Statistics
| Branch: | Revision:

ffmpeg / libswscale / yuv2rgb.c @ 86593486

History | View | Annotate | Download (24.4 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_VIS
43
#include "yuv2rgb_vis.c"
44
#endif
45

    
46
#ifdef HAVE_MLIB
47
#include "yuv2rgb_mlib.c"
48
#endif
49

    
50
#define DITHER1XBPP // only for mmx
51

    
52
const uint8_t  __attribute__((aligned(8))) dither_2x2_4[2][8]={
53
{  1,   3,   1,   3,   1,   3,   1,   3, },
54
{  2,   0,   2,   0,   2,   0,   2,   0, },
55
};
56

    
57
const uint8_t  __attribute__((aligned(8))) dither_2x2_8[2][8]={
58
{  6,   2,   6,   2,   6,   2,   6,   2, },
59
{  0,   4,   0,   4,   0,   4,   0,   4, },
60
};
61

    
62
const uint8_t  __attribute__((aligned(8))) dither_8x8_32[8][8]={
63
{ 17,   9,  23,  15,  16,   8,  22,  14, },
64
{  5,  29,   3,  27,   4,  28,   2,  26, },
65
{ 21,  13,  19,  11,  20,  12,  18,  10, },
66
{  0,  24,   6,  30,   1,  25,   7,  31, },
67
{ 16,   8,  22,  14,  17,   9,  23,  15, },
68
{  4,  28,   2,  26,   5,  29,   3,  27, },
69
{ 20,  12,  18,  10,  21,  13,  19,  11, },
70
{  1,  25,   7,  31,   0,  24,   6,  30, },
71
};
72

    
73
#if 0
74
const uint8_t  __attribute__((aligned(8))) dither_8x8_64[8][8]={
75
{  0,  48,  12,  60,   3,  51,  15,  63, },
76
{ 32,  16,  44,  28,  35,  19,  47,  31, },
77
{  8,  56,   4,  52,  11,  59,   7,  55, },
78
{ 40,  24,  36,  20,  43,  27,  39,  23, },
79
{  2,  50,  14,  62,   1,  49,  13,  61, },
80
{ 34,  18,  46,  30,  33,  17,  45,  29, },
81
{ 10,  58,   6,  54,   9,  57,   5,  53, },
82
{ 42,  26,  38,  22,  41,  25,  37,  21, },
83
};
84
#endif
85

    
86
const uint8_t  __attribute__((aligned(8))) dither_8x8_73[8][8]={
87
{  0,  55,  14,  68,   3,  58,  17,  72, },
88
{ 37,  18,  50,  32,  40,  22,  54,  35, },
89
{  9,  64,   5,  59,  13,  67,   8,  63, },
90
{ 46,  27,  41,  23,  49,  31,  44,  26, },
91
{  2,  57,  16,  71,   1,  56,  15,  70, },
92
{ 39,  21,  52,  34,  38,  19,  51,  33, },
93
{ 11,  66,   7,  62,  10,  65,   6,  60, },
94
{ 48,  30,  43,  25,  47,  29,  42,  24, },
95
};
96

    
97
#if 0
98
const uint8_t  __attribute__((aligned(8))) dither_8x8_128[8][8]={
99
{ 68,  36,  92,  60,  66,  34,  90,  58, },
100
{ 20, 116,  12, 108,  18, 114,  10, 106, },
101
{ 84,  52,  76,  44,  82,  50,  74,  42, },
102
{  0,  96,  24, 120,   6, 102,  30, 126, },
103
{ 64,  32,  88,  56,  70,  38,  94,  62, },
104
{ 16, 112,   8, 104,  22, 118,  14, 110, },
105
{ 80,  48,  72,  40,  86,  54,  78,  46, },
106
{  4, 100,  28, 124,   2,  98,  26, 122, },
107
};
108
#endif
109

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

    
159
#ifdef HAVE_MMX
160

    
161
/* hope these constant values are cache line aligned */
162
DECLARE_ASM_CONST(8, uint64_t, mmx_00ffw)   = 0x00ff00ff00ff00ffULL;
163
DECLARE_ASM_CONST(8, uint64_t, mmx_redmask) = 0xf8f8f8f8f8f8f8f8ULL;
164
DECLARE_ASM_CONST(8, uint64_t, mmx_grnmask) = 0xfcfcfcfcfcfcfcfcULL;
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
#undef HAVE_MMX
174

    
175
//MMX versions
176
#undef RENAME
177
#define HAVE_MMX
178
#undef HAVE_MMX2
179
#undef HAVE_3DNOW
180
#define RENAME(a) a ## _MMX
181
#include "yuv2rgb_template.c"
182

    
183
//MMX2 versions
184
#undef RENAME
185
#define HAVE_MMX
186
#define HAVE_MMX2
187
#undef HAVE_3DNOW
188
#define RENAME(a) a ## _MMX2
189
#include "yuv2rgb_template.c"
190

    
191
#endif /* defined(ARCH_X86) */
192

    
193
const int32_t Inverse_Table_6_9[8][4] = {
194
    {117504, 138453, 13954, 34903}, /* no sequence_display_extension */
195
    {117504, 138453, 13954, 34903}, /* ITU-R Rec. 709 (1990) */
196
    {104597, 132201, 25675, 53279}, /* unspecified */
197
    {104597, 132201, 25675, 53279}, /* reserved */
198
    {104448, 132798, 24759, 53109}, /* FCC */
199
    {104597, 132201, 25675, 53279}, /* ITU-R Rec. 624-4 System B, G */
200
    {104597, 132201, 25675, 53279}, /* SMPTE 170M */
201
    {117579, 136230, 16907, 35559}  /* SMPTE 240M (1987) */
202
};
203

    
204
#define RGB(i)                                      \
205
    U = pu[i];                                      \
206
    V = pv[i];                                      \
207
    r = (void *)c->table_rV[V];                     \
208
    g = (void *)(c->table_gU[U] + c->table_gV[V]);  \
209
    b = (void *)c->table_bU[U];
210

    
211
#define DST1(i)                         \
212
    Y = py_1[2*i];                      \
213
    dst_1[2*i] = r[Y] + g[Y] + b[Y];    \
214
    Y = py_1[2*i+1];                    \
215
    dst_1[2*i+1] = r[Y] + g[Y] + b[Y];
216

    
217
#define DST2(i)                         \
218
    Y = py_2[2*i];                      \
219
    dst_2[2*i] = r[Y] + g[Y] + b[Y];    \
220
    Y = py_2[2*i+1];                    \
221
    dst_2[2*i+1] = r[Y] + g[Y] + b[Y];
222

    
223
#define DST1RGB(i)                                                \
224
    Y = py_1[2*i];                                                \
225
    dst_1[6*i] = r[Y]; dst_1[6*i+1] = g[Y]; dst_1[6*i+2] = b[Y];  \
226
    Y = py_1[2*i+1];                                              \
227
    dst_1[6*i+3] = r[Y]; dst_1[6*i+4] = g[Y]; dst_1[6*i+5] = b[Y];
228

    
229
#define DST2RGB(i)                                                \
230
    Y = py_2[2*i];                                                \
231
    dst_2[6*i] = r[Y]; dst_2[6*i+1] = g[Y]; dst_2[6*i+2] = b[Y];  \
232
    Y = py_2[2*i+1];                                              \
233
    dst_2[6*i+3] = r[Y]; dst_2[6*i+4] = g[Y]; dst_2[6*i+5] = b[Y];
234

    
235
#define DST1BGR(i)                                                \
236
    Y = py_1[2*i];                                                \
237
    dst_1[6*i] = b[Y]; dst_1[6*i+1] = g[Y]; dst_1[6*i+2] = r[Y];  \
238
    Y = py_1[2*i+1];                                              \
239
    dst_1[6*i+3] = b[Y]; dst_1[6*i+4] = g[Y]; dst_1[6*i+5] = r[Y];
240

    
241
#define DST2BGR(i)                                                \
242
    Y = py_2[2*i];                                                \
243
    dst_2[6*i] = b[Y]; dst_2[6*i+1] = g[Y]; dst_2[6*i+2] = r[Y];  \
244
    Y = py_2[2*i+1];                                              \
245
    dst_2[6*i+3] = b[Y]; dst_2[6*i+4] = g[Y]; dst_2[6*i+5] = r[Y];
246

    
247
#define PROLOG(func_name, dst_type) \
248
static int func_name(SwsContext *c, uint8_t* src[], int srcStride[], int srcSliceY, \
249
                     int srcSliceH, uint8_t* dst[], int dstStride[]){\
250
    int y;\
251
\
252
    if (c->srcFormat == PIX_FMT_YUV422P){\
253
        srcStride[1] *= 2;\
254
        srcStride[2] *= 2;\
255
    }\
256
    for (y=0; y<srcSliceH; y+=2){\
257
        dst_type *dst_1= (dst_type*)(dst[0] + (y+srcSliceY  )*dstStride[0]);\
258
        dst_type *dst_2= (dst_type*)(dst[0] + (y+srcSliceY+1)*dstStride[0]);\
259
        dst_type av_unused *r, *b;\
260
        dst_type *g;\
261
        uint8_t *py_1= src[0] + y*srcStride[0];\
262
        uint8_t *py_2= py_1 + srcStride[0];\
263
        uint8_t *pu= src[1] + (y>>1)*srcStride[1];\
264
        uint8_t *pv= src[2] + (y>>1)*srcStride[2];\
265
        unsigned int h_size= c->dstW>>3;\
266
        while (h_size--) {\
267
            int av_unused U, V;\
268
            int Y;\
269

    
270
#define EPILOG1(dst_delta)\
271
            pu += 4;\
272
            pv += 4;\
273
            py_1 += 8;\
274
            py_2 += 8;\
275
            dst_1 += dst_delta;\
276
            dst_2 += dst_delta;\
277
        }\
278
        if (c->dstW & 4) {\
279
            int av_unused Y, U, V;\
280

    
281
#define EPILOG2()\
282
        }\
283
    }\
284
    return srcSliceH;\
285
}
286

    
287
#define EPILOG(dst_delta)\
288
    EPILOG1(dst_delta)\
289
    EPILOG2()
290

    
291
PROLOG(yuv2rgb_c_32, uint32_t)
292
    RGB(0);
293
    DST1(0);
294
    DST2(0);
295

    
296
    RGB(1);
297
    DST2(1);
298
    DST1(1);
299

    
300
    RGB(2);
301
    DST1(2);
302
    DST2(2);
303

    
304
    RGB(3);
305
    DST2(3);
306
    DST1(3);
307
EPILOG1(8)
308
    RGB(0);
309
    DST1(0);
310
    DST2(0);
311

    
312
    RGB(1);
313
    DST2(1);
314
    DST1(1);
315
EPILOG2()
316

    
317
PROLOG(yuv2rgb_c_24_rgb, uint8_t)
318
    RGB(0);
319
    DST1RGB(0);
320
    DST2RGB(0);
321

    
322
    RGB(1);
323
    DST2RGB(1);
324
    DST1RGB(1);
325

    
326
    RGB(2);
327
    DST1RGB(2);
328
    DST2RGB(2);
329

    
330
    RGB(3);
331
    DST2RGB(3);
332
    DST1RGB(3);
333
EPILOG1(24)
334
    RGB(0);
335
    DST1RGB(0);
336
    DST2RGB(0);
337

    
338
    RGB(1);
339
    DST2RGB(1);
340
    DST1RGB(1);
341
EPILOG2()
342

    
343
// only trivial mods from yuv2rgb_c_24_rgb
344
PROLOG(yuv2rgb_c_24_bgr, uint8_t)
345
    RGB(0);
346
    DST1BGR(0);
347
    DST2BGR(0);
348

    
349
    RGB(1);
350
    DST2BGR(1);
351
    DST1BGR(1);
352

    
353
    RGB(2);
354
    DST1BGR(2);
355
    DST2BGR(2);
356

    
357
    RGB(3);
358
    DST2BGR(3);
359
    DST1BGR(3);
360
EPILOG1(24)
361
    RGB(0);
362
    DST1BGR(0);
363
    DST2BGR(0);
364

    
365
    RGB(1);
366
    DST2BGR(1);
367
    DST1BGR(1);
368
EPILOG2()
369

    
370
// This is exactly the same code as yuv2rgb_c_32 except for the types of
371
// r, g, b, dst_1, dst_2
372
PROLOG(yuv2rgb_c_16, uint16_t)
373
    RGB(0);
374
    DST1(0);
375
    DST2(0);
376

    
377
    RGB(1);
378
    DST2(1);
379
    DST1(1);
380

    
381
    RGB(2);
382
    DST1(2);
383
    DST2(2);
384

    
385
    RGB(3);
386
    DST2(3);
387
    DST1(3);
388
EPILOG(8)
389

    
390
// This is exactly the same code as yuv2rgb_c_32 except for the types of
391
// r, g, b, dst_1, dst_2
392
PROLOG(yuv2rgb_c_8, uint8_t)
393
    RGB(0);
394
    DST1(0);
395
    DST2(0);
396

    
397
    RGB(1);
398
    DST2(1);
399
    DST1(1);
400

    
401
    RGB(2);
402
    DST1(2);
403
    DST2(2);
404

    
405
    RGB(3);
406
    DST2(3);
407
    DST1(3);
408
EPILOG(8)
409

    
410
// r, g, b, dst_1, dst_2
411
PROLOG(yuv2rgb_c_8_ordered_dither, uint8_t)
412
    const uint8_t *d32= dither_8x8_32[y&7];
413
    const uint8_t *d64= dither_8x8_73[y&7];
414
#define DST1bpp8(i,o)                                               \
415
    Y = py_1[2*i];                                                  \
416
    dst_1[2*i]   = r[Y+d32[0+o]] + g[Y+d32[0+o]] + b[Y+d64[0+o]];   \
417
    Y = py_1[2*i+1];                                                \
418
    dst_1[2*i+1] = r[Y+d32[1+o]] + g[Y+d32[1+o]] + b[Y+d64[1+o]];
419

    
420
#define DST2bpp8(i,o)                                               \
421
    Y = py_2[2*i];                                                  \
422
    dst_2[2*i]   =  r[Y+d32[8+o]] + g[Y+d32[8+o]] + b[Y+d64[8+o]];  \
423
    Y = py_2[2*i+1];                                                \
424
    dst_2[2*i+1] =  r[Y+d32[9+o]] + g[Y+d32[9+o]] + b[Y+d64[9+o]];
425

    
426

    
427
    RGB(0);
428
    DST1bpp8(0,0);
429
    DST2bpp8(0,0);
430

    
431
    RGB(1);
432
    DST2bpp8(1,2);
433
    DST1bpp8(1,2);
434

    
435
    RGB(2);
436
    DST1bpp8(2,4);
437
    DST2bpp8(2,4);
438

    
439
    RGB(3);
440
    DST2bpp8(3,6);
441
    DST1bpp8(3,6);
442
EPILOG(8)
443

    
444

    
445
// This is exactly the same code as yuv2rgb_c_32 except for the types of
446
// r, g, b, dst_1, dst_2
447
PROLOG(yuv2rgb_c_4, uint8_t)
448
    int acc;
449
#define DST1_4(i)                   \
450
    Y = py_1[2*i];                  \
451
    acc = r[Y] + g[Y] + b[Y];       \
452
    Y = py_1[2*i+1];                \
453
    acc |= (r[Y] + g[Y] + b[Y])<<4; \
454
    dst_1[i] = acc;
455

    
456
#define DST2_4(i)                   \
457
    Y = py_2[2*i];                  \
458
    acc = r[Y] + g[Y] + b[Y];       \
459
    Y = py_2[2*i+1];                \
460
    acc |= (r[Y] + g[Y] + b[Y])<<4; \
461
    dst_2[i] = acc;
462

    
463
    RGB(0);
464
    DST1_4(0);
465
    DST2_4(0);
466

    
467
    RGB(1);
468
    DST2_4(1);
469
    DST1_4(1);
470

    
471
    RGB(2);
472
    DST1_4(2);
473
    DST2_4(2);
474

    
475
    RGB(3);
476
    DST2_4(3);
477
    DST1_4(3);
478
EPILOG(4)
479

    
480
PROLOG(yuv2rgb_c_4_ordered_dither, uint8_t)
481
    const uint8_t *d64= dither_8x8_73[y&7];
482
    const uint8_t *d128=dither_8x8_220[y&7];
483
    int acc;
484

    
485
#define DST1bpp4(i,o)                                             \
486
    Y = py_1[2*i];                                                \
487
    acc = r[Y+d128[0+o]] + g[Y+d64[0+o]] + b[Y+d128[0+o]];        \
488
    Y = py_1[2*i+1];                                              \
489
    acc |= (r[Y+d128[1+o]] + g[Y+d64[1+o]] + b[Y+d128[1+o]])<<4;  \
490
    dst_1[i]= acc;
491

    
492
#define DST2bpp4(i,o)                                             \
493
    Y = py_2[2*i];                                                \
494
    acc =  r[Y+d128[8+o]] + g[Y+d64[8+o]] + b[Y+d128[8+o]];       \
495
    Y = py_2[2*i+1];                                              \
496
    acc |=  (r[Y+d128[9+o]] + g[Y+d64[9+o]] + b[Y+d128[9+o]])<<4; \
497
    dst_2[i]= acc;
498

    
499

    
500
    RGB(0);
501
    DST1bpp4(0,0);
502
    DST2bpp4(0,0);
503

    
504
    RGB(1);
505
    DST2bpp4(1,2);
506
    DST1bpp4(1,2);
507

    
508
    RGB(2);
509
    DST1bpp4(2,4);
510
    DST2bpp4(2,4);
511

    
512
    RGB(3);
513
    DST2bpp4(3,6);
514
    DST1bpp4(3,6);
515
EPILOG(4)
516

    
517
// This is exactly the same code as yuv2rgb_c_32 except for the types of
518
// r, g, b, dst_1, dst_2
519
PROLOG(yuv2rgb_c_4b, uint8_t)
520
    RGB(0);
521
    DST1(0);
522
    DST2(0);
523

    
524
    RGB(1);
525
    DST2(1);
526
    DST1(1);
527

    
528
    RGB(2);
529
    DST1(2);
530
    DST2(2);
531

    
532
    RGB(3);
533
    DST2(3);
534
    DST1(3);
535
EPILOG(8)
536

    
537
PROLOG(yuv2rgb_c_4b_ordered_dither, uint8_t)
538
    const uint8_t *d64= dither_8x8_73[y&7];
539
    const uint8_t *d128=dither_8x8_220[y&7];
540

    
541
#define DST1bpp4b(i,o)                                                \
542
    Y = py_1[2*i];                                                    \
543
    dst_1[2*i]   = r[Y+d128[0+o]] + g[Y+d64[0+o]] + b[Y+d128[0+o]];   \
544
    Y = py_1[2*i+1];                                                  \
545
    dst_1[2*i+1] = r[Y+d128[1+o]] + g[Y+d64[1+o]] + b[Y+d128[1+o]];
546

    
547
#define DST2bpp4b(i,o)                                                \
548
    Y = py_2[2*i];                                                    \
549
    dst_2[2*i]   =  r[Y+d128[8+o]] + g[Y+d64[8+o]] + b[Y+d128[8+o]];  \
550
    Y = py_2[2*i+1];                                                  \
551
    dst_2[2*i+1] =  r[Y+d128[9+o]] + g[Y+d64[9+o]] + b[Y+d128[9+o]];
552

    
553

    
554
    RGB(0);
555
    DST1bpp4b(0,0);
556
    DST2bpp4b(0,0);
557

    
558
    RGB(1);
559
    DST2bpp4b(1,2);
560
    DST1bpp4b(1,2);
561

    
562
    RGB(2);
563
    DST1bpp4b(2,4);
564
    DST2bpp4b(2,4);
565

    
566
    RGB(3);
567
    DST2bpp4b(3,6);
568
    DST1bpp4b(3,6);
569
EPILOG(8)
570

    
571
PROLOG(yuv2rgb_c_1_ordered_dither, uint8_t)
572
        const uint8_t *d128=dither_8x8_220[y&7];
573
        char out_1=0, out_2=0;
574
        g= c->table_gU[128] + c->table_gV[128];
575

    
576
#define DST1bpp1(i,o)               \
577
    Y = py_1[2*i];                  \
578
    out_1+= out_1 + g[Y+d128[0+o]]; \
579
    Y = py_1[2*i+1];                \
580
    out_1+= out_1 + g[Y+d128[1+o]];
581

    
582
#define DST2bpp1(i,o)               \
583
    Y = py_2[2*i];                  \
584
    out_2+= out_2 + g[Y+d128[8+o]]; \
585
    Y = py_2[2*i+1];                \
586
    out_2+= out_2 + g[Y+d128[9+o]];
587

    
588
    DST1bpp1(0,0);
589
    DST2bpp1(0,0);
590

    
591
    DST2bpp1(1,2);
592
    DST1bpp1(1,2);
593

    
594
    DST1bpp1(2,4);
595
    DST2bpp1(2,4);
596

    
597
    DST2bpp1(3,6);
598
    DST1bpp1(3,6);
599

    
600
    dst_1[0]= out_1;
601
    dst_2[0]= out_2;
602
EPILOG(1)
603

    
604
SwsFunc yuv2rgb_get_func_ptr (SwsContext *c)
605
{
606
#if defined(HAVE_MMX2) || defined(HAVE_MMX)
607
    if (c->flags & SWS_CPU_CAPS_MMX2){
608
        switch(c->dstFormat){
609
        case PIX_FMT_RGB32:  return yuv420_rgb32_MMX2;
610
        case PIX_FMT_BGR24:  return yuv420_rgb24_MMX2;
611
        case PIX_FMT_BGR565: return yuv420_rgb16_MMX2;
612
        case PIX_FMT_BGR555: return yuv420_rgb15_MMX2;
613
        }
614
    }
615
    if (c->flags & SWS_CPU_CAPS_MMX){
616
        switch(c->dstFormat){
617
        case PIX_FMT_RGB32:  return yuv420_rgb32_MMX;
618
        case PIX_FMT_BGR24:  return yuv420_rgb24_MMX;
619
        case PIX_FMT_BGR565: return yuv420_rgb16_MMX;
620
        case PIX_FMT_BGR555: return yuv420_rgb15_MMX;
621
        }
622
    }
623
#endif
624
#ifdef HAVE_VIS
625
    {
626
        SwsFunc t= yuv2rgb_init_vis(c);
627
        if (t) return t;
628
    }
629
#endif
630
#ifdef HAVE_MLIB
631
    {
632
        SwsFunc t= yuv2rgb_init_mlib(c);
633
        if (t) return t;
634
    }
635
#endif
636
#ifdef HAVE_ALTIVEC
637
    if (c->flags & SWS_CPU_CAPS_ALTIVEC)
638
    {
639
        SwsFunc t = yuv2rgb_init_altivec(c);
640
        if (t) return t;
641
    }
642
#endif
643

    
644
#ifdef ARCH_BFIN
645
    if (c->flags & SWS_CPU_CAPS_BFIN)
646
    {
647
        SwsFunc t = ff_bfin_yuv2rgb_get_func_ptr (c);
648
        if (t) return t;
649
    }
650
#endif
651

    
652
    av_log(c, AV_LOG_WARNING, "No accelerated colorspace conversion found\n");
653

    
654
    switch(c->dstFormat){
655
    case PIX_FMT_BGR32:
656
    case PIX_FMT_RGB32: return yuv2rgb_c_32;
657
    case PIX_FMT_RGB24: return yuv2rgb_c_24_rgb;
658
    case PIX_FMT_BGR24: return yuv2rgb_c_24_bgr;
659
    case PIX_FMT_RGB565:
660
    case PIX_FMT_BGR565:
661
    case PIX_FMT_RGB555:
662
    case PIX_FMT_BGR555: return yuv2rgb_c_16;
663
    case PIX_FMT_RGB8:
664
    case PIX_FMT_BGR8:  return yuv2rgb_c_8_ordered_dither;
665
    case PIX_FMT_RGB4:
666
    case PIX_FMT_BGR4:  return yuv2rgb_c_4_ordered_dither;
667
    case PIX_FMT_RGB4_BYTE:
668
    case PIX_FMT_BGR4_BYTE:  return yuv2rgb_c_4b_ordered_dither;
669
    case PIX_FMT_MONOBLACK:  return yuv2rgb_c_1_ordered_dither;
670
    default:
671
        assert(0);
672
    }
673
    return NULL;
674
}
675

    
676
static int div_round (int dividend, int divisor)
677
{
678
    if (dividend > 0)
679
        return (dividend + (divisor>>1)) / divisor;
680
    else
681
        return -((-dividend + (divisor>>1)) / divisor);
682
}
683

    
684
int yuv2rgb_c_init_tables (SwsContext *c, const int inv_table[4], int fullRange, int brightness, int contrast, int saturation)
685
{
686
    const int isRgb = isBGR(c->dstFormat);
687
    const int bpp = fmt_depth(c->dstFormat);
688
    int i;
689
    uint8_t table_Y[1024];
690
    uint32_t *table_32 = 0;
691
    uint16_t *table_16 = 0;
692
    uint8_t *table_8 = 0;
693
    uint8_t *table_332 = 0;
694
    uint8_t *table_121 = 0;
695
    uint8_t *table_1 = 0;
696
    int entry_size = 0;
697
    void *table_r = 0, *table_g = 0, *table_b = 0;
698
    void *table_start;
699

    
700
    int64_t crv =  inv_table[0];
701
    int64_t cbu =  inv_table[1];
702
    int64_t cgu = -inv_table[2];
703
    int64_t cgv = -inv_table[3];
704
    int64_t cy  = 1<<16;
705
    int64_t oy  = 0;
706

    
707
//printf("%lld %lld %lld %lld %lld\n", cy, crv, cbu, cgu, cgv);
708
    if (!fullRange){
709
        cy= (cy*255) / 219;
710
        oy= 16<<16;
711
    }else{
712
        crv= (crv*224) / 255;
713
        cbu= (cbu*224) / 255;
714
        cgu= (cgu*224) / 255;
715
        cgv= (cgv*224) / 255;
716
    }
717

    
718
    cy = (cy *contrast             )>>16;
719
    crv= (crv*contrast * saturation)>>32;
720
    cbu= (cbu*contrast * saturation)>>32;
721
    cgu= (cgu*contrast * saturation)>>32;
722
    cgv= (cgv*contrast * saturation)>>32;
723
//printf("%lld %lld %lld %lld %lld\n", cy, crv, cbu, cgu, cgv);
724
    oy -= 256*brightness;
725

    
726
    for (i = 0; i < 1024; i++) {
727
        int j;
728

    
729
        j= (cy*(((i - 384)<<16) - oy) + (1<<31))>>32;
730
        j = (j < 0) ? 0 : ((j > 255) ? 255 : j);
731
        table_Y[i] = j;
732
    }
733

    
734
    switch (bpp) {
735
    case 32:
736
        table_start= table_32 = av_malloc ((197 + 2*682 + 256 + 132) * sizeof (uint32_t));
737

    
738
        entry_size = sizeof (uint32_t);
739
        table_r = table_32 + 197;
740
        table_b = table_32 + 197 + 685;
741
        table_g = table_32 + 197 + 2*682;
742

    
743
        for (i = -197; i < 256+197; i++)
744
            ((uint32_t *)table_r)[i] = table_Y[i+384] << (isRgb ? 16 : 0);
745
        for (i = -132; i < 256+132; i++)
746
            ((uint32_t *)table_g)[i] = table_Y[i+384] << 8;
747
        for (i = -232; i < 256+232; i++)
748
            ((uint32_t *)table_b)[i] = table_Y[i+384] << (isRgb ? 0 : 16);
749
        break;
750

    
751
    case 24:
752
        table_start= table_8 = av_malloc ((256 + 2*232) * sizeof (uint8_t));
753

    
754
        entry_size = sizeof (uint8_t);
755
        table_r = table_g = table_b = table_8 + 232;
756

    
757
        for (i = -232; i < 256+232; i++)
758
            ((uint8_t * )table_b)[i] = table_Y[i+384];
759
        break;
760

    
761
    case 15:
762
    case 16:
763
        table_start= table_16 = av_malloc ((197 + 2*682 + 256 + 132) * sizeof (uint16_t));
764

    
765
        entry_size = sizeof (uint16_t);
766
        table_r = table_16 + 197;
767
        table_b = table_16 + 197 + 685;
768
        table_g = table_16 + 197 + 2*682;
769

    
770
        for (i = -197; i < 256+197; i++) {
771
            int j = table_Y[i+384] >> 3;
772

    
773
            if (isRgb)
774
                j <<= ((bpp==16) ? 11 : 10);
775

    
776
            ((uint16_t *)table_r)[i] = j;
777
        }
778
        for (i = -132; i < 256+132; i++) {
779
            int j = table_Y[i+384] >> ((bpp==16) ? 2 : 3);
780

    
781
            ((uint16_t *)table_g)[i] = j << 5;
782
        }
783
        for (i = -232; i < 256+232; i++) {
784
            int j = table_Y[i+384] >> 3;
785

    
786
            if (!isRgb)
787
                j <<= ((bpp==16) ? 11 : 10);
788

    
789
            ((uint16_t *)table_b)[i] = j;
790
        }
791
        break;
792

    
793
    case 8:
794
        table_start= table_332 = av_malloc ((197 + 2*682 + 256 + 132) * sizeof (uint8_t));
795

    
796
        entry_size = sizeof (uint8_t);
797
        table_r = table_332 + 197;
798
        table_b = table_332 + 197 + 685;
799
        table_g = table_332 + 197 + 2*682;
800

    
801
        for (i = -197; i < 256+197; i++) {
802
            int j = (table_Y[i+384 - 16] + 18)/36;
803

    
804
            if (isRgb)
805
                j <<= 5;
806

    
807
            ((uint8_t *)table_r)[i] = j;
808
        }
809
        for (i = -132; i < 256+132; i++) {
810
            int j = (table_Y[i+384 - 16] + 18)/36;
811

    
812
            if (!isRgb)
813
                j <<= 1;
814

    
815
            ((uint8_t *)table_g)[i] = j << 2;
816
        }
817
        for (i = -232; i < 256+232; i++) {
818
            int j = (table_Y[i+384 - 37] + 43)/85;
819

    
820
            if (!isRgb)
821
                j <<= 6;
822

    
823
            ((uint8_t *)table_b)[i] = j;
824
        }
825
        break;
826
    case 4:
827
    case 4|128:
828
        table_start= table_121 = av_malloc ((197 + 2*682 + 256 + 132) * sizeof (uint8_t));
829

    
830
        entry_size = sizeof (uint8_t);
831
        table_r = table_121 + 197;
832
        table_b = table_121 + 197 + 685;
833
        table_g = table_121 + 197 + 2*682;
834

    
835
        for (i = -197; i < 256+197; i++) {
836
            int j = table_Y[i+384 - 110] >> 7;
837

    
838
            if (isRgb)
839
                j <<= 3;
840

    
841
            ((uint8_t *)table_r)[i] = j;
842
        }
843
        for (i = -132; i < 256+132; i++) {
844
            int j = (table_Y[i+384 - 37]+ 43)/85;
845

    
846
            ((uint8_t *)table_g)[i] = j << 1;
847
        }
848
        for (i = -232; i < 256+232; i++) {
849
            int j =table_Y[i+384 - 110] >> 7;
850

    
851
            if (!isRgb)
852
                j <<= 3;
853

    
854
            ((uint8_t *)table_b)[i] = j;
855
        }
856
        break;
857

    
858
    case 1:
859
        table_start= table_1 = av_malloc (256*2 * sizeof (uint8_t));
860

    
861
        entry_size = sizeof (uint8_t);
862
        table_g = table_1;
863
        table_r = table_b = NULL;
864

    
865
        for (i = 0; i < 256+256; i++) {
866
            int j = table_Y[i + 384 - 110]>>7;
867

    
868
            ((uint8_t *)table_g)[i] = j;
869
        }
870
        break;
871

    
872
    default:
873
        table_start= NULL;
874
        av_log(c, AV_LOG_ERROR, "%ibpp not supported by yuv2rgb\n", bpp);
875
        //free mem?
876
        return -1;
877
    }
878

    
879
    for (i = 0; i < 256; i++) {
880
        c->table_rV[i] = (uint8_t *)table_r + entry_size * div_round (crv * (i-128), 76309);
881
        c->table_gU[i] = (uint8_t *)table_g + entry_size * div_round (cgu * (i-128), 76309);
882
        c->table_gV[i] = entry_size * div_round (cgv * (i-128), 76309);
883
        c->table_bU[i] = (uint8_t *)table_b + entry_size * div_round (cbu * (i-128), 76309);
884
    }
885

    
886
    av_free(c->yuvTable);
887
    c->yuvTable= table_start;
888
    return 0;
889
}