r[Y]; dst_2[6*i+4] = g[Y]; dst_2[6*i+5] = b[Y];
242

    
243
#define DST1BGR(i)                                                \
244
    Y = py_1[2*i];                                                \
245
    dst_1[6*i] = b[Y]; dst_1[6*i+1] = g[Y]; dst_1[6*i+2] = r[Y];  \
246
    Y = py_1[2*i+1];                                              \
247
    dst_1[6*i+3] = b[Y]; dst_1[6*i+4] = g[Y]; dst_1[6*i+5] = r[Y];
248

    
249
#define DST2BGR(i)                                                \
250
    Y = py_2[2*i];                                                \
251
    dst_2[6*i] = b[Y]; dst_2[6*i+1] = g[Y]; dst_2[6*i+2] = r[Y];  \
252
    Y = py_2[2*i+1];                                              \
253
    dst_2[6*i+3] = b[Y]; dst_2[6*i+4] = g[Y]; dst_2[6*i+5] = r[Y];
254

    
255
#define PROLOG(func_name, dst_type) \
256
static int func_name(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, \
257
                     int srcSliceH, uint8_t* dst[], int dstStride[]){\
258
    int y;\
259
\
260
    if (c->srcFormat == PIX_FMT_YUV422P){\
261
        srcStride[1] *= 2;\
262
        srcStride[2] *= 2;\
263
    }\
264
    for (y=0; y<srcSliceH; y+=2){\
265
        dst_type *dst_1= (dst_type*)(dst[0] + (y+srcSliceY  )*dstStride[0]);\
266
        dst_type *dst_2= (dst_type*)(dst[0] + (y+srcSliceY+1)*dstStride[0]);\
267
        dst_type av_unused *r, *b;\
268
        dst_type *g;\
269
        uint8_t *py_1= src[0] + y*srcStride[0];\
270
        uint8_t *py_2= py_1 + srcStride[0];\
271
        uint8_t *pu= src[1] + (y>>1)*srcStride[1];\
272
        uint8_t *pv= src[2] + (y>>1)*srcStride[2];\
273
        unsigned int h_size= c->dstW>>3;\
274
        while (h_size--) {\
275
            int av_unused U, V;\
276
            int Y;\
277

    
278
#define EPILOG1(dst_delta)\
279
            pu += 4;\
280
            pv += 4;\
281
            py_1 += 8;\
282
            py_2 += 8;\
283
            dst_1 += dst_delta;\
284
            dst_2 += dst_delta;\
285
        }\
286
        if (c->dstW & 4) {\
287
            int av_unused U, V;\
288
            int Y;\
289

    
290
#define EPILOG2()\
291
        }\
292
    }\
293
    return srcSliceH;\
294
}
295

    
296
#define EPILOG(dst_delta)\
297
    EPILOG1(dst_delta)\
298
    EPILOG2()
299

    
300
PROLOG(yuv2rgb_c_32, uint32_t)
301
    RGB(0);
302
    DST1(0);
303
    DST2(0);
304

    
305
    RGB(1);
306
    DST2(1);
307
    DST1(1);
308

    
309
    RGB(2);
310
    DST1(2);
311
    DST2(2);
312

    
313
    RGB(3);
314
    DST2(3);
315
    DST1(3);
316
EPILOG1(8)
317
    RGB(0);
318
    DST1(0);
319
    DST2(0);
320

    
321
    RGB(1);
322
    DST2(1);
323
    DST1(1);
324
EPILOG2()
325

    
326
PROLOG(yuv2rgb_c_24_rgb, uint8_t)
327
    RGB(0);
328
    DST1RGB(0);
329
    DST2RGB(0);
330

    
331
    RGB(1);
332
    DST2RGB(1);
333
    DST1RGB(1);
334

    
335
    RGB(2);
336
    DST1RGB(2);
337
    DST2RGB(2);
338

    
339
    RGB(3);
340
    DST2RGB(3);
341
    DST1RGB(3);
342
EPILOG1(24)
343
    RGB(0);
344
    DST1RGB(0);
345
    DST2RGB(0);
346

    
347
    RGB(1);
348
    DST2RGB(1);
349
    DST1RGB(1);
350
EPILOG2()
351

    
352
// only trivial mods from yuv2rgb_c_24_rgb
353
PROLOG(yuv2rgb_c_24_bgr, uint8_t)
354
    RGB(0);
355
    DST1BGR(0);
356
    DST2BGR(0);
357

    
358
    RGB(1);
359
    DST2BGR(1);
360
    DST1BGR(1);
361

    
362
    RGB(2);
363
    DST1BGR(2);
364
    DST2BGR(2);
365

    
366
    RGB(3);
367
    DST2BGR(3);
368
    DST1BGR(3);
369
EPILOG1(24)
370
    RGB(0);
371
    DST1BGR(0);
372
    DST2BGR(0);
373

    
374
    RGB(1);
375
    DST2BGR(1);
376
    DST1BGR(1);
377
EPILOG2()
378

    
379
// This is exactly the same code as yuv2rgb_c_32 except for the types of
380
// r, g, b, dst_1, dst_2
381
PROLOG(yuv2rgb_c_16, uint16_t)
382
    RGB(0);
383
    DST1(0);
384
    DST2(0);
385

    
386
    RGB(1);
387
    DST2(1);
388
    DST1(1);
389

    
390
    RGB(2);
391
    DST1(2);
392
    DST2(2);
393

    
394
    RGB(3);
395
    DST2(3);
396
    DST1(3);
397
EPILOG(8)
398

    
399
// This is exactly the same code as yuv2rgb_c_32 except for the types of
400
// r, g, b, dst_1, dst_2
401
PROLOG(yuv2rgb_c_8, uint8_t)
402
    RGB(0);
403
    DST1(0);
404
    DST2(0);
405

    
406
    RGB(1);
407
    DST2(1);
408
    DST1(1);
409

    
410
    RGB(2);
411
    DST1(2);
412
    DST2(2);
413

    
414
    RGB(3);
415
    DST2(3);
416
    DST1(3);
417
EPILOG(8)
418

    
419
// r, g, b, dst_1, dst_2
420
PROLOG(yuv2rgb_c_8_ordered_dither, uint8_t)
421
    const uint8_t *d32= dither_8x8_32[y&7];
422
    const uint8_t *d64= dither_8x8_73[y&7];
423
#define DST1bpp8(i,o)                                               \
424
    Y = py_1[2*i];                                                  \
425
    dst_1[2*i]   = r[Y+d32[0+o]] + g[Y+d32[0+o]] + b[Y+d64[0+o]];   \
426
    Y = py_1[2*i+1];                                                \
427
    dst_1[2*i+1] = r[Y+d32[1+o]] + g[Y+d32[1+o]] + b[Y+d64[1+o]];
428

    
429
#define DST2bpp8(i,o)                                               \
430
    Y = py_2[2*i];                                                  \
431
    dst_2[2*i]   =  r[Y+d32[8+o]] + g[Y+d32[8+o]] + b[Y+d64[8+o]];  \
432
    Y = py_2[2*i+1];                                                \
433
    dst_2[2*i+1] =  r[Y+d32[9+o]] + g[Y+d32[9+o]] + b[Y+d64[9+o]];
434

    
435

    
436
    RGB(0);
437
    DST1bpp8(0,0);
438
    DST2bpp8(0,0);
439

    
440
    RGB(1);
441
    DST2bpp8(1,2);
442
    DST1bpp8(1,2);
443

    
444
    RGB(2);
445
    DST1bpp8(2,4);
446
    DST2bpp8(2,4);
447

    
448
    RGB(3);
449
    DST2bpp8(3,6);
450
    DST1bpp8(3,6);
451
EPILOG(8)
452

    
453

    
454
// This is exactly the same code as yuv2rgb_c_32 except for the types of
455
// r, g, b, dst_1, dst_2
456
PROLOG(yuv2rgb_c_4, uint8_t)
457
    int acc;
458
#define DST1_4(i)                   \
459
    Y = py_1[2*i];                  \
460
    acc = r[Y] + g[Y] + b[Y];       \
461
    Y = py_1[2*i+1];                \
462
    acc |= (r[Y] + g[Y] + b[Y])<<4; \
463
    dst_1[i] = acc;
464

    
465
#define DST2_4(i)                   \
466
    Y = py_2[2*i];                  \
467
    acc = r[Y] + g[Y] + b[Y];       \
468
    Y = py_2[2*i+1];                \
469
    acc |= (r[Y] + g[Y] + b[Y])<<4; \
470
    dst_2[i] = acc;
471

    
472
    RGB(0);
473
    DST1_4(0);
474
    DST2_4(0);
475

    
476
    RGB(1);
477
    DST2_4(1);
478
    DST1_4(1);
479

    
480
    RGB(2);
481
    DST1_4(2);
482
    DST2_4(2);
483

    
484
    RGB(3);
485
    DST2_4(3);
486
    DST1_4(3);
487
EPILOG(4)
488

    
489
PROLOG(yuv2rgb_c_4_ordered_dither, uint8_t)
490
    const uint8_t *d64= dither_8x8_73[y&7];
491
    const uint8_t *d128=dither_8x8_220[y&7];
492
    int acc;
493

    
494
#define DST1bpp4(i,o)                                             \
495
    Y = py_1[2*i];                                                \
496
    acc = r[Y+d128[0+o]] + g[Y+d64[0+o]] + b[Y+d128[0+o]];        \
497
    Y = py_1[2*i+1];                                              \
498
    acc |= (r[Y+d128[1+o]] + g[Y+d64[1+o]] + b[Y+d128[1+o]])<<4;  \
499
    dst_1[i]= acc;
500

    
501
#define DST2bpp4(i,o)                                             \
502
    Y = py_2[2*i];                                                \
503
    acc =  r[Y+d128[8+o]] + g[Y+d64[8+o]] + b[Y+d128[8+o]];       \
504
    Y = py_2[2*i+1];                                              \
505
    acc |=  (r[Y+d128[9+o]] + g[Y+d64[9+o]] + b[Y+d128[9+o]])<<4; \
506
    dst_2[i]= acc;
507

    
508

    
509
    RGB(0);
510
    DST1bpp4(0,0);
511
    DST2bpp4(0,0);
512

    
513
    RGB(1);
514
    DST2bpp4(1,2);
515
    DST1bpp4(1,2);
516

    
517
    RGB(2);
518
    DST1bpp4(2,4);
519
    DST2bpp4(2,4);
520

    
521
    RGB(3);
522
    DST2bpp4(3,6);
523
    DST1bpp4(3,6);
524
EPILOG(4)
525

    
526
// This is exactly the same code as yuv2rgb_c_32 except for the types of
527
// r, g, b, dst_1, dst_2
528
PROLOG(yuv2rgb_c_4b, uint8_t)
529
    RGB(0);
530
    DST1(0);
531
    DST2(0);
532

    
533
    RGB(1);
534
    DST2(1);
535
    DST1(1);
536

    
537
    RGB(2);
538
    DST1(2);
539
    DST2(2);
540

    
541
    RGB(3);
542
    DST2(3);
543
    DST1(3);
544
EPILOG(8)
545

    
546
PROLOG(yuv2rgb_c_4b_ordered_dither, uint8_t)
547
    const uint8_t *d64= dither_8x8_73[y&7];
548
    const uint8_t *d128=dither_8x8_220[y&7];
549

    
550
#define DST1bpp4b(i,o)                                                \
551
    Y = py_1[2*i];                                                    \
552
    dst_1[2*i]   = r[Y+d128[0+o]] + g[Y+d64[0+o]] + b[Y+d128[0+o]];   \
553
    Y = py_1[2*i+1];                                                  \
554
    dst_1[2*i+1] = r[Y+d128[1+o]] + g[Y+d64[1+o]] + b[Y+d128[1+o]];
555

    
556
#define DST2bpp4b(i,o)                                                \
557
    Y = py_2[2*i];                                                    \
558
    dst_2[2*i]   =  r[Y+d128[8+o]] + g[Y+d64[8+o]] + b[Y+d128[8+o]];  \
559
    Y = py_2[2*i+1];                                                  \
560
    dst_2[2*i+1] =  r[Y+d128[9+o]] + g[Y+d64[9+o]] + b[Y+d128[9+o]];
561

    
562

    
563
    RGB(0);
564
    DST1bpp4b(0,0);
565
    DST2bpp4b(0,0);
566

    
567
    RGB(1);
568
    DST2bpp4b(1,2);
569
    DST1bpp4b(1,2);
570

    
571
    RGB(2);
572
    DST1bpp4b(2,4);
573
    DST2bpp4b(2,4);
574

    
575
    RGB(3);
576
    DST2bpp4b(3,6);
577
    DST1bpp4b(3,6);
578
EPILOG(8)
579

    
580
PROLOG(yuv2rgb_c_1_ordered_dither, uint8_t)
581
        const uint8_t *d128=dither_8x8_220[y&7];
582
        char out_1=0, out_2=0;
583
        g= c->table_gU[128] + c->table_gV[128];
584

    
585
#define DST1bpp1(i,o)               \
586
    Y = py_1[2*i];                  \
587
    out_1+= out_1 + g[Y+d128[0+o]]; \
588
    Y = py_1[2*i+1];                \
589
    out_1+= out_1 + g[Y+d128[1+o]];
590

    
591
#define DST2bpp1(i,o)               \
592
    Y = py_2[2*i];                  \
593
    out_2+= out_2 + g[Y+d128[8+o]]; \
594
    Y = py_2[2*i+1];                \
595
    out_2+= out_2 + g[Y+d128[9+o]];
596

    
597
    DST1bpp1(0,0);
598
    DST2bpp1(0,0);
599

    
600
    DST2bpp1(1,2);
601
    DST1bpp1(1,2);
602

    
603
    DST1bpp1(2,4);
604
    DST2bpp1(2,4);
605

    
606
    DST2bpp1(3,6);
607
    DST1bpp1(3,6);
608

    
609
    dst_1[0]= out_1;
610
    dst_2[0]= out_2;
611
EPILOG(1)
612

    
613
SwsFunc yuv2rgb_get_func_ptr (SwsContext *c)
614
{
615
#if defined(HAVE_MMX2) || defined(HAVE_MMX)
616
    if (c->flags & SWS_CPU_CAPS_MMX2){
617
        switch(c->dstFormat){
618
        case PIX_FMT_RGB32:  return yuv420_rgb32_MMX2;
619
        case PIX_FMT_BGR24:  return yuv420_rgb24_MMX2;
620
        case PIX_FMT_BGR565: return yuv420_rgb16_MMX2;
621
        case PIX_FMT_BGR555: return yuv420_rgb15_MMX2;
622
        }
623
    }
624
    if (c->flags & SWS_CPU_CAPS_MMX){
625
        switch(c->dstFormat){
626
        case PIX_FMT_RGB32:  return yuv420_rgb32_MMX;
627
        case PIX_FMT_BGR24:  return yuv420_rgb24_MMX;
628
        case PIX_FMT_BGR565: return yuv420_rgb16_MMX;
629
        case PIX_FMT_BGR555: return yuv420_rgb15_MMX;
630
        }
631
    }
632
#endif
633
#ifdef HAVE_MLIB
634
    {
635
        SwsFunc t= yuv2rgb_init_mlib(c);
636
        if (t) return t;
637
    }
638
#endif
639
#ifdef HAVE_ALTIVEC
640
    if (c->flags & SWS_CPU_CAPS_ALTIVEC)
641
    {
642
        SwsFunc t = yuv2rgb_init_altivec(c);
643
        if (t) return t;
644
    }
645
#endif
646

    
647
#ifdef ARCH_BFIN
648
    if (c->flags & SWS_CPU_CAPS_BFIN)
649
    {
650
        SwsFunc t = ff_bfin_yuv2rgb_get_func_ptr (c);
651
        if (t) return t;
652
    }
653
#endif
654

    
655
    av_log(c, AV_LOG_WARNING, "No accelerated colorspace conversion found\n");
656

    
657
    switch(c->dstFormat){
658
    case PIX_FMT_BGR32:
659
    case PIX_FMT_RGB32: return yuv2rgb_c_32;
660
    case PIX_FMT_RGB24: return yuv2rgb_c_24_rgb;
661
    case PIX_FMT_BGR24: return yuv2rgb_c_24_bgr;
662
    case PIX_FMT_RGB565:
663
    case PIX_FMT_BGR565:
664
    case PIX_FMT_RGB555:
665
    case PIX_FMT_BGR555: return yuv2rgb_c_16;
666
    case PIX_FMT_RGB8:
667
    case PIX_FMT_BGR8:  return yuv2rgb_c_8_ordered_dither;
668
    case PIX_FMT_RGB4:
669
    case PIX_FMT_BGR4:  return yuv2rgb_c_4_ordered_dither;
670
    case PIX_FMT_RGB4_BYTE:
671
    case PIX_FMT_BGR4_BYTE:  return yuv2rgb_c_4b_ordered_dither;
672
    case PIX_FMT_MONOBLACK:  return yuv2rgb_c_1_ordered_dither;
673
    default:
674
        assert(0);
675
    }
676
    return NULL;
677
}
678

    
679
static int div_round (int dividend, int divisor)
680
{
681
    if (dividend > 0)
682
        return (dividend + (divisor>>1)) / divisor;
683
    else
684
        return -((-dividend + (divisor>>1)) / divisor);
685
}
686

    
687
int yuv2rgb_c_init_tables (SwsContext *c, const int inv_table[4], int fullRange, int brightness, int contrast, int saturation)
688
{
689
    const int isRgb = isBGR(c->dstFormat);
690
    const int bpp = fmt_depth(c->dstFormat);
691
    int i;
692
    uint8_t table_Y[1024];
693
    uint32_t *table_32 = 0;
694
    uint16_t *table_16 = 0;
695
    uint8_t *table_8 = 0;
696
    uint8_t *table_332 = 0;
697
    uint8_t *table_121 = 0;
698
    uint8_t *table_1 = 0;
699
    int entry_size = 0;
700
    void *table_r = 0, *table_g = 0, *table_b = 0;
701
    void *table_start;
702

    
703
    int64_t crv =  inv_table[0];
704
    int64_t cbu =  inv_table[1];
705
    int64_t cgu = -inv_table[2];
706
    int64_t cgv = -inv_table[3];
707
    int64_t cy  = 1<<16;
708
    int64_t oy  = 0;
709

    
710
//printf("%lld %lld %lld %lld %lld\n", cy, crv, cbu, cgu, cgv);
711
    if (!fullRange){
712
        cy= (cy*255) / 219;
713
        oy= 16<<16;
714
    }else{
715
        crv= (crv*224) / 255;
716
        cbu= (cbu*224) / 255;
717
        cgu= (cgu*224) / 255;
718
        cgv= (cgv*224) / 255;
719
    }
720

    
721
    cy = (cy *contrast             )>>16;
722
    crv= (crv*contrast * saturation)>>32;
723
    cbu= (cbu*contrast * saturation)>>32;
724
    cgu= (cgu*contrast * saturation)>>32;
725
    cgv= (cgv*contrast * saturation)>>32;
726
//printf("%lld %lld %lld %lld %lld\n", cy, crv, cbu, cgu, cgv);
727
    oy -= 256*brightness;
728

    
729
    for (i = 0; i < 1024; i++) {
730
        int j;
731

    
732
        j= (cy*(((i - 384)<<16) - oy) + (1<<31))>>32;
733
        j = (j < 0) ? 0 : ((j > 255) ? 255 : j);
734
        table_Y[i] = j;
735
    }
736

    
737
    switch (bpp) {
738
    case 32:
739
        table_start= table_32 = av_malloc ((197 + 2*682 + 256 + 132) * sizeof (uint32_t));
740

    
741
        entry_size = sizeof (uint32_t);
742
        table_r = table_32 + 197;
743
        table_b = table_32 + 197 + 685;
744
        table_g = table_32 + 197 + 2*682;
745

    
746
        for (i = -197; i < 256+197; i++)
747
            ((uint32_t *)table_r)[i] = table_Y[i+384] << (isRgb ? 16 : 0);
748
        for (i = -132; i < 256+132; i++)
749
            ((uint32_t *)table_g)[i] = table_Y[i+384] << 8;
750
        for (i = -232; i < 256+232; i++)
751
            ((uint32_t *)table_b)[i] = table_Y[i+384] << (isRgb ? 0 : 16);
752
        break;
753

    
754
    case 24:
755
        table_start= table_8 = av_malloc ((256 + 2*232) * sizeof (uint8_t));
756

    
757
        entry_size = sizeof (uint8_t);
758
        table_r = table_g = table_b = table_8 + 232;
759

    
760
        for (i = -232; i < 256+232; i++)
761
            ((uint8_t * )table_b)[i] = table_Y[i+384];
762
        break;
763

    
764
    case 15:
765
    case 16:
766
        table_start= table_16 = av_malloc ((197 + 2*682 + 256 + 132) * sizeof (uint16_t));
767

    
768
        entry_size = sizeof (uint16_t);
769
        table_r = table_16 + 197;
770
        table_b = table_16 + 197 + 685;
771
        table_g = table_16 + 197 + 2*682;
772

    
773
        for (i = -197; i < 256+197; i++) {
774
            int j = table_Y[i+384] >> 3;
775

    
776
            if (isRgb)
777
                j <<= ((bpp==16) ? 11 : 10);
778

    
779
            ((uint16_t *)table_r)[i] = j;
780
        }
781
        for (i = -132; i < 256+132; i++) {
782
            int j = table_Y[i+384] >> ((bpp==16) ? 2 : 3);
783

    
784
            ((uint16_t *)table_g)[i] = j << 5;
785
        }
786
        for (i = -232; i < 256+232; i++) {
787
            int j = table_Y[i+384] >> 3;
788

    
789
            if (!isRgb)
790
                j <<= ((bpp==16) ? 11 : 10);
791

    
792
            ((uint16_t *)table_b)[i] = j;
793
        }
794
        break;
795

    
796
    case 8:
797
        table_start= table_332 = av_malloc ((197 + 2*682 + 256 + 132) * sizeof (uint8_t));
798

    
799
        entry_size = sizeof (uint8_t);
800
        table_r = table_332 + 197;
801
        table_b = table_332 + 197 + 685;
802
        table_g = table_332 + 197 + 2*682;
803

    
804
        for (i = -197; i < 256+197; i++) {
805
            int j = (table_Y[i+384 - 16] + 18)/36;
806

    
807
            if (isRgb)
808
                j <<= 5;
809

    
810
            ((uint8_t *)table_r)[i] = j;
811
        }
812
        for (i = -132; i < 256+132; i++) {
813
            int j = (table_Y[i+384 - 16] + 18)/36;
814

    
815
            if (!isRgb)
816
                j <<= 1;
817

    
818
            ((uint8_t *)table_g)[i] = j << 2;
819
        }
820
        for (i = -232; i < 256+232; i++) {
821
            int j = (table_Y[i+384 - 37] + 43)/85;
822

    
823
            if (!isRgb)
824
                j <<= 6;
825

    
826
            ((uint8_t *)table_b)[i] = j;
827
        }
828
        break;
829
    case 4:
830
    case 4|128:
831
        table_start= table_121 = av_malloc ((197 + 2*682 + 256 + 132) * sizeof (uint8_t));
832

    
833
        entry_size = sizeof (uint8_t);
834
        table_r = table_121 + 197;
835
        table_b = table_121 + 197 + 685;
836
        table_g = table_121 + 197 + 2*682;
837

    
838
        for (i = -197; i < 256+197; i++) {
839
            int j = table_Y[i+384 - 110] >> 7;
840

    
841
            if (isRgb)
842
                j <<= 3;
843

    
844
            ((uint8_t *)table_r)[i] = j;
845
        }
846
        for (i = -132; i < 256+132; i++) {
847
            int j = (table_Y[i+384 - 37]+ 43)/85;
848

    
849
            ((uint8_t *)table_g)[i] = j << 1;
850
        }
851
        for (i = -232; i < 256+232; i++) {
852
            int j =table_Y[i+384 - 110] >> 7;
853

    
854
            if (!isRgb)
855
                j <<= 3;
856

    
857
            ((uint8_t *)table_b)[i] = j;
858
        }
859
        break;
860

    
861
    case 1:
862
        table_start= table_1 = av_malloc (256*2 * sizeof (uint8_t));
863

    
864
        entry_size = sizeof (uint8_t);
865
        table_g = table_1;
866
        table_r = table_b = NULL;
867

    
868
        for (i = 0; i < 256+256; i++) {
869
            int j = table_Y[i + 384 - 110]>>7;
870

    
871
            ((uint8_t *)table_g)[i] = j;
872
        }
873
        break;
874

    
875
    default:
876
        table_start= NULL;
877
        av_log(c, AV_LOG_ERROR, "%ibpp not supported by yuv2rgb\n", bpp);
878
        //free mem?
879
        return -1;
880
    }
881

    
882
    for (i = 0; i < 256; i++) {
883
        c->table_rV[i] = (uint8_t *)table_r + entry_size * div_round (crv * (i-128), 76309);
884
        c->table_gU[i] = (uint8_t *)table_g + entry_size * div_round (cgu * (i-128), 76309);
885
        c->table_gV[i] = entry_size * div_round (cgv * (i-128), 76309);
886
        c->table_bU[i] = (uint8_t *)table_b + entry_size * div_round (cbu * (i-128), 76309);
887
    }
888

    
889
    av_free(c->yuvTable);
890
    c->yuvTable= table_start;
891
    return 0;
892
}