aboutsummaryrefslogtreecommitdiffstats
path: root/libavcodec/hevc_mvs.c
blob: b2c0bcad0ae29efc3458ae1797f2829af6d05b1f (plain) (blame)
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
36
37
38
39
40
41
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
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
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
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
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
652
653
654
655
656
657
658
659
660
661
662
663
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
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
/*
 * HEVC video decoder
 *
 * Copyright (C) 2012 - 2013 Guillaume Martres
 * Copyright (C) 2013 Anand Meher Kotra
 *
 * This file is part of FFmpeg.
 *
 * FFmpeg is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * FFmpeg 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
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
 * License along with FFmpeg; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */

#include "hevc.h"

static const uint8_t l0_l1_cand_idx[12][2] = {
    { 0, 1, },
    { 1, 0, },
    { 0, 2, },
    { 2, 0, },
    { 1, 2, },
    { 2, 1, },
    { 0, 3, },
    { 3, 0, },
    { 1, 3, },
    { 3, 1, },
    { 2, 3, },
    { 3, 2, },
};

void ff_hevc_set_neighbour_available(HEVCContext *s, int x0, int y0,
                                     int nPbW, int nPbH)
{
    HEVCLocalContext *lc = s->HEVClc;
    int x0b = x0 & ((1 << s->sps->log2_ctb_size) - 1);
    int y0b = y0 & ((1 << s->sps->log2_ctb_size) - 1);

    lc->na.cand_up       = (lc->ctb_up_flag   || y0b);
    lc->na.cand_left     = (lc->ctb_left_flag || x0b);
    lc->na.cand_up_left  = (!x0b && !y0b) ? lc->ctb_up_left_flag : lc->na.cand_left && lc->na.cand_up;
    lc->na.cand_up_right_sap =
            ((x0b + nPbW) == (1 << s->sps->log2_ctb_size)) ?
                    lc->ctb_up_right_flag && !y0b : lc->na.cand_up;
    lc->na.cand_up_right =
            ((x0b + nPbW) == (1 << s->sps->log2_ctb_size) ?
                    lc->ctb_up_right_flag && !y0b : lc->na.cand_up )
                     && (x0 + nPbW) < lc->end_of_tiles_x;
    lc->na.cand_bottom_left = ((y0 + nPbH) >= lc->end_of_tiles_y) ? 0 : lc->na.cand_left;
}

/*
 * 6.4.1 Derivation process for z-scan order block availability
 */
static int z_scan_block_avail(HEVCContext *s, int xCurr, int yCurr,
                              int xN, int yN)
{
#define MIN_TB_ADDR_ZS(x, y)                                            \
    s->pps->min_tb_addr_zs[(y) * (s->sps->tb_mask+2) + (x)]

    int xCurr_ctb = xCurr >> s->sps->log2_ctb_size;
    int yCurr_ctb = yCurr >> s->sps->log2_ctb_size;
    int xN_ctb    = xN    >> s->sps->log2_ctb_size;
    int yN_ctb    = yN    >> s->sps->log2_ctb_size;

    if (xN < 0 || yN < 0 ||
        xN >= s->sps->width ||
        yN >= s->sps->height)
        return 0;

    if( yN_ctb < yCurr_ctb || xN_ctb < xCurr_ctb )
        return 1;
    else {
        int Curr = MIN_TB_ADDR_ZS((xCurr >> s->sps->log2_min_tb_size) & s->sps->tb_mask,
                (yCurr >> s->sps->log2_min_tb_size) & s->sps->tb_mask);
        int N    = MIN_TB_ADDR_ZS((xN >> s->sps->log2_min_tb_size) & s->sps->tb_mask,
                (yN >> s->sps->log2_min_tb_size) & s->sps->tb_mask);
        return N <= Curr;
    }
}

//check if the two luma locations belong to the same mostion estimation region
static int isDiffMER(HEVCContext *s, int xN, int yN, int xP, int yP)
{
    uint8_t plevel = s->pps->log2_parallel_merge_level;

    return xN >> plevel == xP >> plevel &&
           yN >> plevel == yP >> plevel;
}

#define MATCH_MV(x) (AV_RN32A(&A.x) == AV_RN32A(&B.x))
#define MATCH(x) (A.x == B.x)

// check if the mv's and refidx are the same between A and B
static int compareMVrefidx(struct MvField A, struct MvField B)
{
    int a_pf = A.pred_flag;
    int b_pf = B.pred_flag;
    if (a_pf == b_pf) {
        if (a_pf == PF_BI) {
            return MATCH(ref_idx[0]) && MATCH_MV(mv[0]) &&
                   MATCH(ref_idx[1]) && MATCH_MV(mv[1]);
        } else if (a_pf == PF_L0) {
            return MATCH(ref_idx[0]) && MATCH_MV(mv[0]);
        } else if (a_pf == PF_L1) {
            return MATCH(ref_idx[1]) && MATCH_MV(mv[1]);
        }
    }
    return 0;
}

static av_always_inline void mv_scale(Mv *dst, Mv *src, int td, int tb)
{
    int tx, scale_factor;

    td = av_clip_int8(td);
    tb = av_clip_int8(tb);
    tx = (0x4000 + abs(td / 2)) / td;
    scale_factor = av_clip((tb * tx + 32) >> 6, -4096, 4095);
    dst->x = av_clip_int16((scale_factor * src->x + 127 +
                           (scale_factor * src->x < 0)) >> 8);
    dst->y = av_clip_int16((scale_factor * src->y + 127 +
                           (scale_factor * src->y < 0)) >> 8);
}

static int check_mvset(Mv *mvLXCol, Mv *mvCol,
                       int colPic, int poc,
                       RefPicList *refPicList, int X, int refIdxLx,
                       RefPicList *refPicList_col, int listCol, int refidxCol)
{
    int cur_lt = refPicList[X].isLongTerm[refIdxLx];
    int col_lt = refPicList_col[listCol].isLongTerm[refidxCol];
    int col_poc_diff, cur_poc_diff;

    if (cur_lt != col_lt) {
        mvLXCol->x = 0;
        mvLXCol->y = 0;
        return 0;
    }

    col_poc_diff = colPic - refPicList_col[listCol].list[refidxCol];
    cur_poc_diff = poc    - refPicList[X].list[refIdxLx];

    if (cur_lt || col_poc_diff == cur_poc_diff || !col_poc_diff) {
        mvLXCol->x = mvCol->x;
        mvLXCol->y = mvCol->y;
    } else {
        mv_scale(mvLXCol, mvCol, col_poc_diff, cur_poc_diff);
    }
    return 1;
}

#define CHECK_MVSET(l)                                          \
    check_mvset(mvLXCol, temp_col.mv + l,                       \
                colPic, s->poc,                                 \
                refPicList, X, refIdxLx,                        \
                refPicList_col, L ## l, temp_col.ref_idx[l])

// derive the motion vectors section 8.5.3.1.8
static int derive_temporal_colocated_mvs(HEVCContext *s, MvField temp_col,
                                         int refIdxLx, Mv *mvLXCol, int X,
                                         int colPic, RefPicList *refPicList_col)
{
    RefPicList *refPicList = s->ref->refPicList;

    if (temp_col.pred_flag == PF_INTRA)
        return 0;

    if (!(temp_col.pred_flag & PF_L0))
        return CHECK_MVSET(1);
    else if (temp_col.pred_flag == PF_L0)
        return CHECK_MVSET(0);
    else if (temp_col.pred_flag == PF_BI) {
        int check_diffpicount = 0;
        int i, j;
        for (j = 0; j < 2; j++) {
            for (i = 0; i < refPicList[j].nb_refs; i++) {
                if (refPicList[j].list[i] > s->poc) {
                    check_diffpicount++;
                    break;
                }
            }
        }
        if (!check_diffpicount) {
            if (X==0)
                return CHECK_MVSET(0);
            else
                return CHECK_MVSET(1);
        } else {
            if (s->sh.collocated_list == L1)
                return CHECK_MVSET(0);
            else
                return CHECK_MVSET(1);
        }
    }

    return 0;
}

#define TAB_MVF(x, y)                                                   \
    tab_mvf[(y) * min_pu_width + x]

#define TAB_MVF_PU(v)                                                   \
    TAB_MVF(x ## v ## _pu, y ## v ## _pu)

#define DERIVE_TEMPORAL_COLOCATED_MVS                                   \
    derive_temporal_colocated_mvs(s, temp_col,                          \
                                  refIdxLx, mvLXCol, X, colPic,         \
                                  ff_hevc_get_ref_list(s, ref, x, y))

/*
 * 8.5.3.1.7  temporal luma motion vector prediction
 */
static int temporal_luma_motion_vector(HEVCContext *s, int x0, int y0,
                                       int nPbW, int nPbH, int refIdxLx,
                                       Mv *mvLXCol, int X)
{
    MvField *tab_mvf;
    MvField temp_col;
    int x, y, x_pu, y_pu;
    int min_pu_width = s->sps->min_pu_width;
    int availableFlagLXCol = 0;
    int colPic;

    HEVCFrame *ref = s->ref->collocated_ref;

    if (!ref)
        return 0;

    tab_mvf = ref->tab_mvf;
    colPic  = ref->poc;

    //bottom right collocated motion vector
    x = x0 + nPbW;
    y = y0 + nPbH;

    if (s->threads_type == FF_THREAD_FRAME )
        ff_thread_await_progress(&ref->tf, y, 0);
    if (tab_mvf &&
        (y0 >> s->sps->log2_ctb_size) == (y >> s->sps->log2_ctb_size) &&
        y < s->sps->height &&
        x < s->sps->width) {
        x                 &= -16;
        y                 &= -16;
        x_pu               = x >> s->sps->log2_min_pu_size;
        y_pu               = y >> s->sps->log2_min_pu_size;
        temp_col           = TAB_MVF(x_pu, y_pu);
        availableFlagLXCol = DERIVE_TEMPORAL_COLOCATED_MVS;
    }

    // derive center collocated motion vector
    if (tab_mvf && !availableFlagLXCol) {
        x                  = x0 + (nPbW >> 1);
        y                  = y0 + (nPbH >> 1);
        x                 &= -16;
        y                 &= -16;
        x_pu               = x >> s->sps->log2_min_pu_size;
        y_pu               = y >> s->sps->log2_min_pu_size;
        temp_col           = TAB_MVF(x_pu, y_pu);
        availableFlagLXCol = DERIVE_TEMPORAL_COLOCATED_MVS;
    }
    return availableFlagLXCol;
}

#define AVAILABLE(cand, v)                                      \
    (cand && !(TAB_MVF_PU(v).pred_flag == PF_INTRA))

#define PRED_BLOCK_AVAILABLE(v)                                 \
    z_scan_block_avail(s, x0, y0, x ## v, y ## v)

#define COMPARE_MV_REFIDX(a, b)                                 \
    compareMVrefidx(TAB_MVF_PU(a), TAB_MVF_PU(b))

/*
 * 8.5.3.1.2  Derivation process for spatial merging candidates
 */
static void derive_spatial_merge_candidates(HEVCContext *s, int x0, int y0,
                                            int nPbW, int nPbH,
                                            int log2_cb_size,
                                            int singleMCLFlag, int part_idx,
                                            int merge_idx,
                                            struct MvField mergecandlist[])
{
    HEVCLocalContext *lc   = s->HEVClc;
    RefPicList *refPicList = s->ref->refPicList;
    MvField *tab_mvf       = s->ref->tab_mvf;

    const int min_pu_width = s->sps->min_pu_width;

    const int cand_bottom_left = lc->na.cand_bottom_left;
    const int cand_left        = lc->na.cand_left;
    const int cand_up_left     = lc->na.cand_up_left;
    const int cand_up          = lc->na.cand_up;
    const int cand_up_right    = lc->na.cand_up_right_sap;

    const int xA1    = x0 - 1;
    const int yA1    = y0 + nPbH - 1;
    const int xA1_pu = xA1 >> s->sps->log2_min_pu_size;
    const int yA1_pu = yA1 >> s->sps->log2_min_pu_size;

    const int xB1    = x0 + nPbW - 1;
    const int yB1    = y0 - 1;
    const int xB1_pu = xB1 >> s->sps->log2_min_pu_size;
    const int yB1_pu = yB1 >> s->sps->log2_min_pu_size;

    const int xB0    = x0 + nPbW;
    const int yB0    = y0 - 1;
    const int xB0_pu = xB0 >> s->sps->log2_min_pu_size;
    const int yB0_pu = yB0 >> s->sps->log2_min_pu_size;

    const int xA0    = x0 - 1;
    const int yA0    = y0 + nPbH;
    const int xA0_pu = xA0 >> s->sps->log2_min_pu_size;
    const int yA0_pu = yA0 >> s->sps->log2_min_pu_size;

    const int xB2    = x0 - 1;
    const int yB2    = y0 - 1;
    const int xB2_pu = xB2 >> s->sps->log2_min_pu_size;
    const int yB2_pu = yB2 >> s->sps->log2_min_pu_size;

    const int nb_refs = (s->sh.slice_type == P_SLICE) ?
                        s->sh.nb_refs[0] : FFMIN(s->sh.nb_refs[0], s->sh.nb_refs[1]);

    int zero_idx = 0;

    int nb_merge_cand = 0;
    int nb_orig_merge_cand = 0;

    int is_available_a0;
    int is_available_a1;
    int is_available_b0;
    int is_available_b1;
    int is_available_b2;


    if (!singleMCLFlag && part_idx == 1 &&
        (lc->cu.part_mode == PART_Nx2N ||
         lc->cu.part_mode == PART_nLx2N ||
         lc->cu.part_mode == PART_nRx2N) ||
        isDiffMER(s, xA1, yA1, x0, y0)) {
        is_available_a1 = 0;
    } else {
        is_available_a1 = AVAILABLE(cand_left, A1);
        if (is_available_a1) {
            mergecandlist[nb_merge_cand] = TAB_MVF_PU(A1);
            if (merge_idx == 0) return;
            nb_merge_cand++;
        }
    }

    if (!singleMCLFlag && part_idx == 1 &&
        (lc->cu.part_mode == PART_2NxN ||
         lc->cu.part_mode == PART_2NxnU ||
         lc->cu.part_mode == PART_2NxnD) ||
        isDiffMER(s, xB1, yB1, x0, y0)) {
        is_available_b1 = 0;
    } else {
        is_available_b1 = AVAILABLE(cand_up, B1);
        if (is_available_b1 &&
            !(is_available_a1 && COMPARE_MV_REFIDX(B1, A1))) {
            mergecandlist[nb_merge_cand] = TAB_MVF_PU(B1);
            if (merge_idx == nb_merge_cand) return;
            nb_merge_cand++;
        }
    }

    // above right spatial merge candidate
    is_available_b0 = AVAILABLE(cand_up_right, B0) &&
                      xB0 < s->sps->width &&
                      PRED_BLOCK_AVAILABLE(B0) &&
                      !isDiffMER(s, xB0, yB0, x0, y0);

    if (is_available_b0 &&
        !(is_available_b1 && COMPARE_MV_REFIDX(B0, B1))) {
        mergecandlist[nb_merge_cand] = TAB_MVF_PU(B0);
        if (merge_idx == nb_merge_cand) return;
        nb_merge_cand++;
    }

    // left bottom spatial merge candidate
    is_available_a0 = AVAILABLE(cand_bottom_left, A0) &&
                      yA0 < s->sps->height &&
                      PRED_BLOCK_AVAILABLE(A0) &&
                      !isDiffMER(s, xA0, yA0, x0, y0);

    if (is_available_a0 &&
        !(is_available_a1 && COMPARE_MV_REFIDX(A0, A1))) {
        mergecandlist[nb_merge_cand] = TAB_MVF_PU(A0);
        if (merge_idx == nb_merge_cand) return;
        nb_merge_cand++;
    }

    // above left spatial merge candidate
    is_available_b2 = AVAILABLE(cand_up_left, B2) &&
                      !isDiffMER(s, xB2, yB2, x0, y0);

    if (is_available_b2 &&
        !(is_available_a1 && COMPARE_MV_REFIDX(B2, A1)) &&
        !(is_available_b1 && COMPARE_MV_REFIDX(B2, B1)) &&
        nb_merge_cand != 4) {
        mergecandlist[nb_merge_cand] = TAB_MVF_PU(B2);
        if (merge_idx == nb_merge_cand) return;
        nb_merge_cand++;
    }

    // temporal motion vector candidate
    if (s->sh.slice_temporal_mvp_enabled_flag &&
        nb_merge_cand < s->sh.max_num_merge_cand) {
        Mv mv_l0_col, mv_l1_col;
        int available_l0 = temporal_luma_motion_vector(s, x0, y0, nPbW, nPbH,
                                                       0, &mv_l0_col, 0);
        int available_l1 = (s->sh.slice_type == B_SLICE) ?
                           temporal_luma_motion_vector(s, x0, y0, nPbW, nPbH,
                                                       0, &mv_l1_col, 1) : 0;

        if (available_l0 || available_l1) {
            mergecandlist[nb_merge_cand].pred_flag = available_l0 + (available_l1 << 1);
            if (available_l0) {
                mergecandlist[nb_merge_cand].mv[0]      = mv_l0_col;
                mergecandlist[nb_merge_cand].ref_idx[0] = 0;
            }
            if (available_l1) {
                mergecandlist[nb_merge_cand].mv[1]      = mv_l1_col;
                mergecandlist[nb_merge_cand].ref_idx[1] = 0;
            }
            if (merge_idx == nb_merge_cand) return;
            nb_merge_cand++;
        }
    }

    nb_orig_merge_cand = nb_merge_cand;

    // combined bi-predictive merge candidates  (applies for B slices)
    if (s->sh.slice_type == B_SLICE && nb_orig_merge_cand > 1 &&
        nb_orig_merge_cand < s->sh.max_num_merge_cand) {
        int comb_idx = 0;

        for (comb_idx = 0; nb_merge_cand < s->sh.max_num_merge_cand &&
                           comb_idx < nb_orig_merge_cand * (nb_orig_merge_cand - 1); comb_idx++) {
            int l0_cand_idx = l0_l1_cand_idx[comb_idx][0];
            int l1_cand_idx = l0_l1_cand_idx[comb_idx][1];
            MvField l0_cand = mergecandlist[l0_cand_idx];
            MvField l1_cand = mergecandlist[l1_cand_idx];

            if ((l0_cand.pred_flag & PF_L0) && (l1_cand.pred_flag & PF_L1) &&
                (refPicList[0].list[l0_cand.ref_idx[0]] !=
                 refPicList[1].list[l1_cand.ref_idx[1]] ||
                 AV_RN32A(&l0_cand.mv[0]) != AV_RN32A(&l1_cand.mv[1]))) {
                mergecandlist[nb_merge_cand].ref_idx[0]   = l0_cand.ref_idx[0];
                mergecandlist[nb_merge_cand].ref_idx[1]   = l1_cand.ref_idx[1];
                mergecandlist[nb_merge_cand].pred_flag    = PF_BI;
                AV_COPY32(&mergecandlist[nb_merge_cand].mv[0], &l0_cand.mv[0]);
                AV_COPY32(&mergecandlist[nb_merge_cand].mv[1], &l1_cand.mv[1]);
                if (merge_idx == nb_merge_cand) return;
                nb_merge_cand++;
            }
        }
    }

    // append Zero motion vector candidates
    while (nb_merge_cand < s->sh.max_num_merge_cand) {
        mergecandlist[nb_merge_cand].pred_flag    = PF_L0 + ((s->sh.slice_type == B_SLICE) << 1);
        AV_ZERO32(mergecandlist[nb_merge_cand].mv+0);
        AV_ZERO32(mergecandlist[nb_merge_cand].mv+1);
        mergecandlist[nb_merge_cand].ref_idx[0]   = zero_idx < nb_refs ? zero_idx : 0;
        mergecandlist[nb_merge_cand].ref_idx[1]   = zero_idx < nb_refs ? zero_idx : 0;

        if (merge_idx == nb_merge_cand) return;
        nb_merge_cand++;
        zero_idx++;
    }
}

/*
 * 8.5.3.1.1 Derivation process of luma Mvs for merge mode
 */
void ff_hevc_luma_mv_merge_mode(HEVCContext *s, int x0, int y0, int nPbW,
                                int nPbH, int log2_cb_size, int part_idx,
                                int merge_idx, MvField *mv)
{
    int singleMCLFlag = 0;
    int nCS = 1 << log2_cb_size;
    LOCAL_ALIGNED(4, MvField, mergecand_list, [MRG_MAX_NUM_CANDS]);
    int nPbW2 = nPbW;
    int nPbH2 = nPbH;
    HEVCLocalContext *lc = s->HEVClc;

    if (s->pps->log2_parallel_merge_level > 2 && nCS == 8) {
        singleMCLFlag = 1;
        x0            = lc->cu.x;
        y0            = lc->cu.y;
        nPbW          = nCS;
        nPbH          = nCS;
        part_idx      = 0;
    }

    ff_hevc_set_neighbour_available(s, x0, y0, nPbW, nPbH);
    derive_spatial_merge_candidates(s, x0, y0, nPbW, nPbH, log2_cb_size,
                                    singleMCLFlag, part_idx,
                                    merge_idx, mergecand_list);

    if (mergecand_list[merge_idx].pred_flag == PF_BI &&
        (nPbW2 + nPbH2) == 12) {
        mergecand_list[merge_idx].pred_flag = PF_L0;
    }

    *mv = mergecand_list[merge_idx];
}

static av_always_inline void dist_scale(HEVCContext *s, Mv *mv,
                                        int min_pu_width, int x, int y,
                                        int elist, int ref_idx_curr, int ref_idx)
{
    RefPicList *refPicList = s->ref->refPicList;
    MvField *tab_mvf       = s->ref->tab_mvf;
    int ref_pic_elist      = refPicList[elist].list[TAB_MVF(x, y).ref_idx[elist]];
    int ref_pic_curr       = refPicList[ref_idx_curr].list[ref_idx];

    if (ref_pic_elist != ref_pic_curr) {
        int poc_diff = s->poc - ref_pic_elist;
        if (!poc_diff)
            poc_diff = 1;
        mv_scale(mv, mv, poc_diff, s->poc - ref_pic_curr);
    }
}

static int mv_mp_mode_mx(HEVCContext *s, int x, int y, int pred_flag_index,
                         Mv *mv, int ref_idx_curr, int ref_idx)
{
    MvField *tab_mvf = s->ref->tab_mvf;
    int min_pu_width = s->sps->min_pu_width;

    RefPicList *refPicList = s->ref->refPicList;

    if (((TAB_MVF(x, y).pred_flag) & (1 << pred_flag_index)) &&
        refPicList[pred_flag_index].list[TAB_MVF(x, y).ref_idx[pred_flag_index]] == refPicList[ref_idx_curr].list[ref_idx]) {
        *mv = TAB_MVF(x, y).mv[pred_flag_index];
        return 1;
    }
    return 0;
}

static int mv_mp_mode_mx_lt(HEVCContext *s, int x, int y, int pred_flag_index,
                            Mv *mv, int ref_idx_curr, int ref_idx)
{
    MvField *tab_mvf = s->ref->tab_mvf;
    int min_pu_width = s->sps->min_pu_width;

    RefPicList *refPicList = s->ref->refPicList;

    if ((TAB_MVF(x, y).pred_flag) & (1 << pred_flag_index)) {
        int currIsLongTerm     = refPicList[ref_idx_curr].isLongTerm[ref_idx];

        int colIsLongTerm =
            refPicList[pred_flag_index].isLongTerm[(TAB_MVF(x, y).ref_idx[pred_flag_index])];

        if (colIsLongTerm == currIsLongTerm) {
            *mv = TAB_MVF(x, y).mv[pred_flag_index];
            if (!currIsLongTerm)
                dist_scale(s, mv, min_pu_width, x, y,
                           pred_flag_index, ref_idx_curr, ref_idx);
            return 1;
        }
    }
    return 0;
}

#define MP_MX(v, pred, mx)                                      \
    mv_mp_mode_mx(s, x ## v ## _pu, y ## v ## _pu, pred,        \
                  &mx, ref_idx_curr, ref_idx)

#define MP_MX_LT(v, pred, mx)                                   \
    mv_mp_mode_mx_lt(s, x ## v ## _pu, y ## v ## _pu, pred,     \
                     &mx, ref_idx_curr, ref_idx)

void ff_hevc_luma_mv_mvp_mode(HEVCContext *s, int x0, int y0, int nPbW,
                              int nPbH, int log2_cb_size, int part_idx,
                              int merge_idx, MvField *mv,
                              int mvp_lx_flag, int LX)
{
    HEVCLocalContext *lc = s->HEVClc;
    MvField *tab_mvf = s->ref->tab_mvf;
    int isScaledFlag_L0 = 0;
    int availableFlagLXA0 = 0;
    int availableFlagLXB0 = 0;
    int numMVPCandLX = 0;
    int min_pu_width = s->sps->min_pu_width;

    int xA0, yA0;
    int xA0_pu, yA0_pu;
    int is_available_a0;

    int xA1, yA1;
    int xA1_pu, yA1_pu;
    int is_available_a1;

    int xB0, yB0;
    int xB0_pu, yB0_pu;
    int is_available_b0;

    int xB1, yB1;
    int xB1_pu = 0, yB1_pu = 0;
    int is_available_b1 = 0;

    int xB2, yB2;
    int xB2_pu = 0, yB2_pu = 0;
    int is_available_b2 = 0;
    Mv mvpcand_list[2] = { { 0 } };
    Mv mxA;
    Mv mxB;
    int ref_idx_curr = 0;
    int ref_idx = 0;
    int pred_flag_index_l0;
    int pred_flag_index_l1;
    int x0b = x0 & ((1 << s->sps->log2_ctb_size) - 1);
    int y0b = y0 & ((1 << s->sps->log2_ctb_size) - 1);

    int cand_up = (lc->ctb_up_flag || y0b);
    int cand_left = (lc->ctb_left_flag || x0b);
    int cand_up_left =
            (!x0b && !y0b) ? lc->ctb_up_left_flag : cand_left && cand_up;
    int cand_up_right =
            (x0b + nPbW == (1 << s->sps->log2_ctb_size) ||
             x0  + nPbW >= lc->end_of_tiles_x) ? lc->ctb_up_right_flag && !y0b
                                               : cand_up;
    int cand_bottom_left = (y0 + nPbH >= lc->end_of_tiles_y) ? 0 : cand_left;

    ref_idx_curr       = LX;
    ref_idx            = mv->ref_idx[LX];
    pred_flag_index_l0 = LX;
    pred_flag_index_l1 = !LX;

    // left bottom spatial candidate
    xA0 = x0 - 1;
    yA0 = y0 + nPbH;
    xA0_pu = xA0 >> s->sps->log2_min_pu_size;
    yA0_pu = yA0 >> s->sps->log2_min_pu_size;

    is_available_a0 = AVAILABLE(cand_bottom_left, A0) &&
                      yA0 < s->sps->height &&
                      PRED_BLOCK_AVAILABLE(A0);

    //left spatial merge candidate
    xA1    = x0 - 1;
    yA1    = y0 + nPbH - 1;
    xA1_pu = xA1 >> s->sps->log2_min_pu_size;
    yA1_pu = yA1 >> s->sps->log2_min_pu_size;

    is_available_a1 = AVAILABLE(cand_left, A1);
    if (is_available_a0 || is_available_a1)
        isScaledFlag_L0 = 1;

    if (is_available_a0) {
        availableFlagLXA0 = MP_MX(A0, pred_flag_index_l0, mxA);
        if (!availableFlagLXA0)
            availableFlagLXA0 = MP_MX(A0, pred_flag_index_l1, mxA);
    }

    if (is_available_a1 && !availableFlagLXA0) {
        availableFlagLXA0 = MP_MX(A1, pred_flag_index_l0, mxA);
        if (!availableFlagLXA0)
            availableFlagLXA0 = MP_MX(A1, pred_flag_index_l1, mxA);
    }

    if (is_available_a0 && !availableFlagLXA0) {
        availableFlagLXA0 = MP_MX_LT(A0, pred_flag_index_l0, mxA);
        if (!availableFlagLXA0)
            availableFlagLXA0 = MP_MX_LT(A0, pred_flag_index_l1, mxA);
    }

    if (is_available_a1 && !availableFlagLXA0) {
        availableFlagLXA0 = MP_MX_LT(A1, pred_flag_index_l0, mxA);
        if (!availableFlagLXA0)
            availableFlagLXA0 = MP_MX_LT(A1, pred_flag_index_l1, mxA);
    }

    if(availableFlagLXA0 && !mvp_lx_flag) {
        mv->mv[LX] = mxA;
        return;
    }

    // B candidates
    // above right spatial merge candidate
    xB0    = x0 + nPbW;
    yB0    = y0 - 1;
    xB0_pu = xB0 >> s->sps->log2_min_pu_size;
    yB0_pu = yB0 >> s->sps->log2_min_pu_size;

    is_available_b0 = PRED_BLOCK_AVAILABLE(B0) && AVAILABLE(cand_up_right, B0);

    if (is_available_b0) {
        availableFlagLXB0 = MP_MX(B0, pred_flag_index_l0, mxB);
        if (!availableFlagLXB0)
            availableFlagLXB0 = MP_MX(B0, pred_flag_index_l1, mxB);
    }

    if (!availableFlagLXB0) {
        // above spatial merge candidate
        xB1    = x0 + nPbW - 1;
        yB1    = y0 - 1;
        xB1_pu = xB1 >> s->sps->log2_min_pu_size;
        yB1_pu = yB1 >> s->sps->log2_min_pu_size;

        is_available_b1 = AVAILABLE(cand_up, B1);

        if (is_available_b1) {
            availableFlagLXB0 = MP_MX(B1, pred_flag_index_l0, mxB);
            if (!availableFlagLXB0)
                availableFlagLXB0 = MP_MX(B1, pred_flag_index_l1, mxB);
        }
    }

    if (!availableFlagLXB0) {
        // above left spatial merge candidate
        xB2 = x0 - 1;
        yB2 = y0 - 1;
        xB2_pu = xB2 >> s->sps->log2_min_pu_size;
        yB2_pu = yB2 >> s->sps->log2_min_pu_size;
        is_available_b2 = AVAILABLE(cand_up_left, B2);

        if (is_available_b2) {
            availableFlagLXB0 = MP_MX(B2, pred_flag_index_l0, mxB);
            if (!availableFlagLXB0)
                availableFlagLXB0 = MP_MX(B2, pred_flag_index_l1, mxB);
        }
    }

    if (isScaledFlag_L0 == 0) {
        if (availableFlagLXB0) {
            availableFlagLXA0 = 1;
            mxA = mxB;
        }
        availableFlagLXB0 = 0;

        // XB0 and L1
        if (is_available_b0) {
            availableFlagLXB0 = MP_MX_LT(B0, pred_flag_index_l0, mxB);
            if (!availableFlagLXB0)
                availableFlagLXB0 = MP_MX_LT(B0, pred_flag_index_l1, mxB);
        }

        if (is_available_b1 && !availableFlagLXB0) {
            availableFlagLXB0 = MP_MX_LT(B1, pred_flag_index_l0, mxB);
            if (!availableFlagLXB0)
                availableFlagLXB0 = MP_MX_LT(B1, pred_flag_index_l1, mxB);
        }

        if (is_available_b2 && !availableFlagLXB0) {
            availableFlagLXB0 = MP_MX_LT(B2, pred_flag_index_l0, mxB);
            if (!availableFlagLXB0)
                availableFlagLXB0 = MP_MX_LT(B2, pred_flag_index_l1, mxB);
        }
    }

    if (availableFlagLXA0)
        mvpcand_list[numMVPCandLX++] = mxA;

    if (availableFlagLXB0 && (!availableFlagLXA0 || mxA.x != mxB.x || mxA.y != mxB.y))
        mvpcand_list[numMVPCandLX++] = mxB;

    //temporal motion vector prediction candidate
    if (numMVPCandLX < 2 && s->sh.slice_temporal_mvp_enabled_flag) {
        Mv mv_col;
        int available_col = temporal_luma_motion_vector(s, x0, y0, nPbW,
                                                        nPbH, ref_idx,
                                                        &mv_col, LX);
        if (available_col)
            mvpcand_list[numMVPCandLX++] = mv_col;
    }

    mv->mv[LX] = mvpcand_list[mvp_lx_flag];
}