Statistics
| Branch: | Revision:

ffmpeg / libswscale / yuv2rgb.c @ c1578479

History | View | Annotate | Download (24.9 KB)

1
/*
2
 * yuv2rgb.c, Software YUV to RGB coverter
3
 *
4
 *  Copyright (C) 1999, Aaron Holtzman <aholtzma@ess.engr.uvic.ca>
5
 *  All Rights Reserved.
6
 *
7
 *  Functions broken out from display_x11.c and several new modes
8
 *  added by H?kan Hjort <d95hjort@dtek.chalmers.se>
9
 *
10
 *  15 & 16 bpp support by Franck Sicard <Franck.Sicard@solsoft.fr>
11
 *
12
 *  MMX/MMX2 template stuff (needed for fast movntq support),
13
 *  1,4,8bpp support and context / deglobalize stuff
14
 *  by Michael Niedermayer (michaelni@gmx.at)
15
 *
16
 *  This file is part of mpeg2dec, a free MPEG-2 video decoder
17
 *
18
 *  mpeg2dec is free software; you can redistribute it and/or modify
19
 *  it under the terms of the GNU General Public License as published by
20
 *  the Free Software Foundation; either version 2, or (at your option)
21
 *  any later version.
22
 *
23
 *  mpeg2dec is distributed in the hope that it will be useful,
24
 *  but WITHOUT ANY WARRANTY; without even the implied warranty of
25
 *  MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
26
 *  GNU General Public License for more details.
27
 *
28
 *  You should have received a copy of the GNU General Public License
29
 *  along with mpeg2dec; if not, write to the Free Software
30
 *  Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
31
 */
32

    
33
#include <stdio.h>
34
#include <stdlib.h>
35
#include <inttypes.h>
36
#include <assert.h>
37

    
38
#include "config.h"
39
#include "rgb2rgb.h"
40
#include "swscale.h"
41
#include "swscale_internal.h"
42

    
43
#ifdef HAVE_MLIB
44
#include "yuv2rgb_mlib.c"
45
#endif
46

    
47
#define DITHER1XBPP // only for mmx
48

    
49
const uint8_t  __attribute__((aligned(8))) dither_2x2_4[2][8]={
50
{  1,   3,   1,   3,   1,   3,   1,   3, },
51
{  2,   0,   2,   0,   2,   0,   2,   0, },
52
};
53

    
54
const uint8_t  __attribute__((aligned(8))) dither_2x2_8[2][8]={
55
{  6,   2,   6,   2,   6,   2,   6,   2, },
56
{  0,   4,   0,   4,   0,   4,   0,   4, },
57
};
58

    
59
const uint8_t  __attribute__((aligned(8))) dither_8x8_32[8][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
};
69

    
70
#if 0
71
const uint8_t  __attribute__((aligned(8))) dither_8x8_64[8][8]={
72
{  0,  48,  12,  60,   3,  51,  15,  63, },
73
{ 32,  16,  44,  28,  35,  19,  47,  31, },
74
{  8,  56,   4,  52,  11,  59,   7,  55, },
75
{ 40,  24,  36,  20,  43,  27,  39,  23, },
76
{  2,  50,  14,  62,   1,  49,  13,  61, },
77
{ 34,  18,  46,  30,  33,  17,  45,  29, },
78
{ 10,  58,   6,  54,   9,  57,   5,  53, },
79
{ 42,  26,  38,  22,  41,  25,  37,  21, },
80
};
81
#endif
82

    
83
const uint8_t  __attribute__((aligned(8))) dither_8x8_73[8][8]={
84
{  0,  55,  14,  68,   3,  58,  17,  72, },
85
{ 37,  18,  50,  32,  40,  22,  54,  35, },
86
{  9,  64,   5,  59,  13,  67,   8,  63, },
87
{ 46,  27,  41,  23,  49,  31,  44,  26, },
88
{  2,  57,  16,  71,   1,  56,  15,  70, },
89
{ 39,  21,  52,  34,  38,  19,  51,  33, },
90
{ 11,  66,   7,  62,  10,  65,   6,  60, },
91
{ 48,  30,  43,  25,  47,  29,  42,  24, },
92
};
93

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

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

    
156
#ifdef HAVE_MMX
157

    
158
/* hope these constant values are cache line aligned */
159
static uint64_t attribute_used __attribute__((aligned(8))) mmx_00ffw   = 0x00ff00ff00ff00ffULL;
160
static uint64_t attribute_used __attribute__((aligned(8))) mmx_redmask = 0xf8f8f8f8f8f8f8f8ULL;
161
static uint64_t attribute_used __attribute__((aligned(8))) mmx_grnmask = 0xfcfcfcfcfcfcfcfcULL;
162

    
163
static uint64_t attribute_used __attribute__((aligned(8))) M24A=   0x00FF0000FF0000FFULL;
164
static uint64_t attribute_used __attribute__((aligned(8))) M24B=   0xFF0000FF0000FF00ULL;
165
static uint64_t attribute_used __attribute__((aligned(8))) M24C=   0x0000FF0000FF0000ULL;
166

    
167
// the volatile is required because gcc otherwise optimizes some writes away not knowing that these
168
// are read in the asm block
169
static volatile uint64_t attribute_used __attribute__((aligned(8))) b5Dither;
170
static volatile uint64_t attribute_used __attribute__((aligned(8))) g5Dither;
171
static volatile uint64_t attribute_used __attribute__((aligned(8))) g6Dither;
172
static volatile uint64_t attribute_used __attribute__((aligned(8))) r5Dither;
173

    
174
static uint64_t __attribute__((aligned(8))) dither4[2]={
175
    0x0103010301030103LL,
176
    0x0200020002000200LL,};
177

    
178
static uint64_t __attribute__((aligned(8))) dither8[2]={
179
    0x0602060206020602LL,
180
    0x0004000400040004LL,};
181

    
182
#undef HAVE_MMX
183

    
184
//MMX versions
185
#undef RENAME
186
#define HAVE_MMX
187
#undef HAVE_MMX2
188
#undef HAVE_3DNOW
189
#define RENAME(a) a ## _MMX
190
#include "yuv2rgb_template.c"
191

    
192
//MMX2 versions
193
#undef RENAME
194
#define HAVE_MMX
195
#define HAVE_MMX2
196
#undef HAVE_3DNOW
197
#define RENAME(a) a ## _MMX2
198
#include "yuv2rgb_template.c"
199

    
200
#endif /* defined(ARCH_X86) */
201

    
202
const int32_t Inverse_Table_6_9[8][4] = {
203
    {117504, 138453, 13954, 34903}, /* no sequence_display_extension */
204
    {117504, 138453, 13954, 34903}, /* ITU-R Rec. 709 (1990) */
205
    {104597, 132201, 25675, 53279}, /* unspecified */
206
    {104597, 132201, 25675, 53279}, /* reserved */
207
    {104448, 132798, 24759, 53109}, /* FCC */
208
    {104597, 132201, 25675, 53279}, /* ITU-R Rec. 624-4 System B, G */
209
    {104597, 132201, 25675, 53279}, /* SMPTE 170M */
210
    {117579, 136230, 16907, 35559}  /* SMPTE 240M (1987) */
211
};
212

    
213
#define RGB(i)                                      \
214
    U = pu[i];                                      \
215
    V = pv[i];                                      \
216
    r = (void *)c->table_rV[V];                     \
217
    g = (void *)(c->table_gU[U] + c->table_gV[V]);  \
218
    b = (void *)c->table_bU[U];
219

    
220
#define DST1(i)                         \
221
    Y = py_1[2*i];                      \
222
    dst_1[2*i] = r[Y] + g[Y] + b[Y];    \
223
    Y = py_1[2*i+1];                    \
224
    dst_1[2*i+1] = r[Y] + g[Y] + b[Y];
225

    
226
#define DST2(i)                         \
227
    Y = py_2[2*i];                      \
228
    dst_2[2*i] = r[Y] + g[Y] + b[Y];    \
229
    Y = py_2[2*i+1];                    \
230
    dst_2[2*i+1] = r[Y] + g[Y] + b[Y];
231

    
232
#define DST1RGB(i)                                                \
233
    Y = py_1[2*i];                                                \
234
    dst_1[6*i] = r[Y]; dst_1[6*i+1] = g[Y]; dst_1[6*i+2] = b[Y];  \
235
    Y = py_1[2*i+1];                                              \
236
    dst_1[6*i+3] = r[Y]; dst_1[6*i+4] = g[Y]; dst_1[6*i+5] = b[Y];
237

    
238
#define DST2RGB(i)                                                \
239
    Y = py_2[2*i];                                                \
240
    dst_2[6*i] = r[Y]; dst_2[6*i+1] = g[Y]; dst_2[6*i+2] = b[Y];  \
241
    Y = py_2[2*i+1];                                              \
242
    dst_2[6*i+3] = r[Y]; dst_2[6*i+4] = g[Y]; dst_2[6*i+5] = b[Y];
243

    
244
#define DST1BGR(i)                                                \
245
    Y = py_1[2*i];                                                \
246
    dst_1[6*i] = b[Y]; dst_1[6*i+1] = g[Y]; dst_1[6*i+2] = r[Y];  \
247
    Y = py_1[2*i+1];                                              \
248
    dst_1[6*i+3] = b[Y]; dst_1[6*i+4] = g[Y]; dst_1[6*i+5] = r[Y];
249

    
250
#define DST2BGR(i)                                                \
251
    Y = py_2[2*i];                                                \
252
    dst_2[6*i] = b[Y]; dst_2[6*i+1] = g[Y]; dst_2[6*i+2] = r[Y];  \
253
    Y = py_2[2*i+1];                                              \
254
    dst_2[6*i+3] = b[Y]; dst_2[6*i+4] = g[Y]; dst_2[6*i+5] = r[Y];
255

    
256
#define PROLOG(func_name, dst_type) \
257
static int func_name(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, \
258
                     int srcSliceH, uint8_t* dst[], int dstStride[]){\
259
    int y;\
260
\
261
    if (c->srcFormat == PIX_FMT_YUV422P){\
262
        srcStride[1] *= 2;\
263
        srcStride[2] *= 2;\
264
    }\
265
    for (y=0; y<srcSliceH; y+=2){\
266
        dst_type *dst_1= (dst_type*)(dst[0] + (y+srcSliceY  )*dstStride[0]);\
267
        dst_type *dst_2= (dst_type*)(dst[0] + (y+srcSliceY+1)*dstStride[0]);\
268
        dst_type av_unused *r, *b;\
269
        dst_type *g;\
270
        uint8_t *py_1= src[0] + y*srcStride[0];\
271
        uint8_t *py_2= py_1 + srcStride[0];\
272
        uint8_t *pu= src[1] + (y>>1)*srcStride[1];\
273
        uint8_t *pv= src[2] + (y>>1)*srcStride[2];\
274
        unsigned int h_size= c->dstW>>3;\
275
        while (h_size--) {\
276
            int av_unused U, V;\
277
            int Y;\
278

    
279
#define EPILOG1(dst_delta)\
280
            pu += 4;\
281
            pv += 4;\
282
            py_1 += 8;\
283
            py_2 += 8;\
284
            dst_1 += dst_delta;\
285
            dst_2 += dst_delta;\
286
        }\
287
        if (c->dstW & 4) {\
288
            int av_unused U, V;\
289
            int Y;\
290

    
291
#define EPILOG2()\
292
        }\
293
    }\
294
    return srcSliceH;\
295
}
296

    
297
#define EPILOG(dst_delta)\
298
    EPILOG1(dst_delta)\
299
    EPILOG2()
300

    
301
PROLOG(yuv2rgb_c_32, uint32_t)
302
    RGB(0);
303
    DST1(0);
304
    DST2(0);
305

    
306
    RGB(1);
307
    DST2(1);
308
    DST1(1);
309

    
310
    RGB(2);
311
    DST1(2);
312
    DST2(2);
313

    
314
    RGB(3);
315
    DST2(3);
316
    DST1(3);
317
EPILOG1(8)
318
    RGB(0);
319
    DST1(0);
320
    DST2(0);
321

    
322
    RGB(1);
323
    DST2(1);
324
    DST1(1);
325
EPILOG2()
326

    
327
PROLOG(yuv2rgb_c_24_rgb, uint8_t)
328
    RGB(0);
329
    DST1RGB(0);
330
    DST2RGB(0);
331

    
332
    RGB(1);
333
    DST2RGB(1);
334
    DST1RGB(1);
335

    
336
    RGB(2);
337
    DST1RGB(2);
338
    DST2RGB(2);
339

    
340
    RGB(3);
341
    DST2RGB(3);
342
    DST1RGB(3);
343
EPILOG1(24)
344
    RGB(0);
345
    DST1RGB(0);
346
    DST2RGB(0);
347

    
348
    RGB(1);
349
    DST2RGB(1);
350
    DST1RGB(1);
351
EPILOG2()
352

    
353
// only trivial mods from yuv2rgb_c_24_rgb
354
PROLOG(yuv2rgb_c_24_bgr, uint8_t)
355
    RGB(0);
356
    DST1BGR(0);
357
    DST2BGR(0);
358

    
359
    RGB(1);
360
    DST2BGR(1);
361
    DST1BGR(1);
362

    
363
    RGB(2);
364
    DST1BGR(2);
365
    DST2BGR(2);
366

    
367
    RGB(3);
368
    DST2BGR(3);
369
    DST1BGR(3);
370
EPILOG1(24)
371
    RGB(0);
372
    DST1BGR(0);
373
    DST2BGR(0);
374

    
375
    RGB(1);
376
    DST2BGR(1);
377
    DST1BGR(1);
378
EPILOG2()
379

    
380
// This is exactly the same code as yuv2rgb_c_32 except for the types of
381
// r, g, b, dst_1, dst_2
382
PROLOG(yuv2rgb_c_16, uint16_t)
383
    RGB(0);
384
    DST1(0);
385
    DST2(0);
386

    
387
    RGB(1);
388
    DST2(1);
389
    DST1(1);
390

    
391
    RGB(2);
392
    DST1(2);
393
    DST2(2);
394

    
395
    RGB(3);
396
    DST2(3);
397
    DST1(3);
398
EPILOG(8)
399

    
400
// This is exactly the same code as yuv2rgb_c_32 except for the types of
401
// r, g, b, dst_1, dst_2
402
PROLOG(yuv2rgb_c_8, uint8_t)
403
    RGB(0);
404
    DST1(0);
405
    DST2(0);
406

    
407
    RGB(1);
408
    DST2(1);
409
    DST1(1);
410

    
411
    RGB(2);
412
    DST1(2);
413
    DST2(2);
414

    
415
    RGB(3);
416
    DST2(3);
417
    DST1(3);
418
EPILOG(8)
419

    
420
// r, g, b, dst_1, dst_2
421
PROLOG(yuv2rgb_c_8_ordered_dither, uint8_t)
422
    const uint8_t *d32= dither_8x8_32[y&7];
423
    const uint8_t *d64= dither_8x8_73[y&7];
424
#define DST1bpp8(i,o)                                               \
425
    Y = py_1[2*i];                                                  \
426
    dst_1[2*i]   = r[Y+d32[0+o]] + g[Y+d32[0+o]] + b[Y+d64[0+o]];   \
427
    Y = py_1[2*i+1];                                                \
428
    dst_1[2*i+1] = r[Y+d32[1+o]] + g[Y+d32[1+o]] + b[Y+d64[1+o]];
429

    
430
#define DST2bpp8(i,o)                                               \
431
    Y = py_2[2*i];                                                  \
432
    dst_2[2*i]   =  r[Y+d32[8+o]] + g[Y+d32[8+o]] + b[Y+d64[8+o]];  \
433
    Y = py_2[2*i+1];                                                \
434
    dst_2[2*i+1] =  r[Y+d32[9+o]] + g[Y+d32[9+o]] + b[Y+d64[9+o]];
435

    
436

    
437
    RGB(0);
438
    DST1bpp8(0,0);
439
    DST2bpp8(0,0);
440

    
441
    RGB(1);
442
    DST2bpp8(1,2);
443
    DST1bpp8(1,2);
444

    
445
    RGB(2);
446
    DST1bpp8(2,4);
447
    DST2bpp8(2,4);
448

    
449
    RGB(3);
450
    DST2bpp8(3,6);
451
    DST1bpp8(3,6);
452
EPILOG(8)
453

    
454

    
455
// This is exactly the same code as yuv2rgb_c_32 except for the types of
456
// r, g, b, dst_1, dst_2
457
PROLOG(yuv2rgb_c_4, uint8_t)
458
    int acc;
459
#define DST1_4(i)                   \
460
    Y = py_1[2*i];                  \
461
    acc = r[Y] + g[Y] + b[Y];       \
462
    Y = py_1[2*i+1];                \
463
    acc |= (r[Y] + g[Y] + b[Y])<<4; \
464
    dst_1[i] = acc;
465

    
466
#define DST2_4(i)                   \
467
    Y = py_2[2*i];                  \
468
    acc = r[Y] + g[Y] + b[Y];       \
469
    Y = py_2[2*i+1];                \
470
    acc |= (r[Y] + g[Y] + b[Y])<<4; \
471
    dst_2[i] = acc;
472

    
473
    RGB(0);
474
    DST1_4(0);
475
    DST2_4(0);
476

    
477
    RGB(1);
478
    DST2_4(1);
479
    DST1_4(1);
480

    
481
    RGB(2);
482
    DST1_4(2);
483
    DST2_4(2);
484

    
485
    RGB(3);
486
    DST2_4(3);
487
    DST1_4(3);
488
EPILOG(4)
489

    
490
PROLOG(yuv2rgb_c_4_ordered_dither, uint8_t)
491
    const uint8_t *d64= dither_8x8_73[y&7];
492
    const uint8_t *d128=dither_8x8_220[y&7];
493
    int acc;
494

    
495
#define DST1bpp4(i,o)                                             \
496
    Y = py_1[2*i];                                                \
497
    acc = r[Y+d128[0+o]] + g[Y+d64[0+o]] + b[Y+d128[0+o]];        \
498
    Y = py_1[2*i+1];                                              \
499
    acc |= (r[Y+d128[1+o]] + g[Y+d64[1+o]] + b[Y+d128[1+o]])<<4;  \
500
    dst_1[i]= acc;
501

    
502
#define DST2bpp4(i,o)                                             \
503
    Y = py_2[2*i];                                                \
504
    acc =  r[Y+d128[8+o]] + g[Y+d64[8+o]] + b[Y+d128[8+o]];       \
505
    Y = py_2[2*i+1];                                              \
506
    acc |=  (r[Y+d128[9+o]] + g[Y+d64[9+o]] + b[Y+d128[9+o]])<<4; \
507
    dst_2[i]= acc;
508

    
509

    
510
    RGB(0);
511
    DST1bpp4(0,0);
512
    DST2bpp4(0,0);
513

    
514
    RGB(1);
515
    DST2bpp4(1,2);
516
    DST1bpp4(1,2);
517

    
518
    RGB(2);
519
    DST1bpp4(2,4);
520
    DST2bpp4(2,4);
521

    
522
    RGB(3);
523
    DST2bpp4(3,6);
524
    DST1bpp4(3,6);
525
EPILOG(4)
526

    
527
// This is exactly the same code as yuv2rgb_c_32 except for the types of
528
// r, g, b, dst_1, dst_2
529
PROLOG(yuv2rgb_c_4b, uint8_t)
530
    RGB(0);
531
    DST1(0);
532
    DST2(0);
533

    
534
    RGB(1);
535
    DST2(1);
536
    DST1(1);
537

    
538
    RGB(2);
539
    DST1(2);
540
    DST2(2);
541

    
542
    RGB(3);
543
    DST2(3);
544
    DST1(3);
545
EPILOG(8)
546

    
547
PROLOG(yuv2rgb_c_4b_ordered_dither, uint8_t)
548
    const uint8_t *d64= dither_8x8_73[y&7];
549
    const uint8_t *d128=dither_8x8_220[y&7];
550

    
551
#define DST1bpp4b(i,o)                                                \
552
    Y = py_1[2*i];                                                    \
553
    dst_1[2*i]   = r[Y+d128[0+o]] + g[Y+d64[0+o]] + b[Y+d128[0+o]];   \
554
    Y = py_1[2*i+1];                                                  \
555
    dst_1[2*i+1] = r[Y+d128[1+o]] + g[Y+d64[1+o]] + b[Y+d128[1+o]];
556

    
557
#define DST2bpp4b(i,o)                                                \
558
    Y = py_2[2*i];                                                    \
559
    dst_2[2*i]   =  r[Y+d128[8+o]] + g[Y+d64[8+o]] + b[Y+d128[8+o]];  \
560
    Y = py_2[2*i+1];                                                  \
561
    dst_2[2*i+1] =  r[Y+d128[9+o]] + g[Y+d64[9+o]] + b[Y+d128[9+o]];
562

    
563

    
564
    RGB(0);
565
    DST1bpp4b(0,0);
566
    DST2bpp4b(0,0);
567

    
568
    RGB(1);
569
    DST2bpp4b(1,2);
570
    DST1bpp4b(1,2);
571

    
572
    RGB(2);
573
    DST1bpp4b(2,4);
574
    DST2bpp4b(2,4);
575

    
576
    RGB(3);
577
    DST2bpp4b(3,6);
578
    DST1bpp4b(3,6);
579
EPILOG(8)
580

    
581
PROLOG(yuv2rgb_c_1_ordered_dither, uint8_t)
582
        const uint8_t *d128=dither_8x8_220[y&7];
583
        char out_1=0, out_2=0;
584
        g= c->table_gU[128] + c->table_gV[128];
585

    
586
#define DST1bpp1(i,o)               \
587
    Y = py_1[2*i];                  \
588
    out_1+= out_1 + g[Y+d128[0+o]]; \
589
    Y = py_1[2*i+1];                \
590
    out_1+= out_1 + g[Y+d128[1+o]];
591

    
592
#define DST2bpp1(i,o)               \
593
    Y = py_2[2*i];                  \
594
    out_2+= out_2 + g[Y+d128[8+o]]; \
595
    Y = py_2[2*i+1];                \
596
    out_2+= out_2 + g[Y+d128[9+o]];
597

    
598
    DST1bpp1(0,0);
599
    DST2bpp1(0,0);
600

    
601
    DST2bpp1(1,2);
602
    DST1bpp1(1,2);
603

    
604
    DST1bpp1(2,4);
605
    DST2bpp1(2,4);
606

    
607
    DST2bpp1(3,6);
608
    DST1bpp1(3,6);
609

    
610
    dst_1[0]= out_1;
611
    dst_2[0]= out_2;
612
EPILOG(1)
613

    
614
SwsFunc yuv2rgb_get_func_ptr (SwsContext *c)
615
{
616
#if defined(HAVE_MMX2) || defined(HAVE_MMX)
617
    if (c->flags & SWS_CPU_CAPS_MMX2){
618
        switch(c->dstFormat){
619
        case PIX_FMT_RGB32:  return yuv420_rgb32_MMX2;
620
        case PIX_FMT_BGR24:  return yuv420_rgb24_MMX2;
621
        case PIX_FMT_BGR565: return yuv420_rgb16_MMX2;
622
        case PIX_FMT_BGR555: return yuv420_rgb15_MMX2;
623
        }
624
    }
625
    if (c->flags & SWS_CPU_CAPS_MMX){
626
        switch(c->dstFormat){
627
        case PIX_FMT_RGB32:  return yuv420_rgb32_MMX;
628
        case PIX_FMT_BGR24:  return yuv420_rgb24_MMX;
629
        case PIX_FMT_BGR565: return yuv420_rgb16_MMX;
630
        case PIX_FMT_BGR555: return yuv420_rgb15_MMX;
631
        }
632
    }
633
#endif
634
#ifdef HAVE_MLIB
635
    {
636
        SwsFunc t= yuv2rgb_init_mlib(c);
637
        if (t) return t;
638
    }
639
#endif
640
#ifdef HAVE_ALTIVEC
641
    if (c->flags & SWS_CPU_CAPS_ALTIVEC)
642
    {
643
        SwsFunc t = yuv2rgb_init_altivec(c);
644
        if (t) return t;
645
    }
646
#endif
647

    
648
#ifdef ARCH_BFIN
649
    if (c->flags & SWS_CPU_CAPS_BFIN)
650
    {
651
        SwsFunc t = ff_bfin_yuv2rgb_get_func_ptr (c);
652
        if (t) return t;
653
    }
654
#endif
655

    
656
    av_log(c, AV_LOG_WARNING, "No accelerated colorspace conversion found\n");
657

    
658
    switch(c->dstFormat){
659
    case PIX_FMT_BGR32:
660
    case PIX_FMT_RGB32: return yuv2rgb_c_32;
661
    case PIX_FMT_RGB24: return yuv2rgb_c_24_rgb;
662
    case PIX_FMT_BGR24: return yuv2rgb_c_24_bgr;
663
    case PIX_FMT_RGB565:
664
    case PIX_FMT_BGR565:
665
    case PIX_FMT_RGB555:
666
    case PIX_FMT_BGR555: return yuv2rgb_c_16;
667
    case PIX_FMT_RGB8:
668
    case PIX_FMT_BGR8:  return yuv2rgb_c_8_ordered_dither;
669
    case PIX_FMT_RGB4:
670
    case PIX_FMT_BGR4:  return yuv2rgb_c_4_ordered_dither;
671
    case PIX_FMT_RGB4_BYTE:
672
    case PIX_FMT_BGR4_BYTE:  return yuv2rgb_c_4b_ordered_dither;
673
    case PIX_FMT_MONOBLACK:  return yuv2rgb_c_1_ordered_dither;
674
    default:
675
        assert(0);
676
    }
677
    return NULL;
678
}
679

    
680
static int div_round (int dividend, int divisor)
681
{
682
    if (dividend > 0)
683
        return (dividend + (divisor>>1)) / divisor;
684
    else
685
        return -((-dividend + (divisor>>1)) / divisor);
686
}
687

    
688
int yuv2rgb_c_init_tables (SwsContext *c, const int inv_table[4], int fullRange, int brightness, int contrast, int saturation)
689
{
690
    const int isRgb = isBGR(c->dstFormat);
691
    const int bpp = fmt_depth(c->dstFormat);
692
    int i;
693
    uint8_t table_Y[1024];
694
    uint32_t *table_32 = 0;
695
    uint16_t *table_16 = 0;
696
    uint8_t *table_8 = 0;
697
    uint8_t *table_332 = 0;
698
    uint8_t *table_121 = 0;
699
    uint8_t *table_1 = 0;
700
    int entry_size = 0;
701
    void *table_r = 0, *table_g = 0, *table_b = 0;
702
    void *table_start;
703

    
704
    int64_t crv =  inv_table[0];
705
    int64_t cbu =  inv_table[1];
706
    int64_t cgu = -inv_table[2];
707
    int64_t cgv = -inv_table[3];
708
    int64_t cy  = 1<<16;
709
    int64_t oy  = 0;
710

    
711
//printf("%lld %lld %lld %lld %lld\n", cy, crv, cbu, cgu, cgv);
712
    if (!fullRange){
713
        cy= (cy*255) / 219;
714
        oy= 16<<16;
715
    }else{
716
        crv= (crv*224) / 255;
717
        cbu= (cbu*224) / 255;
718
        cgu= (cgu*224) / 255;
719
        cgv= (cgv*224) / 255;
720
    }
721

    
722
    cy = (cy *contrast             )>>16;
723
    crv= (crv*contrast * saturation)>>32;
724
    cbu= (cbu*contrast * saturation)>>32;
725
    cgu= (cgu*contrast * saturation)>>32;
726
    cgv= (cgv*contrast * saturation)>>32;
727
//printf("%lld %lld %lld %lld %lld\n", cy, crv, cbu, cgu, cgv);
728
    oy -= 256*brightness;
729

    
730
    for (i = 0; i < 1024; i++) {
731
        int j;
732

    
733
        j= (cy*(((i - 384)<<16) - oy) + (1<<31))>>32;
734
        j = (j < 0) ? 0 : ((j > 255) ? 255 : j);
735
        table_Y[i] = j;
736
    }
737

    
738
    switch (bpp) {
739
    case 32:
740
        table_start= table_32 = av_malloc ((197 + 2*682 + 256 + 132) * sizeof (uint32_t));
741

    
742
        entry_size = sizeof (uint32_t);
743
        table_r = table_32 + 197;
744
        table_b = table_32 + 197 + 685;
745
        table_g = table_32 + 197 + 2*682;
746

    
747
        for (i = -197; i < 256+197; i++)
748
            ((uint32_t *)table_r)[i] = table_Y[i+384] << (isRgb ? 16 : 0);
749
        for (i = -132; i < 256+132; i++)
750
            ((uint32_t *)table_g)[i] = table_Y[i+384] << 8;
751
        for (i = -232; i < 256+232; i++)
752
            ((uint32_t *)table_b)[i] = table_Y[i+384] << (isRgb ? 0 : 16);
753
        break;
754

    
755
    case 24:
756
        table_start= table_8 = av_malloc ((256 + 2*232) * sizeof (uint8_t));
757

    
758
        entry_size = sizeof (uint8_t);
759
        table_r = table_g = table_b = table_8 + 232;
760

    
761
        for (i = -232; i < 256+232; i++)
762
            ((uint8_t * )table_b)[i] = table_Y[i+384];
763
        break;
764

    
765
    case 15:
766
    case 16:
767
        table_start= table_16 = av_malloc ((197 + 2*682 + 256 + 132) * sizeof (uint16_t));
768

    
769
        entry_size = sizeof (uint16_t);
770
        table_r = table_16 + 197;
771
        table_b = table_16 + 197 + 685;
772
        table_g = table_16 + 197 + 2*682;
773

    
774
        for (i = -197; i < 256+197; i++) {
775
            int j = table_Y[i+384] >> 3;
776

    
777
            if (isRgb)
778
                j <<= ((bpp==16) ? 11 : 10);
779

    
780
            ((uint16_t *)table_r)[i] = j;
781
        }
782
        for (i = -132; i < 256+132; i++) {
783
            int j = table_Y[i+384] >> ((bpp==16) ? 2 : 3);
784

    
785
            ((uint16_t *)table_g)[i] = j << 5;
786
        }
787
        for (i = -232; i < 256+232; i++) {
788
            int j = table_Y[i+384] >> 3;
789

    
790
            if (!isRgb)
791
                j <<= ((bpp==16) ? 11 : 10);
792

    
793
            ((uint16_t *)table_b)[i] = j;
794
        }
795
        break;
796

    
797
    case 8:
798
        table_start= table_332 = av_malloc ((197 + 2*682 + 256 + 132) * sizeof (uint8_t));
799

    
800
        entry_size = sizeof (uint8_t);
801
        table_r = table_332 + 197;
802
        table_b = table_332 + 197 + 685;
803
        table_g = table_332 + 197 + 2*682;
804

    
805
        for (i = -197; i < 256+197; i++) {
806
            int j = (table_Y[i+384 - 16] + 18)/36;
807

    
808
            if (isRgb)
809
                j <<= 5;
810

    
811
            ((uint8_t *)table_r)[i] = j;
812
        }
813
        for (i = -132; i < 256+132; i++) {
814
            int j = (table_Y[i+384 - 16] + 18)/36;
815

    
816
            if (!isRgb)
817
                j <<= 1;
818

    
819
            ((uint8_t *)table_g)[i] = j << 2;
820
        }
821
        for (i = -232; i < 256+232; i++) {
822
            int j = (table_Y[i+384 - 37] + 43)/85;
823

    
824
            if (!isRgb)
825
                j <<= 6;
826

    
827
            ((uint8_t *)table_b)[i] = j;
828
        }
829
        break;
830
    case 4:
831
    case 4|128:
832
        table_start= table_121 = av_malloc ((197 + 2*682 + 256 + 132) * sizeof (uint8_t));
833

    
834
        entry_size = sizeof (uint8_t);
835
        table_r = table_121 + 197;
836
        table_b = table_121 + 197 + 685;
837
        table_g = table_121 + 197 + 2*682;
838

    
839
        for (i = -197; i < 256+197; i++) {
840
            int j = table_Y[i+384 - 110] >> 7;
841

    
842
            if (isRgb)
843
                j <<= 3;
844

    
845
            ((uint8_t *)table_r)[i] = j;
846
        }
847
        for (i = -132; i < 256+132; i++) {
848
            int j = (table_Y[i+384 - 37]+ 43)/85;
849

    
850
            ((uint8_t *)table_g)[i] = j << 1;
851
        }
852
        for (i = -232; i < 256+232; i++) {
853
            int j =table_Y[i+384 - 110] >> 7;
854

    
855
            if (!isRgb)
856
                j <<= 3;
857

    
858
            ((uint8_t *)table_b)[i] = j;
859
        }
860
        break;
861

    
862
    case 1:
863
        table_start= table_1 = av_malloc (256*2 * sizeof (uint8_t));
864

    
865
        entry_size = sizeof (uint8_t);
866
        table_g = table_1;
867
        table_r = table_b = NULL;
868

    
869
        for (i = 0; i < 256+256; i++) {
870
            int j = table_Y[i + 384 - 110]>>7;
871

    
872
            ((uint8_t *)table_g)[i] = j;
873
        }
874
        break;
875

    
876
    default:
877
        table_start= NULL;
878
        av_log(c, AV_LOG_ERROR, "%ibpp not supported by yuv2rgb\n", bpp);
879
        //free mem?
880
        return -1;
881
    }
882

    
883
    for (i = 0; i < 256; i++) {
884
        c->table_rV[i] = (uint8_t *)table_r + entry_size * div_round (crv * (i-128), 76309);
885
        c->table_gU[i] = (uint8_t *)table_g + entry_size * div_round (cgu * (i-128), 76309);
886
        c->table_gV[i] = entry_size * div_round (cgv * (i-128), 76309);
887
        c->table_bU[i] = (uint8_t *)table_b + entry_size * div_round (cbu * (i-128), 76309);
888
    }
889

    
890
    av_free(c->yuvTable);
891
    c->yuvTable= table_start;
892
    return 0;
893
}