CoCalc -- output.c

05. Matplotlib / ffmpeg-3.0 / libswscale / output.c
⁵²⁸⁶⁸ views
1
/*
2
 * Copyright (C) 2001-2012 Michael Niedermayer <[email protected]>
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
#undef output_pixel
184

185
#define output_pixel(pos, val) \
186
    if (big_endian) { \
187
        AV_WB16(pos, av_clip_uintp2(val >> shift, output_bits)); \
188
    } else { \
189
        AV_WL16(pos, av_clip_uintp2(val >> shift, output_bits)); \
190
    }
191

192
static av_always_inline void
193
yuv2plane1_10_c_template(const int16_t *src, uint16_t *dest, int dstW,
194
                         int big_endian, int output_bits)
195
{
196
    int i;
197
    int shift = 15 - output_bits;
198

199
    for (i = 0; i < dstW; i++) {
200
        int val = src[i] + (1 << (shift - 1));
201
        output_pixel(&dest[i], val);
202
    }
203
}
204

205
static av_always_inline void
206
yuv2planeX_10_c_template(const int16_t *filter, int filterSize,
207
                         const int16_t **src, uint16_t *dest, int dstW,
208
                         int big_endian, int output_bits)
209
{
210
    int i;
211
    int shift = 11 + 16 - output_bits;
212

213
    for (i = 0; i < dstW; i++) {
214
        int val = 1 << (shift - 1);
215
        int j;
216

217
        for (j = 0; j < filterSize; j++)
218
            val += src[j][i] * filter[j];
219

220
        output_pixel(&dest[i], val);
221
    }
222
}
223

224
#undef output_pixel
225

226
#define yuv2NBPS(bits, BE_LE, is_be, template_size, typeX_t) \
227
static void yuv2plane1_ ## bits ## BE_LE ## _c(const int16_t *src, \
228
                              uint8_t *dest, int dstW, \
229
                              const uint8_t *dither, int offset)\
230
{ \
231
    yuv2plane1_ ## template_size ## _c_template((const typeX_t *) src, \
232
                         (uint16_t *) dest, dstW, is_be, bits); \
233
}\
234
static void yuv2planeX_ ## bits ## BE_LE ## _c(const int16_t *filter, int filterSize, \
235
                              const int16_t **src, uint8_t *dest, int dstW, \
236
                              const uint8_t *dither, int offset)\
237
{ \
238
    yuv2planeX_## template_size ## _c_template(filter, \
239
                         filterSize, (const typeX_t **) src, \
240
                         (uint16_t *) dest, dstW, is_be, bits); \
241
}
242
yuv2NBPS( 9, BE, 1, 10, int16_t)
243
yuv2NBPS( 9, LE, 0, 10, int16_t)
244
yuv2NBPS(10, BE, 1, 10, int16_t)
245
yuv2NBPS(10, LE, 0, 10, int16_t)
246
yuv2NBPS(12, BE, 1, 10, int16_t)
247
yuv2NBPS(12, LE, 0, 10, int16_t)
248
yuv2NBPS(14, BE, 1, 10, int16_t)
249
yuv2NBPS(14, LE, 0, 10, int16_t)
250
yuv2NBPS(16, BE, 1, 16, int32_t)
251
yuv2NBPS(16, LE, 0, 16, int32_t)
252

253
static void yuv2planeX_8_c(const int16_t *filter, int filterSize,
254
                           const int16_t **src, uint8_t *dest, int dstW,
255
                           const uint8_t *dither, int offset)
256
{
257
    int i;
258
    for (i=0; i<dstW; i++) {
259
        int val = dither[(i + offset) & 7] << 12;
260
        int j;
261
        for (j=0; j<filterSize; j++)
262
            val += src[j][i] * filter[j];
263

264
        dest[i]= av_clip_uint8(val>>19);
265
    }
266
}
267

268
static void yuv2plane1_8_c(const int16_t *src, uint8_t *dest, int dstW,
269
                           const uint8_t *dither, int offset)
270
{
271
    int i;
272
    for (i=0; i<dstW; i++) {
273
        int val = (src[i] + dither[(i + offset) & 7]) >> 7;
274
        dest[i]= av_clip_uint8(val);
275
    }
276
}
277

278
static void yuv2nv12cX_c(SwsContext *c, const int16_t *chrFilter, int chrFilterSize,
279
                        const int16_t **chrUSrc, const int16_t **chrVSrc,
280
                        uint8_t *dest, int chrDstW)
281
{
282
    enum AVPixelFormat dstFormat = c->dstFormat;
283
    const uint8_t *chrDither = c->chrDither8;
284
    int i;
285

286
    if (dstFormat == AV_PIX_FMT_NV12)
287
        for (i=0; i<chrDstW; i++) {
288
            int u = chrDither[i & 7] << 12;
289
            int v = chrDither[(i + 3) & 7] << 12;
290
            int j;
291
            for (j=0; j<chrFilterSize; j++) {
292
                u += chrUSrc[j][i] * chrFilter[j];
293
                v += chrVSrc[j][i] * chrFilter[j];
294
            }
295

296
            dest[2*i]= av_clip_uint8(u>>19);
297
            dest[2*i+1]= av_clip_uint8(v>>19);
298
        }
299
    else
300
        for (i=0; i<chrDstW; i++) {
301
            int u = chrDither[i & 7] << 12;
302
            int v = chrDither[(i + 3) & 7] << 12;
303
            int j;
304
            for (j=0; j<chrFilterSize; j++) {
305
                u += chrUSrc[j][i] * chrFilter[j];
306
                v += chrVSrc[j][i] * chrFilter[j];
307
            }
308

309
            dest[2*i]= av_clip_uint8(v>>19);
310
            dest[2*i+1]= av_clip_uint8(u>>19);
311
        }
312
}
313

314
#define accumulate_bit(acc, val) \
315
    acc <<= 1; \
316
    acc |= (val) >= 234
317
#define output_pixel(pos, acc) \
318
    if (target == AV_PIX_FMT_MONOBLACK) { \
319
        pos = acc; \
320
    } else { \
321
        pos = ~acc; \
322
    }
323

324
static av_always_inline void
325
yuv2mono_X_c_template(SwsContext *c, const int16_t *lumFilter,
326
                      const int16_t **lumSrc, int lumFilterSize,
327
                      const int16_t *chrFilter, const int16_t **chrUSrc,
328
                      const int16_t **chrVSrc, int chrFilterSize,
329
                      const int16_t **alpSrc, uint8_t *dest, int dstW,
330
                      int y, enum AVPixelFormat target)
331
{
332
    const uint8_t * const d128 = ff_dither_8x8_220[y&7];
333
    int i;
334
    unsigned acc = 0;
335
    int err = 0;
336

337
    for (i = 0; i < dstW; i += 2) {
338
        int j;
339
        int Y1 = 1 << 18;
340
        int Y2 = 1 << 18;
341

342
        for (j = 0; j < lumFilterSize; j++) {
343
            Y1 += lumSrc[j][i]   * lumFilter[j];
344
            Y2 += lumSrc[j][i+1] * lumFilter[j];
345
        }
346
        Y1 >>= 19;
347
        Y2 >>= 19;
348
        if ((Y1 | Y2) & 0x100) {
349
            Y1 = av_clip_uint8(Y1);
350
            Y2 = av_clip_uint8(Y2);
351
        }
352
        if (c->dither == SWS_DITHER_ED) {
353
            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;
354
            c->dither_error[0][i] = err;
355
            acc = 2*acc + (Y1 >= 128);
356
            Y1 -= 220*(acc&1);
357

358
            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);
359
            c->dither_error[0][i+1] = Y1;
360
            acc = 2*acc + (err >= 128);
361
            err -= 220*(acc&1);
362
        } else {
363
            accumulate_bit(acc, Y1 + d128[(i + 0) & 7]);
364
            accumulate_bit(acc, Y2 + d128[(i + 1) & 7]);
365
        }
366
        if ((i & 7) == 6) {
367
            output_pixel(*dest++, acc);
368
        }
369
    }
370
    c->dither_error[0][i] = err;
371

372
    if (i & 6) {
373
        output_pixel(*dest, acc);
374
    }
375
}
376

377
static av_always_inline void
378
yuv2mono_2_c_template(SwsContext *c, const int16_t *buf[2],
379
                      const int16_t *ubuf[2], const int16_t *vbuf[2],
380
                      const int16_t *abuf[2], uint8_t *dest, int dstW,
381
                      int yalpha, int uvalpha, int y,
382
                      enum AVPixelFormat target)
383
{
384
    const int16_t *buf0  = buf[0],  *buf1  = buf[1];
385
    const uint8_t * const d128 = ff_dither_8x8_220[y & 7];
386
    int  yalpha1 = 4096 - yalpha;
387
    int i;
388
    av_assert2(yalpha  <= 4096U);
389

390
    if (c->dither == SWS_DITHER_ED) {
391
        int err = 0;
392
        int acc = 0;
393
        for (i = 0; i < dstW; i +=2) {
394
            int Y;
395

396
            Y = (buf0[i + 0] * yalpha1 + buf1[i + 0] * yalpha) >> 19;
397
            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;
398
            c->dither_error[0][i] = err;
399
            acc = 2*acc + (Y >= 128);
400
            Y -= 220*(acc&1);
401

402
            err = (buf0[i + 1] * yalpha1 + buf1[i + 1] * yalpha) >> 19;
403
            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;
404
            c->dither_error[0][i+1] = Y;
405
            acc = 2*acc + (err >= 128);
406
            err -= 220*(acc&1);
407

408
            if ((i & 7) == 6)
409
                output_pixel(*dest++, acc);
410
        }
411
        c->dither_error[0][i] = err;
412
    } else {
413
    for (i = 0; i < dstW; i += 8) {
414
        int Y, acc = 0;
415

416
        Y = (buf0[i + 0] * yalpha1 + buf1[i + 0] * yalpha) >> 19;
417
        accumulate_bit(acc, Y + d128[0]);
418
        Y = (buf0[i + 1] * yalpha1 + buf1[i + 1] * yalpha) >> 19;
419
        accumulate_bit(acc, Y + d128[1]);
420
        Y = (buf0[i + 2] * yalpha1 + buf1[i + 2] * yalpha) >> 19;
421
        accumulate_bit(acc, Y + d128[2]);
422
        Y = (buf0[i + 3] * yalpha1 + buf1[i + 3] * yalpha) >> 19;
423
        accumulate_bit(acc, Y + d128[3]);
424
        Y = (buf0[i + 4] * yalpha1 + buf1[i + 4] * yalpha) >> 19;
425
        accumulate_bit(acc, Y + d128[4]);
426
        Y = (buf0[i + 5] * yalpha1 + buf1[i + 5] * yalpha) >> 19;
427
        accumulate_bit(acc, Y + d128[5]);
428
        Y = (buf0[i + 6] * yalpha1 + buf1[i + 6] * yalpha) >> 19;
429
        accumulate_bit(acc, Y + d128[6]);
430
        Y = (buf0[i + 7] * yalpha1 + buf1[i + 7] * yalpha) >> 19;
431
        accumulate_bit(acc, Y + d128[7]);
432

433
        output_pixel(*dest++, acc);
434
    }
435
    }
436
}
437

438
static av_always_inline void
439
yuv2mono_1_c_template(SwsContext *c, const int16_t *buf0,
440
                      const int16_t *ubuf[2], const int16_t *vbuf[2],
441
                      const int16_t *abuf0, uint8_t *dest, int dstW,
442
                      int uvalpha, int y, enum AVPixelFormat target)
443
{
444
    const uint8_t * const d128 = ff_dither_8x8_220[y & 7];
445
    int i;
446

447
    if (c->dither == SWS_DITHER_ED) {
448
        int err = 0;
449
        int acc = 0;
450
        for (i = 0; i < dstW; i +=2) {
451
            int Y;
452

453
            Y = ((buf0[i + 0] + 64) >> 7);
454
            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;
455
            c->dither_error[0][i] = err;
456
            acc = 2*acc + (Y >= 128);
457
            Y -= 220*(acc&1);
458

459
            err = ((buf0[i + 1] + 64) >> 7);
460
            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;
461
            c->dither_error[0][i+1] = Y;
462
            acc = 2*acc + (err >= 128);
463
            err -= 220*(acc&1);
464

465
            if ((i & 7) == 6)
466
                output_pixel(*dest++, acc);
467
        }
468
        c->dither_error[0][i] = err;
469
    } else {
470
    for (i = 0; i < dstW; i += 8) {
471
        int acc = 0;
472
        accumulate_bit(acc, ((buf0[i + 0] + 64) >> 7) + d128[0]);
473
        accumulate_bit(acc, ((buf0[i + 1] + 64) >> 7) + d128[1]);
474
        accumulate_bit(acc, ((buf0[i + 2] + 64) >> 7) + d128[2]);
475
        accumulate_bit(acc, ((buf0[i + 3] + 64) >> 7) + d128[3]);
476
        accumulate_bit(acc, ((buf0[i + 4] + 64) >> 7) + d128[4]);
477
        accumulate_bit(acc, ((buf0[i + 5] + 64) >> 7) + d128[5]);
478
        accumulate_bit(acc, ((buf0[i + 6] + 64) >> 7) + d128[6]);
479
        accumulate_bit(acc, ((buf0[i + 7] + 64) >> 7) + d128[7]);
480

481
        output_pixel(*dest++, acc);
482
    }
483
    }
484
}
485

486
#undef output_pixel
487
#undef accumulate_bit
488

489
#define YUV2PACKEDWRAPPER(name, base, ext, fmt) \
490
static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
491
                                const int16_t **lumSrc, int lumFilterSize, \
492
                                const int16_t *chrFilter, const int16_t **chrUSrc, \
493
                                const int16_t **chrVSrc, int chrFilterSize, \
494
                                const int16_t **alpSrc, uint8_t *dest, int dstW, \
495
                                int y) \
496
{ \
497
    name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
498
                                  chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
499
                                  alpSrc, dest, dstW, y, fmt); \
500
} \
501
 \
502
static void name ## ext ## _2_c(SwsContext *c, const int16_t *buf[2], \
503
                                const int16_t *ubuf[2], const int16_t *vbuf[2], \
504
                                const int16_t *abuf[2], uint8_t *dest, int dstW, \
505
                                int yalpha, int uvalpha, int y) \
506
{ \
507
    name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
508
                                  dest, dstW, yalpha, uvalpha, y, fmt); \
509
} \
510
 \
511
static void name ## ext ## _1_c(SwsContext *c, const int16_t *buf0, \
512
                                const int16_t *ubuf[2], const int16_t *vbuf[2], \
513
                                const int16_t *abuf0, uint8_t *dest, int dstW, \
514
                                int uvalpha, int y) \
515
{ \
516
    name ## base ## _1_c_template(c, buf0, ubuf, vbuf, \
517
                                  abuf0, dest, dstW, uvalpha, \
518
                                  y, fmt); \
519
}
520

521
YUV2PACKEDWRAPPER(yuv2mono,, white, AV_PIX_FMT_MONOWHITE)
522
YUV2PACKEDWRAPPER(yuv2mono,, black, AV_PIX_FMT_MONOBLACK)
523

524
#define output_pixels(pos, Y1, U, Y2, V) \
525
    if (target == AV_PIX_FMT_YUYV422) { \
526
        dest[pos + 0] = Y1; \
527
        dest[pos + 1] = U;  \
528
        dest[pos + 2] = Y2; \
529
        dest[pos + 3] = V;  \
530
    } else if (target == AV_PIX_FMT_YVYU422) { \
531
        dest[pos + 0] = Y1; \
532
        dest[pos + 1] = V;  \
533
        dest[pos + 2] = Y2; \
534
        dest[pos + 3] = U;  \
535
    } else { /* AV_PIX_FMT_UYVY422 */ \
536
        dest[pos + 0] = U;  \
537
        dest[pos + 1] = Y1; \
538
        dest[pos + 2] = V;  \
539
        dest[pos + 3] = Y2; \
540
    }
541

542
static av_always_inline void
543
yuv2422_X_c_template(SwsContext *c, const int16_t *lumFilter,
544
                     const int16_t **lumSrc, int lumFilterSize,
545
                     const int16_t *chrFilter, const int16_t **chrUSrc,
546
                     const int16_t **chrVSrc, int chrFilterSize,
547
                     const int16_t **alpSrc, uint8_t *dest, int dstW,
548
                     int y, enum AVPixelFormat target)
549
{
550
    int i;
551

552
    for (i = 0; i < ((dstW + 1) >> 1); i++) {
553
        int j;
554
        int Y1 = 1 << 18;
555
        int Y2 = 1 << 18;
556
        int U  = 1 << 18;
557
        int V  = 1 << 18;
558

559
        for (j = 0; j < lumFilterSize; j++) {
560
            Y1 += lumSrc[j][i * 2]     * lumFilter[j];
561
            Y2 += lumSrc[j][i * 2 + 1] * lumFilter[j];
562
        }
563
        for (j = 0; j < chrFilterSize; j++) {
564
            U += chrUSrc[j][i] * chrFilter[j];
565
            V += chrVSrc[j][i] * chrFilter[j];
566
        }
567
        Y1 >>= 19;
568
        Y2 >>= 19;
569
        U  >>= 19;
570
        V  >>= 19;
571
        if ((Y1 | Y2 | U | V) & 0x100) {
572
            Y1 = av_clip_uint8(Y1);
573
            Y2 = av_clip_uint8(Y2);
574
            U  = av_clip_uint8(U);
575
            V  = av_clip_uint8(V);
576
        }
577
        output_pixels(4*i, Y1, U, Y2, V);
578
    }
579
}
580

581
static av_always_inline void
582
yuv2422_2_c_template(SwsContext *c, const int16_t *buf[2],
583
                     const int16_t *ubuf[2], const int16_t *vbuf[2],
584
                     const int16_t *abuf[2], uint8_t *dest, int dstW,
585
                     int yalpha, int uvalpha, int y,
586
                     enum AVPixelFormat target)
587
{
588
    const int16_t *buf0  = buf[0],  *buf1  = buf[1],
589
                  *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
590
                  *vbuf0 = vbuf[0], *vbuf1 = vbuf[1];
591
    int  yalpha1 = 4096 - yalpha;
592
    int uvalpha1 = 4096 - uvalpha;
593
    int i;
594
    av_assert2(yalpha  <= 4096U);
595
    av_assert2(uvalpha <= 4096U);
596

597
    for (i = 0; i < ((dstW + 1) >> 1); i++) {
598
        int Y1 = (buf0[i * 2]     * yalpha1  + buf1[i * 2]     * yalpha)  >> 19;
599
        int Y2 = (buf0[i * 2 + 1] * yalpha1  + buf1[i * 2 + 1] * yalpha)  >> 19;
600
        int U  = (ubuf0[i]        * uvalpha1 + ubuf1[i]        * uvalpha) >> 19;
601
        int V  = (vbuf0[i]        * uvalpha1 + vbuf1[i]        * uvalpha) >> 19;
602

603
        if ((Y1 | Y2 | U | V) & 0x100) {
604
            Y1 = av_clip_uint8(Y1);
605
            Y2 = av_clip_uint8(Y2);
606
            U  = av_clip_uint8(U);
607
            V  = av_clip_uint8(V);
608
        }
609

610
        output_pixels(i * 4, Y1, U, Y2, V);
611
    }
612
}
613

614
static av_always_inline void
615
yuv2422_1_c_template(SwsContext *c, const int16_t *buf0,
616
                     const int16_t *ubuf[2], const int16_t *vbuf[2],
617
                     const int16_t *abuf0, uint8_t *dest, int dstW,
618
                     int uvalpha, int y, enum AVPixelFormat target)
619
{
620
    const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
621
    int i;
622

623
    if (uvalpha < 2048) {
624
        for (i = 0; i < ((dstW + 1) >> 1); i++) {
625
            int Y1 = (buf0[i * 2    ]+64) >> 7;
626
            int Y2 = (buf0[i * 2 + 1]+64) >> 7;
627
            int U  = (ubuf0[i]       +64) >> 7;
628
            int V  = (vbuf0[i]       +64) >> 7;
629

630
            if ((Y1 | Y2 | U | V) & 0x100) {
631
                Y1 = av_clip_uint8(Y1);
632
                Y2 = av_clip_uint8(Y2);
633
                U  = av_clip_uint8(U);
634
                V  = av_clip_uint8(V);
635
            }
636

637
            Y1 = av_clip_uint8(Y1);
638
            Y2 = av_clip_uint8(Y2);
639
            U  = av_clip_uint8(U);
640
            V  = av_clip_uint8(V);
641

642
            output_pixels(i * 4, Y1, U, Y2, V);
643
        }
644
    } else {
645
        const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
646
        for (i = 0; i < ((dstW + 1) >> 1); i++) {
647
            int Y1 = (buf0[i * 2    ]    + 64) >> 7;
648
            int Y2 = (buf0[i * 2 + 1]    + 64) >> 7;
649
            int U  = (ubuf0[i] + ubuf1[i]+128) >> 8;
650
            int V  = (vbuf0[i] + vbuf1[i]+128) >> 8;
651

652
            if ((Y1 | Y2 | U | V) & 0x100) {
653
                Y1 = av_clip_uint8(Y1);
654
                Y2 = av_clip_uint8(Y2);
655
                U  = av_clip_uint8(U);
656
                V  = av_clip_uint8(V);
657
            }
658

659
            Y1 = av_clip_uint8(Y1);
660
            Y2 = av_clip_uint8(Y2);
661
            U  = av_clip_uint8(U);
662
            V  = av_clip_uint8(V);
663

664
            output_pixels(i * 4, Y1, U, Y2, V);
665
        }
666
    }
667
}
668

669
#undef output_pixels
670

671
YUV2PACKEDWRAPPER(yuv2, 422, yuyv422, AV_PIX_FMT_YUYV422)
672
YUV2PACKEDWRAPPER(yuv2, 422, yvyu422, AV_PIX_FMT_YVYU422)
673
YUV2PACKEDWRAPPER(yuv2, 422, uyvy422, AV_PIX_FMT_UYVY422)
674

675
#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)
676
#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)
677
#define output_pixel(pos, val) \
678
    if (isBE(target)) { \
679
        AV_WB16(pos, val); \
680
    } else { \
681
        AV_WL16(pos, val); \
682
    }
683

684
static av_always_inline void
685
yuv2rgba64_X_c_template(SwsContext *c, const int16_t *lumFilter,
686
                       const int32_t **lumSrc, int lumFilterSize,
687
                       const int16_t *chrFilter, const int32_t **chrUSrc,
688
                       const int32_t **chrVSrc, int chrFilterSize,
689
                       const int32_t **alpSrc, uint16_t *dest, int dstW,
690
                       int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
691
{
692
    int i;
693
    int A1 = 0xffff<<14, A2 = 0xffff<<14;
694

695
    for (i = 0; i < ((dstW + 1) >> 1); i++) {
696
        int j;
697
        int Y1 = -0x40000000;
698
        int Y2 = -0x40000000;
699
        int U  = -(128 << 23); // 19
700
        int V  = -(128 << 23);
701
        int R, G, B;
702

703
        for (j = 0; j < lumFilterSize; j++) {
704
            Y1 += lumSrc[j][i * 2]     * (unsigned)lumFilter[j];
705
            Y2 += lumSrc[j][i * 2 + 1] * (unsigned)lumFilter[j];
706
        }
707
        for (j = 0; j < chrFilterSize; j++) {;
708
            U += chrUSrc[j][i] * (unsigned)chrFilter[j];
709
            V += chrVSrc[j][i] * (unsigned)chrFilter[j];
710
        }
711

712
        if (hasAlpha) {
713
            A1 = -0x40000000;
714
            A2 = -0x40000000;
715
            for (j = 0; j < lumFilterSize; j++) {
716
                A1 += alpSrc[j][i * 2]     * (unsigned)lumFilter[j];
717
                A2 += alpSrc[j][i * 2 + 1] * (unsigned)lumFilter[j];
718
            }
719
            A1 >>= 1;
720
            A1 += 0x20002000;
721
            A2 >>= 1;
722
            A2 += 0x20002000;
723
        }
724

725
        // 8bit: 12+15=27; 16-bit: 12+19=31
726
        Y1 >>= 14; // 10
727
        Y1 += 0x10000;
728
        Y2 >>= 14;
729
        Y2 += 0x10000;
730
        U  >>= 14;
731
        V  >>= 14;
732

733
        // 8bit: 27 -> 17bit, 16bit: 31 - 14 = 17bit
734
        Y1 -= c->yuv2rgb_y_offset;
735
        Y2 -= c->yuv2rgb_y_offset;
736
        Y1 *= c->yuv2rgb_y_coeff;
737
        Y2 *= c->yuv2rgb_y_coeff;
738
        Y1 += 1 << 13; // 21
739
        Y2 += 1 << 13;
740
        // 8bit: 17 + 13bit = 30bit, 16bit: 17 + 13bit = 30bit
741

742
        R = V * c->yuv2rgb_v2r_coeff;
743
        G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
744
        B =                            U * c->yuv2rgb_u2b_coeff;
745

746
        // 8bit: 30 - 22 = 8bit, 16bit: 30bit - 14 = 16bit
747
        output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
748
        output_pixel(&dest[1], av_clip_uintp2(  G + Y1, 30) >> 14);
749
        output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
750
        if (eightbytes) {
751
            output_pixel(&dest[3], av_clip_uintp2(A1      , 30) >> 14);
752
            output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
753
            output_pixel(&dest[5], av_clip_uintp2(  G + Y2, 30) >> 14);
754
            output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
755
            output_pixel(&dest[7], av_clip_uintp2(A2      , 30) >> 14);
756
            dest += 8;
757
        } else {
758
            output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
759
            output_pixel(&dest[4], av_clip_uintp2(  G + Y2, 30) >> 14);
760
            output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
761
            dest += 6;
762
        }
763
    }
764
}
765

766
static av_always_inline void
767
yuv2rgba64_2_c_template(SwsContext *c, const int32_t *buf[2],
768
                       const int32_t *ubuf[2], const int32_t *vbuf[2],
769
                       const int32_t *abuf[2], uint16_t *dest, int dstW,
770
                       int yalpha, int uvalpha, int y,
771
                       enum AVPixelFormat target, int hasAlpha, int eightbytes)
772
{
773
    const int32_t *buf0  = buf[0],  *buf1  = buf[1],
774
                  *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
775
                  *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
776
                  *abuf0 = hasAlpha ? abuf[0] : NULL,
777
                  *abuf1 = hasAlpha ? abuf[1] : NULL;
778
    int  yalpha1 = 4096 - yalpha;
779
    int uvalpha1 = 4096 - uvalpha;
780
    int i;
781
    int A1 = 0xffff<<14, A2 = 0xffff<<14;
782

783
    av_assert2(yalpha  <= 4096U);
784
    av_assert2(uvalpha <= 4096U);
785

786
    for (i = 0; i < ((dstW + 1) >> 1); i++) {
787
        int Y1 = (buf0[i * 2]     * yalpha1  + buf1[i * 2]     * yalpha) >> 14;
788
        int Y2 = (buf0[i * 2 + 1] * yalpha1  + buf1[i * 2 + 1] * yalpha) >> 14;
789
        int U  = (ubuf0[i]        * uvalpha1 + ubuf1[i]        * uvalpha - (128 << 23)) >> 14;
790
        int V  = (vbuf0[i]        * uvalpha1 + vbuf1[i]        * uvalpha - (128 << 23)) >> 14;
791
        int R, G, B;
792

793
        Y1 -= c->yuv2rgb_y_offset;
794
        Y2 -= c->yuv2rgb_y_offset;
795
        Y1 *= c->yuv2rgb_y_coeff;
796
        Y2 *= c->yuv2rgb_y_coeff;
797
        Y1 += 1 << 13;
798
        Y2 += 1 << 13;
799

800
        R = V * c->yuv2rgb_v2r_coeff;
801
        G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
802
        B =                            U * c->yuv2rgb_u2b_coeff;
803

804
        if (hasAlpha) {
805
            A1 = (abuf0[i * 2    ] * yalpha1 + abuf1[i * 2    ] * yalpha) >> 1;
806
            A2 = (abuf0[i * 2 + 1] * yalpha1 + abuf1[i * 2 + 1] * yalpha) >> 1;
807

808
            A1 += 1 << 13;
809
            A2 += 1 << 13;
810
        }
811

812
        output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
813
        output_pixel(&dest[1], av_clip_uintp2(  G + Y1, 30) >> 14);
814
        output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
815
        if (eightbytes) {
816
            output_pixel(&dest[3], av_clip_uintp2(A1      , 30) >> 14);
817
            output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
818
            output_pixel(&dest[5], av_clip_uintp2(  G + Y2, 30) >> 14);
819
            output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
820
            output_pixel(&dest[7], av_clip_uintp2(A2      , 30) >> 14);
821
            dest += 8;
822
        } else {
823
            output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
824
            output_pixel(&dest[4], av_clip_uintp2(  G + Y2, 30) >> 14);
825
            output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
826
            dest += 6;
827
        }
828
    }
829
}
830

831
static av_always_inline void
832
yuv2rgba64_1_c_template(SwsContext *c, const int32_t *buf0,
833
                       const int32_t *ubuf[2], const int32_t *vbuf[2],
834
                       const int32_t *abuf0, uint16_t *dest, int dstW,
835
                       int uvalpha, int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
836
{
837
    const int32_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
838
    int i;
839
    int A1 = 0xffff<<14, A2= 0xffff<<14;
840

841
    if (uvalpha < 2048) {
842
        for (i = 0; i < ((dstW + 1) >> 1); i++) {
843
            int Y1 = (buf0[i * 2]    ) >> 2;
844
            int Y2 = (buf0[i * 2 + 1]) >> 2;
845
            int U  = (ubuf0[i] - (128 << 11)) >> 2;
846
            int V  = (vbuf0[i] - (128 << 11)) >> 2;
847
            int R, G, B;
848

849
            Y1 -= c->yuv2rgb_y_offset;
850
            Y2 -= c->yuv2rgb_y_offset;
851
            Y1 *= c->yuv2rgb_y_coeff;
852
            Y2 *= c->yuv2rgb_y_coeff;
853
            Y1 += 1 << 13;
854
            Y2 += 1 << 13;
855

856
            if (hasAlpha) {
857
                A1 = abuf0[i * 2    ] << 11;
858
                A2 = abuf0[i * 2 + 1] << 11;
859

860
                A1 += 1 << 13;
861
                A2 += 1 << 13;
862
            }
863

864
            R = V * c->yuv2rgb_v2r_coeff;
865
            G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
866
            B =                            U * c->yuv2rgb_u2b_coeff;
867

868
            output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
869
            output_pixel(&dest[1], av_clip_uintp2(  G + Y1, 30) >> 14);
870
            output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
871
            if (eightbytes) {
872
                output_pixel(&dest[3], av_clip_uintp2(A1      , 30) >> 14);
873
                output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
874
                output_pixel(&dest[5], av_clip_uintp2(  G + Y2, 30) >> 14);
875
                output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
876
                output_pixel(&dest[7], av_clip_uintp2(A2      , 30) >> 14);
877
                dest += 8;
878
            } else {
879
                output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
880
                output_pixel(&dest[4], av_clip_uintp2(  G + Y2, 30) >> 14);
881
                output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
882
                dest += 6;
883
            }
884
        }
885
    } else {
886
        const int32_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
887
        int A1 = 0xffff<<14, A2 = 0xffff<<14;
888
        for (i = 0; i < ((dstW + 1) >> 1); i++) {
889
            int Y1 = (buf0[i * 2]    ) >> 2;
890
            int Y2 = (buf0[i * 2 + 1]) >> 2;
891
            int U  = (ubuf0[i] + ubuf1[i] - (128 << 12)) >> 3;
892
            int V  = (vbuf0[i] + vbuf1[i] - (128 << 12)) >> 3;
893
            int R, G, B;
894

895
            Y1 -= c->yuv2rgb_y_offset;
896
            Y2 -= c->yuv2rgb_y_offset;
897
            Y1 *= c->yuv2rgb_y_coeff;
898
            Y2 *= c->yuv2rgb_y_coeff;
899
            Y1 += 1 << 13;
900
            Y2 += 1 << 13;
901

902
            if (hasAlpha) {
903
                A1 = abuf0[i * 2    ] << 11;
904
                A2 = abuf0[i * 2 + 1] << 11;
905

906
                A1 += 1 << 13;
907
                A2 += 1 << 13;
908
            }
909

910
            R = V * c->yuv2rgb_v2r_coeff;
911
            G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
912
            B =                            U * c->yuv2rgb_u2b_coeff;
913

914
            output_pixel(&dest[0], av_clip_uintp2(R_B + Y1, 30) >> 14);
915
            output_pixel(&dest[1], av_clip_uintp2(  G + Y1, 30) >> 14);
916
            output_pixel(&dest[2], av_clip_uintp2(B_R + Y1, 30) >> 14);
917
            if (eightbytes) {
918
                output_pixel(&dest[3], av_clip_uintp2(A1      , 30) >> 14);
919
                output_pixel(&dest[4], av_clip_uintp2(R_B + Y2, 30) >> 14);
920
                output_pixel(&dest[5], av_clip_uintp2(  G + Y2, 30) >> 14);
921
                output_pixel(&dest[6], av_clip_uintp2(B_R + Y2, 30) >> 14);
922
                output_pixel(&dest[7], av_clip_uintp2(A2      , 30) >> 14);
923
                dest += 8;
924
            } else {
925
                output_pixel(&dest[3], av_clip_uintp2(R_B + Y2, 30) >> 14);
926
                output_pixel(&dest[4], av_clip_uintp2(  G + Y2, 30) >> 14);
927
                output_pixel(&dest[5], av_clip_uintp2(B_R + Y2, 30) >> 14);
928
                dest += 6;
929
            }
930
        }
931
    }
932
}
933

934
static av_always_inline void
935
yuv2rgba64_full_X_c_template(SwsContext *c, const int16_t *lumFilter,
936
                       const int32_t **lumSrc, int lumFilterSize,
937
                       const int16_t *chrFilter, const int32_t **chrUSrc,
938
                       const int32_t **chrVSrc, int chrFilterSize,
939
                       const int32_t **alpSrc, uint16_t *dest, int dstW,
940
                       int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
941
{
942
    int i;
943
    int A = 0xffff<<14;
944

945
    for (i = 0; i < dstW; i++) {
946
        int j;
947
        int Y  = -0x40000000;
948
        int U  = -(128 << 23); // 19
949
        int V  = -(128 << 23);
950
        int R, G, B;
951

952
        for (j = 0; j < lumFilterSize; j++) {
953
            Y += lumSrc[j][i]  * (unsigned)lumFilter[j];
954
        }
955
        for (j = 0; j < chrFilterSize; j++) {;
956
            U += chrUSrc[j][i] * (unsigned)chrFilter[j];
957
            V += chrVSrc[j][i] * (unsigned)chrFilter[j];
958
        }
959

960
        if (hasAlpha) {
961
            A = -0x40000000;
962
            for (j = 0; j < lumFilterSize; j++) {
963
                A += alpSrc[j][i] * (unsigned)lumFilter[j];
964
            }
965
            A >>= 1;
966
            A += 0x20002000;
967
        }
968

969
        // 8bit: 12+15=27; 16-bit: 12+19=31
970
        Y  >>= 14; // 10
971
        Y += 0x10000;
972
        U  >>= 14;
973
        V  >>= 14;
974

975
        // 8bit: 27 -> 17bit, 16bit: 31 - 14 = 17bit
976
        Y -= c->yuv2rgb_y_offset;
977
        Y *= c->yuv2rgb_y_coeff;
978
        Y += 1 << 13; // 21
979
        // 8bit: 17 + 13bit = 30bit, 16bit: 17 + 13bit = 30bit
980

981
        R = V * c->yuv2rgb_v2r_coeff;
982
        G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
983
        B =                            U * c->yuv2rgb_u2b_coeff;
984

985
        // 8bit: 30 - 22 = 8bit, 16bit: 30bit - 14 = 16bit
986
        output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
987
        output_pixel(&dest[1], av_clip_uintp2(  G + Y, 30) >> 14);
988
        output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
989
        if (eightbytes) {
990
            output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
991
            dest += 4;
992
        } else {
993
            dest += 3;
994
        }
995
    }
996
}
997

998
static av_always_inline void
999
yuv2rgba64_full_2_c_template(SwsContext *c, const int32_t *buf[2],
1000
                       const int32_t *ubuf[2], const int32_t *vbuf[2],
1001
                       const int32_t *abuf[2], uint16_t *dest, int dstW,
1002
                       int yalpha, int uvalpha, int y,
1003
                       enum AVPixelFormat target, int hasAlpha, int eightbytes)
1004
{
1005
    const int32_t *buf0  = buf[0],  *buf1  = buf[1],
1006
                  *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
1007
                  *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
1008
                  *abuf0 = hasAlpha ? abuf[0] : NULL,
1009
                  *abuf1 = hasAlpha ? abuf[1] : NULL;
1010
    int  yalpha1 = 4096 - yalpha;
1011
    int uvalpha1 = 4096 - uvalpha;
1012
    int i;
1013
    int A = 0xffff<<14;
1014

1015
    av_assert2(yalpha  <= 4096U);
1016
    av_assert2(uvalpha <= 4096U);
1017

1018
    for (i = 0; i < dstW; i++) {
1019
        int Y  = (buf0[i]     * yalpha1  + buf1[i]     * yalpha) >> 14;
1020
        int U  = (ubuf0[i]   * uvalpha1 + ubuf1[i]     * uvalpha - (128 << 23)) >> 14;
1021
        int V  = (vbuf0[i]   * uvalpha1 + vbuf1[i]     * uvalpha - (128 << 23)) >> 14;
1022
        int R, G, B;
1023

1024
        Y -= c->yuv2rgb_y_offset;
1025
        Y *= c->yuv2rgb_y_coeff;
1026
        Y += 1 << 13;
1027

1028
        R = V * c->yuv2rgb_v2r_coeff;
1029
        G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1030
        B =                            U * c->yuv2rgb_u2b_coeff;
1031

1032
        if (hasAlpha) {
1033
            A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha) >> 1;
1034

1035
            A += 1 << 13;
1036
        }
1037

1038
        output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1039
        output_pixel(&dest[1], av_clip_uintp2(  G + Y, 30) >> 14);
1040
        output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1041
        if (eightbytes) {
1042
            output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1043
            dest += 4;
1044
        } else {
1045
            dest += 3;
1046
        }
1047
    }
1048
}
1049

1050
static av_always_inline void
1051
yuv2rgba64_full_1_c_template(SwsContext *c, const int32_t *buf0,
1052
                       const int32_t *ubuf[2], const int32_t *vbuf[2],
1053
                       const int32_t *abuf0, uint16_t *dest, int dstW,
1054
                       int uvalpha, int y, enum AVPixelFormat target, int hasAlpha, int eightbytes)
1055
{
1056
    const int32_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1057
    int i;
1058
    int A = 0xffff<<14;
1059

1060
    if (uvalpha < 2048) {
1061
        for (i = 0; i < dstW; i++) {
1062
            int Y  = (buf0[i]) >> 2;
1063
            int U  = (ubuf0[i] - (128 << 11)) >> 2;
1064
            int V  = (vbuf0[i] - (128 << 11)) >> 2;
1065
            int R, G, B;
1066

1067
            Y -= c->yuv2rgb_y_offset;
1068
            Y *= c->yuv2rgb_y_coeff;
1069
            Y += 1 << 13;
1070

1071
            if (hasAlpha) {
1072
                A = abuf0[i] << 11;
1073

1074
                A += 1 << 13;
1075
            }
1076

1077
            R = V * c->yuv2rgb_v2r_coeff;
1078
            G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1079
            B =                            U * c->yuv2rgb_u2b_coeff;
1080

1081
            output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1082
            output_pixel(&dest[1], av_clip_uintp2(  G + Y, 30) >> 14);
1083
            output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1084
            if (eightbytes) {
1085
                output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1086
                dest += 4;
1087
            } else {
1088
                dest += 3;
1089
            }
1090
        }
1091
    } else {
1092
        const int32_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1093
        int A = 0xffff<<14;
1094
        for (i = 0; i < dstW; i++) {
1095
            int Y  = (buf0[i]    ) >> 2;
1096
            int U  = (ubuf0[i] + ubuf1[i] - (128 << 12)) >> 3;
1097
            int V  = (vbuf0[i] + vbuf1[i] - (128 << 12)) >> 3;
1098
            int R, G, B;
1099

1100
            Y -= c->yuv2rgb_y_offset;
1101
            Y *= c->yuv2rgb_y_coeff;
1102
            Y += 1 << 13;
1103

1104
            if (hasAlpha) {
1105
                A = abuf0[i] << 11;
1106

1107
                A += 1 << 13;
1108
            }
1109

1110
            R = V * c->yuv2rgb_v2r_coeff;
1111
            G = V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1112
            B =                            U * c->yuv2rgb_u2b_coeff;
1113

1114
            output_pixel(&dest[0], av_clip_uintp2(R_B + Y, 30) >> 14);
1115
            output_pixel(&dest[1], av_clip_uintp2(  G + Y, 30) >> 14);
1116
            output_pixel(&dest[2], av_clip_uintp2(B_R + Y, 30) >> 14);
1117
            if (eightbytes) {
1118
                output_pixel(&dest[3], av_clip_uintp2(A, 30) >> 14);
1119
                dest += 4;
1120
            } else {
1121
                dest += 3;
1122
            }
1123
        }
1124
    }
1125
}
1126

1127
#undef output_pixel
1128
#undef r_b
1129
#undef b_r
1130

1131
#define YUV2PACKED16WRAPPER(name, base, ext, fmt, hasAlpha, eightbytes) \
1132
static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
1133
                        const int16_t **_lumSrc, int lumFilterSize, \
1134
                        const int16_t *chrFilter, const int16_t **_chrUSrc, \
1135
                        const int16_t **_chrVSrc, int chrFilterSize, \
1136
                        const int16_t **_alpSrc, uint8_t *_dest, int dstW, \
1137
                        int y) \
1138
{ \
1139
    const int32_t **lumSrc  = (const int32_t **) _lumSrc, \
1140
                  **chrUSrc = (const int32_t **) _chrUSrc, \
1141
                  **chrVSrc = (const int32_t **) _chrVSrc, \
1142
                  **alpSrc  = (const int32_t **) _alpSrc; \
1143
    uint16_t *dest = (uint16_t *) _dest; \
1144
    name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
1145
                          chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
1146
                          alpSrc, dest, dstW, y, fmt, hasAlpha, eightbytes); \
1147
} \
1148
 \
1149
static void name ## ext ## _2_c(SwsContext *c, const int16_t *_buf[2], \
1150
                        const int16_t *_ubuf[2], const int16_t *_vbuf[2], \
1151
                        const int16_t *_abuf[2], uint8_t *_dest, int dstW, \
1152
                        int yalpha, int uvalpha, int y) \
1153
{ \
1154
    const int32_t **buf  = (const int32_t **) _buf, \
1155
                  **ubuf = (const int32_t **) _ubuf, \
1156
                  **vbuf = (const int32_t **) _vbuf, \
1157
                  **abuf = (const int32_t **) _abuf; \
1158
    uint16_t *dest = (uint16_t *) _dest; \
1159
    name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
1160
                          dest, dstW, yalpha, uvalpha, y, fmt, hasAlpha, eightbytes); \
1161
} \
1162
 \
1163
static void name ## ext ## _1_c(SwsContext *c, const int16_t *_buf0, \
1164
                        const int16_t *_ubuf[2], const int16_t *_vbuf[2], \
1165
                        const int16_t *_abuf0, uint8_t *_dest, int dstW, \
1166
                        int uvalpha, int y) \
1167
{ \
1168
    const int32_t *buf0  = (const int32_t *)  _buf0, \
1169
                 **ubuf  = (const int32_t **) _ubuf, \
1170
                 **vbuf  = (const int32_t **) _vbuf, \
1171
                  *abuf0 = (const int32_t *)  _abuf0; \
1172
    uint16_t *dest = (uint16_t *) _dest; \
1173
    name ## base ## _1_c_template(c, buf0, ubuf, vbuf, abuf0, dest, \
1174
                                  dstW, uvalpha, y, fmt, hasAlpha, eightbytes); \
1175
}
1176

1177
YUV2PACKED16WRAPPER(yuv2, rgba64, rgb48be, AV_PIX_FMT_RGB48BE, 0, 0)
1178
YUV2PACKED16WRAPPER(yuv2, rgba64, rgb48le, AV_PIX_FMT_RGB48LE, 0, 0)
1179
YUV2PACKED16WRAPPER(yuv2, rgba64, bgr48be, AV_PIX_FMT_BGR48BE, 0, 0)
1180
YUV2PACKED16WRAPPER(yuv2, rgba64, bgr48le, AV_PIX_FMT_BGR48LE, 0, 0)
1181
YUV2PACKED16WRAPPER(yuv2, rgba64, rgba64be, AV_PIX_FMT_RGBA64BE, 1, 1)
1182
YUV2PACKED16WRAPPER(yuv2, rgba64, rgba64le, AV_PIX_FMT_RGBA64LE, 1, 1)
1183
YUV2PACKED16WRAPPER(yuv2, rgba64, rgbx64be, AV_PIX_FMT_RGBA64BE, 0, 1)
1184
YUV2PACKED16WRAPPER(yuv2, rgba64, rgbx64le, AV_PIX_FMT_RGBA64LE, 0, 1)
1185
YUV2PACKED16WRAPPER(yuv2, rgba64, bgra64be, AV_PIX_FMT_BGRA64BE, 1, 1)
1186
YUV2PACKED16WRAPPER(yuv2, rgba64, bgra64le, AV_PIX_FMT_BGRA64LE, 1, 1)
1187
YUV2PACKED16WRAPPER(yuv2, rgba64, bgrx64be, AV_PIX_FMT_BGRA64BE, 0, 1)
1188
YUV2PACKED16WRAPPER(yuv2, rgba64, bgrx64le, AV_PIX_FMT_BGRA64LE, 0, 1)
1189

1190
YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgb48be_full, AV_PIX_FMT_RGB48BE, 0, 0)
1191
YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgb48le_full, AV_PIX_FMT_RGB48LE, 0, 0)
1192
YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgr48be_full, AV_PIX_FMT_BGR48BE, 0, 0)
1193
YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgr48le_full, AV_PIX_FMT_BGR48LE, 0, 0)
1194
YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgba64be_full, AV_PIX_FMT_RGBA64BE, 1, 1)
1195
YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgba64le_full, AV_PIX_FMT_RGBA64LE, 1, 1)
1196
YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgbx64be_full, AV_PIX_FMT_RGBA64BE, 0, 1)
1197
YUV2PACKED16WRAPPER(yuv2, rgba64_full, rgbx64le_full, AV_PIX_FMT_RGBA64LE, 0, 1)
1198
YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgra64be_full, AV_PIX_FMT_BGRA64BE, 1, 1)
1199
YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgra64le_full, AV_PIX_FMT_BGRA64LE, 1, 1)
1200
YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgrx64be_full, AV_PIX_FMT_BGRA64BE, 0, 1)
1201
YUV2PACKED16WRAPPER(yuv2, rgba64_full, bgrx64le_full, AV_PIX_FMT_BGRA64LE, 0, 1)
1202

1203
/*
1204
 * Write out 2 RGB pixels in the target pixel format. This function takes a
1205
 * R/G/B LUT as generated by ff_yuv2rgb_c_init_tables(), which takes care of
1206
 * things like endianness conversion and shifting. The caller takes care of
1207
 * setting the correct offset in these tables from the chroma (U/V) values.
1208
 * This function then uses the luminance (Y1/Y2) values to write out the
1209
 * correct RGB values into the destination buffer.
1210
 */
1211
static av_always_inline void
1212
yuv2rgb_write(uint8_t *_dest, int i, int Y1, int Y2,
1213
              unsigned A1, unsigned A2,
1214
              const void *_r, const void *_g, const void *_b, int y,
1215
              enum AVPixelFormat target, int hasAlpha)
1216
{
1217
    if (target == AV_PIX_FMT_ARGB || target == AV_PIX_FMT_RGBA ||
1218
        target == AV_PIX_FMT_ABGR || target == AV_PIX_FMT_BGRA) {
1219
        uint32_t *dest = (uint32_t *) _dest;
1220
        const uint32_t *r = (const uint32_t *) _r;
1221
        const uint32_t *g = (const uint32_t *) _g;
1222
        const uint32_t *b = (const uint32_t *) _b;
1223

1224
#if CONFIG_SMALL
1225
        int sh = hasAlpha ? ((target == AV_PIX_FMT_RGB32_1 || target == AV_PIX_FMT_BGR32_1) ? 0 : 24) : 0;
1226

1227
        dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1] + (hasAlpha ? A1 << sh : 0);
1228
        dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2] + (hasAlpha ? A2 << sh : 0);
1229
#else
1230
        if (hasAlpha) {
1231
            int sh = (target == AV_PIX_FMT_RGB32_1 || target == AV_PIX_FMT_BGR32_1) ? 0 : 24;
1232

1233
            av_assert2((((r[Y1] + g[Y1] + b[Y1]) >> sh) & 0xFF) == 0);
1234
            dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1] + (A1 << sh);
1235
            dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2] + (A2 << sh);
1236
        } else {
1237
#if defined(ASSERT_LEVEL) && ASSERT_LEVEL > 1
1238
            int sh = (target == AV_PIX_FMT_RGB32_1 || target == AV_PIX_FMT_BGR32_1) ? 0 : 24;
1239

1240
            av_assert2((((r[Y1] + g[Y1] + b[Y1]) >> sh) & 0xFF) == 0xFF);
1241
#endif
1242
            dest[i * 2 + 0] = r[Y1] + g[Y1] + b[Y1];
1243
            dest[i * 2 + 1] = r[Y2] + g[Y2] + b[Y2];
1244
        }
1245
#endif
1246
    } else if (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) {
1247
        uint8_t *dest = (uint8_t *) _dest;
1248
        const uint8_t *r = (const uint8_t *) _r;
1249
        const uint8_t *g = (const uint8_t *) _g;
1250
        const uint8_t *b = (const uint8_t *) _b;
1251

1252
#define r_b ((target == AV_PIX_FMT_RGB24) ? r : b)
1253
#define b_r ((target == AV_PIX_FMT_RGB24) ? b : r)
1254

1255
        dest[i * 6 + 0] = r_b[Y1];
1256
        dest[i * 6 + 1] =   g[Y1];
1257
        dest[i * 6 + 2] = b_r[Y1];
1258
        dest[i * 6 + 3] = r_b[Y2];
1259
        dest[i * 6 + 4] =   g[Y2];
1260
        dest[i * 6 + 5] = b_r[Y2];
1261
#undef r_b
1262
#undef b_r
1263
    } else if (target == AV_PIX_FMT_RGB565 || target == AV_PIX_FMT_BGR565 ||
1264
               target == AV_PIX_FMT_RGB555 || target == AV_PIX_FMT_BGR555 ||
1265
               target == AV_PIX_FMT_RGB444 || target == AV_PIX_FMT_BGR444) {
1266
        uint16_t *dest = (uint16_t *) _dest;
1267
        const uint16_t *r = (const uint16_t *) _r;
1268
        const uint16_t *g = (const uint16_t *) _g;
1269
        const uint16_t *b = (const uint16_t *) _b;
1270
        int dr1, dg1, db1, dr2, dg2, db2;
1271

1272
        if (target == AV_PIX_FMT_RGB565 || target == AV_PIX_FMT_BGR565) {
1273
            dr1 = ff_dither_2x2_8[ y & 1     ][0];
1274
            dg1 = ff_dither_2x2_4[ y & 1     ][0];
1275
            db1 = ff_dither_2x2_8[(y & 1) ^ 1][0];
1276
            dr2 = ff_dither_2x2_8[ y & 1     ][1];
1277
            dg2 = ff_dither_2x2_4[ y & 1     ][1];
1278
            db2 = ff_dither_2x2_8[(y & 1) ^ 1][1];
1279
        } else if (target == AV_PIX_FMT_RGB555 || target == AV_PIX_FMT_BGR555) {
1280
            dr1 = ff_dither_2x2_8[ y & 1     ][0];
1281
            dg1 = ff_dither_2x2_8[ y & 1     ][1];
1282
            db1 = ff_dither_2x2_8[(y & 1) ^ 1][0];
1283
            dr2 = ff_dither_2x2_8[ y & 1     ][1];
1284
            dg2 = ff_dither_2x2_8[ y & 1     ][0];
1285
            db2 = ff_dither_2x2_8[(y & 1) ^ 1][1];
1286
        } else {
1287
            dr1 = ff_dither_4x4_16[ y & 3     ][0];
1288
            dg1 = ff_dither_4x4_16[ y & 3     ][1];
1289
            db1 = ff_dither_4x4_16[(y & 3) ^ 3][0];
1290
            dr2 = ff_dither_4x4_16[ y & 3     ][1];
1291
            dg2 = ff_dither_4x4_16[ y & 3     ][0];
1292
            db2 = ff_dither_4x4_16[(y & 3) ^ 3][1];
1293
        }
1294

1295
        dest[i * 2 + 0] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1];
1296
        dest[i * 2 + 1] = r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2];
1297
    } else /* 8/4-bit */ {
1298
        uint8_t *dest = (uint8_t *) _dest;
1299
        const uint8_t *r = (const uint8_t *) _r;
1300
        const uint8_t *g = (const uint8_t *) _g;
1301
        const uint8_t *b = (const uint8_t *) _b;
1302
        int dr1, dg1, db1, dr2, dg2, db2;
1303

1304
        if (target == AV_PIX_FMT_RGB8 || target == AV_PIX_FMT_BGR8) {
1305
            const uint8_t * const d64 = ff_dither_8x8_73[y & 7];
1306
            const uint8_t * const d32 = ff_dither_8x8_32[y & 7];
1307
            dr1 = dg1 = d32[(i * 2 + 0) & 7];
1308
            db1 =       d64[(i * 2 + 0) & 7];
1309
            dr2 = dg2 = d32[(i * 2 + 1) & 7];
1310
            db2 =       d64[(i * 2 + 1) & 7];
1311
        } else {
1312
            const uint8_t * const d64  = ff_dither_8x8_73 [y & 7];
1313
            const uint8_t * const d128 = ff_dither_8x8_220[y & 7];
1314
            dr1 = db1 = d128[(i * 2 + 0) & 7];
1315
            dg1 =        d64[(i * 2 + 0) & 7];
1316
            dr2 = db2 = d128[(i * 2 + 1) & 7];
1317
            dg2 =        d64[(i * 2 + 1) & 7];
1318
        }
1319

1320
        if (target == AV_PIX_FMT_RGB4 || target == AV_PIX_FMT_BGR4) {
1321
            dest[i] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1] +
1322
                    ((r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2]) << 4);
1323
        } else {
1324
            dest[i * 2 + 0] = r[Y1 + dr1] + g[Y1 + dg1] + b[Y1 + db1];
1325
            dest[i * 2 + 1] = r[Y2 + dr2] + g[Y2 + dg2] + b[Y2 + db2];
1326
        }
1327
    }
1328
}
1329

1330
static av_always_inline void
1331
yuv2rgb_X_c_template(SwsContext *c, const int16_t *lumFilter,
1332
                     const int16_t **lumSrc, int lumFilterSize,
1333
                     const int16_t *chrFilter, const int16_t **chrUSrc,
1334
                     const int16_t **chrVSrc, int chrFilterSize,
1335
                     const int16_t **alpSrc, uint8_t *dest, int dstW,
1336
                     int y, enum AVPixelFormat target, int hasAlpha)
1337
{
1338
    int i;
1339

1340
    for (i = 0; i < ((dstW + 1) >> 1); i++) {
1341
        int j, A1, A2;
1342
        int Y1 = 1 << 18;
1343
        int Y2 = 1 << 18;
1344
        int U  = 1 << 18;
1345
        int V  = 1 << 18;
1346
        const void *r, *g, *b;
1347

1348
        for (j = 0; j < lumFilterSize; j++) {
1349
            Y1 += lumSrc[j][i * 2]     * lumFilter[j];
1350
            Y2 += lumSrc[j][i * 2 + 1] * lumFilter[j];
1351
        }
1352
        for (j = 0; j < chrFilterSize; j++) {
1353
            U += chrUSrc[j][i] * chrFilter[j];
1354
            V += chrVSrc[j][i] * chrFilter[j];
1355
        }
1356
        Y1 >>= 19;
1357
        Y2 >>= 19;
1358
        U  >>= 19;
1359
        V  >>= 19;
1360
        if (hasAlpha) {
1361
            A1 = 1 << 18;
1362
            A2 = 1 << 18;
1363
            for (j = 0; j < lumFilterSize; j++) {
1364
                A1 += alpSrc[j][i * 2    ] * lumFilter[j];
1365
                A2 += alpSrc[j][i * 2 + 1] * lumFilter[j];
1366
            }
1367
            A1 >>= 19;
1368
            A2 >>= 19;
1369
            if ((A1 | A2) & 0x100) {
1370
                A1 = av_clip_uint8(A1);
1371
                A2 = av_clip_uint8(A2);
1372
            }
1373
        }
1374

1375
        r =  c->table_rV[V + YUVRGB_TABLE_HEADROOM];
1376
        g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]);
1377
        b =  c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1378

1379
        yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1380
                      r, g, b, y, target, hasAlpha);
1381
    }
1382
}
1383

1384
static av_always_inline void
1385
yuv2rgb_2_c_template(SwsContext *c, const int16_t *buf[2],
1386
                     const int16_t *ubuf[2], const int16_t *vbuf[2],
1387
                     const int16_t *abuf[2], uint8_t *dest, int dstW,
1388
                     int yalpha, int uvalpha, int y,
1389
                     enum AVPixelFormat target, int hasAlpha)
1390
{
1391
    const int16_t *buf0  = buf[0],  *buf1  = buf[1],
1392
                  *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
1393
                  *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
1394
                  *abuf0 = hasAlpha ? abuf[0] : NULL,
1395
                  *abuf1 = hasAlpha ? abuf[1] : NULL;
1396
    int  yalpha1 = 4096 - yalpha;
1397
    int uvalpha1 = 4096 - uvalpha;
1398
    int i;
1399
    av_assert2(yalpha  <= 4096U);
1400
    av_assert2(uvalpha <= 4096U);
1401

1402
    for (i = 0; i < ((dstW + 1) >> 1); i++) {
1403
        int Y1 = (buf0[i * 2]     * yalpha1  + buf1[i * 2]     * yalpha)  >> 19;
1404
        int Y2 = (buf0[i * 2 + 1] * yalpha1  + buf1[i * 2 + 1] * yalpha)  >> 19;
1405
        int U  = (ubuf0[i]        * uvalpha1 + ubuf1[i]        * uvalpha) >> 19;
1406
        int V  = (vbuf0[i]        * uvalpha1 + vbuf1[i]        * uvalpha) >> 19;
1407
        int A1, A2;
1408
        const void *r =  c->table_rV[V + YUVRGB_TABLE_HEADROOM],
1409
                   *g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]),
1410
                   *b =  c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1411

1412
        if (hasAlpha) {
1413
            A1 = (abuf0[i * 2    ] * yalpha1 + abuf1[i * 2    ] * yalpha) >> 19;
1414
            A2 = (abuf0[i * 2 + 1] * yalpha1 + abuf1[i * 2 + 1] * yalpha) >> 19;
1415
            A1 = av_clip_uint8(A1);
1416
            A2 = av_clip_uint8(A2);
1417
        }
1418

1419
        yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1420
                      r, g, b, y, target, hasAlpha);
1421
    }
1422
}
1423

1424
static av_always_inline void
1425
yuv2rgb_1_c_template(SwsContext *c, const int16_t *buf0,
1426
                     const int16_t *ubuf[2], const int16_t *vbuf[2],
1427
                     const int16_t *abuf0, uint8_t *dest, int dstW,
1428
                     int uvalpha, int y, enum AVPixelFormat target,
1429
                     int hasAlpha)
1430
{
1431
    const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1432
    int i;
1433

1434
    if (uvalpha < 2048) {
1435
        for (i = 0; i < ((dstW + 1) >> 1); i++) {
1436
            int Y1 = (buf0[i * 2    ] + 64) >> 7;
1437
            int Y2 = (buf0[i * 2 + 1] + 64) >> 7;
1438
            int U  = (ubuf0[i]        + 64) >> 7;
1439
            int V  = (vbuf0[i]        + 64) >> 7;
1440
            int A1, A2;
1441
            const void *r =  c->table_rV[V + YUVRGB_TABLE_HEADROOM],
1442
                       *g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]),
1443
                       *b =  c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1444

1445
            if (hasAlpha) {
1446
                A1 = abuf0[i * 2    ] * 255 + 16384 >> 15;
1447
                A2 = abuf0[i * 2 + 1] * 255 + 16384 >> 15;
1448
                A1 = av_clip_uint8(A1);
1449
                A2 = av_clip_uint8(A2);
1450
            }
1451

1452
            yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1453
                          r, g, b, y, target, hasAlpha);
1454
        }
1455
    } else {
1456
        const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1457
        for (i = 0; i < ((dstW + 1) >> 1); i++) {
1458
            int Y1 = (buf0[i * 2    ]     +  64) >> 7;
1459
            int Y2 = (buf0[i * 2 + 1]     +  64) >> 7;
1460
            int U  = (ubuf0[i] + ubuf1[i] + 128) >> 8;
1461
            int V  = (vbuf0[i] + vbuf1[i] + 128) >> 8;
1462
            int A1, A2;
1463
            const void *r =  c->table_rV[V + YUVRGB_TABLE_HEADROOM],
1464
                       *g = (c->table_gU[U + YUVRGB_TABLE_HEADROOM] + c->table_gV[V + YUVRGB_TABLE_HEADROOM]),
1465
                       *b =  c->table_bU[U + YUVRGB_TABLE_HEADROOM];
1466

1467
            if (hasAlpha) {
1468
                A1 = (abuf0[i * 2    ] + 64) >> 7;
1469
                A2 = (abuf0[i * 2 + 1] + 64) >> 7;
1470
                A1 = av_clip_uint8(A1);
1471
                A2 = av_clip_uint8(A2);
1472
            }
1473

1474
            yuv2rgb_write(dest, i, Y1, Y2, hasAlpha ? A1 : 0, hasAlpha ? A2 : 0,
1475
                          r, g, b, y, target, hasAlpha);
1476
        }
1477
    }
1478
}
1479

1480
#define YUV2RGBWRAPPERX(name, base, ext, fmt, hasAlpha) \
1481
static void name ## ext ## _X_c(SwsContext *c, const int16_t *lumFilter, \
1482
                                const int16_t **lumSrc, int lumFilterSize, \
1483
                                const int16_t *chrFilter, const int16_t **chrUSrc, \
1484
                                const int16_t **chrVSrc, int chrFilterSize, \
1485
                                const int16_t **alpSrc, uint8_t *dest, int dstW, \
1486
                                int y) \
1487
{ \
1488
    name ## base ## _X_c_template(c, lumFilter, lumSrc, lumFilterSize, \
1489
                                  chrFilter, chrUSrc, chrVSrc, chrFilterSize, \
1490
                                  alpSrc, dest, dstW, y, fmt, hasAlpha); \
1491
}
1492

1493
#define YUV2RGBWRAPPERX2(name, base, ext, fmt, hasAlpha) \
1494
YUV2RGBWRAPPERX(name, base, ext, fmt, hasAlpha) \
1495
static void name ## ext ## _2_c(SwsContext *c, const int16_t *buf[2], \
1496
                                const int16_t *ubuf[2], const int16_t *vbuf[2], \
1497
                                const int16_t *abuf[2], uint8_t *dest, int dstW, \
1498
                                int yalpha, int uvalpha, int y) \
1499
{ \
1500
    name ## base ## _2_c_template(c, buf, ubuf, vbuf, abuf, \
1501
                                  dest, dstW, yalpha, uvalpha, y, fmt, hasAlpha); \
1502
}
1503

1504
#define YUV2RGBWRAPPER(name, base, ext, fmt, hasAlpha) \
1505
YUV2RGBWRAPPERX2(name, base, ext, fmt, hasAlpha) \
1506
static void name ## ext ## _1_c(SwsContext *c, const int16_t *buf0, \
1507
                                const int16_t *ubuf[2], const int16_t *vbuf[2], \
1508
                                const int16_t *abuf0, uint8_t *dest, int dstW, \
1509
                                int uvalpha, int y) \
1510
{ \
1511
    name ## base ## _1_c_template(c, buf0, ubuf, vbuf, abuf0, dest, \
1512
                                  dstW, uvalpha, y, fmt, hasAlpha); \
1513
}
1514

1515
#if CONFIG_SMALL
1516
YUV2RGBWRAPPER(yuv2rgb,,  32_1,  AV_PIX_FMT_RGB32_1,   CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1517
YUV2RGBWRAPPER(yuv2rgb,,  32,    AV_PIX_FMT_RGB32,     CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1518
#else
1519
#if CONFIG_SWSCALE_ALPHA
1520
YUV2RGBWRAPPER(yuv2rgb,, a32_1,  AV_PIX_FMT_RGB32_1,   1)
1521
YUV2RGBWRAPPER(yuv2rgb,, a32,    AV_PIX_FMT_RGB32,     1)
1522
#endif
1523
YUV2RGBWRAPPER(yuv2rgb,, x32_1,  AV_PIX_FMT_RGB32_1,   0)
1524
YUV2RGBWRAPPER(yuv2rgb,, x32,    AV_PIX_FMT_RGB32,     0)
1525
#endif
1526
YUV2RGBWRAPPER(yuv2, rgb, rgb24, AV_PIX_FMT_RGB24,   0)
1527
YUV2RGBWRAPPER(yuv2, rgb, bgr24, AV_PIX_FMT_BGR24,   0)
1528
YUV2RGBWRAPPER(yuv2rgb,,  16,    AV_PIX_FMT_RGB565,    0)
1529
YUV2RGBWRAPPER(yuv2rgb,,  15,    AV_PIX_FMT_RGB555,    0)
1530
YUV2RGBWRAPPER(yuv2rgb,,  12,    AV_PIX_FMT_RGB444,    0)
1531
YUV2RGBWRAPPER(yuv2rgb,,   8,    AV_PIX_FMT_RGB8,      0)
1532
YUV2RGBWRAPPER(yuv2rgb,,   4,    AV_PIX_FMT_RGB4,      0)
1533
YUV2RGBWRAPPER(yuv2rgb,,   4b,   AV_PIX_FMT_RGB4_BYTE, 0)
1534

1535
static av_always_inline void yuv2rgb_write_full(SwsContext *c,
1536
    uint8_t *dest, int i, int Y, int A, int U, int V,
1537
    int y, enum AVPixelFormat target, int hasAlpha, int err[4])
1538
{
1539
    int R, G, B;
1540
    int isrgb8 = target == AV_PIX_FMT_BGR8 || target == AV_PIX_FMT_RGB8;
1541

1542
    Y -= c->yuv2rgb_y_offset;
1543
    Y *= c->yuv2rgb_y_coeff;
1544
    Y += 1 << 21;
1545
    R = Y + V*c->yuv2rgb_v2r_coeff;
1546
    G = Y + V*c->yuv2rgb_v2g_coeff + U*c->yuv2rgb_u2g_coeff;
1547
    B = Y +                          U*c->yuv2rgb_u2b_coeff;
1548
    if ((R | G | B) & 0xC0000000) {
1549
        R = av_clip_uintp2(R, 30);
1550
        G = av_clip_uintp2(G, 30);
1551
        B = av_clip_uintp2(B, 30);
1552
    }
1553

1554
    switch(target) {
1555
    case AV_PIX_FMT_ARGB:
1556
        dest[0] = hasAlpha ? A : 255;
1557
        dest[1] = R >> 22;
1558
        dest[2] = G >> 22;
1559
        dest[3] = B >> 22;
1560
        break;
1561
    case AV_PIX_FMT_RGB24:
1562
        dest[0] = R >> 22;
1563
        dest[1] = G >> 22;
1564
        dest[2] = B >> 22;
1565
        break;
1566
    case AV_PIX_FMT_RGBA:
1567
        dest[0] = R >> 22;
1568
        dest[1] = G >> 22;
1569
        dest[2] = B >> 22;
1570
        dest[3] = hasAlpha ? A : 255;
1571
        break;
1572
    case AV_PIX_FMT_ABGR:
1573
        dest[0] = hasAlpha ? A : 255;
1574
        dest[1] = B >> 22;
1575
        dest[2] = G >> 22;
1576
        dest[3] = R >> 22;
1577
        break;
1578
    case AV_PIX_FMT_BGR24:
1579
        dest[0] = B >> 22;
1580
        dest[1] = G >> 22;
1581
        dest[2] = R >> 22;
1582
        break;
1583
    case AV_PIX_FMT_BGRA:
1584
        dest[0] = B >> 22;
1585
        dest[1] = G >> 22;
1586
        dest[2] = R >> 22;
1587
        dest[3] = hasAlpha ? A : 255;
1588
        break;
1589
    case AV_PIX_FMT_BGR4_BYTE:
1590
    case AV_PIX_FMT_RGB4_BYTE:
1591
    case AV_PIX_FMT_BGR8:
1592
    case AV_PIX_FMT_RGB8:
1593
    {
1594
        int r,g,b;
1595

1596
        switch (c->dither) {
1597
        default:
1598
        case SWS_DITHER_AUTO:
1599
        case SWS_DITHER_ED:
1600
            R >>= 22;
1601
            G >>= 22;
1602
            B >>= 22;
1603
            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;
1604
            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;
1605
            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;
1606
            c->dither_error[0][i] = err[0];
1607
            c->dither_error[1][i] = err[1];
1608
            c->dither_error[2][i] = err[2];
1609
            r = R >> (isrgb8 ? 5 : 7);
1610
            g = G >> (isrgb8 ? 5 : 6);
1611
            b = B >> (isrgb8 ? 6 : 7);
1612
            r = av_clip(r, 0, isrgb8 ? 7 : 1);
1613
            g = av_clip(g, 0, isrgb8 ? 7 : 3);
1614
            b = av_clip(b, 0, isrgb8 ? 3 : 1);
1615
            err[0] = R - r*(isrgb8 ? 36 : 255);
1616
            err[1] = G - g*(isrgb8 ? 36 : 85);
1617
            err[2] = B - b*(isrgb8 ? 85 : 255);
1618
            break;
1619
        case SWS_DITHER_A_DITHER:
1620
            if (isrgb8) {
1621
  /* see http://pippin.gimp.org/a_dither/ for details/origin */
1622
#define A_DITHER(u,v)   (((((u)+((v)*236))*119)&0xff))
1623
                r = (((R >> 19) + A_DITHER(i,y)  -96)>>8);
1624
                g = (((G >> 19) + A_DITHER(i + 17,y) - 96)>>8);
1625
                b = (((B >> 20) + A_DITHER(i + 17*2,y) -96)>>8);
1626
                r = av_clip_uintp2(r, 3);
1627
                g = av_clip_uintp2(g, 3);
1628
                b = av_clip_uintp2(b, 2);
1629
            } else {
1630
                r = (((R >> 21) + A_DITHER(i,y)-256)>>8);
1631
                g = (((G >> 19) + A_DITHER(i + 17,y)-256)>>8);
1632
                b = (((B >> 21) + A_DITHER(i + 17*2,y)-256)>>8);
1633
                r = av_clip_uintp2(r, 1);
1634
                g = av_clip_uintp2(g, 2);
1635
                b = av_clip_uintp2(b, 1);
1636
            }
1637
            break;
1638
        case SWS_DITHER_X_DITHER:
1639
            if (isrgb8) {
1640
  /* see http://pippin.gimp.org/a_dither/ for details/origin */
1641
#define X_DITHER(u,v)   (((((u)^((v)*237))*181)&0x1ff)/2)
1642
                r = (((R >> 19) + X_DITHER(i,y) - 96)>>8);
1643
                g = (((G >> 19) + X_DITHER(i + 17,y) - 96)>>8);
1644
                b = (((B >> 20) + X_DITHER(i + 17*2,y) - 96)>>8);
1645
                r = av_clip_uintp2(r, 3);
1646
                g = av_clip_uintp2(g, 3);
1647
                b = av_clip_uintp2(b, 2);
1648
            } else {
1649
                r = (((R >> 21) + X_DITHER(i,y)-256)>>8);
1650
                g = (((G >> 19) + X_DITHER(i + 17,y)-256)>>8);
1651
                b = (((B >> 21) + X_DITHER(i + 17*2,y)-256)>>8);
1652
                r = av_clip_uintp2(r, 1);
1653
                g = av_clip_uintp2(g, 2);
1654
                b = av_clip_uintp2(b, 1);
1655
            }
1656

1657
            break;
1658
        }
1659

1660
        if(target == AV_PIX_FMT_BGR4_BYTE) {
1661
            dest[0] = r + 2*g + 8*b;
1662
        } else if(target == AV_PIX_FMT_RGB4_BYTE) {
1663
            dest[0] = b + 2*g + 8*r;
1664
        } else if(target == AV_PIX_FMT_BGR8) {
1665
            dest[0] = r + 8*g + 64*b;
1666
        } else if(target == AV_PIX_FMT_RGB8) {
1667
            dest[0] = b + 4*g + 32*r;
1668
        } else
1669
            av_assert2(0);
1670
        break;}
1671
    }
1672
}
1673

1674
static av_always_inline void
1675
yuv2rgb_full_X_c_template(SwsContext *c, const int16_t *lumFilter,
1676
                          const int16_t **lumSrc, int lumFilterSize,
1677
                          const int16_t *chrFilter, const int16_t **chrUSrc,
1678
                          const int16_t **chrVSrc, int chrFilterSize,
1679
                          const int16_t **alpSrc, uint8_t *dest,
1680
                          int dstW, int y, enum AVPixelFormat target, int hasAlpha)
1681
{
1682
    int i;
1683
    int step = (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) ? 3 : 4;
1684
    int err[4] = {0};
1685
    int A = 0; //init to silence warning
1686

1687
    if(   target == AV_PIX_FMT_BGR4_BYTE || target == AV_PIX_FMT_RGB4_BYTE
1688
       || target == AV_PIX_FMT_BGR8      || target == AV_PIX_FMT_RGB8)
1689
        step = 1;
1690

1691
    for (i = 0; i < dstW; i++) {
1692
        int j;
1693
        int Y = 1<<9;
1694
        int U = (1<<9)-(128 << 19);
1695
        int V = (1<<9)-(128 << 19);
1696

1697
        for (j = 0; j < lumFilterSize; j++) {
1698
            Y += lumSrc[j][i] * lumFilter[j];
1699
        }
1700
        for (j = 0; j < chrFilterSize; j++) {
1701
            U += chrUSrc[j][i] * chrFilter[j];
1702
            V += chrVSrc[j][i] * chrFilter[j];
1703
        }
1704
        Y >>= 10;
1705
        U >>= 10;
1706
        V >>= 10;
1707
        if (hasAlpha) {
1708
            A = 1 << 18;
1709
            for (j = 0; j < lumFilterSize; j++) {
1710
                A += alpSrc[j][i] * lumFilter[j];
1711
            }
1712
            A >>= 19;
1713
            if (A & 0x100)
1714
                A = av_clip_uint8(A);
1715
        }
1716
        yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
1717
        dest += step;
1718
    }
1719
    c->dither_error[0][i] = err[0];
1720
    c->dither_error[1][i] = err[1];
1721
    c->dither_error[2][i] = err[2];
1722
}
1723

1724
static av_always_inline void
1725
yuv2rgb_full_2_c_template(SwsContext *c, const int16_t *buf[2],
1726
                     const int16_t *ubuf[2], const int16_t *vbuf[2],
1727
                     const int16_t *abuf[2], uint8_t *dest, int dstW,
1728
                     int yalpha, int uvalpha, int y,
1729
                     enum AVPixelFormat target, int hasAlpha)
1730
{
1731
    const int16_t *buf0  = buf[0],  *buf1  = buf[1],
1732
                  *ubuf0 = ubuf[0], *ubuf1 = ubuf[1],
1733
                  *vbuf0 = vbuf[0], *vbuf1 = vbuf[1],
1734
                  *abuf0 = hasAlpha ? abuf[0] : NULL,
1735
                  *abuf1 = hasAlpha ? abuf[1] : NULL;
1736
    int  yalpha1 = 4096 - yalpha;
1737
    int uvalpha1 = 4096 - uvalpha;
1738
    int i;
1739
    int step = (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) ? 3 : 4;
1740
    int err[4] = {0};
1741
    int A = 0; // init to silcene warning
1742

1743
    av_assert2(yalpha  <= 4096U);
1744
    av_assert2(uvalpha <= 4096U);
1745

1746
    if(   target == AV_PIX_FMT_BGR4_BYTE || target == AV_PIX_FMT_RGB4_BYTE
1747
       || target == AV_PIX_FMT_BGR8      || target == AV_PIX_FMT_RGB8)
1748
        step = 1;
1749

1750
    for (i = 0; i < dstW; i++) {
1751
        int Y = ( buf0[i] * yalpha1  +  buf1[i] * yalpha             ) >> 10; //FIXME rounding
1752
        int U = (ubuf0[i] * uvalpha1 + ubuf1[i] * uvalpha-(128 << 19)) >> 10;
1753
        int V = (vbuf0[i] * uvalpha1 + vbuf1[i] * uvalpha-(128 << 19)) >> 10;
1754

1755
        if (hasAlpha) {
1756
            A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha + (1<<18)) >> 19;
1757
            if (A & 0x100)
1758
                A = av_clip_uint8(A);
1759
        }
1760

1761
        yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
1762
        dest += step;
1763
    }
1764
    c->dither_error[0][i] = err[0];
1765
    c->dither_error[1][i] = err[1];
1766
    c->dither_error[2][i] = err[2];
1767
}
1768

1769
static av_always_inline void
1770
yuv2rgb_full_1_c_template(SwsContext *c, const int16_t *buf0,
1771
                     const int16_t *ubuf[2], const int16_t *vbuf[2],
1772
                     const int16_t *abuf0, uint8_t *dest, int dstW,
1773
                     int uvalpha, int y, enum AVPixelFormat target,
1774
                     int hasAlpha)
1775
{
1776
    const int16_t *ubuf0 = ubuf[0], *vbuf0 = vbuf[0];
1777
    int i;
1778
    int step = (target == AV_PIX_FMT_RGB24 || target == AV_PIX_FMT_BGR24) ? 3 : 4;
1779
    int err[4] = {0};
1780

1781
    if(   target == AV_PIX_FMT_BGR4_BYTE || target == AV_PIX_FMT_RGB4_BYTE
1782
       || target == AV_PIX_FMT_BGR8      || target == AV_PIX_FMT_RGB8)
1783
        step = 1;
1784

1785
    if (uvalpha < 2048) {
1786
        int A = 0; //init to silence warning
1787
        for (i = 0; i < dstW; i++) {
1788
            int Y = buf0[i] << 2;
1789
            int U = (ubuf0[i] - (128<<7)) * 4;
1790
            int V = (vbuf0[i] - (128<<7)) * 4;
1791

1792
            if (hasAlpha) {
1793
                A = (abuf0[i] + 64) >> 7;
1794
                if (A & 0x100)
1795
                    A = av_clip_uint8(A);
1796
            }
1797

1798
            yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
1799
            dest += step;
1800
        }
1801
    } else {
1802
        const int16_t *ubuf1 = ubuf[1], *vbuf1 = vbuf[1];
1803
        int A = 0; //init to silence warning
1804
        for (i = 0; i < dstW; i++) {
1805
            int Y = buf0[i] << 2;
1806
            int U = (ubuf0[i] + ubuf1[i] - (128<<8)) << 1;
1807
            int V = (vbuf0[i] + vbuf1[i] - (128<<8)) << 1;
1808

1809
            if (hasAlpha) {
1810
                A = (abuf0[i] + 64) >> 7;
1811
                if (A & 0x100)
1812
                    A = av_clip_uint8(A);
1813
            }
1814

1815
            yuv2rgb_write_full(c, dest, i, Y, A, U, V, y, target, hasAlpha, err);
1816
            dest += step;
1817
        }
1818
    }
1819

1820
    c->dither_error[0][i] = err[0];
1821
    c->dither_error[1][i] = err[1];
1822
    c->dither_error[2][i] = err[2];
1823
}
1824

1825
#if CONFIG_SMALL
1826
YUV2RGBWRAPPER(yuv2, rgb_full, bgra32_full, AV_PIX_FMT_BGRA,  CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1827
YUV2RGBWRAPPER(yuv2, rgb_full, abgr32_full, AV_PIX_FMT_ABGR,  CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1828
YUV2RGBWRAPPER(yuv2, rgb_full, rgba32_full, AV_PIX_FMT_RGBA,  CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1829
YUV2RGBWRAPPER(yuv2, rgb_full, argb32_full, AV_PIX_FMT_ARGB,  CONFIG_SWSCALE_ALPHA && c->alpPixBuf)
1830
#else
1831
#if CONFIG_SWSCALE_ALPHA
1832
YUV2RGBWRAPPER(yuv2, rgb_full, bgra32_full, AV_PIX_FMT_BGRA,  1)
1833
YUV2RGBWRAPPER(yuv2, rgb_full, abgr32_full, AV_PIX_FMT_ABGR,  1)
1834
YUV2RGBWRAPPER(yuv2, rgb_full, rgba32_full, AV_PIX_FMT_RGBA,  1)
1835
YUV2RGBWRAPPER(yuv2, rgb_full, argb32_full, AV_PIX_FMT_ARGB,  1)
1836
#endif
1837
YUV2RGBWRAPPER(yuv2, rgb_full, bgrx32_full, AV_PIX_FMT_BGRA,  0)
1838
YUV2RGBWRAPPER(yuv2, rgb_full, xbgr32_full, AV_PIX_FMT_ABGR,  0)
1839
YUV2RGBWRAPPER(yuv2, rgb_full, rgbx32_full, AV_PIX_FMT_RGBA,  0)
1840
YUV2RGBWRAPPER(yuv2, rgb_full, xrgb32_full, AV_PIX_FMT_ARGB,  0)
1841
#endif
1842
YUV2RGBWRAPPER(yuv2, rgb_full, bgr24_full,  AV_PIX_FMT_BGR24, 0)
1843
YUV2RGBWRAPPER(yuv2, rgb_full, rgb24_full,  AV_PIX_FMT_RGB24, 0)
1844

1845
YUV2RGBWRAPPER(yuv2, rgb_full, bgr4_byte_full,  AV_PIX_FMT_BGR4_BYTE, 0)
1846
YUV2RGBWRAPPER(yuv2, rgb_full, rgb4_byte_full,  AV_PIX_FMT_RGB4_BYTE, 0)
1847
YUV2RGBWRAPPER(yuv2, rgb_full, bgr8_full,   AV_PIX_FMT_BGR8,  0)
1848
YUV2RGBWRAPPER(yuv2, rgb_full, rgb8_full,   AV_PIX_FMT_RGB8,  0)
1849

1850
static void
1851
yuv2gbrp_full_X_c(SwsContext *c, const int16_t *lumFilter,
1852
                  const int16_t **lumSrc, int lumFilterSize,
1853
                  const int16_t *chrFilter, const int16_t **chrUSrc,
1854
                  const int16_t **chrVSrc, int chrFilterSize,
1855
                  const int16_t **alpSrc, uint8_t **dest,
1856
                  int dstW, int y)
1857
{
1858
    const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(c->dstFormat);
1859
    int i;
1860
    int hasAlpha = (desc->flags & AV_PIX_FMT_FLAG_ALPHA) && alpSrc;
1861
    uint16_t **dest16 = (uint16_t**)dest;
1862
    int SH = 22 + 8 - desc->comp[0].depth;
1863
    int A = 0; // init to silence warning
1864

1865
    for (i = 0; i < dstW; i++) {
1866
        int j;
1867
        int Y = 1 << 9;
1868
        int U = (1 << 9) - (128 << 19);
1869
        int V = (1 << 9) - (128 << 19);
1870
        int R, G, B;
1871

1872
        for (j = 0; j < lumFilterSize; j++)
1873
            Y += lumSrc[j][i] * lumFilter[j];
1874

1875
        for (j = 0; j < chrFilterSize; j++) {
1876
            U += chrUSrc[j][i] * chrFilter[j];
1877
            V += chrVSrc[j][i] * chrFilter[j];
1878
        }
1879

1880
        Y >>= 10;
1881
        U >>= 10;
1882
        V >>= 10;
1883

1884
        if (hasAlpha) {
1885
            A = 1 << 18;
1886

1887
            for (j = 0; j < lumFilterSize; j++)
1888
                A += alpSrc[j][i] * lumFilter[j];
1889

1890
            A >>= 19;
1891

1892
            if (A & 0x100)
1893
                A = av_clip_uint8(A);
1894
        }
1895

1896
        Y -= c->yuv2rgb_y_offset;
1897
        Y *= c->yuv2rgb_y_coeff;
1898
        Y += 1 << 21;
1899
        R = Y + V * c->yuv2rgb_v2r_coeff;
1900
        G = Y + V * c->yuv2rgb_v2g_coeff + U * c->yuv2rgb_u2g_coeff;
1901
        B = Y +                            U * c->yuv2rgb_u2b_coeff;
1902

1903
        if ((R | G | B) & 0xC0000000) {
1904
            R = av_clip_uintp2(R, 30);
1905
            G = av_clip_uintp2(G, 30);
1906
            B = av_clip_uintp2(B, 30);
1907
        }
1908

1909
        if (SH != 22) {
1910
            dest16[0][i] = G >> SH;
1911
            dest16[1][i] = B >> SH;
1912
            dest16[2][i] = R >> SH;
1913
            if (hasAlpha)
1914
                dest16[3][i] = A;
1915
        } else {
1916
            dest[0][i] = G >> 22;
1917
            dest[1][i] = B >> 22;
1918
            dest[2][i] = R >> 22;
1919
            if (hasAlpha)
1920
                dest[3][i] = A;
1921
        }
1922
    }
1923
    if (SH != 22 && (!isBE(c->dstFormat)) != (!HAVE_BIGENDIAN)) {
1924
        for (i = 0; i < dstW; i++) {
1925
            dest16[0][i] = av_bswap16(dest16[0][i]);
1926
            dest16[1][i] = av_bswap16(dest16[1][i]);
1927
            dest16[2][i] = av_bswap16(dest16[2][i]);
1928
            if (hasAlpha)
1929
                dest16[3][i] = av_bswap16(dest16[3][i]);
1930
        }
1931
    }
1932
}
1933

1934
static void
1935
yuv2ya8_1_c(SwsContext *c, const int16_t *buf0,
1936
            const int16_t *ubuf[2], const int16_t *vbuf[2],
1937
            const int16_t *abuf0, uint8_t *dest, int dstW,
1938
            int uvalpha, int y)
1939
{
1940
    int hasAlpha = !!abuf0;
1941
    int i;
1942

1943
    for (i = 0; i < dstW; i++) {
1944
        int Y = (buf0[i] + 64) >> 7;
1945
        int A;
1946

1947
        Y = av_clip_uint8(Y);
1948

1949
        if (hasAlpha) {
1950
            A = (abuf0[i] + 64) >> 7;
1951
            if (A & 0x100)
1952
                A = av_clip_uint8(A);
1953
        }
1954

1955
        dest[i * 2    ] = Y;
1956
        dest[i * 2 + 1] = hasAlpha ? A : 255;
1957
    }
1958
}
1959

1960
static void
1961
yuv2ya8_2_c(SwsContext *c, const int16_t *buf[2],
1962
            const int16_t *ubuf[2], const int16_t *vbuf[2],
1963
            const int16_t *abuf[2], uint8_t *dest, int dstW,
1964
            int yalpha, int uvalpha, int y)
1965
{
1966
    int hasAlpha = abuf && abuf[0] && abuf[1];
1967
    const int16_t *buf0  = buf[0],  *buf1  = buf[1],
1968
                  *abuf0 = hasAlpha ? abuf[0] : NULL,
1969
                  *abuf1 = hasAlpha ? abuf[1] : NULL;
1970
    int  yalpha1 = 4096 - yalpha;
1971
    int i;
1972

1973
    av_assert2(yalpha  <= 4096U);
1974

1975
    for (i = 0; i < dstW; i++) {
1976
        int Y = (buf0[i] * yalpha1 + buf1[i] * yalpha) >> 19;
1977
        int A;
1978

1979
        Y = av_clip_uint8(Y);
1980

1981
        if (hasAlpha) {
1982
            A = (abuf0[i] * yalpha1 + abuf1[i] * yalpha) >> 19;
1983
            A = av_clip_uint8(A);
1984
        }
1985

1986
        dest[i * 2    ] = Y;
1987
        dest[i * 2 + 1] = hasAlpha ? A : 255;
1988
    }
1989
}
1990

1991
static void
1992
yuv2ya8_X_c(SwsContext *c, const int16_t *lumFilter,
1993
            const int16_t **lumSrc, int lumFilterSize,
1994
            const int16_t *chrFilter, const int16_t **chrUSrc,
1995
            const int16_t **chrVSrc, int chrFilterSize,
1996
            const int16_t **alpSrc, uint8_t *dest, int dstW, int y)
1997
{
1998
    int hasAlpha = !!alpSrc;
1999
    int i;
2000

2001
    for (i = 0; i < dstW; i++) {
2002
        int j;
2003
        int Y = 1 << 18, A = 1 << 18;
2004

2005
        for (j = 0; j < lumFilterSize; j++)
2006
            Y += lumSrc[j][i] * lumFilter[j];
2007

2008
        Y >>= 19;
2009
        if (Y  & 0x100)
2010
            Y = av_clip_uint8(Y);
2011

2012
        if (hasAlpha) {
2013
            for (j = 0; j < lumFilterSize; j++)
2014
                A += alpSrc[j][i] * lumFilter[j];
2015

2016
            A >>= 19;
2017

2018
            if (A & 0x100)
2019
                A = av_clip_uint8(A);
2020
        }
2021

2022
        dest[2 * i    ] = Y;
2023
        dest[2 * i + 1] = hasAlpha ? A : 255;
2024
    }
2025
}
2026

2027
static void
2028
yuv2ayuv64le_X_c(SwsContext *c, const int16_t *lumFilter,
2029
                 const int16_t **_lumSrc, int lumFilterSize,
2030
                 const int16_t *chrFilter, const int16_t **_chrUSrc,
2031
                 const int16_t **_chrVSrc, int chrFilterSize,
2032
                 const int16_t **_alpSrc, uint8_t *dest, int dstW, int y)
2033
{
2034
    const int32_t **lumSrc  = (const int32_t **) _lumSrc,
2035
                  **chrUSrc = (const int32_t **) _chrUSrc,
2036
                  **chrVSrc = (const int32_t **) _chrVSrc,
2037
                  **alpSrc  = (const int32_t **) _alpSrc;
2038
    int hasAlpha = !!alpSrc;
2039
    int i;
2040

2041
    for (i = 0; i < dstW; i++) {
2042
        int Y = 1 << 14, U = 1 << 14;
2043
        int V = 1 << 14, A = 1 << 14;
2044
        int j;
2045

2046
        Y -= 0x40000000;
2047
        U -= 0x40000000;
2048
        V -= 0x40000000;
2049
        A -= 0x40000000;
2050

2051
        for (j = 0; j < lumFilterSize; j++)
2052
            Y += lumSrc[j][i] * (unsigned)lumFilter[j];
2053

2054
        for (j = 0; j < chrFilterSize; j++)
2055
            U += chrUSrc[j][i] * (unsigned)chrFilter[j];
2056

2057
        for (j = 0; j < chrFilterSize; j++)
2058
            V += chrVSrc[j][i] * (unsigned)chrFilter[j];
2059

2060
        if (hasAlpha)
2061
            for (j = 0; j < lumFilterSize; j++)
2062
                A += alpSrc[j][i] * (unsigned)lumFilter[j];
2063

2064
        Y = 0x8000 + av_clip_int16(Y >> 15);
2065
        U = 0x8000 + av_clip_int16(U >> 15);
2066
        V = 0x8000 + av_clip_int16(V >> 15);
2067
        A = 0x8000 + av_clip_int16(A >> 15);
2068

2069
        AV_WL16(dest + 8 * i, hasAlpha ? A : 65535);
2070
        AV_WL16(dest + 8 * i + 2, Y);
2071
        AV_WL16(dest + 8 * i + 4, U);
2072
        AV_WL16(dest + 8 * i + 6, V);
2073
    }
2074
}
2075

2076
av_cold void ff_sws_init_output_funcs(SwsContext *c,
2077
                                      yuv2planar1_fn *yuv2plane1,
2078
                                      yuv2planarX_fn *yuv2planeX,
2079
                                      yuv2interleavedX_fn *yuv2nv12cX,
2080
                                      yuv2packed1_fn *yuv2packed1,
2081
                                      yuv2packed2_fn *yuv2packed2,
2082
                                      yuv2packedX_fn *yuv2packedX,
2083
                                      yuv2anyX_fn *yuv2anyX)
2084
{
2085
    enum AVPixelFormat dstFormat = c->dstFormat;
2086
    const AVPixFmtDescriptor *desc = av_pix_fmt_desc_get(dstFormat);
2087

2088
    if (is16BPS(dstFormat)) {
2089
        *yuv2planeX = isBE(dstFormat) ? yuv2planeX_16BE_c  : yuv2planeX_16LE_c;
2090
        *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_16BE_c  : yuv2plane1_16LE_c;
2091
    } else if (is9_OR_10BPS(dstFormat)) {
2092
        if (desc->comp[0].depth == 9) {
2093
            *yuv2planeX = isBE(dstFormat) ? yuv2planeX_9BE_c  : yuv2planeX_9LE_c;
2094
            *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_9BE_c  : yuv2plane1_9LE_c;
2095
        } else if (desc->comp[0].depth == 10) {
2096
            *yuv2planeX = isBE(dstFormat) ? yuv2planeX_10BE_c  : yuv2planeX_10LE_c;
2097
            *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_10BE_c  : yuv2plane1_10LE_c;
2098
        } else if (desc->comp[0].depth == 12) {
2099
            *yuv2planeX = isBE(dstFormat) ? yuv2planeX_12BE_c  : yuv2planeX_12LE_c;
2100
            *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_12BE_c  : yuv2plane1_12LE_c;
2101
        } else if (desc->comp[0].depth == 14) {
2102
            *yuv2planeX = isBE(dstFormat) ? yuv2planeX_14BE_c  : yuv2planeX_14LE_c;
2103
            *yuv2plane1 = isBE(dstFormat) ? yuv2plane1_14BE_c  : yuv2plane1_14LE_c;
2104
        } else
2105
            av_assert0(0);
2106
    } else {
2107
        *yuv2plane1 = yuv2plane1_8_c;
2108
        *yuv2planeX = yuv2planeX_8_c;
2109
        if (dstFormat == AV_PIX_FMT_NV12 || dstFormat == AV_PIX_FMT_NV21)
2110
            *yuv2nv12cX = yuv2nv12cX_c;
2111
    }
2112

2113
    if(c->flags & SWS_FULL_CHR_H_INT) {
2114
        switch (dstFormat) {
2115
            case AV_PIX_FMT_RGBA:
2116
#if CONFIG_SMALL
2117
                *yuv2packedX = yuv2rgba32_full_X_c;
2118
                *yuv2packed2 = yuv2rgba32_full_2_c;
2119
                *yuv2packed1 = yuv2rgba32_full_1_c;
2120
#else
2121
#if CONFIG_SWSCALE_ALPHA
2122
                if (c->alpPixBuf) {
2123
                    *yuv2packedX = yuv2rgba32_full_X_c;
2124
                    *yuv2packed2 = yuv2rgba32_full_2_c;
2125
                    *yuv2packed1 = yuv2rgba32_full_1_c;
2126
                } else
2127
#endif /* CONFIG_SWSCALE_ALPHA */
2128
                {
2129
                    *yuv2packedX = yuv2rgbx32_full_X_c;
2130
                    *yuv2packed2 = yuv2rgbx32_full_2_c;
2131
                    *yuv2packed1 = yuv2rgbx32_full_1_c;
2132
                }
2133
#endif /* !CONFIG_SMALL */
2134
                break;
2135
            case AV_PIX_FMT_ARGB:
2136
#if CONFIG_SMALL
2137
                *yuv2packedX = yuv2argb32_full_X_c;
2138
                *yuv2packed2 = yuv2argb32_full_2_c;
2139
                *yuv2packed1 = yuv2argb32_full_1_c;
2140
#else
2141
#if CONFIG_SWSCALE_ALPHA
2142
                if (c->alpPixBuf) {
2143
                    *yuv2packedX = yuv2argb32_full_X_c;
2144
                    *yuv2packed2 = yuv2argb32_full_2_c;
2145
                    *yuv2packed1 = yuv2argb32_full_1_c;
2146
                } else
2147
#endif /* CONFIG_SWSCALE_ALPHA */
2148
                {
2149
                    *yuv2packedX = yuv2xrgb32_full_X_c;
2150
                    *yuv2packed2 = yuv2xrgb32_full_2_c;
2151
                    *yuv2packed1 = yuv2xrgb32_full_1_c;
2152
                }
2153
#endif /* !CONFIG_SMALL */
2154
                break;
2155
            case AV_PIX_FMT_BGRA:
2156
#if CONFIG_SMALL
2157
                *yuv2packedX = yuv2bgra32_full_X_c;
2158
                *yuv2packed2 = yuv2bgra32_full_2_c;
2159
                *yuv2packed1 = yuv2bgra32_full_1_c;
2160
#else
2161
#if CONFIG_SWSCALE_ALPHA
2162
                if (c->alpPixBuf) {
2163
                    *yuv2packedX = yuv2bgra32_full_X_c;
2164
                    *yuv2packed2 = yuv2bgra32_full_2_c;
2165
                    *yuv2packed1 = yuv2bgra32_full_1_c;
2166
                } else
2167
#endif /* CONFIG_SWSCALE_ALPHA */
2168
                {
2169
                    *yuv2packedX = yuv2bgrx32_full_X_c;
2170
                    *yuv2packed2 = yuv2bgrx32_full_2_c;
2171
                    *yuv2packed1 = yuv2bgrx32_full_1_c;
2172
                }
2173
#endif /* !CONFIG_SMALL */
2174
                break;
2175
            case AV_PIX_FMT_ABGR:
2176
#if CONFIG_SMALL
2177
                *yuv2packedX = yuv2abgr32_full_X_c;
2178
                *yuv2packed2 = yuv2abgr32_full_2_c;
2179
                *yuv2packed1 = yuv2abgr32_full_1_c;
2180
#else
2181
#if CONFIG_SWSCALE_ALPHA
2182
                if (c->alpPixBuf) {
2183
                    *yuv2packedX = yuv2abgr32_full_X_c;
2184
                    *yuv2packed2 = yuv2abgr32_full_2_c;
2185
                    *yuv2packed1 = yuv2abgr32_full_1_c;
2186
                } else
2187
#endif /* CONFIG_SWSCALE_ALPHA */
2188
                {
2189
                    *yuv2packedX = yuv2xbgr32_full_X_c;
2190
                    *yuv2packed2 = yuv2xbgr32_full_2_c;
2191
                    *yuv2packed1 = yuv2xbgr32_full_1_c;
2192
                }
2193
#endif /* !CONFIG_SMALL */
2194
                break;
2195
        case AV_PIX_FMT_RGBA64LE:
2196
#if CONFIG_SWSCALE_ALPHA
2197
            if (c->alpPixBuf) {
2198
                *yuv2packedX = yuv2rgba64le_full_X_c;
2199
                *yuv2packed2 = yuv2rgba64le_full_2_c;
2200
                *yuv2packed1 = yuv2rgba64le_full_1_c;
2201
            } else
2202
#endif /* CONFIG_SWSCALE_ALPHA */
2203
            {
2204
                *yuv2packedX = yuv2rgbx64le_full_X_c;
2205
                *yuv2packed2 = yuv2rgbx64le_full_2_c;
2206
                *yuv2packed1 = yuv2rgbx64le_full_1_c;
2207
            }
2208
            break;
2209
        case AV_PIX_FMT_RGBA64BE:
2210
#if CONFIG_SWSCALE_ALPHA
2211
            if (c->alpPixBuf) {
2212
                *yuv2packedX = yuv2rgba64be_full_X_c;
2213
                *yuv2packed2 = yuv2rgba64be_full_2_c;
2214
                *yuv2packed1 = yuv2rgba64be_full_1_c;
2215
            } else
2216
#endif /* CONFIG_SWSCALE_ALPHA */
2217
            {
2218
                *yuv2packedX = yuv2rgbx64be_full_X_c;
2219
                *yuv2packed2 = yuv2rgbx64be_full_2_c;
2220
                *yuv2packed1 = yuv2rgbx64be_full_1_c;
2221
            }
2222
            break;
2223
        case AV_PIX_FMT_BGRA64LE:
2224
#if CONFIG_SWSCALE_ALPHA
2225
            if (c->alpPixBuf) {
2226
                *yuv2packedX = yuv2bgra64le_full_X_c;
2227
                *yuv2packed2 = yuv2bgra64le_full_2_c;
2228
                *yuv2packed1 = yuv2bgra64le_full_1_c;
2229
            } else
2230
#endif /* CONFIG_SWSCALE_ALPHA */
2231
            {
2232
                *yuv2packedX = yuv2bgrx64le_full_X_c;
2233
                *yuv2packed2 = yuv2bgrx64le_full_2_c;
2234
                *yuv2packed1 = yuv2bgrx64le_full_1_c;
2235
            }
2236
            break;
2237
        case AV_PIX_FMT_BGRA64BE:
2238
#if CONFIG_SWSCALE_ALPHA
2239
            if (c->alpPixBuf) {
2240
                *yuv2packedX = yuv2bgra64be_full_X_c;
2241
                *yuv2packed2 = yuv2bgra64be_full_2_c;
2242
                *yuv2packed1 = yuv2bgra64be_full_1_c;
2243
            } else
2244
#endif /* CONFIG_SWSCALE_ALPHA */
2245
            {
2246
                *yuv2packedX = yuv2bgrx64be_full_X_c;
2247
                *yuv2packed2 = yuv2bgrx64be_full_2_c;
2248
                *yuv2packed1 = yuv2bgrx64be_full_1_c;
2249
            }
2250
            break;
2251

2252
        case AV_PIX_FMT_RGB24:
2253
            *yuv2packedX = yuv2rgb24_full_X_c;
2254
            *yuv2packed2 = yuv2rgb24_full_2_c;
2255
            *yuv2packed1 = yuv2rgb24_full_1_c;
2256
            break;
2257
        case AV_PIX_FMT_BGR24:
2258
            *yuv2packedX = yuv2bgr24_full_X_c;
2259
            *yuv2packed2 = yuv2bgr24_full_2_c;
2260
            *yuv2packed1 = yuv2bgr24_full_1_c;
2261
            break;
2262
        case AV_PIX_FMT_RGB48LE:
2263
            *yuv2packedX = yuv2rgb48le_full_X_c;
2264
            *yuv2packed2 = yuv2rgb48le_full_2_c;
2265
            *yuv2packed1 = yuv2rgb48le_full_1_c;
2266
            break;
2267
        case AV_PIX_FMT_BGR48LE:
2268
            *yuv2packedX = yuv2bgr48le_full_X_c;
2269
            *yuv2packed2 = yuv2bgr48le_full_2_c;
2270
            *yuv2packed1 = yuv2bgr48le_full_1_c;
2271
            break;
2272
        case AV_PIX_FMT_RGB48BE:
2273
            *yuv2packedX = yuv2rgb48be_full_X_c;
2274
            *yuv2packed2 = yuv2rgb48be_full_2_c;
2275
            *yuv2packed1 = yuv2rgb48be_full_1_c;
2276
            break;
2277
        case AV_PIX_FMT_BGR48BE:
2278
            *yuv2packedX = yuv2bgr48be_full_X_c;
2279
            *yuv2packed2 = yuv2bgr48be_full_2_c;
2280
            *yuv2packed1 = yuv2bgr48be_full_1_c;
2281
            break;
2282
        case AV_PIX_FMT_BGR4_BYTE:
2283
            *yuv2packedX = yuv2bgr4_byte_full_X_c;
2284
            *yuv2packed2 = yuv2bgr4_byte_full_2_c;
2285
            *yuv2packed1 = yuv2bgr4_byte_full_1_c;
2286
            break;
2287
        case AV_PIX_FMT_RGB4_BYTE:
2288
            *yuv2packedX = yuv2rgb4_byte_full_X_c;
2289
            *yuv2packed2 = yuv2rgb4_byte_full_2_c;
2290
            *yuv2packed1 = yuv2rgb4_byte_full_1_c;
2291
            break;
2292
        case AV_PIX_FMT_BGR8:
2293
            *yuv2packedX = yuv2bgr8_full_X_c;
2294
            *yuv2packed2 = yuv2bgr8_full_2_c;
2295
            *yuv2packed1 = yuv2bgr8_full_1_c;
2296
            break;
2297
        case AV_PIX_FMT_RGB8:
2298
            *yuv2packedX = yuv2rgb8_full_X_c;
2299
            *yuv2packed2 = yuv2rgb8_full_2_c;
2300
            *yuv2packed1 = yuv2rgb8_full_1_c;
2301
            break;
2302
        case AV_PIX_FMT_GBRP:
2303
        case AV_PIX_FMT_GBRP9BE:
2304
        case AV_PIX_FMT_GBRP9LE:
2305
        case AV_PIX_FMT_GBRP10BE:
2306
        case AV_PIX_FMT_GBRP10LE:
2307
        case AV_PIX_FMT_GBRP12BE:
2308
        case AV_PIX_FMT_GBRP12LE:
2309
        case AV_PIX_FMT_GBRP14BE:
2310
        case AV_PIX_FMT_GBRP14LE:
2311
        case AV_PIX_FMT_GBRP16BE:
2312
        case AV_PIX_FMT_GBRP16LE:
2313
        case AV_PIX_FMT_GBRAP:
2314
            *yuv2anyX = yuv2gbrp_full_X_c;
2315
            break;
2316
        }
2317
        if (!*yuv2packedX && !*yuv2anyX)
2318
            goto YUV_PACKED;
2319
    } else {
2320
        YUV_PACKED:
2321
        switch (dstFormat) {
2322
        case AV_PIX_FMT_RGBA64LE:
2323
#if CONFIG_SWSCALE_ALPHA
2324
            if (c->alpPixBuf) {
2325
                *yuv2packed1 = yuv2rgba64le_1_c;
2326
                *yuv2packed2 = yuv2rgba64le_2_c;
2327
                *yuv2packedX = yuv2rgba64le_X_c;
2328
            } else
2329
#endif /* CONFIG_SWSCALE_ALPHA */
2330
            {
2331
                *yuv2packed1 = yuv2rgbx64le_1_c;
2332
                *yuv2packed2 = yuv2rgbx64le_2_c;
2333
                *yuv2packedX = yuv2rgbx64le_X_c;
2334
            }
2335
            break;
2336
        case AV_PIX_FMT_RGBA64BE:
2337
#if CONFIG_SWSCALE_ALPHA
2338
            if (c->alpPixBuf) {
2339
                *yuv2packed1 = yuv2rgba64be_1_c;
2340
                *yuv2packed2 = yuv2rgba64be_2_c;
2341
                *yuv2packedX = yuv2rgba64be_X_c;
2342
            } else
2343
#endif /* CONFIG_SWSCALE_ALPHA */
2344
            {
2345
                *yuv2packed1 = yuv2rgbx64be_1_c;
2346
                *yuv2packed2 = yuv2rgbx64be_2_c;
2347
                *yuv2packedX = yuv2rgbx64be_X_c;
2348
            }
2349
            break;
2350
        case AV_PIX_FMT_BGRA64LE:
2351
#if CONFIG_SWSCALE_ALPHA
2352
            if (c->alpPixBuf) {
2353
                *yuv2packed1 = yuv2bgra64le_1_c;
2354
                *yuv2packed2 = yuv2bgra64le_2_c;
2355
                *yuv2packedX = yuv2bgra64le_X_c;
2356
            } else
2357
#endif /* CONFIG_SWSCALE_ALPHA */
2358
            {
2359
                *yuv2packed1 = yuv2bgrx64le_1_c;
2360
                *yuv2packed2 = yuv2bgrx64le_2_c;
2361
                *yuv2packedX = yuv2bgrx64le_X_c;
2362
            }
2363
            break;
2364
        case AV_PIX_FMT_BGRA64BE:
2365
#if CONFIG_SWSCALE_ALPHA
2366
            if (c->alpPixBuf) {
2367
                *yuv2packed1 = yuv2bgra64be_1_c;
2368
                *yuv2packed2 = yuv2bgra64be_2_c;
2369
                *yuv2packedX = yuv2bgra64be_X_c;
2370
            } else
2371
#endif /* CONFIG_SWSCALE_ALPHA */
2372
            {
2373
                *yuv2packed1 = yuv2bgrx64be_1_c;
2374
                *yuv2packed2 = yuv2bgrx64be_2_c;
2375
                *yuv2packedX = yuv2bgrx64be_X_c;
2376
            }
2377
            break;
2378
        case AV_PIX_FMT_RGB48LE:
2379
            *yuv2packed1 = yuv2rgb48le_1_c;
2380
            *yuv2packed2 = yuv2rgb48le_2_c;
2381
            *yuv2packedX = yuv2rgb48le_X_c;
2382
            break;
2383
        case AV_PIX_FMT_RGB48BE:
2384
            *yuv2packed1 = yuv2rgb48be_1_c;
2385
            *yuv2packed2 = yuv2rgb48be_2_c;
2386
            *yuv2packedX = yuv2rgb48be_X_c;
2387
            break;
2388
        case AV_PIX_FMT_BGR48LE:
2389
            *yuv2packed1 = yuv2bgr48le_1_c;
2390
            *yuv2packed2 = yuv2bgr48le_2_c;
2391
            *yuv2packedX = yuv2bgr48le_X_c;
2392
            break;
2393
        case AV_PIX_FMT_BGR48BE:
2394
            *yuv2packed1 = yuv2bgr48be_1_c;
2395
            *yuv2packed2 = yuv2bgr48be_2_c;
2396
            *yuv2packedX = yuv2bgr48be_X_c;
2397
            break;
2398
        case AV_PIX_FMT_RGB32:
2399
        case AV_PIX_FMT_BGR32:
2400
#if CONFIG_SMALL
2401
            *yuv2packed1 = yuv2rgb32_1_c;
2402
            *yuv2packed2 = yuv2rgb32_2_c;
2403
            *yuv2packedX = yuv2rgb32_X_c;
2404
#else
2405
#if CONFIG_SWSCALE_ALPHA
2406
                if (c->alpPixBuf) {
2407
                    *yuv2packed1 = yuv2rgba32_1_c;
2408
                    *yuv2packed2 = yuv2rgba32_2_c;
2409
                    *yuv2packedX = yuv2rgba32_X_c;
2410
                } else
2411
#endif /* CONFIG_SWSCALE_ALPHA */
2412
                {
2413
                    *yuv2packed1 = yuv2rgbx32_1_c;
2414
                    *yuv2packed2 = yuv2rgbx32_2_c;
2415
                    *yuv2packedX = yuv2rgbx32_X_c;
2416
                }
2417
#endif /* !CONFIG_SMALL */
2418
            break;
2419
        case AV_PIX_FMT_RGB32_1:
2420
        case AV_PIX_FMT_BGR32_1:
2421
#if CONFIG_SMALL
2422
                *yuv2packed1 = yuv2rgb32_1_1_c;
2423
                *yuv2packed2 = yuv2rgb32_1_2_c;
2424
                *yuv2packedX = yuv2rgb32_1_X_c;
2425
#else
2426
#if CONFIG_SWSCALE_ALPHA
2427
                if (c->alpPixBuf) {
2428
                    *yuv2packed1 = yuv2rgba32_1_1_c;
2429
                    *yuv2packed2 = yuv2rgba32_1_2_c;
2430
                    *yuv2packedX = yuv2rgba32_1_X_c;
2431
                } else
2432
#endif /* CONFIG_SWSCALE_ALPHA */
2433
                {
2434
                    *yuv2packed1 = yuv2rgbx32_1_1_c;
2435
                    *yuv2packed2 = yuv2rgbx32_1_2_c;
2436
                    *yuv2packedX = yuv2rgbx32_1_X_c;
2437
                }
2438
#endif /* !CONFIG_SMALL */
2439
                break;
2440
        case AV_PIX_FMT_RGB24:
2441
            *yuv2packed1 = yuv2rgb24_1_c;
2442
            *yuv2packed2 = yuv2rgb24_2_c;
2443
            *yuv2packedX = yuv2rgb24_X_c;
2444
            break;
2445
        case AV_PIX_FMT_BGR24:
2446
            *yuv2packed1 = yuv2bgr24_1_c;
2447
            *yuv2packed2 = yuv2bgr24_2_c;
2448
            *yuv2packedX = yuv2bgr24_X_c;
2449
            break;
2450
        case AV_PIX_FMT_RGB565LE:
2451
        case AV_PIX_FMT_RGB565BE:
2452
        case AV_PIX_FMT_BGR565LE:
2453
        case AV_PIX_FMT_BGR565BE:
2454
            *yuv2packed1 = yuv2rgb16_1_c;
2455
            *yuv2packed2 = yuv2rgb16_2_c;
2456
            *yuv2packedX = yuv2rgb16_X_c;
2457
            break;
2458
        case AV_PIX_FMT_RGB555LE:
2459
        case AV_PIX_FMT_RGB555BE:
2460
        case AV_PIX_FMT_BGR555LE:
2461
        case AV_PIX_FMT_BGR555BE:
2462
            *yuv2packed1 = yuv2rgb15_1_c;
2463
            *yuv2packed2 = yuv2rgb15_2_c;
2464
            *yuv2packedX = yuv2rgb15_X_c;
2465
            break;
2466
        case AV_PIX_FMT_RGB444LE:
2467
        case AV_PIX_FMT_RGB444BE:
2468
        case AV_PIX_FMT_BGR444LE:
2469
        case AV_PIX_FMT_BGR444BE:
2470
            *yuv2packed1 = yuv2rgb12_1_c;
2471
            *yuv2packed2 = yuv2rgb12_2_c;
2472
            *yuv2packedX = yuv2rgb12_X_c;
2473
            break;
2474
        case AV_PIX_FMT_RGB8:
2475
        case AV_PIX_FMT_BGR8:
2476
            *yuv2packed1 = yuv2rgb8_1_c;
2477
            *yuv2packed2 = yuv2rgb8_2_c;
2478
            *yuv2packedX = yuv2rgb8_X_c;
2479
            break;
2480
        case AV_PIX_FMT_RGB4:
2481
        case AV_PIX_FMT_BGR4:
2482
            *yuv2packed1 = yuv2rgb4_1_c;
2483
            *yuv2packed2 = yuv2rgb4_2_c;
2484
            *yuv2packedX = yuv2rgb4_X_c;
2485
            break;
2486
        case AV_PIX_FMT_RGB4_BYTE:
2487
        case AV_PIX_FMT_BGR4_BYTE:
2488
            *yuv2packed1 = yuv2rgb4b_1_c;
2489
            *yuv2packed2 = yuv2rgb4b_2_c;
2490
            *yuv2packedX = yuv2rgb4b_X_c;
2491
            break;
2492
        }
2493
    }
2494
    switch (dstFormat) {
2495
    case AV_PIX_FMT_MONOWHITE:
2496
        *yuv2packed1 = yuv2monowhite_1_c;
2497
        *yuv2packed2 = yuv2monowhite_2_c;
2498
        *yuv2packedX = yuv2monowhite_X_c;
2499
        break;
2500
    case AV_PIX_FMT_MONOBLACK:
2501
        *yuv2packed1 = yuv2monoblack_1_c;
2502
        *yuv2packed2 = yuv2monoblack_2_c;
2503
        *yuv2packedX = yuv2monoblack_X_c;
2504
        break;
2505
    case AV_PIX_FMT_YUYV422:
2506
        *yuv2packed1 = yuv2yuyv422_1_c;
2507
        *yuv2packed2 = yuv2yuyv422_2_c;
2508
        *yuv2packedX = yuv2yuyv422_X_c;
2509
        break;
2510
    case AV_PIX_FMT_YVYU422:
2511
        *yuv2packed1 = yuv2yvyu422_1_c;
2512
        *yuv2packed2 = yuv2yvyu422_2_c;
2513
        *yuv2packedX = yuv2yvyu422_X_c;
2514
        break;
2515
    case AV_PIX_FMT_UYVY422:
2516
        *yuv2packed1 = yuv2uyvy422_1_c;
2517
        *yuv2packed2 = yuv2uyvy422_2_c;
2518
        *yuv2packedX = yuv2uyvy422_X_c;
2519
        break;
2520
    case AV_PIX_FMT_YA8:
2521
        *yuv2packed1 = yuv2ya8_1_c;
2522
        *yuv2packed2 = yuv2ya8_2_c;
2523
        *yuv2packedX = yuv2ya8_X_c;
2524
        break;
2525
    case AV_PIX_FMT_AYUV64LE:
2526
        *yuv2packedX = yuv2ayuv64le_X_c;
2527
        break;
2528
    }
2529
}
2530

2531
Product

Resources

Company
