frame.c 27.9 KB
Newer Older
Laurent Aimar's avatar
Laurent Aimar committed
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
/*****************************************************************************
 * frame.c: h264 encoder library
 *****************************************************************************
 * Copyright (C) 2003 Laurent Aimar
 * $Id: frame.c,v 1.1 2004/06/03 19:27:06 fenrir Exp $
 *
 * Authors: Laurent Aimar <fenrir@via.ecp.fr>
 *
 * This program is free software; you can redistribute it and/or modify
 * it under the terms of the GNU General Public License as published by
 * the Free Software Foundation; either version 2 of the License, or
 * (at your option) any later version.
 *
 * This program is distributed in the hope that it will be useful,
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
 * GNU General Public License for more details.
 *
 * You should have received a copy of the GNU General Public License
 * along with this program; if not, write to the Free Software
 * Foundation, Inc., 59 Temple Place - Suite 330, Boston, MA  02111, USA.
 *****************************************************************************/

#include <stdlib.h>
#include <stdio.h>
#include <string.h>

#include "common.h"
#include "macroblock.h"

x264_frame_t *x264_frame_new( x264_t *h )
{
    x264_frame_t   *frame = x264_malloc( sizeof( x264_frame_t ) );
    int i;

36
    int i_mb_count = h->mb.i_mb_count;
Laurent Aimar's avatar
Laurent Aimar committed
37
38
39
    int i_stride;
    int i_lines;

40
41
    memset( frame, 0, sizeof(x264_frame_t) );

Laurent Aimar's avatar
Laurent Aimar committed
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
    /* allocate frame data (+64 for extra data for me) */
    i_stride = ( ( h->param.i_width  + 15 )&0xfffff0 )+ 64;
    i_lines  = ( ( h->param.i_height + 15 )&0xfffff0 );

    frame->i_plane = 3;
    for( i = 0; i < 3; i++ )
    {
        int i_divh = 1;
        int i_divw = 1;
        if( i > 0 )
        {
            if( h->param.i_csp == X264_CSP_I420 )
                i_divh = i_divw = 2;
            else if( h->param.i_csp == X264_CSP_I422 )
                i_divw = 2;
        }
        frame->i_stride[i] = i_stride / i_divw;
        frame->i_lines[i] = i_lines / i_divh;
        frame->buffer[i] = x264_malloc( frame->i_stride[i] *
                                        ( frame->i_lines[i] + 64 / i_divh ) );

        frame->plane[i] = ((uint8_t*)frame->buffer[i]) +
                          frame->i_stride[i] * 32 / i_divh + 32 / i_divw;
    }
    frame->i_stride[3] = 0;
    frame->i_lines[3] = 0;
    frame->buffer[3] = NULL;
    frame->plane[3] = NULL;

71
72
73
74
75
76
77
    frame->filtered[0] = frame->plane[0];
    for( i = 0; i < 3; i++ )
    {
        frame->buffer[4+i] = x264_malloc( frame->i_stride[0] *
                                        ( frame->i_lines[0] + 64 ) );

        frame->filtered[i+1] = ((uint8_t*)frame->buffer[4+i]) +
78
                                frame->i_stride[0] * 32 + 32;
79
80
    }

81
    if( h->frames.b_have_lowres )
82
    {
83
84
85
86
87
88
89
90
91
        frame->i_stride_lowres = frame->i_stride[0]/2 + 32;
        frame->i_lines_lowres = frame->i_lines[0]/2;
        for( i = 0; i < 4; i++ )
        {
            frame->buffer[7+i] = x264_malloc( frame->i_stride_lowres *
                                            ( frame->i_lines[0]/2 + 64 ) );
            frame->lowres[i] = ((uint8_t*)frame->buffer[7+i]) +
                                frame->i_stride_lowres * 32 + 32;
        }
92
    }
93

Laurent Aimar's avatar
Laurent Aimar committed
94
95
96
    frame->i_poc = -1;
    frame->i_type = X264_TYPE_AUTO;
    frame->i_qpplus1 = 0;
97
98
    frame->i_pts = -1;
    frame->i_frame = -1;
99
    frame->i_frame_num = -1;
Laurent Aimar's avatar
Laurent Aimar committed
100

101
    frame->mb_type= x264_malloc( i_mb_count * sizeof( int8_t) );
102
103
104
105
106
107
108
109
110
111
112
113
114
    frame->mv[0]  = x264_malloc( 2*16 * i_mb_count * sizeof( int16_t ) );
    frame->ref[0] = x264_malloc( 4 * i_mb_count * sizeof( int8_t ) );
    if( h->param.i_bframe )
    {
        frame->mv[1]  = x264_malloc( 2*16 * i_mb_count * sizeof( int16_t ) );
        frame->ref[1] = x264_malloc( 4 * i_mb_count * sizeof( int8_t ) );
    }
    else
    {
        frame->mv[1]  = NULL;
        frame->ref[1] = NULL;
    }

Laurent Aimar's avatar
Laurent Aimar committed
115
116
117
118
119
120
121
122
123
124
    return frame;
}

void x264_frame_delete( x264_frame_t *frame )
{
    int i;
    for( i = 0; i < frame->i_plane; i++ )
    {
        x264_free( frame->buffer[i] );
    }
125
    for( i = 4; i < 11; i++ ) /* filtered planes */
126
127
128
    {
        x264_free( frame->buffer[i] );
    }
129
    x264_free( frame->mb_type );
130
131
132
133
    x264_free( frame->mv[0] );
    x264_free( frame->mv[1] );
    x264_free( frame->ref[0] );
    x264_free( frame->ref[1] );
Laurent Aimar's avatar
Laurent Aimar committed
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
    x264_free( frame );
}

void x264_frame_copy_picture( x264_t *h, x264_frame_t *dst, x264_picture_t *src )
{
    dst->i_type     = src->i_type;
    dst->i_qpplus1  = src->i_qpplus1;
    dst->i_pts      = src->i_pts;

    switch( src->img.i_csp & X264_CSP_MASK )
    {
        case X264_CSP_I420:
            h->csp.i420( dst, &src->img, h->param.i_width, h->param.i_height );
            break;
        case X264_CSP_YV12:
            h->csp.yv12( dst, &src->img, h->param.i_width, h->param.i_height );
            break;
        case X264_CSP_I422:
            h->csp.i422( dst, &src->img, h->param.i_width, h->param.i_height );
            break;
        case X264_CSP_I444:
            h->csp.i444( dst, &src->img, h->param.i_width, h->param.i_height );
            break;
        case X264_CSP_YUYV:
            h->csp.yuyv( dst, &src->img, h->param.i_width, h->param.i_height );
            break;
        case X264_CSP_RGB:
            h->csp.rgb( dst, &src->img, h->param.i_width, h->param.i_height );
            break;
        case X264_CSP_BGR:
            h->csp.bgr( dst, &src->img, h->param.i_width, h->param.i_height );
            break;
        case X264_CSP_BGRA:
            h->csp.bgra( dst, &src->img, h->param.i_width, h->param.i_height );
            break;

        default:
171
            x264_log( h, X264_LOG_ERROR, "Arg invalid CSP\n" );
Laurent Aimar's avatar
Laurent Aimar committed
172
173
174
175
176
177
            break;
    }
}



Loren Merritt's avatar
Loren Merritt committed
178
static void plane_expand_border( uint8_t *pix, int i_stride, int i_height, int i_pad )
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
{
#define PPIXEL(x, y) ( pix + (x) + (y)*i_stride )
    const int i_width = i_stride - 2*i_pad;
    int y;

    for( y = 0; y < i_height; y++ )
    {
        /* left band */
        memset( PPIXEL(-i_pad, y), PPIXEL(0, y)[0], i_pad );
        /* right band */
        memset( PPIXEL(i_width, y), PPIXEL(i_width-1, y)[0], i_pad );
    }
    /* upper band */
    for( y = 0; y < i_pad; y++ )
        memcpy( PPIXEL(-i_pad, -y-1), PPIXEL(-i_pad, 0), i_stride );
    /* lower band */
    for( y = 0; y < i_pad; y++ )
        memcpy( PPIXEL(-i_pad, i_height+y), PPIXEL(-i_pad, i_height-1), i_stride );
#undef PPIXEL
}

Laurent Aimar's avatar
Laurent Aimar committed
200
201
void x264_frame_expand_border( x264_frame_t *frame )
{
202
    int i;
Laurent Aimar's avatar
Laurent Aimar committed
203
204
    for( i = 0; i < frame->i_plane; i++ )
    {
205
206
        int i_pad = i ? 16 : 32;
        plane_expand_border( frame->plane[i], frame->i_stride[i], frame->i_lines[i], i_pad );
Laurent Aimar's avatar
Laurent Aimar committed
207
208
209
    }
}

210
211
212
213
void x264_frame_expand_border_filtered( x264_frame_t *frame )
{
    /* during filtering, 8 extra pixels were filtered on each edge. 
       we want to expand border from the last filtered pixel */
214
    int i;
215
    for( i = 1; i < 4; i++ )
216
        plane_expand_border( frame->filtered[i] - 8*frame->i_stride[0] - 8, frame->i_stride[0], frame->i_lines[0]+2*8, 24 );
217
218
}

219
void x264_frame_expand_border_lowres( x264_frame_t *frame )
220
{
221
    int i;
222
    for( i = 0; i < 4; i++ )
223
        plane_expand_border( frame->lowres[i], frame->i_stride_lowres, frame->i_lines_lowres, 32 );
224
225
}

226

Laurent Aimar's avatar
Laurent Aimar committed
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
/* FIXME theses tables are duplicated with the ones in macroblock.c */
static const uint8_t block_idx_xy[4][4] =
{
    { 0, 2, 8,  10},
    { 1, 3, 9,  11},
    { 4, 6, 12, 14},
    { 5, 7, 13, 15}
};
static const int i_chroma_qp_table[52] =
{
     0,  1,  2,  3,  4,  5,  6,  7,  8,  9,
    10, 11, 12, 13, 14, 15, 16, 17, 18, 19,
    20, 21, 22, 23, 24, 25, 26, 27, 28, 29,
    29, 30, 31, 32, 32, 33, 34, 34, 35, 35,
    36, 36, 37, 37, 37, 38, 38, 38, 39, 39,
    39, 39
};

/* Deblocking filter (p153) */
static const int i_alpha_table[52] =
{
     0,  0,  0,  0,  0,  0,  0,  0,  0,  0,
     0,  0,  0,  0,  0,  0,  4,  4,  5,  6,
     7,  8,  9, 10, 12, 13, 15, 17, 20, 22,
    25, 28, 32, 36, 40, 45, 50, 56, 63, 71,
    80, 90,101,113,127,144,162,182,203,226,
    255, 255
};
static const int i_beta_table[52] =
{
     0,  0,  0,  0,  0,  0,  0,  0,  0,  0,
     0,  0,  0,  0,  0,  0,  2,  2,  2,  3,
     3,  3,  3,  4,  4,  4,  6,  6,  7,  7,
     8,  8,  9,  9, 10, 10, 11, 11, 12, 12,
    13, 13, 14, 14, 15, 15, 16, 16, 17, 17,
    18, 18
};
static const int i_tc0_table[52][3] =
{
    { 0, 0, 0 }, { 0, 0, 0 }, { 0, 0, 0 }, { 0, 0, 0 }, { 0, 0, 0 }, { 0, 0, 0 },
    { 0, 0, 0 }, { 0, 0, 0 }, { 0, 0, 0 }, { 0, 0, 0 }, { 0, 0, 0 }, { 0, 0, 0 },
    { 0, 0, 0 }, { 0, 0, 0 }, { 0, 0, 0 }, { 0, 0, 0 }, { 0, 0, 0 }, { 0, 0, 1 },
    { 0, 0, 1 }, { 0, 0, 1 }, { 0, 0, 1 }, { 0, 1, 1 }, { 0, 1, 1 }, { 1, 1, 1 },
    { 1, 1, 1 }, { 1, 1, 1 }, { 1, 1, 1 }, { 1, 1, 2 }, { 1, 1, 2 }, { 1, 1, 2 },
    { 1, 1, 2 }, { 1, 2, 3 }, { 1, 2, 3 }, { 2, 2, 3 }, { 2, 2, 4 }, { 2, 3, 4 },
    { 2, 3, 4 }, { 3, 3, 5 }, { 3, 4, 6 }, { 3, 4, 6 }, { 4, 5, 7 }, { 4, 5, 8 },
    { 4, 6, 9 }, { 5, 7,10 }, { 6, 8,11 }, { 6, 8,13 }, { 7,10,14 }, { 8,11,16 },
    { 9,12,18 }, {10,13,20 }, {11,15,23 }, {13,17,25 }
};

/* From ffmpeg */
static inline int clip_uint8( int a )
{
    if (a&(~255))
        return (-a)>>31;
    else
        return a;
}

static inline void deblocking_filter_edgev( x264_t *h, uint8_t *pix, int i_pix_stride, int bS[4], int i_QP )
{
    int i, d;
    const int i_index_a = x264_clip3( i_QP + h->sh.i_alpha_c0_offset, 0, 51 );
    const int alpha = i_alpha_table[i_index_a];
    const int beta  = i_beta_table[x264_clip3( i_QP + h->sh.i_beta_offset, 0, 51 )];

    for( i = 0; i < 4; i++ )
    {
        if( bS[i] == 0 )
        {
            pix += 4 * i_pix_stride;
            continue;
        }

        if( bS[i] < 4 )
        {
            const int tc0 = i_tc0_table[i_index_a][bS[i] - 1];

            /* 4px edge length */
            for( d = 0; d < 4; d++ )
            {
                const int p0 = pix[-1];
                const int p1 = pix[-2];
                const int p2 = pix[-3];
                const int q0 = pix[0];
                const int q1 = pix[1];
                const int q2 = pix[2];

                if( abs( p0 - q0 ) < alpha &&
                    abs( p1 - p0 ) < beta &&
                    abs( q1 - q0 ) < beta )
                {
                    int tc = tc0;
                    int i_delta;

                    if( abs( p2 - p0 ) < beta )
                    {
                        pix[-2] = p1 + x264_clip3( ( p2 + ( ( p0 + q0 + 1 ) >> 1 ) - ( p1 << 1 ) ) >> 1, -tc0, tc0 );
                        tc++;
                    }
                    if( abs( q2 - q0 ) < beta )
                    {
                        pix[1] = q1 + x264_clip3( ( q2 + ( ( p0 + q0 + 1 ) >> 1 ) - ( q1 << 1 ) ) >> 1, -tc0, tc0 );
                        tc++;
                    }

                    i_delta = x264_clip3( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
                    pix[-1] = clip_uint8( p0 + i_delta );    /* p0' */
                    pix[0]  = clip_uint8( q0 - i_delta );    /* q0' */
                }
                pix += i_pix_stride;
            }
        }
        else
        {
            /* 4px edge length */
            for( d = 0; d < 4; d++ )
            {
                const int p0 = pix[-1];
                const int p1 = pix[-2];
                const int p2 = pix[-3];

                const int q0 = pix[0];
                const int q1 = pix[1];
                const int q2 = pix[2];

                if( abs( p0 - q0 ) < alpha &&
                    abs( p1 - p0 ) < beta &&
                    abs( q1 - q0 ) < beta )
                {
                    if( abs( p0 - q0 ) < (( alpha >> 2 ) + 2 ) )
                    {
                        if( abs( p2 - p0 ) < beta )
                        {
                            const int p3 = pix[-4];
                            /* p0', p1', p2' */
                            pix[-1] = ( p2 + 2*p1 + 2*p0 + 2*q0 + q1 + 4 ) >> 3;
                            pix[-2] = ( p2 + p1 + p0 + q0 + 2 ) >> 2;
                            pix[-3] = ( 2*p3 + 3*p2 + p1 + p0 + q0 + 4 ) >> 3;
                        }
                        else
                        {
                            /* p0' */
                            pix[-1] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
                        }
                        if( abs( q2 - q0 ) < beta )
                        {
                            const int q3 = pix[3];
                            /* q0', q1', q2' */
                            pix[0] = ( p1 + 2*p0 + 2*q0 + 2*q1 + q2 + 4 ) >> 3;
                            pix[1] = ( p0 + q0 + q1 + q2 + 2 ) >> 2;
                            pix[2] = ( 2*q3 + 3*q2 + q1 + q0 + p0 + 4 ) >> 3;
                        }
                        else
                        {
                            /* q0' */
                            pix[0] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
                        }
                    }
                    else
                    {
                        /* p0', q0' */
                        pix[-1] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
                        pix[0] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
                    }
                }
                pix += i_pix_stride;
            }
        }
    }
}

static inline void deblocking_filter_edgecv( x264_t *h, uint8_t *pix, int i_pix_stride, int bS[4], int i_QP )
{
    int i, d;
    const int i_index_a = x264_clip3( i_QP + h->sh.i_alpha_c0_offset, 0, 51 );
    const int alpha = i_alpha_table[i_index_a];
    const int beta  = i_beta_table[x264_clip3( i_QP + h->sh.i_beta_offset, 0, 51 )];

    for( i = 0; i < 4; i++ )
    {
        if( bS[i] == 0 )
        {
            pix += 2 * i_pix_stride;
            continue;
        }

        if( bS[i] < 4 )
        {
            const int tc = i_tc0_table[i_index_a][bS[i] - 1] + 1;
            /* 2px edge length (because we use same bS than the one for luma) */
            for( d = 0; d < 2; d++ )
            {
                const int p0 = pix[-1];
                const int p1 = pix[-2];
                const int q0 = pix[0];
                const int q1 = pix[1];

                if( abs( p0 - q0 ) < alpha &&
                    abs( p1 - p0 ) < beta &&
                    abs( q1 - q0 ) < beta )
                {
                    const int i_delta = x264_clip3( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );

                    pix[-1] = clip_uint8( p0 + i_delta );    /* p0' */
                    pix[0]  = clip_uint8( q0 - i_delta );    /* q0' */
                }
                pix += i_pix_stride;
            }
        }
        else
        {
            /* 2px edge length (because we use same bS than the one for luma) */
            for( d = 0; d < 2; d++ )
            {
                const int p0 = pix[-1];
                const int p1 = pix[-2];
                const int q0 = pix[0];
                const int q1 = pix[1];

                if( abs( p0 - q0 ) < alpha &&
                    abs( p1 - p0 ) < beta &&
                    abs( q1 - q0 ) < beta )
                {
                    pix[-1] = ( 2*p1 + p0 + q1 + 2 ) >> 2;   /* p0' */
                    pix[0]  = ( 2*q1 + q0 + p1 + 2 ) >> 2;   /* q0' */
                }
                pix += i_pix_stride;
            }
        }
    }
}

static inline void deblocking_filter_edgeh( x264_t *h, uint8_t *pix, int i_pix_stride, int bS[4], int i_QP )
{
    int i, d;
    const int i_index_a = x264_clip3( i_QP + h->sh.i_alpha_c0_offset, 0, 51 );
    const int alpha = i_alpha_table[i_index_a];
    const int beta  = i_beta_table[x264_clip3( i_QP + h->sh.i_beta_offset, 0, 51 )];

    int i_pix_next  = i_pix_stride;

    for( i = 0; i < 4; i++ )
    {
        if( bS[i] == 0 )
        {
            pix += 4;
            continue;
        }

        if( bS[i] < 4 )
        {
            const int tc0 = i_tc0_table[i_index_a][bS[i] - 1];
            /* 4px edge length */
            for( d = 0; d < 4; d++ )
            {
                const int p0 = pix[-i_pix_next];
                const int p1 = pix[-2*i_pix_next];
                const int p2 = pix[-3*i_pix_next];
                const int q0 = pix[0];
                const int q1 = pix[1*i_pix_next];
                const int q2 = pix[2*i_pix_next];

                if( abs( p0 - q0 ) < alpha &&
                    abs( p1 - p0 ) < beta &&
                    abs( q1 - q0 ) < beta )
                {
                    int tc = tc0;
                    int i_delta;

                    if( abs( p2 - p0 ) < beta )
                    {
                        pix[-2*i_pix_next] = p1 + x264_clip3( ( p2 + ( ( p0 + q0 + 1 ) >> 1 ) - ( p1 << 1 ) ) >> 1, -tc0, tc0 );
                        tc++;
                    }
                    if( abs( q2 - q0 ) < beta )
                    {
                        pix[i_pix_next] = q1 + x264_clip3( ( q2 + ( ( p0 + q0 + 1 ) >> 1 ) - ( q1 << 1 ) ) >> 1, -tc0, tc0 );
                        tc++;
                    }

                    i_delta = x264_clip3( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );
                    pix[-i_pix_next] = clip_uint8( p0 + i_delta );    /* p0' */
                    pix[0]           = clip_uint8( q0 - i_delta );    /* q0' */
                }
                pix++;
            }
        }
        else
        {
            /* 4px edge length */
            for( d = 0; d < 4; d++ )
            {
                const int p0 = pix[-i_pix_next];
                const int p1 = pix[-2*i_pix_next];
                const int p2 = pix[-3*i_pix_next];
                const int q0 = pix[0];
                const int q1 = pix[1*i_pix_next];
                const int q2 = pix[2*i_pix_next];

                if( abs( p0 - q0 ) < alpha &&
                    abs( p1 - p0 ) < beta &&
                    abs( q1 - q0 ) < beta )
                {
                    const int p3 = pix[-4*i_pix_next];
                    const int q3 = pix[ 3*i_pix_next];

                    if( abs( p0 - q0 ) < (( alpha >> 2 ) + 2 ) )
                    {
                        if( abs( p2 - p0 ) < beta )
                        {
                            /* p0', p1', p2' */
                            pix[-1*i_pix_next] = ( p2 + 2*p1 + 2*p0 + 2*q0 + q1 + 4 ) >> 3;
                            pix[-2*i_pix_next] = ( p2 + p1 + p0 + q0 + 2 ) >> 2;
                            pix[-3*i_pix_next] = ( 2*p3 + 3*p2 + p1 + p0 + q0 + 4 ) >> 3;
                        }
                        else
                        {
                            /* p0' */
                            pix[-1*i_pix_next] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
                        }
                        if( abs( q2 - q0 ) < beta )
                        {
                            /* q0', q1', q2' */
                            pix[0*i_pix_next] = ( p1 + 2*p0 + 2*q0 + 2*q1 + q2 + 4 ) >> 3;
                            pix[1*i_pix_next] = ( p0 + q0 + q1 + q2 + 2 ) >> 2;
                            pix[2*i_pix_next] = ( 2*q3 + 3*q2 + q1 + q0 + p0 + 4 ) >> 3;
                        }
                        else
                        {
                            /* q0' */
                            pix[0*i_pix_next] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
                        }
                    }
                    else
                    {
                        /* p0' */
                        pix[-1*i_pix_next] = ( 2*p1 + p0 + q1 + 2 ) >> 2;
                        /* q0' */
                        pix[0*i_pix_next] = ( 2*q1 + q0 + p1 + 2 ) >> 2;
                    }
                }
                pix++;
            }

        }
    }
}

static inline void deblocking_filter_edgech( x264_t *h, uint8_t *pix, int i_pix_stride, int bS[4], int i_QP )
{
    int i, d;
    const int i_index_a = x264_clip3( i_QP + h->sh.i_alpha_c0_offset, 0, 51 );
    const int alpha = i_alpha_table[i_index_a];
    const int beta  = i_beta_table[x264_clip3( i_QP + h->sh.i_beta_offset, 0, 51 )];

    int i_pix_next  = i_pix_stride;

    for( i = 0; i < 4; i++ )
    {
        if( bS[i] == 0 )
        {
            pix += 2;
            continue;
        }
        if( bS[i] < 4 )
        {
            int tc = i_tc0_table[i_index_a][bS[i] - 1] + 1;
            /* 2px edge length (see deblocking_filter_edgecv) */
            for( d = 0; d < 2; d++ )
            {
                const int p0 = pix[-1*i_pix_next];
                const int p1 = pix[-2*i_pix_next];
                const int q0 = pix[0];
                const int q1 = pix[1*i_pix_next];

                if( abs( p0 - q0 ) < alpha &&
                    abs( p1 - p0 ) < beta &&
                    abs( q1 - q0 ) < beta )
                {
                    int i_delta = x264_clip3( (((q0 - p0 ) << 2) + (p1 - q1) + 4) >> 3, -tc, tc );

                    pix[-i_pix_next] = clip_uint8( p0 + i_delta );    /* p0' */
                    pix[0]           = clip_uint8( q0 - i_delta );    /* q0' */
                }
                pix++;
            }
        }
        else
        {
            /* 2px edge length (see deblocking_filter_edgecv) */
            for( d = 0; d < 2; d++ )
            {
                const int p0 = pix[-1*i_pix_next];
                const int p1 = pix[-2*i_pix_next];
                const int q0 = pix[0];
                const int q1 = pix[1*i_pix_next];

                if( abs( p0 - q0 ) < alpha &&
                    abs( p1 - p0 ) < beta &&
                    abs( q1 - q0 ) < beta )
                {
                    pix[-i_pix_next] = ( 2*p1 + p0 + q1 + 2 ) >> 2;   /* p0' */
                    pix[0]           = ( 2*q1 + q0 + p1 + 2 ) >> 2;   /* q0' */
                }
                pix++;
            }
        }
    }
}

void x264_frame_deblocking_filter( x264_t *h, int i_slice_type )
{
    const int s8x8 = 2 * h->mb.i_mb_stride;
    const int s4x4 = 4 * h->mb.i_mb_stride;

    int mb_y, mb_x;

    for( mb_y = 0, mb_x = 0; mb_y < h->sps->i_mb_height; )
    {
        const int mb_xy  = mb_y * h->mb.i_mb_stride + mb_x;
        const int mb_8x8 = 2 * s8x8 * mb_y + 2 * mb_x;
        const int mb_4x4 = 4 * s4x4 * mb_y + 4 * mb_x;
        int i_edge;
        int i_dir;
652
653
654
655
656
657
658
659
660
661
662
663
        const int b_8x8_transform = h->mb.mb_transform_size[mb_xy];

        /* cavlc + 8x8 transform stores nnz per 16 coeffs for the purpose of
         * entropy coding, but per 64 coeffs for the purpose of deblocking */
        if( !h->param.b_cabac && b_8x8_transform )
        {
            uint32_t *nnz = (uint32_t*)h->mb.non_zero_count[mb_xy];
            if( nnz[0] ) nnz[0] = 0x01010101;
            if( nnz[1] ) nnz[1] = 0x01010101;
            if( nnz[2] ) nnz[2] = 0x01010101;
            if( nnz[3] ) nnz[3] = 0x01010101;
        }
Laurent Aimar's avatar
Laurent Aimar committed
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701

        /* i_dir == 0 -> vertical edge
         * i_dir == 1 -> horizontal edge */
        for( i_dir = 0; i_dir < 2; i_dir++ )
        {
            int i_start;
            int i_qp, i_qpn;

            i_start = (( i_dir == 0 && mb_x != 0 ) || ( i_dir == 1 && mb_y != 0 ) ) ? 0 : 1;

            for( i_edge = i_start; i_edge < 4; i_edge++ )
            {
                int mbn_xy  = i_edge > 0 ? mb_xy  : ( i_dir == 0 ? mb_xy  - 1 : mb_xy - h->mb.i_mb_stride );
                int mbn_8x8 = i_edge > 0 ? mb_8x8 : ( i_dir == 0 ? mb_8x8 - 2 : mb_8x8 - 2 * s8x8 );
                int mbn_4x4 = i_edge > 0 ? mb_4x4 : ( i_dir == 0 ? mb_4x4 - 4 : mb_4x4 - 4 * s4x4 );

                int bS[4];  /* filtering strength */

                /* *** Get bS for each 4px for the current edge *** */
                if( IS_INTRA( h->mb.type[mb_xy] ) || IS_INTRA( h->mb.type[mbn_xy] ) )
                {
                    bS[0] = bS[1] = bS[2] = bS[3] = ( i_edge == 0 ? 4 : 3 );
                }
                else
                {
                    int i;
                    for( i = 0; i < 4; i++ )
                    {
                        int x  = i_dir == 0 ? i_edge : i;
                        int y  = i_dir == 0 ? i      : i_edge;
                        int xn = (x - (i_dir == 0 ? 1 : 0 ))&0x03;
                        int yn = (y - (i_dir == 0 ? 0 : 1 ))&0x03;

                        if( h->mb.non_zero_count[mb_xy][block_idx_xy[x][y]] != 0 ||
                            h->mb.non_zero_count[mbn_xy][block_idx_xy[xn][yn]] != 0 )
                        {
                            bS[i] = 2;
                        }
702
                        else
Laurent Aimar's avatar
Laurent Aimar committed
703
                        {
704
705
706
707
708
709
710
711
712
713
714
715
716
717
                            /* FIXME: A given frame may occupy more than one position in
                             * the reference list. So we should compare the frame numbers,
                             * not the indices in the ref list.
                             * No harm yet, as we don't generate that case.*/

                            int i8p= mb_8x8+(x/2)+(y/2)*s8x8;
                            int i8q= mbn_8x8+(xn/2)+(yn/2)*s8x8;
                            int i4p= mb_4x4+x+y*s4x4;
                            int i4q= mbn_4x4+xn+yn*s4x4;
                            int l;

                            bS[i] = 0;

                            for( l = 0; l < 1 + (i_slice_type == SLICE_TYPE_B); l++ )
Laurent Aimar's avatar
Laurent Aimar committed
718
                            {
719
720
721
722
723
724
725
                                if( h->mb.ref[l][i8p] != h->mb.ref[l][i8q] ||
                                    abs( h->mb.mv[l][i4p][0] - h->mb.mv[l][i4q][0] ) >= 4 ||
                                    abs( h->mb.mv[l][i4p][1] - h->mb.mv[l][i4q][1] ) >= 4 )
                                {
                                    bS[i] = 1;
                                    break;
                                }
Laurent Aimar's avatar
Laurent Aimar committed
726
727
728
729
730
731
732
733
734
735
736
737
738
                            }
                        }
                    }
                }

                /* *** filter *** */
                /* Y plane */
                i_qp = h->mb.qp[mb_xy];
                i_qpn= h->mb.qp[mbn_xy];

                if( i_dir == 0 )
                {
                    /* vertical edge */
739
740
741
742
743
744
                    if( !b_8x8_transform || !(i_edge & 1) )
                    {
                        deblocking_filter_edgev( h, &h->fdec->plane[0][16 * mb_y * h->fdec->i_stride[0]+ 16 * mb_x + 4 * i_edge],
                                                 h->fdec->i_stride[0], bS, (i_qp+i_qpn+1) >> 1);
                    }
                    if( !(i_edge & 1) )
Laurent Aimar's avatar
Laurent Aimar committed
745
746
747
748
749
750
751
752
753
754
755
756
757
                    {
                        /* U/V planes */
                        int i_qpc = ( i_chroma_qp_table[x264_clip3( i_qp + h->pps->i_chroma_qp_index_offset, 0, 51 )] +
                                      i_chroma_qp_table[x264_clip3( i_qpn + h->pps->i_chroma_qp_index_offset, 0, 51 )] + 1 ) >> 1;
                        deblocking_filter_edgecv( h, &h->fdec->plane[1][8*(mb_y*h->fdec->i_stride[1]+mb_x)+i_edge*2],
                                                      h->fdec->i_stride[1], bS, i_qpc );
                        deblocking_filter_edgecv( h, &h->fdec->plane[2][8*(mb_y*h->fdec->i_stride[2]+mb_x)+i_edge*2],
                                                  h->fdec->i_stride[2], bS, i_qpc );
                    }
                }
                else
                {
                    /* horizontal edge */
758
759
760
761
762
                    if( !b_8x8_transform || !(i_edge & 1) )
                    {
                        deblocking_filter_edgeh( h, &h->fdec->plane[0][(16*mb_y + 4 * i_edge) * h->fdec->i_stride[0]+ 16 * mb_x],
                                                 h->fdec->i_stride[0], bS, (i_qp+i_qpn+1) >> 1 );
                    }
Laurent Aimar's avatar
Laurent Aimar committed
763
                    /* U/V planes */
764
                    if( !(i_edge & 1) )
Laurent Aimar's avatar
Laurent Aimar committed
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
                    {
                        int i_qpc = ( i_chroma_qp_table[x264_clip3( i_qp + h->pps->i_chroma_qp_index_offset, 0, 51 )] +
                                      i_chroma_qp_table[x264_clip3( i_qpn + h->pps->i_chroma_qp_index_offset, 0, 51 )] + 1 ) >> 1;
                        deblocking_filter_edgech( h, &h->fdec->plane[1][8*(mb_y*h->fdec->i_stride[1]+mb_x)+i_edge*2*h->fdec->i_stride[1]],
                                                 h->fdec->i_stride[1], bS, i_qpc );
                        deblocking_filter_edgech( h, &h->fdec->plane[2][8*(mb_y*h->fdec->i_stride[2]+mb_x)+i_edge*2*h->fdec->i_stride[2]],
                                                 h->fdec->i_stride[2], bS, i_qpc );
                    }
                }
            }
        }

        /* newt mb */
        mb_x++;
        if( mb_x >= h->sps->i_mb_width )
        {
            mb_x = 0;
            mb_y++;
        }
    }
}