summaryrefslogtreecommitdiffstats
path: root/third_party/aom/av1/common/resize.c
blob: 17e6823b1e8c2e10dfe75869ae5a2c762c284e2a (plain)
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
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
1279
1280
1281
1282
1283
1284
1285
1286
1287
1288
1289
1290
1291
1292
1293
1294
1295
1296
1297
1298
1299
1300
1301
1302
1303
1304
1305
1306
1307
1308
1309
1310
/*
 * Copyright (c) 2016, Alliance for Open Media. All rights reserved
 *
 * This source code is subject to the terms of the BSD 2 Clause License and
 * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
 * was not distributed with this source code in the LICENSE file, you can
 * obtain it at www.aomedia.org/license/software. If the Alliance for Open
 * Media Patent License 1.0 was not distributed with this source code in the
 * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
 */

#include <assert.h>
#include <limits.h>
#include <math.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>

#include "config/aom_config.h"

#include "aom_dsp/aom_dsp_common.h"
#include "aom_ports/mem.h"
#include "aom_scale/aom_scale.h"
#include "av1/common/common.h"
#include "av1/common/resize.h"

#include "config/aom_scale_rtcd.h"

// Filters for interpolation (0.5-band) - note this also filters integer pels.
static const InterpKernel filteredinterp_filters500[(1 << RS_SUBPEL_BITS)] = {
  { -3, 0, 35, 64, 35, 0, -3, 0 },    { -3, 0, 34, 64, 36, 0, -3, 0 },
  { -3, -1, 34, 64, 36, 1, -3, 0 },   { -3, -1, 33, 64, 37, 1, -3, 0 },
  { -3, -1, 32, 64, 38, 1, -3, 0 },   { -3, -1, 31, 64, 39, 1, -3, 0 },
  { -3, -1, 31, 63, 39, 2, -3, 0 },   { -2, -2, 30, 63, 40, 2, -3, 0 },
  { -2, -2, 29, 63, 41, 2, -3, 0 },   { -2, -2, 29, 63, 41, 3, -4, 0 },
  { -2, -2, 28, 63, 42, 3, -4, 0 },   { -2, -2, 27, 63, 43, 3, -4, 0 },
  { -2, -3, 27, 63, 43, 4, -4, 0 },   { -2, -3, 26, 62, 44, 5, -4, 0 },
  { -2, -3, 25, 62, 45, 5, -4, 0 },   { -2, -3, 25, 62, 45, 5, -4, 0 },
  { -2, -3, 24, 62, 46, 5, -4, 0 },   { -2, -3, 23, 61, 47, 6, -4, 0 },
  { -2, -3, 23, 61, 47, 6, -4, 0 },   { -2, -3, 22, 61, 48, 7, -4, -1 },
  { -2, -3, 21, 60, 49, 7, -4, 0 },   { -1, -4, 20, 60, 49, 8, -4, 0 },
  { -1, -4, 20, 60, 50, 8, -4, -1 },  { -1, -4, 19, 59, 51, 9, -4, -1 },
  { -1, -4, 19, 59, 51, 9, -4, -1 },  { -1, -4, 18, 58, 52, 10, -4, -1 },
  { -1, -4, 17, 58, 52, 11, -4, -1 }, { -1, -4, 16, 58, 53, 11, -4, -1 },
  { -1, -4, 16, 57, 53, 12, -4, -1 }, { -1, -4, 15, 57, 54, 12, -4, -1 },
  { -1, -4, 15, 56, 54, 13, -4, -1 }, { -1, -4, 14, 56, 55, 13, -4, -1 },
  { -1, -4, 14, 55, 55, 14, -4, -1 }, { -1, -4, 13, 55, 56, 14, -4, -1 },
  { -1, -4, 13, 54, 56, 15, -4, -1 }, { -1, -4, 12, 54, 57, 15, -4, -1 },
  { -1, -4, 12, 53, 57, 16, -4, -1 }, { -1, -4, 11, 53, 58, 16, -4, -1 },
  { -1, -4, 11, 52, 58, 17, -4, -1 }, { -1, -4, 10, 52, 58, 18, -4, -1 },
  { -1, -4, 9, 51, 59, 19, -4, -1 },  { -1, -4, 9, 51, 59, 19, -4, -1 },
  { -1, -4, 8, 50, 60, 20, -4, -1 },  { 0, -4, 8, 49, 60, 20, -4, -1 },
  { 0, -4, 7, 49, 60, 21, -3, -2 },   { -1, -4, 7, 48, 61, 22, -3, -2 },
  { 0, -4, 6, 47, 61, 23, -3, -2 },   { 0, -4, 6, 47, 61, 23, -3, -2 },
  { 0, -4, 5, 46, 62, 24, -3, -2 },   { 0, -4, 5, 45, 62, 25, -3, -2 },
  { 0, -4, 5, 45, 62, 25, -3, -2 },   { 0, -4, 5, 44, 62, 26, -3, -2 },
  { 0, -4, 4, 43, 63, 27, -3, -2 },   { 0, -4, 3, 43, 63, 27, -2, -2 },
  { 0, -4, 3, 42, 63, 28, -2, -2 },   { 0, -4, 3, 41, 63, 29, -2, -2 },
  { 0, -3, 2, 41, 63, 29, -2, -2 },   { 0, -3, 2, 40, 63, 30, -2, -2 },
  { 0, -3, 2, 39, 63, 31, -1, -3 },   { 0, -3, 1, 39, 64, 31, -1, -3 },
  { 0, -3, 1, 38, 64, 32, -1, -3 },   { 0, -3, 1, 37, 64, 33, -1, -3 },
  { 0, -3, 1, 36, 64, 34, -1, -3 },   { 0, -3, 0, 36, 64, 34, 0, -3 },
};

// Filters for interpolation (0.625-band) - note this also filters integer pels.
static const InterpKernel filteredinterp_filters625[(1 << RS_SUBPEL_BITS)] = {
  { -1, -8, 33, 80, 33, -8, -1, 0 }, { -1, -8, 31, 80, 34, -8, -1, 1 },
  { -1, -8, 30, 80, 35, -8, -1, 1 }, { -1, -8, 29, 80, 36, -7, -2, 1 },
  { -1, -8, 28, 80, 37, -7, -2, 1 }, { -1, -8, 27, 80, 38, -7, -2, 1 },
  { 0, -8, 26, 79, 39, -7, -2, 1 },  { 0, -8, 25, 79, 40, -7, -2, 1 },
  { 0, -8, 24, 79, 41, -7, -2, 1 },  { 0, -8, 23, 78, 42, -6, -2, 1 },
  { 0, -8, 22, 78, 43, -6, -2, 1 },  { 0, -8, 21, 78, 44, -6, -2, 1 },
  { 0, -8, 20, 78, 45, -5, -3, 1 },  { 0, -8, 19, 77, 47, -5, -3, 1 },
  { 0, -8, 18, 77, 48, -5, -3, 1 },  { 0, -8, 17, 77, 49, -5, -3, 1 },
  { 0, -8, 16, 76, 50, -4, -3, 1 },  { 0, -8, 15, 76, 51, -4, -3, 1 },
  { 0, -8, 15, 75, 52, -3, -4, 1 },  { 0, -7, 14, 74, 53, -3, -4, 1 },
  { 0, -7, 13, 74, 54, -3, -4, 1 },  { 0, -7, 12, 73, 55, -2, -4, 1 },
  { 0, -7, 11, 73, 56, -2, -4, 1 },  { 0, -7, 10, 72, 57, -1, -4, 1 },
  { 1, -7, 10, 71, 58, -1, -5, 1 },  { 0, -7, 9, 71, 59, 0, -5, 1 },
  { 1, -7, 8, 70, 60, 0, -5, 1 },    { 1, -7, 7, 69, 61, 1, -5, 1 },
  { 1, -6, 6, 68, 62, 1, -5, 1 },    { 0, -6, 6, 68, 62, 2, -5, 1 },
  { 1, -6, 5, 67, 63, 2, -5, 1 },    { 1, -6, 5, 66, 64, 3, -6, 1 },
  { 1, -6, 4, 65, 65, 4, -6, 1 },    { 1, -6, 3, 64, 66, 5, -6, 1 },
  { 1, -5, 2, 63, 67, 5, -6, 1 },    { 1, -5, 2, 62, 68, 6, -6, 0 },
  { 1, -5, 1, 62, 68, 6, -6, 1 },    { 1, -5, 1, 61, 69, 7, -7, 1 },
  { 1, -5, 0, 60, 70, 8, -7, 1 },    { 1, -5, 0, 59, 71, 9, -7, 0 },
  { 1, -5, -1, 58, 71, 10, -7, 1 },  { 1, -4, -1, 57, 72, 10, -7, 0 },
  { 1, -4, -2, 56, 73, 11, -7, 0 },  { 1, -4, -2, 55, 73, 12, -7, 0 },
  { 1, -4, -3, 54, 74, 13, -7, 0 },  { 1, -4, -3, 53, 74, 14, -7, 0 },
  { 1, -4, -3, 52, 75, 15, -8, 0 },  { 1, -3, -4, 51, 76, 15, -8, 0 },
  { 1, -3, -4, 50, 76, 16, -8, 0 },  { 1, -3, -5, 49, 77, 17, -8, 0 },
  { 1, -3, -5, 48, 77, 18, -8, 0 },  { 1, -3, -5, 47, 77, 19, -8, 0 },
  { 1, -3, -5, 45, 78, 20, -8, 0 },  { 1, -2, -6, 44, 78, 21, -8, 0 },
  { 1, -2, -6, 43, 78, 22, -8, 0 },  { 1, -2, -6, 42, 78, 23, -8, 0 },
  { 1, -2, -7, 41, 79, 24, -8, 0 },  { 1, -2, -7, 40, 79, 25, -8, 0 },
  { 1, -2, -7, 39, 79, 26, -8, 0 },  { 1, -2, -7, 38, 80, 27, -8, -1 },
  { 1, -2, -7, 37, 80, 28, -8, -1 }, { 1, -2, -7, 36, 80, 29, -8, -1 },
  { 1, -1, -8, 35, 80, 30, -8, -1 }, { 1, -1, -8, 34, 80, 31, -8, -1 },
};

// Filters for interpolation (0.75-band) - note this also filters integer pels.
static const InterpKernel filteredinterp_filters750[(1 << RS_SUBPEL_BITS)] = {
  { 2, -11, 25, 96, 25, -11, 2, 0 }, { 2, -11, 24, 96, 26, -11, 2, 0 },
  { 2, -11, 22, 96, 28, -11, 2, 0 }, { 2, -10, 21, 96, 29, -12, 2, 0 },
  { 2, -10, 19, 96, 31, -12, 2, 0 }, { 2, -10, 18, 95, 32, -11, 2, 0 },
  { 2, -10, 17, 95, 34, -12, 2, 0 }, { 2, -9, 15, 95, 35, -12, 2, 0 },
  { 2, -9, 14, 94, 37, -12, 2, 0 },  { 2, -9, 13, 94, 38, -12, 2, 0 },
  { 2, -8, 12, 93, 40, -12, 1, 0 },  { 2, -8, 11, 93, 41, -12, 1, 0 },
  { 2, -8, 9, 92, 43, -12, 1, 1 },   { 2, -8, 8, 92, 44, -12, 1, 1 },
  { 2, -7, 7, 91, 46, -12, 1, 0 },   { 2, -7, 6, 90, 47, -12, 1, 1 },
  { 2, -7, 5, 90, 49, -12, 1, 0 },   { 2, -6, 4, 89, 50, -12, 1, 0 },
  { 2, -6, 3, 88, 52, -12, 0, 1 },   { 2, -6, 2, 87, 54, -12, 0, 1 },
  { 2, -5, 1, 86, 55, -12, 0, 1 },   { 2, -5, 0, 85, 57, -12, 0, 1 },
  { 2, -5, -1, 84, 58, -11, 0, 1 },  { 2, -5, -2, 83, 60, -11, 0, 1 },
  { 2, -4, -2, 82, 61, -11, -1, 1 }, { 1, -4, -3, 81, 63, -10, -1, 1 },
  { 2, -4, -4, 80, 64, -10, -1, 1 }, { 1, -4, -4, 79, 66, -10, -1, 1 },
  { 1, -3, -5, 77, 67, -9, -1, 1 },  { 1, -3, -6, 76, 69, -9, -1, 1 },
  { 1, -3, -6, 75, 70, -8, -2, 1 },  { 1, -2, -7, 74, 71, -8, -2, 1 },
  { 1, -2, -7, 72, 72, -7, -2, 1 },  { 1, -2, -8, 71, 74, -7, -2, 1 },
  { 1, -2, -8, 70, 75, -6, -3, 1 },  { 1, -1, -9, 69, 76, -6, -3, 1 },
  { 1, -1, -9, 67, 77, -5, -3, 1 },  { 1, -1, -10, 66, 79, -4, -4, 1 },
  { 1, -1, -10, 64, 80, -4, -4, 2 }, { 1, -1, -10, 63, 81, -3, -4, 1 },
  { 1, -1, -11, 61, 82, -2, -4, 2 }, { 1, 0, -11, 60, 83, -2, -5, 2 },
  { 1, 0, -11, 58, 84, -1, -5, 2 },  { 1, 0, -12, 57, 85, 0, -5, 2 },
  { 1, 0, -12, 55, 86, 1, -5, 2 },   { 1, 0, -12, 54, 87, 2, -6, 2 },
  { 1, 0, -12, 52, 88, 3, -6, 2 },   { 0, 1, -12, 50, 89, 4, -6, 2 },
  { 0, 1, -12, 49, 90, 5, -7, 2 },   { 1, 1, -12, 47, 90, 6, -7, 2 },
  { 0, 1, -12, 46, 91, 7, -7, 2 },   { 1, 1, -12, 44, 92, 8, -8, 2 },
  { 1, 1, -12, 43, 92, 9, -8, 2 },   { 0, 1, -12, 41, 93, 11, -8, 2 },
  { 0, 1, -12, 40, 93, 12, -8, 2 },  { 0, 2, -12, 38, 94, 13, -9, 2 },
  { 0, 2, -12, 37, 94, 14, -9, 2 },  { 0, 2, -12, 35, 95, 15, -9, 2 },
  { 0, 2, -12, 34, 95, 17, -10, 2 }, { 0, 2, -11, 32, 95, 18, -10, 2 },
  { 0, 2, -12, 31, 96, 19, -10, 2 }, { 0, 2, -12, 29, 96, 21, -10, 2 },
  { 0, 2, -11, 28, 96, 22, -11, 2 }, { 0, 2, -11, 26, 96, 24, -11, 2 },
};

// Filters for interpolation (0.875-band) - note this also filters integer pels.
static const InterpKernel filteredinterp_filters875[(1 << RS_SUBPEL_BITS)] = {
  { 3, -8, 13, 112, 13, -8, 3, 0 },   { 2, -7, 12, 112, 15, -8, 3, -1 },
  { 3, -7, 10, 112, 17, -9, 3, -1 },  { 2, -6, 8, 112, 19, -9, 3, -1 },
  { 2, -6, 7, 112, 21, -10, 3, -1 },  { 2, -5, 6, 111, 22, -10, 3, -1 },
  { 2, -5, 4, 111, 24, -10, 3, -1 },  { 2, -4, 3, 110, 26, -11, 3, -1 },
  { 2, -4, 1, 110, 28, -11, 3, -1 },  { 2, -4, 0, 109, 30, -12, 4, -1 },
  { 1, -3, -1, 108, 32, -12, 4, -1 }, { 1, -3, -2, 108, 34, -13, 4, -1 },
  { 1, -2, -4, 107, 36, -13, 4, -1 }, { 1, -2, -5, 106, 38, -13, 4, -1 },
  { 1, -1, -6, 105, 40, -14, 4, -1 }, { 1, -1, -7, 104, 42, -14, 4, -1 },
  { 1, -1, -7, 103, 44, -15, 4, -1 }, { 1, 0, -8, 101, 46, -15, 4, -1 },
  { 1, 0, -9, 100, 48, -15, 4, -1 },  { 1, 0, -10, 99, 50, -15, 4, -1 },
  { 1, 1, -11, 97, 53, -16, 4, -1 },  { 0, 1, -11, 96, 55, -16, 4, -1 },
  { 0, 1, -12, 95, 57, -16, 4, -1 },  { 0, 2, -13, 93, 59, -16, 4, -1 },
  { 0, 2, -13, 91, 61, -16, 4, -1 },  { 0, 2, -14, 90, 63, -16, 4, -1 },
  { 0, 2, -14, 88, 65, -16, 4, -1 },  { 0, 2, -15, 86, 67, -16, 4, 0 },
  { 0, 3, -15, 84, 69, -17, 4, 0 },   { 0, 3, -16, 83, 71, -17, 4, 0 },
  { 0, 3, -16, 81, 73, -16, 3, 0 },   { 0, 3, -16, 79, 75, -16, 3, 0 },
  { 0, 3, -16, 77, 77, -16, 3, 0 },   { 0, 3, -16, 75, 79, -16, 3, 0 },
  { 0, 3, -16, 73, 81, -16, 3, 0 },   { 0, 4, -17, 71, 83, -16, 3, 0 },
  { 0, 4, -17, 69, 84, -15, 3, 0 },   { 0, 4, -16, 67, 86, -15, 2, 0 },
  { -1, 4, -16, 65, 88, -14, 2, 0 },  { -1, 4, -16, 63, 90, -14, 2, 0 },
  { -1, 4, -16, 61, 91, -13, 2, 0 },  { -1, 4, -16, 59, 93, -13, 2, 0 },
  { -1, 4, -16, 57, 95, -12, 1, 0 },  { -1, 4, -16, 55, 96, -11, 1, 0 },
  { -1, 4, -16, 53, 97, -11, 1, 1 },  { -1, 4, -15, 50, 99, -10, 0, 1 },
  { -1, 4, -15, 48, 100, -9, 0, 1 },  { -1, 4, -15, 46, 101, -8, 0, 1 },
  { -1, 4, -15, 44, 103, -7, -1, 1 }, { -1, 4, -14, 42, 104, -7, -1, 1 },
  { -1, 4, -14, 40, 105, -6, -1, 1 }, { -1, 4, -13, 38, 106, -5, -2, 1 },
  { -1, 4, -13, 36, 107, -4, -2, 1 }, { -1, 4, -13, 34, 108, -2, -3, 1 },
  { -1, 4, -12, 32, 108, -1, -3, 1 }, { -1, 4, -12, 30, 109, 0, -4, 2 },
  { -1, 3, -11, 28, 110, 1, -4, 2 },  { -1, 3, -11, 26, 110, 3, -4, 2 },
  { -1, 3, -10, 24, 111, 4, -5, 2 },  { -1, 3, -10, 22, 111, 6, -5, 2 },
  { -1, 3, -10, 21, 112, 7, -6, 2 },  { -1, 3, -9, 19, 112, 8, -6, 2 },
  { -1, 3, -9, 17, 112, 10, -7, 3 },  { -1, 3, -8, 15, 112, 12, -7, 2 },
};

// Filters for interpolation (full-band) - no filtering for integer pixels
static const InterpKernel filteredinterp_filters1000[(1 << RS_SUBPEL_BITS)] = {
  { 0, 0, 0, 128, 0, 0, 0, 0 },        { 0, 0, -1, 128, 2, -1, 0, 0 },
  { 0, 1, -3, 127, 4, -2, 1, 0 },      { 0, 1, -4, 127, 6, -3, 1, 0 },
  { 0, 2, -6, 126, 8, -3, 1, 0 },      { 0, 2, -7, 125, 11, -4, 1, 0 },
  { -1, 2, -8, 125, 13, -5, 2, 0 },    { -1, 3, -9, 124, 15, -6, 2, 0 },
  { -1, 3, -10, 123, 18, -6, 2, -1 },  { -1, 3, -11, 122, 20, -7, 3, -1 },
  { -1, 4, -12, 121, 22, -8, 3, -1 },  { -1, 4, -13, 120, 25, -9, 3, -1 },
  { -1, 4, -14, 118, 28, -9, 3, -1 },  { -1, 4, -15, 117, 30, -10, 4, -1 },
  { -1, 5, -16, 116, 32, -11, 4, -1 }, { -1, 5, -16, 114, 35, -12, 4, -1 },
  { -1, 5, -17, 112, 38, -12, 4, -1 }, { -1, 5, -18, 111, 40, -13, 5, -1 },
  { -1, 5, -18, 109, 43, -14, 5, -1 }, { -1, 6, -19, 107, 45, -14, 5, -1 },
  { -1, 6, -19, 105, 48, -15, 5, -1 }, { -1, 6, -19, 103, 51, -16, 5, -1 },
  { -1, 6, -20, 101, 53, -16, 6, -1 }, { -1, 6, -20, 99, 56, -17, 6, -1 },
  { -1, 6, -20, 97, 58, -17, 6, -1 },  { -1, 6, -20, 95, 61, -18, 6, -1 },
  { -2, 7, -20, 93, 64, -18, 6, -2 },  { -2, 7, -20, 91, 66, -19, 6, -1 },
  { -2, 7, -20, 88, 69, -19, 6, -1 },  { -2, 7, -20, 86, 71, -19, 6, -1 },
  { -2, 7, -20, 84, 74, -20, 7, -2 },  { -2, 7, -20, 81, 76, -20, 7, -1 },
  { -2, 7, -20, 79, 79, -20, 7, -2 },  { -1, 7, -20, 76, 81, -20, 7, -2 },
  { -2, 7, -20, 74, 84, -20, 7, -2 },  { -1, 6, -19, 71, 86, -20, 7, -2 },
  { -1, 6, -19, 69, 88, -20, 7, -2 },  { -1, 6, -19, 66, 91, -20, 7, -2 },
  { -2, 6, -18, 64, 93, -20, 7, -2 },  { -1, 6, -18, 61, 95, -20, 6, -1 },
  { -1, 6, -17, 58, 97, -20, 6, -1 },  { -1, 6, -17, 56, 99, -20, 6, -1 },
  { -1, 6, -16, 53, 101, -20, 6, -1 }, { -1, 5, -16, 51, 103, -19, 6, -1 },
  { -1, 5, -15, 48, 105, -19, 6, -1 }, { -1, 5, -14, 45, 107, -19, 6, -1 },
  { -1, 5, -14, 43, 109, -18, 5, -1 }, { -1, 5, -13, 40, 111, -18, 5, -1 },
  { -1, 4, -12, 38, 112, -17, 5, -1 }, { -1, 4, -12, 35, 114, -16, 5, -1 },
  { -1, 4, -11, 32, 116, -16, 5, -1 }, { -1, 4, -10, 30, 117, -15, 4, -1 },
  { -1, 3, -9, 28, 118, -14, 4, -1 },  { -1, 3, -9, 25, 120, -13, 4, -1 },
  { -1, 3, -8, 22, 121, -12, 4, -1 },  { -1, 3, -7, 20, 122, -11, 3, -1 },
  { -1, 2, -6, 18, 123, -10, 3, -1 },  { 0, 2, -6, 15, 124, -9, 3, -1 },
  { 0, 2, -5, 13, 125, -8, 2, -1 },    { 0, 1, -4, 11, 125, -7, 2, 0 },
  { 0, 1, -3, 8, 126, -6, 2, 0 },      { 0, 1, -3, 6, 127, -4, 1, 0 },
  { 0, 1, -2, 4, 127, -3, 1, 0 },      { 0, 0, -1, 2, 128, -1, 0, 0 },
};

const int16_t av1_resize_filter_normative[(
    1 << RS_SUBPEL_BITS)][UPSCALE_NORMATIVE_TAPS] = {
#if UPSCALE_NORMATIVE_TAPS == 8
  { 0, 0, 0, 128, 0, 0, 0, 0 },        { 0, 0, -1, 128, 2, -1, 0, 0 },
  { 0, 1, -3, 127, 4, -2, 1, 0 },      { 0, 1, -4, 127, 6, -3, 1, 0 },
  { 0, 2, -6, 126, 8, -3, 1, 0 },      { 0, 2, -7, 125, 11, -4, 1, 0 },
  { -1, 2, -8, 125, 13, -5, 2, 0 },    { -1, 3, -9, 124, 15, -6, 2, 0 },
  { -1, 3, -10, 123, 18, -6, 2, -1 },  { -1, 3, -11, 122, 20, -7, 3, -1 },
  { -1, 4, -12, 121, 22, -8, 3, -1 },  { -1, 4, -13, 120, 25, -9, 3, -1 },
  { -1, 4, -14, 118, 28, -9, 3, -1 },  { -1, 4, -15, 117, 30, -10, 4, -1 },
  { -1, 5, -16, 116, 32, -11, 4, -1 }, { -1, 5, -16, 114, 35, -12, 4, -1 },
  { -1, 5, -17, 112, 38, -12, 4, -1 }, { -1, 5, -18, 111, 40, -13, 5, -1 },
  { -1, 5, -18, 109, 43, -14, 5, -1 }, { -1, 6, -19, 107, 45, -14, 5, -1 },
  { -1, 6, -19, 105, 48, -15, 5, -1 }, { -1, 6, -19, 103, 51, -16, 5, -1 },
  { -1, 6, -20, 101, 53, -16, 6, -1 }, { -1, 6, -20, 99, 56, -17, 6, -1 },
  { -1, 6, -20, 97, 58, -17, 6, -1 },  { -1, 6, -20, 95, 61, -18, 6, -1 },
  { -2, 7, -20, 93, 64, -18, 6, -2 },  { -2, 7, -20, 91, 66, -19, 6, -1 },
  { -2, 7, -20, 88, 69, -19, 6, -1 },  { -2, 7, -20, 86, 71, -19, 6, -1 },
  { -2, 7, -20, 84, 74, -20, 7, -2 },  { -2, 7, -20, 81, 76, -20, 7, -1 },
  { -2, 7, -20, 79, 79, -20, 7, -2 },  { -1, 7, -20, 76, 81, -20, 7, -2 },
  { -2, 7, -20, 74, 84, -20, 7, -2 },  { -1, 6, -19, 71, 86, -20, 7, -2 },
  { -1, 6, -19, 69, 88, -20, 7, -2 },  { -1, 6, -19, 66, 91, -20, 7, -2 },
  { -2, 6, -18, 64, 93, -20, 7, -2 },  { -1, 6, -18, 61, 95, -20, 6, -1 },
  { -1, 6, -17, 58, 97, -20, 6, -1 },  { -1, 6, -17, 56, 99, -20, 6, -1 },
  { -1, 6, -16, 53, 101, -20, 6, -1 }, { -1, 5, -16, 51, 103, -19, 6, -1 },
  { -1, 5, -15, 48, 105, -19, 6, -1 }, { -1, 5, -14, 45, 107, -19, 6, -1 },
  { -1, 5, -14, 43, 109, -18, 5, -1 }, { -1, 5, -13, 40, 111, -18, 5, -1 },
  { -1, 4, -12, 38, 112, -17, 5, -1 }, { -1, 4, -12, 35, 114, -16, 5, -1 },
  { -1, 4, -11, 32, 116, -16, 5, -1 }, { -1, 4, -10, 30, 117, -15, 4, -1 },
  { -1, 3, -9, 28, 118, -14, 4, -1 },  { -1, 3, -9, 25, 120, -13, 4, -1 },
  { -1, 3, -8, 22, 121, -12, 4, -1 },  { -1, 3, -7, 20, 122, -11, 3, -1 },
  { -1, 2, -6, 18, 123, -10, 3, -1 },  { 0, 2, -6, 15, 124, -9, 3, -1 },
  { 0, 2, -5, 13, 125, -8, 2, -1 },    { 0, 1, -4, 11, 125, -7, 2, 0 },
  { 0, 1, -3, 8, 126, -6, 2, 0 },      { 0, 1, -3, 6, 127, -4, 1, 0 },
  { 0, 1, -2, 4, 127, -3, 1, 0 },      { 0, 0, -1, 2, 128, -1, 0, 0 },
#else
#error "Invalid value of UPSCALE_NORMATIVE_TAPS"
#endif  // UPSCALE_NORMATIVE_TAPS == 8
};

// Filters for factor of 2 downsampling.
static const int16_t av1_down2_symeven_half_filter[] = { 56, 12, -3, -1 };
static const int16_t av1_down2_symodd_half_filter[] = { 64, 35, 0, -3 };

static const InterpKernel *choose_interp_filter(int in_length, int out_length) {
  int out_length16 = out_length * 16;
  if (out_length16 >= in_length * 16)
    return filteredinterp_filters1000;
  else if (out_length16 >= in_length * 13)
    return filteredinterp_filters875;
  else if (out_length16 >= in_length * 11)
    return filteredinterp_filters750;
  else if (out_length16 >= in_length * 9)
    return filteredinterp_filters625;
  else
    return filteredinterp_filters500;
}

static void interpolate_core(const uint8_t *const input, int in_length,
                             uint8_t *output, int out_length,
                             const int16_t *interp_filters, int interp_taps) {
  const int32_t delta =
      (((uint32_t)in_length << RS_SCALE_SUBPEL_BITS) + out_length / 2) /
      out_length;
  const int32_t offset =
      in_length > out_length
          ? (((int32_t)(in_length - out_length) << (RS_SCALE_SUBPEL_BITS - 1)) +
             out_length / 2) /
                out_length
          : -(((int32_t)(out_length - in_length)
               << (RS_SCALE_SUBPEL_BITS - 1)) +
              out_length / 2) /
                out_length;
  uint8_t *optr = output;
  int x, x1, x2, sum, k, int_pel, sub_pel;
  int32_t y;

  x = 0;
  y = offset + RS_SCALE_EXTRA_OFF;
  while ((y >> RS_SCALE_SUBPEL_BITS) < (interp_taps / 2 - 1)) {
    x++;
    y += delta;
  }
  x1 = x;
  x = out_length - 1;
  y = delta * x + offset + RS_SCALE_EXTRA_OFF;
  while ((y >> RS_SCALE_SUBPEL_BITS) + (int32_t)(interp_taps / 2) >=
         in_length) {
    x--;
    y -= delta;
  }
  x2 = x;
  if (x1 > x2) {
    for (x = 0, y = offset + RS_SCALE_EXTRA_OFF; x < out_length;
         ++x, y += delta) {
      int_pel = y >> RS_SCALE_SUBPEL_BITS;
      sub_pel = (y >> RS_SCALE_EXTRA_BITS) & RS_SUBPEL_MASK;
      const int16_t *filter = &interp_filters[sub_pel * interp_taps];
      sum = 0;
      for (k = 0; k < interp_taps; ++k) {
        const int pk = int_pel - interp_taps / 2 + 1 + k;
        sum += filter[k] * input[AOMMAX(AOMMIN(pk, in_length - 1), 0)];
      }
      *optr++ = clip_pixel(ROUND_POWER_OF_TWO(sum, FILTER_BITS));
    }
  } else {
    // Initial part.
    for (x = 0, y = offset + RS_SCALE_EXTRA_OFF; x < x1; ++x, y += delta) {
      int_pel = y >> RS_SCALE_SUBPEL_BITS;
      sub_pel = (y >> RS_SCALE_EXTRA_BITS) & RS_SUBPEL_MASK;
      const int16_t *filter = &interp_filters[sub_pel * interp_taps];
      sum = 0;
      for (k = 0; k < interp_taps; ++k)
        sum += filter[k] * input[AOMMAX(int_pel - interp_taps / 2 + 1 + k, 0)];
      *optr++ = clip_pixel(ROUND_POWER_OF_TWO(sum, FILTER_BITS));
    }
    // Middle part.
    for (; x <= x2; ++x, y += delta) {
      int_pel = y >> RS_SCALE_SUBPEL_BITS;
      sub_pel = (y >> RS_SCALE_EXTRA_BITS) & RS_SUBPEL_MASK;
      const int16_t *filter = &interp_filters[sub_pel * interp_taps];
      sum = 0;
      for (k = 0; k < interp_taps; ++k)
        sum += filter[k] * input[int_pel - interp_taps / 2 + 1 + k];
      *optr++ = clip_pixel(ROUND_POWER_OF_TWO(sum, FILTER_BITS));
    }
    // End part.
    for (; x < out_length; ++x, y += delta) {
      int_pel = y >> RS_SCALE_SUBPEL_BITS;
      sub_pel = (y >> RS_SCALE_EXTRA_BITS) & RS_SUBPEL_MASK;
      const int16_t *filter = &interp_filters[sub_pel * interp_taps];
      sum = 0;
      for (k = 0; k < interp_taps; ++k)
        sum += filter[k] *
               input[AOMMIN(int_pel - interp_taps / 2 + 1 + k, in_length - 1)];
      *optr++ = clip_pixel(ROUND_POWER_OF_TWO(sum, FILTER_BITS));
    }
  }
}

static void interpolate(const uint8_t *const input, int in_length,
                        uint8_t *output, int out_length) {
  const InterpKernel *interp_filters =
      choose_interp_filter(in_length, out_length);

  interpolate_core(input, in_length, output, out_length, &interp_filters[0][0],
                   SUBPEL_TAPS);
}

int32_t av1_get_upscale_convolve_step(int in_length, int out_length) {
  return ((in_length << RS_SCALE_SUBPEL_BITS) + out_length / 2) / out_length;
}

static int32_t get_upscale_convolve_x0(int in_length, int out_length,
                                       int32_t x_step_qn) {
  const int err = out_length * x_step_qn - (in_length << RS_SCALE_SUBPEL_BITS);
  const int32_t x0 =
      (-((out_length - in_length) << (RS_SCALE_SUBPEL_BITS - 1)) +
       out_length / 2) /
          out_length +
      RS_SCALE_EXTRA_OFF - err / 2;
  return (int32_t)((uint32_t)x0 & RS_SCALE_SUBPEL_MASK);
}

#ifndef __clang_analyzer__
static void down2_symeven(const uint8_t *const input, int length,
                          uint8_t *output) {
  // Actual filter len = 2 * filter_len_half.
  const int16_t *filter = av1_down2_symeven_half_filter;
  const int filter_len_half = sizeof(av1_down2_symeven_half_filter) / 2;
  int i, j;
  uint8_t *optr = output;
  int l1 = filter_len_half;
  int l2 = (length - filter_len_half);
  l1 += (l1 & 1);
  l2 += (l2 & 1);
  if (l1 > l2) {
    // Short input length.
    for (i = 0; i < length; i += 2) {
      int sum = (1 << (FILTER_BITS - 1));
      for (j = 0; j < filter_len_half; ++j) {
        sum +=
            (input[AOMMAX(i - j, 0)] + input[AOMMIN(i + 1 + j, length - 1)]) *
            filter[j];
      }
      sum >>= FILTER_BITS;
      *optr++ = clip_pixel(sum);
    }
  } else {
    // Initial part.
    for (i = 0; i < l1; i += 2) {
      int sum = (1 << (FILTER_BITS - 1));
      for (j = 0; j < filter_len_half; ++j) {
        sum += (input[AOMMAX(i - j, 0)] + input[i + 1 + j]) * filter[j];
      }
      sum >>= FILTER_BITS;
      *optr++ = clip_pixel(sum);
    }
    // Middle part.
    for (; i < l2; i += 2) {
      int sum = (1 << (FILTER_BITS - 1));
      for (j = 0; j < filter_len_half; ++j) {
        sum += (input[i - j] + input[i + 1 + j]) * filter[j];
      }
      sum >>= FILTER_BITS;
      *optr++ = clip_pixel(sum);
    }
    // End part.
    for (; i < length; i += 2) {
      int sum = (1 << (FILTER_BITS - 1));
      for (j = 0; j < filter_len_half; ++j) {
        sum +=
            (input[i - j] + input[AOMMIN(i + 1 + j, length - 1)]) * filter[j];
      }
      sum >>= FILTER_BITS;
      *optr++ = clip_pixel(sum);
    }
  }
}
#endif

static void down2_symodd(const uint8_t *const input, int length,
                         uint8_t *output) {
  // Actual filter len = 2 * filter_len_half - 1.
  const int16_t *filter = av1_down2_symodd_half_filter;
  const int filter_len_half = sizeof(av1_down2_symodd_half_filter) / 2;
  int i, j;
  uint8_t *optr = output;
  int l1 = filter_len_half - 1;
  int l2 = (length - filter_len_half + 1);
  l1 += (l1 & 1);
  l2 += (l2 & 1);
  if (l1 > l2) {
    // Short input length.
    for (i = 0; i < length; i += 2) {
      int sum = (1 << (FILTER_BITS - 1)) + input[i] * filter[0];
      for (j = 1; j < filter_len_half; ++j) {
        sum += (input[(i - j < 0 ? 0 : i - j)] +
                input[(i + j >= length ? length - 1 : i + j)]) *
               filter[j];
      }
      sum >>= FILTER_BITS;
      *optr++ = clip_pixel(sum);
    }
  } else {
    // Initial part.
    for (i = 0; i < l1; i += 2) {
      int sum = (1 << (FILTER_BITS - 1)) + input[i] * filter[0];
      for (j = 1; j < filter_len_half; ++j) {
        sum += (input[(i - j < 0 ? 0 : i - j)] + input[i + j]) * filter[j];
      }
      sum >>= FILTER_BITS;
      *optr++ = clip_pixel(sum);
    }
    // Middle part.
    for (; i < l2; i += 2) {
      int sum = (1 << (FILTER_BITS - 1)) + input[i] * filter[0];
      for (j = 1; j < filter_len_half; ++j) {
        sum += (input[i - j] + input[i + j]) * filter[j];
      }
      sum >>= FILTER_BITS;
      *optr++ = clip_pixel(sum);
    }
    // End part.
    for (; i < length; i += 2) {
      int sum = (1 << (FILTER_BITS - 1)) + input[i] * filter[0];
      for (j = 1; j < filter_len_half; ++j) {
        sum += (input[i - j] + input[(i + j >= length ? length - 1 : i + j)]) *
               filter[j];
      }
      sum >>= FILTER_BITS;
      *optr++ = clip_pixel(sum);
    }
  }
}

static int get_down2_length(int length, int steps) {
  for (int s = 0; s < steps; ++s) length = (length + 1) >> 1;
  return length;
}

static int get_down2_steps(int in_length, int out_length) {
  int steps = 0;
  int proj_in_length;
  while ((proj_in_length = get_down2_length(in_length, 1)) >= out_length) {
    ++steps;
    in_length = proj_in_length;
    if (in_length == 1) {
      // Special case: we break because any further calls to get_down2_length()
      // with be with length == 1, which return 1, resulting in an infinite
      // loop.
      break;
    }
  }
  return steps;
}

static void resize_multistep(const uint8_t *const input, int length,
                             uint8_t *output, int olength, uint8_t *otmp) {
  if (length == olength) {
    memcpy(output, input, sizeof(output[0]) * length);
    return;
  }
  const int steps = get_down2_steps(length, olength);

  if (steps > 0) {
    uint8_t *out = NULL;
    int filteredlength = length;

    assert(otmp != NULL);
    uint8_t *otmp2 = otmp + get_down2_length(length, 1);
    for (int s = 0; s < steps; ++s) {
      const int proj_filteredlength = get_down2_length(filteredlength, 1);
      const uint8_t *const in = (s == 0 ? input : out);
      if (s == steps - 1 && proj_filteredlength == olength)
        out = output;
      else
        out = (s & 1 ? otmp2 : otmp);
      if (filteredlength & 1)
        down2_symodd(in, filteredlength, out);
      else
        down2_symeven(in, filteredlength, out);
      filteredlength = proj_filteredlength;
    }
    if (filteredlength != olength) {
      interpolate(out, filteredlength, output, olength);
    }
  } else {
    interpolate(input, length, output, olength);
  }
}

static void fill_col_to_arr(uint8_t *img, int stride, int len, uint8_t *arr) {
  int i;
  uint8_t *iptr = img;
  uint8_t *aptr = arr;
  for (i = 0; i < len; ++i, iptr += stride) {
    *aptr++ = *iptr;
  }
}

static void fill_arr_to_col(uint8_t *img, int stride, int len, uint8_t *arr) {
  int i;
  uint8_t *iptr = img;
  uint8_t *aptr = arr;
  for (i = 0; i < len; ++i, iptr += stride) {
    *iptr = *aptr++;
  }
}

static void resize_plane(const uint8_t *const input, int height, int width,
                         int in_stride, uint8_t *output, int height2,
                         int width2, int out_stride) {
  int i;
  uint8_t *intbuf = (uint8_t *)aom_malloc(sizeof(uint8_t) * width2 * height);
  uint8_t *tmpbuf =
      (uint8_t *)aom_malloc(sizeof(uint8_t) * AOMMAX(width, height));
  uint8_t *arrbuf = (uint8_t *)aom_malloc(sizeof(uint8_t) * height);
  uint8_t *arrbuf2 = (uint8_t *)aom_malloc(sizeof(uint8_t) * height2);
  if (intbuf == NULL || tmpbuf == NULL || arrbuf == NULL || arrbuf2 == NULL)
    goto Error;
  assert(width > 0);
  assert(height > 0);
  assert(width2 > 0);
  assert(height2 > 0);
  for (i = 0; i < height; ++i)
    resize_multistep(input + in_stride * i, width, intbuf + width2 * i, width2,
                     tmpbuf);
  for (i = 0; i < width2; ++i) {
    fill_col_to_arr(intbuf + i, width2, height, arrbuf);
    resize_multistep(arrbuf, height, arrbuf2, height2, tmpbuf);
    fill_arr_to_col(output + i, out_stride, height2, arrbuf2);
  }

Error:
  aom_free(intbuf);
  aom_free(tmpbuf);
  aom_free(arrbuf);
  aom_free(arrbuf2);
}

static void upscale_normative_rect(const uint8_t *const input, int height,
                                   int width, int in_stride, uint8_t *output,
                                   int height2, int width2, int out_stride,
                                   int x_step_qn, int x0_qn, int pad_left,
                                   int pad_right) {
  assert(width > 0);
  assert(height > 0);
  assert(width2 > 0);
  assert(height2 > 0);
  assert(height2 == height);

  // Extend the left/right pixels of the tile column if needed
  // (either because we can't sample from other tiles, or because we're at
  // a frame edge).
  // Save the overwritten pixels into tmp_left and tmp_right.
  // Note: Because we pass input-1 to av1_convolve_horiz_rs, we need one extra
  // column of border pixels compared to what we'd naively think.
  const int border_cols = UPSCALE_NORMATIVE_TAPS / 2 + 1;
  uint8_t *tmp_left =
      NULL;  // Silence spurious "may be used uninitialized" warnings
  uint8_t *tmp_right = NULL;
  uint8_t *const in_tl = (uint8_t *)(input - border_cols);  // Cast off 'const'
  uint8_t *const in_tr = (uint8_t *)(input + width);
  if (pad_left) {
    tmp_left = (uint8_t *)aom_malloc(sizeof(*tmp_left) * border_cols * height);
    for (int i = 0; i < height; i++) {
      memcpy(tmp_left + i * border_cols, in_tl + i * in_stride, border_cols);
      memset(in_tl + i * in_stride, input[i * in_stride], border_cols);
    }
  }
  if (pad_right) {
    tmp_right =
        (uint8_t *)aom_malloc(sizeof(*tmp_right) * border_cols * height);
    for (int i = 0; i < height; i++) {
      memcpy(tmp_right + i * border_cols, in_tr + i * in_stride, border_cols);
      memset(in_tr + i * in_stride, input[i * in_stride + width - 1],
             border_cols);
    }
  }

  av1_convolve_horiz_rs(input - 1, in_stride, output, out_stride, width2,
                        height2, &av1_resize_filter_normative[0][0], x0_qn,
                        x_step_qn);

  // Restore the left/right border pixels
  if (pad_left) {
    for (int i = 0; i < height; i++) {
      memcpy(in_tl + i * in_stride, tmp_left + i * border_cols, border_cols);
    }
    aom_free(tmp_left);
  }
  if (pad_right) {
    for (int i = 0; i < height; i++) {
      memcpy(in_tr + i * in_stride, tmp_right + i * border_cols, border_cols);
    }
    aom_free(tmp_right);
  }
}

static void highbd_interpolate_core(const uint16_t *const input, int in_length,
                                    uint16_t *output, int out_length, int bd,
                                    const int16_t *interp_filters,
                                    int interp_taps) {
  const int32_t delta =
      (((uint32_t)in_length << RS_SCALE_SUBPEL_BITS) + out_length / 2) /
      out_length;
  const int32_t offset =
      in_length > out_length
          ? (((int32_t)(in_length - out_length) << (RS_SCALE_SUBPEL_BITS - 1)) +
             out_length / 2) /
                out_length
          : -(((int32_t)(out_length - in_length)
               << (RS_SCALE_SUBPEL_BITS - 1)) +
              out_length / 2) /
                out_length;
  uint16_t *optr = output;
  int x, x1, x2, sum, k, int_pel, sub_pel;
  int32_t y;

  x = 0;
  y = offset + RS_SCALE_EXTRA_OFF;
  while ((y >> RS_SCALE_SUBPEL_BITS) < (interp_taps / 2 - 1)) {
    x++;
    y += delta;
  }
  x1 = x;
  x = out_length - 1;
  y = delta * x + offset + RS_SCALE_EXTRA_OFF;
  while ((y >> RS_SCALE_SUBPEL_BITS) + (int32_t)(interp_taps / 2) >=
         in_length) {
    x--;
    y -= delta;
  }
  x2 = x;
  if (x1 > x2) {
    for (x = 0, y = offset + RS_SCALE_EXTRA_OFF; x < out_length;
         ++x, y += delta) {
      int_pel = y >> RS_SCALE_SUBPEL_BITS;
      sub_pel = (y >> RS_SCALE_EXTRA_BITS) & RS_SUBPEL_MASK;
      const int16_t *filter = &interp_filters[sub_pel * interp_taps];
      sum = 0;
      for (k = 0; k < interp_taps; ++k) {
        const int pk = int_pel - interp_taps / 2 + 1 + k;
        sum += filter[k] * input[AOMMAX(AOMMIN(pk, in_length - 1), 0)];
      }
      *optr++ = clip_pixel_highbd(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd);
    }
  } else {
    // Initial part.
    for (x = 0, y = offset + RS_SCALE_EXTRA_OFF; x < x1; ++x, y += delta) {
      int_pel = y >> RS_SCALE_SUBPEL_BITS;
      sub_pel = (y >> RS_SCALE_EXTRA_BITS) & RS_SUBPEL_MASK;
      const int16_t *filter = &interp_filters[sub_pel * interp_taps];
      sum = 0;
      for (k = 0; k < interp_taps; ++k)
        sum += filter[k] * input[AOMMAX(int_pel - interp_taps / 2 + 1 + k, 0)];
      *optr++ = clip_pixel_highbd(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd);
    }
    // Middle part.
    for (; x <= x2; ++x, y += delta) {
      int_pel = y >> RS_SCALE_SUBPEL_BITS;
      sub_pel = (y >> RS_SCALE_EXTRA_BITS) & RS_SUBPEL_MASK;
      const int16_t *filter = &interp_filters[sub_pel * interp_taps];
      sum = 0;
      for (k = 0; k < interp_taps; ++k)
        sum += filter[k] * input[int_pel - interp_taps / 2 + 1 + k];
      *optr++ = clip_pixel_highbd(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd);
    }
    // End part.
    for (; x < out_length; ++x, y += delta) {
      int_pel = y >> RS_SCALE_SUBPEL_BITS;
      sub_pel = (y >> RS_SCALE_EXTRA_BITS) & RS_SUBPEL_MASK;
      const int16_t *filter = &interp_filters[sub_pel * interp_taps];
      sum = 0;
      for (k = 0; k < interp_taps; ++k)
        sum += filter[k] *
               input[AOMMIN(int_pel - interp_taps / 2 + 1 + k, in_length - 1)];
      *optr++ = clip_pixel_highbd(ROUND_POWER_OF_TWO(sum, FILTER_BITS), bd);
    }
  }
}

static void highbd_interpolate(const uint16_t *const input, int in_length,
                               uint16_t *output, int out_length, int bd) {
  const InterpKernel *interp_filters =
      choose_interp_filter(in_length, out_length);

  highbd_interpolate_core(input, in_length, output, out_length, bd,
                          &interp_filters[0][0], SUBPEL_TAPS);
}

#ifndef __clang_analyzer__
static void highbd_down2_symeven(const uint16_t *const input, int length,
                                 uint16_t *output, int bd) {
  // Actual filter len = 2 * filter_len_half.
  static const int16_t *filter = av1_down2_symeven_half_filter;
  const int filter_len_half = sizeof(av1_down2_symeven_half_filter) / 2;
  int i, j;
  uint16_t *optr = output;
  int l1 = filter_len_half;
  int l2 = (length - filter_len_half);
  l1 += (l1 & 1);
  l2 += (l2 & 1);
  if (l1 > l2) {
    // Short input length.
    for (i = 0; i < length; i += 2) {
      int sum = (1 << (FILTER_BITS - 1));
      for (j = 0; j < filter_len_half; ++j) {
        sum +=
            (input[AOMMAX(0, i - j)] + input[AOMMIN(i + 1 + j, length - 1)]) *
            filter[j];
      }
      sum >>= FILTER_BITS;
      *optr++ = clip_pixel_highbd(sum, bd);
    }
  } else {
    // Initial part.
    for (i = 0; i < l1; i += 2) {
      int sum = (1 << (FILTER_BITS - 1));
      for (j = 0; j < filter_len_half; ++j) {
        sum += (input[AOMMAX(0, i - j)] + input[i + 1 + j]) * filter[j];
      }
      sum >>= FILTER_BITS;
      *optr++ = clip_pixel_highbd(sum, bd);
    }
    // Middle part.
    for (; i < l2; i += 2) {
      int sum = (1 << (FILTER_BITS - 1));
      for (j = 0; j < filter_len_half; ++j) {
        sum += (input[i - j] + input[i + 1 + j]) * filter[j];
      }
      sum >>= FILTER_BITS;
      *optr++ = clip_pixel_highbd(sum, bd);
    }
    // End part.
    for (; i < length; i += 2) {
      int sum = (1 << (FILTER_BITS - 1));
      for (j = 0; j < filter_len_half; ++j) {
        sum +=
            (input[i - j] + input[AOMMIN(i + 1 + j, length - 1)]) * filter[j];
      }
      sum >>= FILTER_BITS;
      *optr++ = clip_pixel_highbd(sum, bd);
    }
  }
}

static void highbd_down2_symodd(const uint16_t *const input, int length,
                                uint16_t *output, int bd) {
  // Actual filter len = 2 * filter_len_half - 1.
  static const int16_t *filter = av1_down2_symodd_half_filter;
  const int filter_len_half = sizeof(av1_down2_symodd_half_filter) / 2;
  int i, j;
  uint16_t *optr = output;
  int l1 = filter_len_half - 1;
  int l2 = (length - filter_len_half + 1);
  l1 += (l1 & 1);
  l2 += (l2 & 1);
  if (l1 > l2) {
    // Short input length.
    for (i = 0; i < length; i += 2) {
      int sum = (1 << (FILTER_BITS - 1)) + input[i] * filter[0];
      for (j = 1; j < filter_len_half; ++j) {
        sum += (input[AOMMAX(i - j, 0)] + input[AOMMIN(i + j, length - 1)]) *
               filter[j];
      }
      sum >>= FILTER_BITS;
      *optr++ = clip_pixel_highbd(sum, bd);
    }
  } else {
    // Initial part.
    for (i = 0; i < l1; i += 2) {
      int sum = (1 << (FILTER_BITS - 1)) + input[i] * filter[0];
      for (j = 1; j < filter_len_half; ++j) {
        sum += (input[AOMMAX(i - j, 0)] + input[i + j]) * filter[j];
      }
      sum >>= FILTER_BITS;
      *optr++ = clip_pixel_highbd(sum, bd);
    }
    // Middle part.
    for (; i < l2; i += 2) {
      int sum = (1 << (FILTER_BITS - 1)) + input[i] * filter[0];
      for (j = 1; j < filter_len_half; ++j) {
        sum += (input[i - j] + input[i + j]) * filter[j];
      }
      sum >>= FILTER_BITS;
      *optr++ = clip_pixel_highbd(sum, bd);
    }
    // End part.
    for (; i < length; i += 2) {
      int sum = (1 << (FILTER_BITS - 1)) + input[i] * filter[0];
      for (j = 1; j < filter_len_half; ++j) {
        sum += (input[i - j] + input[AOMMIN(i + j, length - 1)]) * filter[j];
      }
      sum >>= FILTER_BITS;
      *optr++ = clip_pixel_highbd(sum, bd);
    }
  }
}
#endif

static void highbd_resize_multistep(const uint16_t *const input, int length,
                                    uint16_t *output, int olength,
                                    uint16_t *otmp, int bd) {
  if (length == olength) {
    memcpy(output, input, sizeof(output[0]) * length);
    return;
  }
  const int steps = get_down2_steps(length, olength);

  if (steps > 0) {
    uint16_t *out = NULL;
    int filteredlength = length;

    assert(otmp != NULL);
    uint16_t *otmp2 = otmp + get_down2_length(length, 1);
    for (int s = 0; s < steps; ++s) {
      const int proj_filteredlength = get_down2_length(filteredlength, 1);
      const uint16_t *const in = (s == 0 ? input : out);
      if (s == steps - 1 && proj_filteredlength == olength)
        out = output;
      else
        out = (s & 1 ? otmp2 : otmp);
      if (filteredlength & 1)
        highbd_down2_symodd(in, filteredlength, out, bd);
      else
        highbd_down2_symeven(in, filteredlength, out, bd);
      filteredlength = proj_filteredlength;
    }
    if (filteredlength != olength) {
      highbd_interpolate(out, filteredlength, output, olength, bd);
    }
  } else {
    highbd_interpolate(input, length, output, olength, bd);
  }
}

static void highbd_fill_col_to_arr(uint16_t *img, int stride, int len,
                                   uint16_t *arr) {
  int i;
  uint16_t *iptr = img;
  uint16_t *aptr = arr;
  for (i = 0; i < len; ++i, iptr += stride) {
    *aptr++ = *iptr;
  }
}

static void highbd_fill_arr_to_col(uint16_t *img, int stride, int len,
                                   uint16_t *arr) {
  int i;
  uint16_t *iptr = img;
  uint16_t *aptr = arr;
  for (i = 0; i < len; ++i, iptr += stride) {
    *iptr = *aptr++;
  }
}

static void highbd_resize_plane(const uint8_t *const input, int height,
                                int width, int in_stride, uint8_t *output,
                                int height2, int width2, int out_stride,
                                int bd) {
  int i;
  uint16_t *intbuf = (uint16_t *)aom_malloc(sizeof(uint16_t) * width2 * height);
  uint16_t *tmpbuf =
      (uint16_t *)aom_malloc(sizeof(uint16_t) * AOMMAX(width, height));
  uint16_t *arrbuf = (uint16_t *)aom_malloc(sizeof(uint16_t) * height);
  uint16_t *arrbuf2 = (uint16_t *)aom_malloc(sizeof(uint16_t) * height2);
  if (intbuf == NULL || tmpbuf == NULL || arrbuf == NULL || arrbuf2 == NULL)
    goto Error;
  for (i = 0; i < height; ++i) {
    highbd_resize_multistep(CONVERT_TO_SHORTPTR(input + in_stride * i), width,
                            intbuf + width2 * i, width2, tmpbuf, bd);
  }
  for (i = 0; i < width2; ++i) {
    highbd_fill_col_to_arr(intbuf + i, width2, height, arrbuf);
    highbd_resize_multistep(arrbuf, height, arrbuf2, height2, tmpbuf, bd);
    highbd_fill_arr_to_col(CONVERT_TO_SHORTPTR(output + i), out_stride, height2,
                           arrbuf2);
  }

Error:
  aom_free(intbuf);
  aom_free(tmpbuf);
  aom_free(arrbuf);
  aom_free(arrbuf2);
}

static void highbd_upscale_normative_rect(const uint8_t *const input,
                                          int height, int width, int in_stride,
                                          uint8_t *output, int height2,
                                          int width2, int out_stride,
                                          int x_step_qn, int x0_qn,
                                          int pad_left, int pad_right, int bd) {
  assert(width > 0);
  assert(height > 0);
  assert(width2 > 0);
  assert(height2 > 0);
  assert(height2 == height);

  // Extend the left/right pixels of the tile column if needed
  // (either because we can't sample from other tiles, or because we're at
  // a frame edge).
  // Save the overwritten pixels into tmp_left and tmp_right.
  // Note: Because we pass input-1 to av1_convolve_horiz_rs, we need one extra
  // column of border pixels compared to what we'd naively think.
  const int border_cols = UPSCALE_NORMATIVE_TAPS / 2 + 1;
  const int border_size = border_cols * sizeof(uint16_t);
  uint16_t *tmp_left =
      NULL;  // Silence spurious "may be used uninitialized" warnings
  uint16_t *tmp_right = NULL;
  uint16_t *const input16 = CONVERT_TO_SHORTPTR(input);
  uint16_t *const in_tl = input16 - border_cols;
  uint16_t *const in_tr = input16 + width;
  if (pad_left) {
    tmp_left = (uint16_t *)aom_malloc(sizeof(*tmp_left) * border_cols * height);
    for (int i = 0; i < height; i++) {
      memcpy(tmp_left + i * border_cols, in_tl + i * in_stride, border_size);
      aom_memset16(in_tl + i * in_stride, input16[i * in_stride], border_cols);
    }
  }
  if (pad_right) {
    tmp_right =
        (uint16_t *)aom_malloc(sizeof(*tmp_right) * border_cols * height);
    for (int i = 0; i < height; i++) {
      memcpy(tmp_right + i * border_cols, in_tr + i * in_stride, border_size);
      aom_memset16(in_tr + i * in_stride, input16[i * in_stride + width - 1],
                   border_cols);
    }
  }

  av1_highbd_convolve_horiz_rs(CONVERT_TO_SHORTPTR(input - 1), in_stride,
                               CONVERT_TO_SHORTPTR(output), out_stride, width2,
                               height2, &av1_resize_filter_normative[0][0],
                               x0_qn, x_step_qn, bd);

  // Restore the left/right border pixels
  if (pad_left) {
    for (int i = 0; i < height; i++) {
      memcpy(in_tl + i * in_stride, tmp_left + i * border_cols, border_size);
    }
    aom_free(tmp_left);
  }
  if (pad_right) {
    for (int i = 0; i < height; i++) {
      memcpy(in_tr + i * in_stride, tmp_right + i * border_cols, border_size);
    }
    aom_free(tmp_right);
  }
}

void av1_resize_frame420(const uint8_t *const y, int y_stride,
                         const uint8_t *const u, const uint8_t *const v,
                         int uv_stride, int height, int width, uint8_t *oy,
                         int oy_stride, uint8_t *ou, uint8_t *ov,
                         int ouv_stride, int oheight, int owidth) {
  resize_plane(y, height, width, y_stride, oy, oheight, owidth, oy_stride);
  resize_plane(u, height / 2, width / 2, uv_stride, ou, oheight / 2, owidth / 2,
               ouv_stride);
  resize_plane(v, height / 2, width / 2, uv_stride, ov, oheight / 2, owidth / 2,
               ouv_stride);
}

void av1_resize_frame422(const uint8_t *const y, int y_stride,
                         const uint8_t *const u, const uint8_t *const v,
                         int uv_stride, int height, int width, uint8_t *oy,
                         int oy_stride, uint8_t *ou, uint8_t *ov,
                         int ouv_stride, int oheight, int owidth) {
  resize_plane(y, height, width, y_stride, oy, oheight, owidth, oy_stride);
  resize_plane(u, height, width / 2, uv_stride, ou, oheight, owidth / 2,
               ouv_stride);
  resize_plane(v, height, width / 2, uv_stride, ov, oheight, owidth / 2,
               ouv_stride);
}

void av1_resize_frame444(const uint8_t *const y, int y_stride,
                         const uint8_t *const u, const uint8_t *const v,
                         int uv_stride, int height, int width, uint8_t *oy,
                         int oy_stride, uint8_t *ou, uint8_t *ov,
                         int ouv_stride, int oheight, int owidth) {
  resize_plane(y, height, width, y_stride, oy, oheight, owidth, oy_stride);
  resize_plane(u, height, width, uv_stride, ou, oheight, owidth, ouv_stride);
  resize_plane(v, height, width, uv_stride, ov, oheight, owidth, ouv_stride);
}

void av1_highbd_resize_frame420(const uint8_t *const y, int y_stride,
                                const uint8_t *const u, const uint8_t *const v,
                                int uv_stride, int height, int width,
                                uint8_t *oy, int oy_stride, uint8_t *ou,
                                uint8_t *ov, int ouv_stride, int oheight,
                                int owidth, int bd) {
  highbd_resize_plane(y, height, width, y_stride, oy, oheight, owidth,
                      oy_stride, bd);
  highbd_resize_plane(u, height / 2, width / 2, uv_stride, ou, oheight / 2,
                      owidth / 2, ouv_stride, bd);
  highbd_resize_plane(v, height / 2, width / 2, uv_stride, ov, oheight / 2,
                      owidth / 2, ouv_stride, bd);
}

void av1_highbd_resize_frame422(const uint8_t *const y, int y_stride,
                                const uint8_t *const u, const uint8_t *const v,
                                int uv_stride, int height, int width,
                                uint8_t *oy, int oy_stride, uint8_t *ou,
                                uint8_t *ov, int ouv_stride, int oheight,
                                int owidth, int bd) {
  highbd_resize_plane(y, height, width, y_stride, oy, oheight, owidth,
                      oy_stride, bd);
  highbd_resize_plane(u, height, width / 2, uv_stride, ou, oheight, owidth / 2,
                      ouv_stride, bd);
  highbd_resize_plane(v, height, width / 2, uv_stride, ov, oheight, owidth / 2,
                      ouv_stride, bd);
}

void av1_highbd_resize_frame444(const uint8_t *const y, int y_stride,
                                const uint8_t *const u, const uint8_t *const v,
                                int uv_stride, int height, int width,
                                uint8_t *oy, int oy_stride, uint8_t *ou,
                                uint8_t *ov, int ouv_stride, int oheight,
                                int owidth, int bd) {
  highbd_resize_plane(y, height, width, y_stride, oy, oheight, owidth,
                      oy_stride, bd);
  highbd_resize_plane(u, height, width, uv_stride, ou, oheight, owidth,
                      ouv_stride, bd);
  highbd_resize_plane(v, height, width, uv_stride, ov, oheight, owidth,
                      ouv_stride, bd);
}

void av1_resize_and_extend_frame(const YV12_BUFFER_CONFIG *src,
                                 YV12_BUFFER_CONFIG *dst, int bd,
                                 const int num_planes) {
  // TODO(dkovalev): replace YV12_BUFFER_CONFIG with aom_image_t

  // We use AOMMIN(num_planes, MAX_MB_PLANE) instead of num_planes to quiet
  // the static analysis warnings.
  for (int i = 0; i < AOMMIN(num_planes, MAX_MB_PLANE); ++i) {
    const int is_uv = i > 0;
    if (src->flags & YV12_FLAG_HIGHBITDEPTH)
      highbd_resize_plane(src->buffers[i], src->crop_heights[is_uv],
                          src->crop_widths[is_uv], src->strides[is_uv],
                          dst->buffers[i], dst->crop_heights[is_uv],
                          dst->crop_widths[is_uv], dst->strides[is_uv], bd);
    else
      resize_plane(src->buffers[i], src->crop_heights[is_uv],
                   src->crop_widths[is_uv], src->strides[is_uv],
                   dst->buffers[i], dst->crop_heights[is_uv],
                   dst->crop_widths[is_uv], dst->strides[is_uv]);
  }
  aom_extend_frame_borders(dst, num_planes);
}

void av1_upscale_normative_rows(const AV1_COMMON *cm, const uint8_t *src,
                                int src_stride, uint8_t *dst, int dst_stride,
                                int plane, int rows) {
  const int is_uv = (plane > 0);
  const int ss_x = is_uv && cm->subsampling_x;
  const int downscaled_plane_width = ROUND_POWER_OF_TWO(cm->width, ss_x);
  const int upscaled_plane_width =
      ROUND_POWER_OF_TWO(cm->superres_upscaled_width, ss_x);
  const int superres_denom = cm->superres_scale_denominator;

  TileInfo tile_col;
  const int32_t x_step_qn = av1_get_upscale_convolve_step(
      downscaled_plane_width, upscaled_plane_width);
  int32_t x0_qn = get_upscale_convolve_x0(downscaled_plane_width,
                                          upscaled_plane_width, x_step_qn);

  for (int j = 0; j < cm->tile_cols; j++) {
    av1_tile_set_col(&tile_col, cm, j);
    // Determine the limits of this tile column in both the source
    // and destination images.
    // Note: The actual location which we start sampling from is
    // (downscaled_x0 - 1 + (x0_qn/2^14)), and this quantity increases
    // by exactly dst_width * (x_step_qn/2^14) pixels each iteration.
    const int downscaled_x0 = tile_col.mi_col_start << (MI_SIZE_LOG2 - ss_x);
    const int downscaled_x1 = tile_col.mi_col_end << (MI_SIZE_LOG2 - ss_x);
    const int src_width = downscaled_x1 - downscaled_x0;

    const int upscaled_x0 = (downscaled_x0 * superres_denom) / SCALE_NUMERATOR;
    int upscaled_x1;
    if (j == cm->tile_cols - 1) {
      // Note that we can't just use AOMMIN here - due to rounding,
      // (downscaled_x1 * superres_denom) / SCALE_NUMERATOR may be less than
      // upscaled_plane_width.
      upscaled_x1 = upscaled_plane_width;
    } else {
      upscaled_x1 = (downscaled_x1 * superres_denom) / SCALE_NUMERATOR;
    }

    const uint8_t *const src_ptr = src + downscaled_x0;
    uint8_t *const dst_ptr = dst + upscaled_x0;
    const int dst_width = upscaled_x1 - upscaled_x0;

    const int pad_left = (j == 0);
    const int pad_right = (j == cm->tile_cols - 1);

    if (cm->use_highbitdepth)
      highbd_upscale_normative_rect(
          src_ptr, rows, src_width, src_stride, dst_ptr, rows, dst_width,
          dst_stride, x_step_qn, x0_qn, pad_left, pad_right, cm->bit_depth);
    else
      upscale_normative_rect(src_ptr, rows, src_width, src_stride, dst_ptr,
                             rows, dst_width, dst_stride, x_step_qn, x0_qn,
                             pad_left, pad_right);

    // Update the fractional pixel offset to prepare for the next tile column.
    x0_qn += (dst_width * x_step_qn) - (src_width << RS_SCALE_SUBPEL_BITS);
  }
}

void av1_upscale_normative_and_extend_frame(const AV1_COMMON *cm,
                                            const YV12_BUFFER_CONFIG *src,
                                            YV12_BUFFER_CONFIG *dst) {
  const int num_planes = av1_num_planes(cm);
  for (int i = 0; i < num_planes; ++i) {
    const int is_uv = (i > 0);
    av1_upscale_normative_rows(cm, src->buffers[i], src->strides[is_uv],
                               dst->buffers[i], dst->strides[is_uv], i,
                               src->crop_heights[is_uv]);
  }

  aom_extend_frame_borders(dst, num_planes);
}

YV12_BUFFER_CONFIG *av1_scale_if_required(AV1_COMMON *cm,
                                          YV12_BUFFER_CONFIG *unscaled,
                                          YV12_BUFFER_CONFIG *scaled) {
  const int num_planes = av1_num_planes(cm);
  if (cm->width != unscaled->y_crop_width ||
      cm->height != unscaled->y_crop_height) {
    av1_resize_and_extend_frame(unscaled, scaled, (int)cm->bit_depth,
                                num_planes);
    return scaled;
  } else {
    return unscaled;
  }
}

// Calculates the scaled dimension given the original dimension and the scale
// denominator.
static void calculate_scaled_size_helper(int *dim, int denom) {
  if (denom != SCALE_NUMERATOR) {
    // Use this version if we need *dim to be even
    // *width = (*width * SCALE_NUMERATOR + denom) / (2 * denom);
    // *width <<= 1;
    *dim = (*dim * SCALE_NUMERATOR + denom / 2) / (denom);
  }
}

void av1_calculate_scaled_size(int *width, int *height, int resize_denom) {
  calculate_scaled_size_helper(width, resize_denom);
  calculate_scaled_size_helper(height, resize_denom);
}

void av1_calculate_scaled_superres_size(int *width, int *height,
                                        int superres_denom) {
  (void)height;
  calculate_scaled_size_helper(width, superres_denom);
}

void av1_calculate_unscaled_superres_size(int *width, int *height, int denom) {
  if (denom != SCALE_NUMERATOR) {
    // Note: av1_calculate_scaled_superres_size() rounds *up* after division
    // when the resulting dimensions are odd. So here, we round *down*.
    *width = *width * denom / SCALE_NUMERATOR;
    (void)height;
  }
}

// Copy only the config data from 'src' to 'dst'.
static void copy_buffer_config(const YV12_BUFFER_CONFIG *const src,
                               YV12_BUFFER_CONFIG *const dst) {
  dst->bit_depth = src->bit_depth;
  dst->color_primaries = src->color_primaries;
  dst->transfer_characteristics = src->transfer_characteristics;
  dst->matrix_coefficients = src->matrix_coefficients;
  dst->monochrome = src->monochrome;
  dst->chroma_sample_position = src->chroma_sample_position;
  dst->color_range = src->color_range;
}

// TODO(afergs): Look for in-place upscaling
// TODO(afergs): aom_ vs av1_ functions? Which can I use?
// Upscale decoded image.
void av1_superres_upscale(AV1_COMMON *cm, BufferPool *const pool) {
  const int num_planes = av1_num_planes(cm);
  if (!av1_superres_scaled(cm)) return;

  YV12_BUFFER_CONFIG copy_buffer;
  memset(&copy_buffer, 0, sizeof(copy_buffer));

  YV12_BUFFER_CONFIG *const frame_to_show = get_frame_new_buffer(cm);

  const int aligned_width = ALIGN_POWER_OF_TWO(cm->width, 3);
  if (aom_alloc_frame_buffer(&copy_buffer, aligned_width, cm->height,
                             cm->subsampling_x, cm->subsampling_y,
                             cm->use_highbitdepth, AOM_BORDER_IN_PIXELS,
                             cm->byte_alignment))
    aom_internal_error(&cm->error, AOM_CODEC_MEM_ERROR,
                       "Failed to allocate copy buffer for superres upscaling");

  // Copy function assumes the frames are the same size.
  // Note that it does not copy YV12_BUFFER_CONFIG config data.
  aom_yv12_copy_frame(frame_to_show, &copy_buffer, num_planes);

  assert(copy_buffer.y_crop_width == aligned_width);
  assert(copy_buffer.y_crop_height == cm->height);

  // Realloc the current frame buffer at a higher resolution in place.
  if (pool != NULL) {
    // Use callbacks if on the decoder.
    aom_codec_frame_buffer_t *fb =
        &pool->frame_bufs[cm->new_fb_idx].raw_frame_buffer;
    aom_release_frame_buffer_cb_fn_t release_fb_cb = pool->release_fb_cb;
    aom_get_frame_buffer_cb_fn_t cb = pool->get_fb_cb;
    void *cb_priv = pool->cb_priv;

    // Realloc with callback does not release the frame buffer - release first.
    if (release_fb_cb(cb_priv, fb))
      aom_internal_error(
          &cm->error, AOM_CODEC_MEM_ERROR,
          "Failed to free current frame buffer before superres upscaling");

    // aom_realloc_frame_buffer() leaves config data for frame_to_show intact
    if (aom_realloc_frame_buffer(frame_to_show, cm->superres_upscaled_width,
                                 cm->superres_upscaled_height,
                                 cm->subsampling_x, cm->subsampling_y,
                                 cm->use_highbitdepth, AOM_BORDER_IN_PIXELS,
                                 cm->byte_alignment, fb, cb, cb_priv))
      aom_internal_error(
          &cm->error, AOM_CODEC_MEM_ERROR,
          "Failed to allocate current frame buffer for superres upscaling");
  } else {
    // Make a copy of the config data for frame_to_show in copy_buffer
    copy_buffer_config(frame_to_show, &copy_buffer);

    // Don't use callbacks on the encoder.
    // aom_alloc_frame_buffer() clears the config data for frame_to_show
    if (aom_alloc_frame_buffer(frame_to_show, cm->superres_upscaled_width,
                               cm->superres_upscaled_height, cm->subsampling_x,
                               cm->subsampling_y, cm->use_highbitdepth,
                               AOM_BORDER_IN_PIXELS, cm->byte_alignment))
      aom_internal_error(
          &cm->error, AOM_CODEC_MEM_ERROR,
          "Failed to reallocate current frame buffer for superres upscaling");

    // Restore config data back to frame_to_show
    copy_buffer_config(&copy_buffer, frame_to_show);
  }
  // TODO(afergs): verify frame_to_show is correct after realloc
  //               encoder:
  //               decoder:

  assert(frame_to_show->y_crop_width == cm->superres_upscaled_width);
  assert(frame_to_show->y_crop_height == cm->superres_upscaled_height);

  // Scale up and back into frame_to_show.
  assert(frame_to_show->y_crop_width != cm->width);
  av1_upscale_normative_and_extend_frame(cm, &copy_buffer, frame_to_show);

  // Free the copy buffer
  aom_free_frame_buffer(&copy_buffer);
}