vpx/vp9/common/vp9_entropy.c

1709 lines
87 KiB
C
Raw Normal View History

/*
* Copyright (c) 2010 The WebM project authors. All Rights Reserved.
2010-05-18 17:58:33 +02:00
*
* Use of this source code is governed by a BSD-style license
* that can be found in the LICENSE file in the root of the source
* tree. An additional intellectual property rights grant can be found
* in the file PATENTS. All contributing project authors may
* be found in the AUTHORS file in the root of the source tree.
2010-05-18 17:58:33 +02:00
*/
#include <stdio.h>
#include "vp9/common/vp9_entropy.h"
#include "vp9/common/vp9_blockd.h"
#include "vp9/common/vp9_onyxc_int.h"
#include "vp9/common/vp9_entropymode.h"
#include "vpx_mem/vpx_mem.h"
#include "vpx/vpx_integer.h"
#include "vp9/common/vp9_coefupdateprobs.h"
2010-05-18 17:58:33 +02:00
const int vp9_i8x8_block[4] = {0, 2, 8, 10};
DECLARE_ALIGNED(16, const uint8_t, vp9_norm[256]) = {
0, 7, 6, 6, 5, 5, 5, 5, 4, 4, 4, 4, 4, 4, 4, 4,
3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3, 3,
2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2, 2,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0,
0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0
};
// Unified coefficient band structure used by all block sizes
DECLARE_ALIGNED(16, const int, vp9_coef_bands8x8[64]) = {
0, 1, 2, 3, 4, 4, 5, 5,
1, 2, 3, 4, 4, 5, 5, 5,
2, 3, 4, 4, 5, 5, 5, 5,
3, 4, 4, 5, 5, 5, 5, 5,
4, 4, 5, 5, 5, 5, 5, 5,
4, 5, 5, 5, 5, 5, 5, 5,
5, 5, 5, 5, 5, 5, 5, 5,
5, 5, 5, 5, 5, 5, 5, 5
};
DECLARE_ALIGNED(16, const int, vp9_coef_bands4x4[16]) = {
0, 1, 2, 3,
1, 2, 3, 4,
2, 3, 4, 5,
3, 4, 5, 5
};
DECLARE_ALIGNED(16, const uint8_t, vp9_pt_energy_class[MAX_ENTROPY_TOKENS]) = {
0, 1, 2, 3, 3, 4, 4, 5, 5, 5, 5, 5
};
#if CONFIG_SCATTERSCAN
DECLARE_ALIGNED(16, const int, vp9_default_zig_zag1d_4x4[16]) = {
0, 4, 1, 5,
8, 2, 12, 9,
3, 6, 13, 10,
7, 14, 11, 15,
};
DECLARE_ALIGNED(16, const int, vp9_col_scan_4x4[16]) = {
0, 4, 8, 1,
12, 5, 9, 2,
13, 6, 10, 3,
7, 14, 11, 15,
};
DECLARE_ALIGNED(16, const int, vp9_row_scan_4x4[16]) = {
0, 1, 4, 2,
5, 3, 6, 8,
9, 7, 12, 10,
13, 11, 14, 15,
};
DECLARE_ALIGNED(64, const int, vp9_default_zig_zag1d_8x8[64]) = {
0, 8, 1, 16, 9, 2, 17, 24,
10, 3, 18, 25, 32, 11, 4, 26,
33, 19, 40, 12, 34, 27, 5, 41,
20, 48, 13, 35, 42, 28, 21, 6,
49, 56, 36, 43, 29, 7, 14, 50,
57, 44, 22, 37, 15, 51, 58, 30,
45, 23, 52, 59, 38, 31, 60, 53,
46, 39, 61, 54, 47, 62, 55, 63,
};
DECLARE_ALIGNED(16, const int, vp9_col_scan_8x8[64]) = {
0, 8, 16, 1, 24, 9, 32, 17,
2, 40, 25, 10, 33, 18, 48, 3,
26, 41, 11, 56, 19, 34, 4, 49,
27, 42, 12, 35, 20, 57, 50, 28,
5, 43, 13, 36, 58, 51, 21, 44,
6, 29, 59, 37, 14, 52, 22, 7,
45, 60, 30, 15, 38, 53, 23, 46,
31, 61, 39, 54, 47, 62, 55, 63,
};
DECLARE_ALIGNED(16, const int, vp9_row_scan_8x8[64]) = {
0, 1, 2, 8, 9, 3, 16, 10,
4, 17, 11, 24, 5, 18, 25, 12,
19, 26, 32, 6, 13, 20, 33, 27,
7, 34, 40, 21, 28, 41, 14, 35,
48, 42, 29, 36, 49, 22, 43, 15,
56, 37, 50, 44, 30, 57, 23, 51,
58, 45, 38, 52, 31, 59, 53, 46,
60, 39, 61, 47, 54, 55, 62, 63,
};
DECLARE_ALIGNED(16, const int, vp9_default_zig_zag1d_16x16[256]) = {
0, 16, 1, 32, 17, 2, 48, 33, 18, 3, 64, 34, 49, 19, 65, 80,
50, 4, 35, 66, 20, 81, 96, 51, 5, 36, 82, 97, 67, 112, 21, 52,
98, 37, 83, 113, 6, 68, 128, 53, 22, 99, 114, 84, 7, 129, 38, 69,
100, 115, 144, 130, 85, 54, 23, 8, 145, 39, 70, 116, 101, 131, 160, 146,
55, 86, 24, 71, 132, 117, 161, 40, 9, 102, 147, 176, 162, 87, 56, 25,
133, 118, 177, 148, 72, 103, 41, 163, 10, 192, 178, 88, 57, 134, 149, 119,
26, 164, 73, 104, 193, 42, 179, 208, 11, 135, 89, 165, 120, 150, 58, 194,
180, 27, 74, 209, 105, 151, 136, 43, 90, 224, 166, 195, 181, 121, 210, 59,
12, 152, 106, 167, 196, 75, 137, 225, 211, 240, 182, 122, 91, 28, 197, 13,
226, 168, 183, 153, 44, 212, 138, 107, 241, 60, 29, 123, 198, 184, 227, 169,
242, 76, 213, 154, 45, 92, 14, 199, 139, 61, 228, 214, 170, 185, 243, 108,
77, 155, 30, 15, 200, 229, 124, 215, 244, 93, 46, 186, 171, 201, 109, 140,
230, 62, 216, 245, 31, 125, 78, 156, 231, 47, 187, 202, 217, 94, 246, 141,
63, 232, 172, 110, 247, 157, 79, 218, 203, 126, 233, 188, 248, 95, 173, 142,
219, 111, 249, 234, 158, 127, 189, 204, 250, 235, 143, 174, 220, 205, 159, 251,
190, 221, 175, 236, 237, 191, 206, 252, 222, 253, 207, 238, 223, 254, 239, 255,
};
DECLARE_ALIGNED(16, const int, vp9_col_scan_16x16[256]) = {
0, 16, 32, 48, 1, 64, 17, 80, 33, 96, 49, 2, 65, 112, 18, 81,
34, 128, 50, 97, 3, 66, 144, 19, 113, 35, 82, 160, 98, 51, 129, 4,
67, 176, 20, 114, 145, 83, 36, 99, 130, 52, 192, 5, 161, 68, 115, 21,
146, 84, 208, 177, 37, 131, 100, 53, 162, 224, 69, 6, 116, 193, 147, 85,
22, 240, 132, 38, 178, 101, 163, 54, 209, 117, 70, 7, 148, 194, 86, 179,
225, 23, 133, 39, 164, 8, 102, 210, 241, 55, 195, 118, 149, 71, 180, 24,
87, 226, 134, 165, 211, 40, 103, 56, 72, 150, 196, 242, 119, 9, 181, 227,
88, 166, 25, 135, 41, 104, 212, 57, 151, 197, 120, 73, 243, 182, 136, 167,
213, 89, 10, 228, 105, 152, 198, 26, 42, 121, 183, 244, 168, 58, 137, 229,
74, 214, 90, 153, 199, 184, 11, 106, 245, 27, 122, 230, 169, 43, 215, 59,
200, 138, 185, 246, 75, 12, 91, 154, 216, 231, 107, 28, 44, 201, 123, 170,
60, 247, 232, 76, 139, 13, 92, 217, 186, 248, 155, 108, 29, 124, 45, 202,
233, 171, 61, 14, 77, 140, 15, 249, 93, 30, 187, 156, 218, 46, 109, 125,
62, 172, 78, 203, 31, 141, 234, 94, 47, 188, 63, 157, 110, 250, 219, 79,
126, 204, 173, 142, 95, 189, 111, 235, 158, 220, 251, 127, 174, 143, 205, 236,
159, 190, 221, 252, 175, 206, 237, 191, 253, 222, 238, 207, 254, 223, 239, 255,
};
DECLARE_ALIGNED(16, const int, vp9_row_scan_16x16[256]) = {
0, 1, 2, 16, 3, 17, 4, 18, 32, 5, 33, 19, 6, 34, 48, 20,
49, 7, 35, 21, 50, 64, 8, 36, 65, 22, 51, 37, 80, 9, 66, 52,
23, 38, 81, 67, 10, 53, 24, 82, 68, 96, 39, 11, 54, 83, 97, 69,
25, 98, 84, 40, 112, 55, 12, 70, 99, 113, 85, 26, 41, 56, 114, 100,
13, 71, 128, 86, 27, 115, 101, 129, 42, 57, 72, 116, 14, 87, 130, 102,
144, 73, 131, 117, 28, 58, 15, 88, 43, 145, 103, 132, 146, 118, 74, 160,
89, 133, 104, 29, 59, 147, 119, 44, 161, 148, 90, 105, 134, 162, 120, 176,
75, 135, 149, 30, 60, 163, 177, 45, 121, 91, 106, 164, 178, 150, 192, 136,
165, 179, 31, 151, 193, 76, 122, 61, 137, 194, 107, 152, 180, 208, 46, 166,
167, 195, 92, 181, 138, 209, 123, 153, 224, 196, 77, 168, 210, 182, 240, 108,
197, 62, 154, 225, 183, 169, 211, 47, 139, 93, 184, 226, 212, 241, 198, 170,
124, 155, 199, 78, 213, 185, 109, 227, 200, 63, 228, 242, 140, 214, 171, 186,
156, 229, 243, 125, 94, 201, 244, 215, 216, 230, 141, 187, 202, 79, 172, 110,
157, 245, 217, 231, 95, 246, 232, 126, 203, 247, 233, 173, 218, 142, 111, 158,
188, 248, 127, 234, 219, 249, 189, 204, 143, 174, 159, 250, 235, 205, 220, 175,
190, 251, 221, 191, 206, 236, 207, 237, 252, 222, 253, 223, 238, 239, 254, 255,
};
DECLARE_ALIGNED(16, const int, vp9_default_zig_zag1d_32x32[1024]) = {
0, 32, 1, 64, 33, 2, 96, 65, 34, 128, 3, 97, 66, 160, 129, 35, 98, 4, 67, 130, 161, 192, 36, 99, 224, 5, 162, 193, 68, 131, 37, 100,
225, 194, 256, 163, 69, 132, 6, 226, 257, 288, 195, 101, 164, 38, 258, 7, 227, 289, 133, 320, 70, 196, 165, 290, 259, 228, 39, 321, 102, 352, 8, 197,
71, 134, 322, 291, 260, 353, 384, 229, 166, 103, 40, 354, 323, 292, 135, 385, 198, 261, 72, 9, 416, 167, 386, 355, 230, 324, 104, 293, 41, 417, 199, 136,
262, 387, 448, 325, 356, 10, 73, 418, 231, 168, 449, 294, 388, 105, 419, 263, 42, 200, 357, 450, 137, 480, 74, 326, 232, 11, 389, 169, 295, 420, 106, 451,
481, 358, 264, 327, 201, 43, 138, 512, 482, 390, 296, 233, 170, 421, 75, 452, 359, 12, 513, 265, 483, 328, 107, 202, 514, 544, 422, 391, 453, 139, 44, 234,
484, 297, 360, 171, 76, 515, 545, 266, 329, 454, 13, 423, 392, 203, 108, 546, 485, 576, 298, 235, 140, 361, 516, 330, 172, 547, 45, 424, 455, 267, 393, 577,
486, 77, 204, 517, 362, 548, 608, 14, 456, 299, 578, 109, 236, 425, 394, 487, 609, 331, 141, 579, 518, 46, 268, 15, 173, 549, 610, 640, 363, 78, 519, 488,
300, 205, 16, 457, 580, 426, 550, 395, 110, 237, 611, 641, 332, 672, 142, 642, 269, 458, 47, 581, 427, 489, 174, 364, 520, 612, 551, 673, 79, 206, 301, 643,
704, 17, 111, 490, 674, 238, 582, 48, 521, 613, 333, 396, 459, 143, 270, 552, 644, 705, 736, 365, 80, 675, 583, 175, 428, 706, 112, 302, 207, 614, 553, 49,
645, 522, 737, 397, 768, 144, 334, 18, 676, 491, 239, 615, 707, 584, 81, 460, 176, 271, 738, 429, 113, 800, 366, 208, 523, 708, 646, 554, 677, 769, 19, 145,
585, 739, 240, 303, 50, 461, 616, 398, 647, 335, 492, 177, 82, 770, 832, 555, 272, 430, 678, 209, 709, 114, 740, 801, 617, 51, 304, 679, 524, 367, 586, 241,
20, 146, 771, 864, 83, 802, 648, 493, 399, 273, 336, 710, 178, 462, 833, 587, 741, 115, 305, 711, 368, 525, 618, 803, 210, 896, 680, 834, 772, 52, 649, 147,
431, 494, 556, 242, 400, 865, 337, 21, 928, 179, 742, 84, 463, 274, 369, 804, 650, 557, 743, 960, 835, 619, 773, 306, 211, 526, 432, 992, 588, 712, 116, 243,
866, 495, 681, 558, 805, 589, 401, 897, 53, 338, 148, 682, 867, 464, 275, 22, 370, 433, 307, 620, 527, 836, 774, 651, 713, 744, 85, 180, 621, 465, 929, 775,
496, 898, 212, 339, 244, 402, 590, 117, 559, 714, 434, 23, 868, 930, 806, 683, 528, 652, 371, 961, 149, 837, 54, 899, 745, 276, 993, 497, 403, 622, 181, 776,
746, 529, 560, 435, 86, 684, 466, 308, 591, 653, 715, 807, 340, 869, 213, 962, 245, 838, 561, 931, 808, 592, 118, 498, 372, 623, 685, 994, 467, 654, 747, 900,
716, 277, 150, 55, 24, 404, 530, 839, 777, 655, 182, 963, 840, 686, 778, 309, 870, 341, 87, 499, 809, 624, 593, 436, 717, 932, 214, 246, 995, 718, 625, 373,
562, 25, 119, 901, 531, 468, 964, 748, 810, 278, 779, 500, 563, 656, 405, 687, 871, 872, 594, 151, 933, 749, 841, 310, 657, 626, 595, 437, 688, 183, 996, 965,
902, 811, 342, 750, 689, 719, 532, 56, 215, 469, 934, 374, 247, 720, 780, 564, 781, 842, 406, 26, 751, 903, 873, 57, 279, 627, 501, 658, 843, 997, 812, 904,
88, 813, 438, 752, 935, 936, 311, 596, 533, 690, 343, 966, 874, 89, 120, 470, 721, 875, 659, 782, 565, 998, 375, 844, 845, 27, 628, 967, 121, 905, 968, 152,
937, 814, 753, 502, 691, 783, 184, 153, 722, 407, 58, 815, 999, 660, 597, 723, 534, 906, 216, 439, 907, 248, 185, 876, 846, 692, 784, 629, 90, 969, 280, 754,
938, 939, 217, 847, 566, 471, 785, 816, 877, 1000, 249, 878, 661, 503, 312, 970, 755, 122, 817, 281, 344, 786, 598, 724, 28, 59, 29, 154, 535, 630, 376, 1001,
313, 908, 186, 91, 848, 849, 345, 909, 940, 879, 408, 818, 693, 1002, 971, 941, 567, 377, 218, 756, 910, 787, 440, 123, 880, 725, 662, 250, 819, 1003, 282, 972,
850, 599, 472, 409, 155, 441, 942, 757, 788, 694, 911, 881, 314, 631, 973, 504, 187, 1004, 346, 473, 851, 943, 820, 726, 60, 505, 219, 378, 912, 974, 30, 31,
536, 882, 1005, 92, 251, 663, 944, 913, 283, 695, 883, 568, 1006, 975, 410, 442, 945, 789, 852, 537, 1007, 124, 315, 61, 758, 821, 600, 914, 976, 569, 474, 347,
156, 1008, 915, 93, 977, 506, 946, 727, 379, 884, 188, 632, 601, 1009, 790, 853, 978, 947, 220, 411, 125, 633, 664, 759, 252, 443, 916, 538, 157, 822, 62, 570,
979, 284, 1010, 885, 948, 189, 475, 94, 316, 665, 696, 1011, 854, 791, 980, 221, 348, 63, 917, 602, 380, 507, 253, 126, 697, 823, 634, 285, 728, 949, 886, 95,
158, 539, 1012, 317, 412, 444, 760, 571, 190, 981, 729, 918, 127, 666, 349, 381, 476, 855, 761, 1013, 603, 222, 159, 698, 950, 508, 254, 792, 286, 635, 887, 793,
413, 191, 982, 445, 540, 318, 730, 667, 223, 824, 919, 1014, 350, 477, 572, 255, 825, 951, 762, 509, 604, 856, 382, 699, 287, 319, 636, 983, 794, 414, 541, 731,
857, 888, 351, 446, 573, 1015, 668, 889, 478, 826, 383, 763, 605, 920, 510, 637, 415, 700, 921, 858, 447, 952, 542, 795, 479, 953, 732, 890, 669, 574, 511, 984,
827, 985, 922, 1016, 764, 606, 543, 701, 859, 638, 1017, 575, 796, 954, 733, 891, 670, 607, 828, 986, 765, 923, 639, 1018, 702, 860, 955, 671, 892, 734, 797, 703,
987, 829, 1019, 766, 924, 735, 861, 956, 988, 893, 767, 798, 830, 1020, 925, 957, 799, 862, 831, 989, 894, 1021, 863, 926, 895, 958, 990, 1022, 927, 959, 991, 1023,
};
#else // CONFIG_SCATTERSCAN
DECLARE_ALIGNED(16, const int, vp9_default_zig_zag1d_4x4[16]) = {
0, 1, 4, 8,
5, 2, 3, 6,
9, 12, 13, 10,
7, 11, 14, 15,
2010-05-18 17:58:33 +02:00
};
DECLARE_ALIGNED(16, const int, vp9_col_scan_4x4[16]) = {
0, 4, 8, 12,
1, 5, 9, 13,
2, 6, 10, 14,
3, 7, 11, 15
};
DECLARE_ALIGNED(16, const int, vp9_row_scan_4x4[16]) = {
0, 1, 2, 3,
4, 5, 6, 7,
8, 9, 10, 11,
12, 13, 14, 15
};
DECLARE_ALIGNED(64, const int, vp9_default_zig_zag1d_8x8[64]) = {
0, 1, 8, 16, 9, 2, 3, 10, 17, 24, 32, 25, 18, 11, 4, 5,
12, 19, 26, 33, 40, 48, 41, 34, 27, 20, 13, 6, 7, 14, 21, 28,
35, 42, 49, 56, 57, 50, 43, 36, 29, 22, 15, 23, 30, 37, 44, 51,
58, 59, 52, 45, 38, 31, 39, 46, 53, 60, 61, 54, 47, 55, 62, 63,
};
2010-05-18 17:58:33 +02:00
DECLARE_ALIGNED(16, const int, vp9_col_scan_8x8[64]) = {
0, 8, 16, 24, 32, 40, 48, 56,
1, 9, 17, 25, 33, 41, 49, 57,
2, 10, 18, 26, 34, 42, 50, 58,
3, 11, 19, 27, 35, 43, 51, 59,
4, 12, 20, 28, 36, 44, 52, 60,
5, 13, 21, 29, 37, 45, 53, 61,
6, 14, 22, 30, 38, 46, 54, 62,
7, 15, 23, 31, 39, 47, 55, 63,
};
DECLARE_ALIGNED(16, const int, vp9_row_scan_8x8[64]) = {
0, 1, 2, 3, 4, 5, 6, 7,
8, 9, 10, 11, 12, 13, 14, 15,
16, 17, 18, 19, 20, 21, 22, 23,
24, 25, 26, 27, 28, 29, 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,
};
DECLARE_ALIGNED(16, const int, vp9_default_zig_zag1d_16x16[256]) = {
0, 1, 16, 32, 17, 2, 3, 18,
33, 48, 64, 49, 34, 19, 4, 5,
20, 35, 50, 65, 80, 96, 81, 66,
51, 36, 21, 6, 7, 22, 37, 52,
67, 82, 97, 112, 128, 113, 98, 83,
68, 53, 38, 23, 8, 9, 24, 39,
54, 69, 84, 99, 114, 129, 144, 160,
145, 130, 115, 100, 85, 70, 55, 40,
25, 10, 11, 26, 41, 56, 71, 86,
101, 116, 131, 146, 161, 176, 192, 177,
162, 147, 132, 117, 102, 87, 72, 57,
42, 27, 12, 13, 28, 43, 58, 73,
88, 103, 118, 133, 148, 163, 178, 193,
208, 224, 209, 194, 179, 164, 149, 134,
119, 104, 89, 74, 59, 44, 29, 14,
15, 30, 45, 60, 75, 90, 105, 120,
135, 150, 165, 180, 195, 210, 225, 240,
241, 226, 211, 196, 181, 166, 151, 136,
121, 106, 91, 76, 61, 46, 31, 47,
62, 77, 92, 107, 122, 137, 152, 167,
182, 197, 212, 227, 242, 243, 228, 213,
198, 183, 168, 153, 138, 123, 108, 93,
78, 63, 79, 94, 109, 124, 139, 154,
169, 184, 199, 214, 229, 244, 245, 230,
215, 200, 185, 170, 155, 140, 125, 110,
95, 111, 126, 141, 156, 171, 186, 201,
216, 231, 246, 247, 232, 217, 202, 187,
172, 157, 142, 127, 143, 158, 173, 188,
203, 218, 233, 248, 249, 234, 219, 204,
189, 174, 159, 175, 190, 205, 220, 235,
250, 251, 236, 221, 206, 191, 207, 222,
237, 252, 253, 238, 223, 239, 254, 255,
};
DECLARE_ALIGNED(16, const int, vp9_col_scan_16x16[256]) = {
0, 16, 32, 48, 64, 80, 96, 112, 128, 144, 160, 176, 192, 208, 224, 240,
1, 17, 33, 49, 65, 81, 97, 113, 129, 145, 161, 177, 193, 209, 225, 241,
2, 18, 34, 50, 66, 82, 98, 114, 130, 146, 162, 178, 194, 210, 226, 242,
3, 19, 35, 51, 67, 83, 99, 115, 131, 147, 163, 179, 195, 211, 227, 243,
4, 20, 36, 52, 68, 84, 100, 116, 132, 148, 164, 180, 196, 212, 228, 244,
5, 21, 37, 53, 69, 85, 101, 117, 133, 149, 165, 181, 197, 213, 229, 245,
6, 22, 38, 54, 70, 86, 102, 118, 134, 150, 166, 182, 198, 214, 230, 246,
7, 23, 39, 55, 71, 87, 103, 119, 135, 151, 167, 183, 199, 215, 231, 247,
8, 24, 40, 56, 72, 88, 104, 120, 136, 152, 168, 184, 200, 216, 232, 248,
9, 25, 41, 57, 73, 89, 105, 121, 137, 153, 169, 185, 201, 217, 233, 249,
10, 26, 42, 58, 74, 90, 106, 122, 138, 154, 170, 186, 202, 218, 234, 250,
11, 27, 43, 59, 75, 91, 107, 123, 139, 155, 171, 187, 203, 219, 235, 251,
12, 28, 44, 60, 76, 92, 108, 124, 140, 156, 172, 188, 204, 220, 236, 252,
13, 29, 45, 61, 77, 93, 109, 125, 141, 157, 173, 189, 205, 221, 237, 253,
14, 30, 46, 62, 78, 94, 110, 126, 142, 158, 174, 190, 206, 222, 238, 254,
15, 31, 47, 63, 79, 95, 111, 127, 143, 159, 175, 191, 207, 223, 239, 255,
};
DECLARE_ALIGNED(16, const int, vp9_row_scan_16x16[256]) = {
0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15,
16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 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,
};
32x32 transform for superblocks. This adds Debargha's DCT/DWT hybrid and a regular 32x32 DCT, and adds code all over the place to wrap that in the bitstream/encoder/decoder/RD. Some implementation notes (these probably need careful review): - token range is extended by 1 bit, since the value range out of this transform is [-16384,16383]. - the coefficients coming out of the FDCT are manually scaled back by 1 bit, or else they won't fit in int16_t (they are 17 bits). Because of this, the RD error scoring does not right-shift the MSE score by two (unlike for 4x4/8x8/16x16). - to compensate for this loss in precision, the quantizer is halved also. This is currently a little hacky. - FDCT and IDCT is double-only right now. Needs a fixed-point impl. - There are no default probabilities for the 32x32 transform yet; I'm simply using the 16x16 luma ones. A future commit will add newly generated probabilities for all transforms. - No ADST version. I don't think we'll add one for this level; if an ADST is desired, transform-size selection can scale back to 16x16 or lower, and use an ADST at that level. Additional notes specific to Debargha's DWT/DCT hybrid: - coefficient scale is different for the top/left 16x16 (DCT-over-DWT) block than for the rest (DWT pixel differences) of the block. Therefore, RD error scoring isn't easily scalable between coefficient and pixel domain. Thus, unfortunately, we need to compute the RD distortion in the pixel domain until we figure out how to scale these appropriately. Change-Id: I00386f20f35d7fabb19aba94c8162f8aee64ef2b
2012-12-07 23:45:05 +01:00
DECLARE_ALIGNED(16, const int, vp9_default_zig_zag1d_32x32[1024]) = {
0, 1, 32, 64, 33, 2, 3, 34, 65, 96, 128, 97, 66, 35, 4, 5, 36, 67, 98, 129, 160, 192, 161, 130, 99, 68, 37, 6, 7, 38, 69, 100,
131, 162, 193, 224, 256, 225, 194, 163, 132, 101, 70, 39, 8, 9, 40, 71, 102, 133, 164, 195, 226, 257, 288, 320, 289, 258, 227, 196, 165, 134, 103, 72,
41, 10, 11, 42, 73, 104, 135, 166, 197, 228, 259, 290, 321, 352, 384, 353, 322, 291, 260, 229, 198, 167, 136, 105, 74, 43, 12, 13, 44, 75, 106, 137,
168, 199, 230, 261, 292, 323, 354, 385, 416, 448, 417, 386, 355, 324, 293, 262, 231, 200, 169, 138, 107, 76, 45, 14, 15, 46, 77, 108, 139, 170, 201, 232,
263, 294, 325, 356, 387, 418, 449, 480, 512, 481, 450, 419, 388, 357, 326, 295, 264, 233, 202, 171, 140, 109, 78, 47, 16, 17, 48, 79, 110, 141, 172, 203,
234, 265, 296, 327, 358, 389, 420, 451, 482, 513, 544, 576, 545, 514, 483, 452, 421, 390, 359, 328, 297, 266, 235, 204, 173, 142, 111, 80, 49, 18, 19, 50,
81, 112, 143, 174, 205, 236, 267, 298, 329, 360, 391, 422, 453, 484, 515, 546, 577, 608, 640, 609, 578, 547, 516, 485, 454, 423, 392, 361, 330, 299, 268, 237,
206, 175, 144, 113, 82, 51, 20, 21, 52, 83, 114, 145, 176, 207, 238, 269, 300, 331, 362, 393, 424, 455, 486, 517, 548, 579, 610, 641, 672, 704, 673, 642,
611, 580, 549, 518, 487, 456, 425, 394, 363, 332, 301, 270, 239, 208, 177, 146, 115, 84, 53, 22, 23, 54, 85, 116, 147, 178, 209, 240, 271, 302, 333, 364,
395, 426, 457, 488, 519, 550, 581, 612, 643, 674, 705, 736, 768, 737, 706, 675, 644, 613, 582, 551, 520, 489, 458, 427, 396, 365, 334, 303, 272, 241, 210, 179,
148, 117, 86, 55, 24, 25, 56, 87, 118, 149, 180, 211, 242, 273, 304, 335, 366, 397, 428, 459, 490, 521, 552, 583, 614, 645, 676, 707, 738, 769, 800, 832,
801, 770, 739, 708, 677, 646, 615, 584, 553, 522, 491, 460, 429, 398, 367, 336, 305, 274, 243, 212, 181, 150, 119, 88, 57, 26, 27, 58, 89, 120, 151, 182,
213, 244, 275, 306, 337, 368, 399, 430, 461, 492, 523, 554, 585, 616, 647, 678, 709, 740, 771, 802, 833, 864, 896, 865, 834, 803, 772, 741, 710, 679, 648, 617,
586, 555, 524, 493, 462, 431, 400, 369, 338, 307, 276, 245, 214, 183, 152, 121, 90, 59, 28, 29, 60, 91, 122, 153, 184, 215, 246, 277, 308, 339, 370, 401,
432, 463, 494, 525, 556, 587, 618, 649, 680, 711, 742, 773, 804, 835, 866, 897, 928, 960, 929, 898, 867, 836, 805, 774, 743, 712, 681, 650, 619, 588, 557, 526,
495, 464, 433, 402, 371, 340, 309, 278, 247, 216, 185, 154, 123, 92, 61, 30, 31, 62, 93, 124, 155, 186, 217, 248, 279, 310, 341, 372, 403, 434, 465, 496,
527, 558, 589, 620, 651, 682, 713, 744, 775, 806, 837, 868, 899, 930, 961, 992, 993, 962, 931, 900, 869, 838, 807, 776, 745, 714, 683, 652, 621, 590, 559, 528,
497, 466, 435, 404, 373, 342, 311, 280, 249, 218, 187, 156, 125, 94, 63, 95, 126, 157, 188, 219, 250, 281, 312, 343, 374, 405, 436, 467, 498, 529, 560, 591,
622, 653, 684, 715, 746, 777, 808, 839, 870, 901, 932, 963, 994, 995, 964, 933, 902, 871, 840, 809, 778, 747, 716, 685, 654, 623, 592, 561, 530, 499, 468, 437,
406, 375, 344, 313, 282, 251, 220, 189, 158, 127, 159, 190, 221, 252, 283, 314, 345, 376, 407, 438, 469, 500, 531, 562, 593, 624, 655, 686, 717, 748, 779, 810,
841, 872, 903, 934, 965, 996, 997, 966, 935, 904, 873, 842, 811, 780, 749, 718, 687, 656, 625, 594, 563, 532, 501, 470, 439, 408, 377, 346, 315, 284, 253, 222,
191, 223, 254, 285, 316, 347, 378, 409, 440, 471, 502, 533, 564, 595, 626, 657, 688, 719, 750, 781, 812, 843, 874, 905, 936, 967, 998, 999, 968, 937, 906, 875,
844, 813, 782, 751, 720, 689, 658, 627, 596, 565, 534, 503, 472, 441, 410, 379, 348, 317, 286, 255, 287, 318, 349, 380, 411, 442, 473, 504, 535, 566, 597, 628,
659, 690, 721, 752, 783, 814, 845, 876, 907, 938, 969, 1000, 1001, 970, 939, 908, 877, 846, 815, 784, 753, 722, 691, 660, 629, 598, 567, 536, 505, 474, 443, 412,
381, 350, 319, 351, 382, 413, 444, 475, 506, 537, 568, 599, 630, 661, 692, 723, 754, 785, 816, 847, 878, 909, 940, 971, 1002, 1003, 972, 941, 910, 879, 848, 817,
786, 755, 724, 693, 662, 631, 600, 569, 538, 507, 476, 445, 414, 383, 415, 446, 477, 508, 539, 570, 601, 632, 663, 694, 725, 756, 787, 818, 849, 880, 911, 942,
973, 1004, 1005, 974, 943, 912, 881, 850, 819, 788, 757, 726, 695, 664, 633, 602, 571, 540, 509, 478, 447, 479, 510, 541, 572, 603, 634, 665, 696, 727, 758, 789,
820, 851, 882, 913, 944, 975, 1006, 1007, 976, 945, 914, 883, 852, 821, 790, 759, 728, 697, 666, 635, 604, 573, 542, 511, 543, 574, 605, 636, 667, 698, 729, 760,
791, 822, 853, 884, 915, 946, 977, 1008, 1009, 978, 947, 916, 885, 854, 823, 792, 761, 730, 699, 668, 637, 606, 575, 607, 638, 669, 700, 731, 762, 793, 824, 855,
886, 917, 948, 979, 1010, 1011, 980, 949, 918, 887, 856, 825, 794, 763, 732, 701, 670, 639, 671, 702, 733, 764, 795, 826, 857, 888, 919, 950, 981, 1012, 1013, 982,
951, 920, 889, 858, 827, 796, 765, 734, 703, 735, 766, 797, 828, 859, 890, 921, 952, 983, 1014, 1015, 984, 953, 922, 891, 860, 829, 798, 767, 799, 830, 861, 892,
923, 954, 985, 1016, 1017, 986, 955, 924, 893, 862, 831, 863, 894, 925, 956, 987, 1018, 1019, 988, 957, 926, 895, 927, 958, 989, 1020, 1021, 990, 959, 991, 1022, 1023,
};
#endif // CONFIG_SCATTERSCAN
2010-05-18 17:58:33 +02:00
/* Array indices are identical to previously-existing CONTEXT_NODE indices */
const vp9_tree_index vp9_coef_tree[ 22] = /* corresponding _CONTEXT_NODEs */
2010-05-18 17:58:33 +02:00
{
-DCT_EOB_TOKEN, 2, /* 0 = EOB */
-ZERO_TOKEN, 4, /* 1 = ZERO */
-ONE_TOKEN, 6, /* 2 = ONE */
8, 12, /* 3 = LOW_VAL */
-TWO_TOKEN, 10, /* 4 = TWO */
-THREE_TOKEN, -FOUR_TOKEN, /* 5 = THREE */
14, 16, /* 6 = HIGH_LOW */
-DCT_VAL_CATEGORY1, -DCT_VAL_CATEGORY2, /* 7 = CAT_ONE */
18, 20, /* 8 = CAT_THREEFOUR */
-DCT_VAL_CATEGORY3, -DCT_VAL_CATEGORY4, /* 9 = CAT_THREE */
-DCT_VAL_CATEGORY5, -DCT_VAL_CATEGORY6 /* 10 = CAT_FIVE */
2010-05-18 17:58:33 +02:00
};
struct vp9_token vp9_coef_encodings[MAX_ENTROPY_TOKENS];
2010-05-18 17:58:33 +02:00
/* Trees for extra bits. Probabilities are constant and
do not depend on previously encoded bits */
static const vp9_prob Pcat1[] = { 159};
static const vp9_prob Pcat2[] = { 165, 145};
static const vp9_prob Pcat3[] = { 173, 148, 140};
static const vp9_prob Pcat4[] = { 176, 155, 140, 135};
static const vp9_prob Pcat5[] = { 180, 157, 141, 134, 130};
static const vp9_prob Pcat6[] = {
32x32 transform for superblocks. This adds Debargha's DCT/DWT hybrid and a regular 32x32 DCT, and adds code all over the place to wrap that in the bitstream/encoder/decoder/RD. Some implementation notes (these probably need careful review): - token range is extended by 1 bit, since the value range out of this transform is [-16384,16383]. - the coefficients coming out of the FDCT are manually scaled back by 1 bit, or else they won't fit in int16_t (they are 17 bits). Because of this, the RD error scoring does not right-shift the MSE score by two (unlike for 4x4/8x8/16x16). - to compensate for this loss in precision, the quantizer is halved also. This is currently a little hacky. - FDCT and IDCT is double-only right now. Needs a fixed-point impl. - There are no default probabilities for the 32x32 transform yet; I'm simply using the 16x16 luma ones. A future commit will add newly generated probabilities for all transforms. - No ADST version. I don't think we'll add one for this level; if an ADST is desired, transform-size selection can scale back to 16x16 or lower, and use an ADST at that level. Additional notes specific to Debargha's DWT/DCT hybrid: - coefficient scale is different for the top/left 16x16 (DCT-over-DWT) block than for the rest (DWT pixel differences) of the block. Therefore, RD error scoring isn't easily scalable between coefficient and pixel domain. Thus, unfortunately, we need to compute the RD distortion in the pixel domain until we figure out how to scale these appropriately. Change-Id: I00386f20f35d7fabb19aba94c8162f8aee64ef2b
2012-12-07 23:45:05 +01:00
254, 254, 254, 252, 249, 243, 230, 196, 177, 153, 140, 133, 130, 129
};
2010-05-18 17:58:33 +02:00
Modeling default coef probs with distribution Replaces the default tables for single coefficient magnitudes with those obtained from an appropriate distribution. The EOB node is left unchanged. The model is represeted as a 256-size codebook where the index corresponds to the probability of the Zero or the One node. Two variations are implemented corresponding to whether the Zero node or the One-node is used as the peg. The main advantage is that the default prob tables will become considerably smaller and manageable. Besides there is substantially less risk of over-fitting for a training set. Various distributions are tried and the one that gives the best results is the family of Generalized Gaussian distributions with shape parameter 0.75. The results are within about 0.2% of fully trained tables for the Zero peg variant, and within 0.1% of the One peg variant. The forward updates are optionally (controlled by a macro) model-based, i.e. restricted to only convey probabilities from the codebook. Backward updates can also be optionally (controlled by another macro) model-based, but is turned off by default. Currently model-based forward updates work about the same as unconstrained updates, but there is a drop in performance with backward-updates being model based. The model based approach also allows the probabilities for the key frames to be adjusted from the defaults based on the base_qindex of the frame. Currently the adjustment function is a placeholder that adjusts the prob of EOB and Zero node from the nominal one at higher quality (lower qindex) or lower quality (higher qindex) ends of the range. The rest of the probabilities are then derived based on the model from the adjusted prob of zero. Change-Id: Iae050f3cbcc6d8b3f204e8dc395ae47b3b2192c9
2013-03-13 19:03:17 +01:00
#if CONFIG_MODELCOEFPROB
#if UNCONSTRAINED_NODES == 2
Modeling default coef probs with distribution Replaces the default tables for single coefficient magnitudes with those obtained from an appropriate distribution. The EOB node is left unchanged. The model is represeted as a 256-size codebook where the index corresponds to the probability of the Zero or the One node. Two variations are implemented corresponding to whether the Zero node or the One-node is used as the peg. The main advantage is that the default prob tables will become considerably smaller and manageable. Besides there is substantially less risk of over-fitting for a training set. Various distributions are tried and the one that gives the best results is the family of Generalized Gaussian distributions with shape parameter 0.75. The results are within about 0.2% of fully trained tables for the Zero peg variant, and within 0.1% of the One peg variant. The forward updates are optionally (controlled by a macro) model-based, i.e. restricted to only convey probabilities from the codebook. Backward updates can also be optionally (controlled by another macro) model-based, but is turned off by default. Currently model-based forward updates work about the same as unconstrained updates, but there is a drop in performance with backward-updates being model based. The model based approach also allows the probabilities for the key frames to be adjusted from the defaults based on the base_qindex of the frame. Currently the adjustment function is a placeholder that adjusts the prob of EOB and Zero node from the nominal one at higher quality (lower qindex) or lower quality (higher qindex) ends of the range. The rest of the probabilities are then derived based on the model from the adjusted prob of zero. Change-Id: Iae050f3cbcc6d8b3f204e8dc395ae47b3b2192c9
2013-03-13 19:03:17 +01:00
const vp9_prob vp9_modelcoefprobs_gg75[COEFPROB_MODELS][ENTROPY_NODES - 1] = {
// Probs generated with a Generalized Gaussian (with shape parameter 0.75)
// source model with varying quantizer step size for a uniform quantizer
{0, 0, 0, 0, 0, 0, 0, 0, 0, 0,}, // do not use
{1, 2, 6, 87, 129, 11, 88, 39, 93, 47,},
{2, 4, 11, 88, 130, 21, 89, 68, 98, 79,},
{3, 6, 16, 89, 131, 30, 91, 92, 103, 105,},
{4, 8, 21, 90, 131, 38, 92, 112, 107, 126,},
{5, 10, 26, 90, 132, 46, 94, 129, 111, 143,},
{6, 11, 31, 91, 133, 54, 95, 143, 115, 157,},
{7, 13, 35, 92, 133, 61, 96, 156, 119, 170,},
{8, 15, 40, 93, 134, 68, 97, 167, 123, 180,},
{9, 17, 44, 94, 134, 74, 98, 177, 126, 189,},
{10, 19, 48, 94, 135, 80, 100, 185, 130, 197,},
{11, 20, 52, 95, 135, 86, 101, 192, 133, 204,},
{12, 22, 56, 96, 136, 92, 102, 199, 137, 210,},
{13, 24, 60, 96, 136, 97, 103, 205, 140, 215,},
{14, 26, 64, 97, 137, 103, 104, 210, 143, 219,},
{15, 27, 68, 98, 137, 108, 105, 215, 146, 223,},
{16, 29, 71, 98, 138, 112, 106, 219, 149, 227,},
{17, 31, 75, 99, 138, 117, 107, 223, 152, 230,},
{18, 32, 78, 100, 139, 121, 108, 226, 155, 233,},
{19, 34, 82, 100, 139, 126, 109, 229, 158, 235,},
{20, 36, 85, 101, 140, 130, 110, 231, 161, 238,},
{21, 37, 88, 102, 140, 134, 111, 234, 164, 239,},
{22, 39, 91, 102, 141, 138, 112, 236, 167, 241,},
{23, 40, 94, 103, 141, 141, 113, 238, 169, 243,},
{24, 42, 97, 104, 142, 145, 114, 240, 172, 244,},
{25, 44, 100, 104, 142, 149, 115, 241, 174, 245,},
{26, 45, 103, 105, 143, 152, 116, 243, 177, 246,},
{27, 47, 106, 105, 143, 155, 117, 244, 179, 247,},
{28, 48, 109, 106, 143, 158, 118, 245, 182, 248,},
{29, 50, 112, 107, 144, 161, 119, 246, 184, 249,},
{30, 52, 115, 107, 144, 164, 120, 247, 186, 250,},
{31, 53, 117, 108, 145, 167, 121, 248, 188, 250,},
{32, 55, 120, 109, 145, 170, 122, 249, 190, 251,},
{33, 56, 122, 109, 146, 173, 123, 249, 192, 252,},
{34, 58, 125, 110, 146, 175, 124, 250, 194, 252,},
{35, 59, 127, 110, 147, 178, 125, 251, 196, 252,},
{36, 61, 130, 111, 147, 180, 126, 251, 198, 253,},
{37, 62, 132, 112, 147, 183, 127, 251, 200, 253,},
{38, 64, 135, 112, 148, 185, 128, 252, 202, 253,},
{39, 65, 137, 113, 148, 187, 129, 252, 204, 254,},
{40, 67, 139, 114, 149, 189, 130, 253, 205, 254,},
{41, 68, 141, 114, 149, 191, 131, 253, 207, 254,},
{42, 70, 144, 115, 150, 193, 132, 253, 209, 254,},
{43, 71, 146, 115, 150, 195, 133, 254, 210, 254,},
{44, 72, 148, 116, 151, 197, 134, 254, 212, 255,},
{45, 74, 150, 117, 151, 199, 135, 254, 213, 255,},
{46, 75, 152, 117, 151, 201, 136, 254, 215, 255,},
{47, 77, 154, 118, 152, 202, 137, 254, 216, 255,},
{48, 78, 156, 119, 152, 204, 138, 254, 217, 255,},
{49, 80, 158, 119, 153, 206, 139, 255, 219, 255,},
{50, 81, 160, 120, 153, 207, 140, 255, 220, 255,},
{51, 82, 162, 120, 154, 209, 141, 255, 221, 255,},
{52, 84, 164, 121, 154, 210, 142, 255, 222, 255,},
{53, 85, 165, 122, 155, 212, 143, 255, 224, 255,},
{54, 87, 167, 122, 155, 213, 144, 255, 225, 255,},
{55, 88, 169, 123, 155, 215, 145, 255, 226, 255,},
{56, 89, 171, 124, 156, 216, 146, 255, 227, 255,},
{57, 91, 172, 124, 156, 217, 146, 255, 228, 255,},
{58, 92, 174, 125, 157, 218, 147, 255, 229, 255,},
{59, 93, 176, 126, 157, 220, 148, 255, 230, 255,},
{60, 95, 177, 126, 158, 221, 149, 255, 231, 255,},
{61, 96, 179, 127, 158, 222, 150, 255, 232, 255,},
{62, 97, 180, 127, 159, 223, 151, 255, 232, 255,},
{63, 99, 182, 128, 159, 224, 152, 255, 233, 255,},
{64, 100, 183, 129, 159, 225, 153, 255, 234, 255,},
{65, 101, 185, 129, 160, 226, 154, 255, 235, 255,},
{66, 103, 186, 130, 160, 227, 155, 255, 236, 255,},
{67, 104, 188, 131, 161, 228, 156, 255, 236, 255,},
{68, 105, 189, 131, 161, 229, 157, 255, 237, 255,},
{69, 106, 190, 132, 162, 230, 158, 255, 238, 255,},
{70, 108, 192, 133, 162, 231, 159, 255, 238, 255,},
{71, 109, 193, 133, 162, 231, 159, 255, 239, 255,},
{72, 110, 194, 134, 163, 232, 160, 255, 240, 255,},
{73, 111, 196, 134, 163, 233, 161, 255, 240, 255,},
{74, 113, 197, 135, 164, 234, 162, 255, 241, 255,},
{75, 114, 198, 136, 164, 235, 163, 255, 241, 255,},
{76, 115, 199, 136, 165, 235, 164, 255, 242, 255,},
{77, 116, 200, 137, 165, 236, 165, 255, 243, 255,},
{78, 118, 202, 138, 166, 237, 166, 255, 243, 255,},
{79, 119, 203, 138, 166, 237, 167, 255, 244, 255,},
{80, 120, 204, 139, 167, 238, 168, 255, 244, 255,},
{81, 121, 205, 140, 167, 239, 168, 255, 244, 255,},
{82, 123, 206, 140, 167, 239, 169, 255, 245, 255,},
{83, 124, 207, 141, 168, 240, 170, 255, 245, 255,},
{84, 125, 208, 142, 168, 240, 171, 255, 246, 255,},
{85, 126, 209, 142, 169, 241, 172, 255, 246, 255,},
{86, 127, 210, 143, 169, 241, 173, 255, 247, 255,},
{87, 129, 211, 144, 170, 242, 174, 255, 247, 255,},
{88, 130, 212, 144, 170, 242, 175, 255, 247, 255,},
{89, 131, 213, 145, 171, 243, 175, 255, 248, 255,},
{90, 132, 214, 146, 171, 243, 176, 255, 248, 255,},
{91, 133, 215, 146, 171, 244, 177, 255, 248, 255,},
{92, 134, 216, 147, 172, 244, 178, 255, 249, 255,},
{93, 136, 217, 148, 172, 245, 179, 255, 249, 255,},
{94, 137, 218, 148, 173, 245, 180, 255, 249, 255,},
{95, 138, 219, 149, 173, 245, 181, 255, 249, 255,},
{96, 139, 220, 150, 174, 246, 181, 255, 250, 255,},
{97, 140, 220, 150, 174, 246, 182, 255, 250, 255,},
{98, 141, 221, 151, 175, 247, 183, 255, 250, 255,},
{99, 142, 222, 152, 175, 247, 184, 255, 250, 255,},
{100, 144, 223, 152, 176, 247, 185, 255, 251, 255,},
{101, 145, 224, 153, 176, 248, 186, 255, 251, 255,},
{102, 146, 224, 154, 177, 248, 186, 255, 251, 255,},
{103, 147, 225, 154, 177, 248, 187, 255, 251, 255,},
{104, 148, 226, 155, 177, 248, 188, 255, 252, 255,},
{105, 149, 226, 156, 178, 249, 189, 255, 252, 255,},
{106, 150, 227, 156, 178, 249, 190, 255, 252, 255,},
{107, 151, 228, 157, 179, 249, 191, 255, 252, 255,},
{108, 152, 229, 158, 179, 250, 191, 255, 252, 255,},
{109, 153, 229, 158, 180, 250, 192, 255, 252, 255,},
{110, 154, 230, 159, 180, 250, 193, 255, 253, 255,},
{111, 155, 231, 160, 181, 250, 194, 255, 253, 255,},
{112, 157, 231, 160, 181, 251, 195, 255, 253, 255,},
{113, 158, 232, 161, 182, 251, 195, 255, 253, 255,},
{114, 159, 232, 162, 182, 251, 196, 255, 253, 255,},
{115, 160, 233, 162, 183, 251, 197, 255, 253, 255,},
{116, 161, 234, 163, 183, 251, 198, 255, 253, 255,},
{117, 162, 234, 164, 184, 252, 198, 255, 254, 255,},
{118, 163, 235, 165, 184, 252, 199, 255, 254, 255,},
{119, 164, 235, 165, 185, 252, 200, 255, 254, 255,},
{120, 165, 236, 166, 185, 252, 201, 255, 254, 255,},
{121, 166, 236, 167, 186, 252, 201, 255, 254, 255,},
{122, 167, 237, 167, 186, 252, 202, 255, 254, 255,},
{123, 168, 237, 168, 186, 253, 203, 255, 254, 255,},
{124, 169, 238, 169, 187, 253, 204, 255, 254, 255,},
{125, 170, 238, 169, 187, 253, 204, 255, 254, 255,},
{126, 171, 239, 170, 188, 253, 205, 255, 254, 255,},
{127, 172, 239, 171, 188, 253, 206, 255, 254, 255,},
{128, 173, 240, 171, 189, 253, 207, 255, 255, 255,},
{129, 174, 240, 172, 189, 253, 207, 255, 255, 255,},
{130, 175, 241, 173, 190, 253, 208, 255, 255, 255,},
{131, 176, 241, 174, 190, 254, 209, 255, 255, 255,},
{132, 177, 241, 174, 191, 254, 209, 255, 255, 255,},
{133, 178, 242, 175, 191, 254, 210, 255, 255, 255,},
{134, 179, 242, 176, 192, 254, 211, 255, 255, 255,},
{135, 180, 243, 176, 192, 254, 212, 255, 255, 255,},
{136, 180, 243, 177, 193, 254, 212, 255, 255, 255,},
{137, 181, 243, 178, 193, 254, 213, 255, 255, 255,},
{138, 182, 244, 179, 194, 254, 214, 255, 255, 255,},
{139, 183, 244, 179, 194, 254, 214, 255, 255, 255,},
{140, 184, 244, 180, 195, 254, 215, 255, 255, 255,},
{141, 185, 245, 181, 195, 254, 216, 255, 255, 255,},
{142, 186, 245, 181, 196, 255, 216, 255, 255, 255,},
{143, 187, 245, 182, 196, 255, 217, 255, 255, 255,},
{144, 188, 246, 183, 197, 255, 218, 255, 255, 255,},
{145, 189, 246, 183, 197, 255, 218, 255, 255, 255,},
{146, 190, 246, 184, 198, 255, 219, 255, 255, 255,},
{147, 191, 247, 185, 198, 255, 220, 255, 255, 255,},
{148, 191, 247, 186, 199, 255, 220, 255, 255, 255,},
{149, 192, 247, 186, 199, 255, 221, 255, 255, 255,},
{150, 193, 248, 187, 200, 255, 221, 255, 255, 255,},
{151, 194, 248, 188, 200, 255, 222, 255, 255, 255,},
{152, 195, 248, 188, 201, 255, 223, 255, 255, 255,},
{153, 196, 248, 189, 201, 255, 223, 255, 255, 255,},
{154, 197, 249, 190, 202, 255, 224, 255, 255, 255,},
{155, 198, 249, 191, 202, 255, 224, 255, 255, 255,},
{156, 198, 249, 191, 203, 255, 225, 255, 255, 255,},
{157, 199, 249, 192, 203, 255, 226, 255, 255, 255,},
{158, 200, 250, 193, 204, 255, 226, 255, 255, 255,},
{159, 201, 250, 193, 204, 255, 227, 255, 255, 255,},
{160, 202, 250, 194, 205, 255, 227, 255, 255, 255,},
{161, 203, 250, 195, 206, 255, 228, 255, 255, 255,},
{162, 203, 250, 196, 206, 255, 228, 255, 255, 255,},
{163, 204, 251, 196, 207, 255, 229, 255, 255, 255,},
{164, 205, 251, 197, 207, 255, 229, 255, 255, 255,},
{165, 206, 251, 198, 208, 255, 230, 255, 255, 255,},
{166, 207, 251, 198, 208, 255, 231, 255, 255, 255,},
{167, 207, 251, 199, 209, 255, 231, 255, 255, 255,},
{168, 208, 252, 200, 209, 255, 232, 255, 255, 255,},
{169, 209, 252, 201, 210, 255, 232, 255, 255, 255,},
{170, 210, 252, 201, 210, 255, 233, 255, 255, 255,},
{171, 211, 252, 202, 211, 255, 233, 255, 255, 255,},
{172, 211, 252, 203, 211, 255, 234, 255, 255, 255,},
{173, 212, 252, 203, 212, 255, 234, 255, 255, 255,},
{174, 213, 252, 204, 212, 255, 235, 255, 255, 255,},
{175, 214, 253, 205, 213, 255, 235, 255, 255, 255,},
{176, 214, 253, 206, 213, 255, 236, 255, 255, 255,},
{177, 215, 253, 206, 214, 255, 236, 255, 255, 255,},
{178, 216, 253, 207, 214, 255, 237, 255, 255, 255,},
{179, 217, 253, 208, 215, 255, 237, 255, 255, 255,},
{180, 217, 253, 208, 216, 255, 237, 255, 255, 255,},
{181, 218, 253, 209, 216, 255, 238, 255, 255, 255,},
{182, 219, 254, 210, 217, 255, 238, 255, 255, 255,},
{183, 220, 254, 211, 217, 255, 239, 255, 255, 255,},
{184, 220, 254, 211, 218, 255, 239, 255, 255, 255,},
{185, 221, 254, 212, 218, 255, 240, 255, 255, 255,},
{186, 222, 254, 213, 219, 255, 240, 255, 255, 255,},
{187, 222, 254, 213, 219, 255, 241, 255, 255, 255,},
{188, 223, 254, 214, 220, 255, 241, 255, 255, 255,},
{189, 224, 254, 215, 220, 255, 241, 255, 255, 255,},
{190, 225, 254, 215, 221, 255, 242, 255, 255, 255,},
{191, 225, 254, 216, 221, 255, 242, 255, 255, 255,},
{192, 226, 254, 217, 222, 255, 243, 255, 255, 255,},
{193, 227, 255, 218, 223, 255, 243, 255, 255, 255,},
{194, 227, 255, 218, 223, 255, 243, 255, 255, 255,},
{195, 228, 255, 219, 224, 255, 244, 255, 255, 255,},
{196, 229, 255, 220, 224, 255, 244, 255, 255, 255,},
{197, 229, 255, 220, 225, 255, 244, 255, 255, 255,},
{198, 230, 255, 221, 225, 255, 245, 255, 255, 255,},
{199, 230, 255, 222, 226, 255, 245, 255, 255, 255,},
{200, 231, 255, 222, 226, 255, 246, 255, 255, 255,},
{201, 232, 255, 223, 227, 255, 246, 255, 255, 255,},
{202, 232, 255, 224, 228, 255, 246, 255, 255, 255,},
{203, 233, 255, 224, 228, 255, 247, 255, 255, 255,},
{204, 234, 255, 225, 229, 255, 247, 255, 255, 255,},
{205, 234, 255, 226, 229, 255, 247, 255, 255, 255,},
{206, 235, 255, 227, 230, 255, 248, 255, 255, 255,},
{207, 235, 255, 227, 230, 255, 248, 255, 255, 255,},
{208, 236, 255, 228, 231, 255, 248, 255, 255, 255,},
{209, 237, 255, 229, 231, 255, 248, 255, 255, 255,},
{210, 237, 255, 229, 232, 255, 249, 255, 255, 255,},
{211, 238, 255, 230, 233, 255, 249, 255, 255, 255,},
{212, 238, 255, 231, 233, 255, 249, 255, 255, 255,},
{213, 239, 255, 231, 234, 255, 250, 255, 255, 255,},
{214, 239, 255, 232, 234, 255, 250, 255, 255, 255,},
{215, 240, 255, 233, 235, 255, 250, 255, 255, 255,},
{216, 241, 255, 233, 235, 255, 250, 255, 255, 255,},
{217, 241, 255, 234, 236, 255, 251, 255, 255, 255,},
{218, 242, 255, 235, 236, 255, 251, 255, 255, 255,},
{219, 242, 255, 235, 237, 255, 251, 255, 255, 255,},
{220, 243, 255, 236, 237, 255, 251, 255, 255, 255,},
{221, 243, 255, 236, 238, 255, 252, 255, 255, 255,},
{222, 244, 255, 237, 239, 255, 252, 255, 255, 255,},
{223, 244, 255, 238, 239, 255, 252, 255, 255, 255,},
{224, 245, 255, 238, 240, 255, 252, 255, 255, 255,},
{225, 245, 255, 239, 240, 255, 252, 255, 255, 255,},
{226, 246, 255, 240, 241, 255, 253, 255, 255, 255,},
{227, 246, 255, 240, 241, 255, 253, 255, 255, 255,},
{228, 247, 255, 241, 242, 255, 253, 255, 255, 255,},
{229, 247, 255, 242, 242, 255, 253, 255, 255, 255,},
{230, 248, 255, 242, 243, 255, 253, 255, 255, 255,},
{231, 248, 255, 243, 244, 255, 254, 255, 255, 255,},
{232, 248, 255, 243, 244, 255, 254, 255, 255, 255,},
{233, 249, 255, 244, 245, 255, 254, 255, 255, 255,},
{234, 249, 255, 245, 245, 255, 254, 255, 255, 255,},
{235, 250, 255, 245, 246, 255, 254, 255, 255, 255,},
{236, 250, 255, 246, 246, 255, 254, 255, 255, 255,},
{237, 251, 255, 246, 247, 255, 255, 255, 255, 255,},
{238, 251, 255, 247, 247, 255, 255, 255, 255, 255,},
{239, 251, 255, 248, 248, 255, 255, 255, 255, 255,},
{240, 252, 255, 248, 248, 255, 255, 255, 255, 255,},
{241, 252, 255, 249, 249, 255, 255, 255, 255, 255,},
{242, 252, 255, 249, 249, 255, 255, 255, 255, 255,},
{243, 253, 255, 250, 250, 255, 255, 255, 255, 255,},
{244, 253, 255, 250, 250, 255, 255, 255, 255, 255,},
{245, 253, 255, 251, 251, 255, 255, 255, 255, 255,},
{246, 254, 255, 251, 251, 255, 255, 255, 255, 255,},
{247, 254, 255, 252, 252, 255, 255, 255, 255, 255,},
{248, 254, 255, 252, 252, 255, 255, 255, 255, 255,},
{249, 255, 255, 253, 253, 255, 255, 255, 255, 255,},
{250, 255, 255, 253, 253, 255, 255, 255, 255, 255,},
{251, 255, 255, 254, 254, 255, 255, 255, 255, 255,},
{252, 255, 255, 254, 254, 255, 255, 255, 255, 255,},
{253, 255, 255, 255, 255, 255, 255, 255, 255, 255,},
{254, 255, 255, 255, 255, 255, 255, 255, 255, 255,},
{255, 255, 255, 255, 255, 255, 255, 255, 255, 255,}
};
const vp9_prob vp9_modelcoefprobs_gg625[COEFPROB_MODELS][ENTROPY_NODES - 1] = {
// Probs generated with a Generalized Gaussian (with shape parameter 0.625)
// source model with varying quantizer step size for a uniform quantizer
{0, 0, 0, 0, 0, 0, 0, 0, 0, 0,}, // do not use
{1, 2, 6, 88, 130, 10, 88, 35, 94, 40,},
{2, 4, 11, 89, 131, 19, 90, 60, 99, 67,},
{3, 6, 15, 90, 132, 27, 92, 80, 103, 88,},
{4, 7, 20, 91, 132, 34, 93, 97, 107, 105,},
{5, 9, 24, 92, 133, 41, 94, 112, 110, 120,},
{6, 11, 28, 93, 134, 48, 95, 125, 113, 132,},
{7, 13, 33, 93, 134, 54, 97, 136, 116, 143,},
{8, 14, 36, 94, 135, 60, 98, 146, 119, 152,},
{9, 16, 40, 95, 135, 65, 99, 155, 122, 161,},
{10, 18, 44, 95, 136, 70, 100, 163, 125, 168,},
{11, 19, 48, 96, 136, 75, 101, 170, 127, 175,},
{12, 21, 51, 97, 137, 80, 102, 176, 130, 181,},
{13, 23, 55, 97, 137, 85, 102, 182, 132, 187,},
{14, 24, 58, 98, 138, 89, 103, 188, 135, 192,},
{15, 26, 61, 99, 138, 94, 104, 193, 137, 196,},
{16, 27, 64, 99, 139, 98, 105, 197, 140, 201,},
{17, 29, 67, 100, 139, 102, 106, 201, 142, 205,},
{18, 30, 70, 101, 140, 106, 107, 205, 144, 208,},
{19, 32, 73, 101, 140, 109, 108, 209, 146, 211,},
{20, 34, 76, 102, 140, 113, 109, 212, 148, 214,},
{21, 35, 79, 102, 141, 116, 109, 215, 151, 217,},
{22, 37, 82, 103, 141, 120, 110, 218, 153, 220,},
{23, 38, 85, 103, 142, 123, 111, 220, 155, 222,},
{24, 40, 87, 104, 142, 126, 112, 223, 157, 224,},
{25, 41, 90, 105, 143, 129, 113, 225, 159, 226,},
{26, 42, 93, 105, 143, 132, 113, 227, 161, 228,},
{27, 44, 95, 106, 143, 135, 114, 229, 162, 230,},
{28, 45, 98, 106, 144, 138, 115, 230, 164, 232,},
{29, 47, 100, 107, 144, 141, 116, 232, 166, 233,},
{30, 48, 103, 107, 145, 144, 117, 234, 168, 235,},
{31, 50, 105, 108, 145, 146, 117, 235, 170, 236,},
{32, 51, 107, 108, 145, 149, 118, 236, 171, 237,},
{33, 52, 110, 109, 146, 151, 119, 238, 173, 238,},
{34, 54, 112, 110, 146, 154, 120, 239, 175, 239,},
{35, 55, 114, 110, 147, 156, 120, 240, 176, 240,},
{36, 57, 116, 111, 147, 158, 121, 241, 178, 241,},
{37, 58, 119, 111, 147, 161, 122, 242, 180, 242,},
{38, 59, 121, 112, 148, 163, 123, 243, 181, 243,},
{39, 61, 123, 112, 148, 165, 123, 244, 183, 244,},
{40, 62, 125, 113, 148, 167, 124, 244, 184, 245,},
{41, 63, 127, 113, 149, 169, 125, 245, 186, 245,},
{42, 65, 129, 114, 149, 171, 126, 246, 187, 246,},
{43, 66, 131, 114, 150, 173, 126, 246, 188, 247,},
{44, 67, 133, 115, 150, 175, 127, 247, 190, 247,},
{45, 69, 135, 115, 150, 177, 128, 247, 191, 248,},
{46, 70, 136, 116, 151, 178, 129, 248, 193, 248,},
{47, 71, 138, 116, 151, 180, 129, 248, 194, 249,},
{48, 73, 140, 117, 151, 182, 130, 249, 195, 249,},
{49, 74, 142, 118, 152, 184, 131, 249, 197, 250,},
{50, 75, 144, 118, 152, 185, 131, 250, 198, 250,},
{51, 76, 145, 119, 153, 187, 132, 250, 199, 250,},
{52, 78, 147, 119, 153, 188, 133, 251, 200, 251,},
{53, 79, 149, 120, 153, 190, 134, 251, 201, 251,},
{54, 80, 151, 120, 154, 192, 134, 251, 203, 251,},
{55, 82, 152, 121, 154, 193, 135, 251, 204, 252,},
{56, 83, 154, 121, 154, 194, 136, 252, 205, 252,},
{57, 84, 155, 122, 155, 196, 136, 252, 206, 252,},
{58, 85, 157, 122, 155, 197, 137, 252, 207, 252,},
{59, 86, 158, 123, 156, 199, 138, 252, 208, 252,},
{60, 88, 160, 123, 156, 200, 139, 253, 209, 253,},
{61, 89, 162, 124, 156, 201, 139, 253, 210, 253,},
{62, 90, 163, 124, 157, 202, 140, 253, 211, 253,},
{63, 91, 164, 125, 157, 204, 141, 253, 212, 253,},
{64, 93, 166, 125, 157, 205, 141, 253, 213, 253,},
{65, 94, 167, 126, 158, 206, 142, 254, 214, 254,},
{66, 95, 169, 126, 158, 207, 143, 254, 215, 254,},
{67, 96, 170, 127, 158, 208, 143, 254, 216, 254,},
{68, 97, 172, 127, 159, 209, 144, 254, 217, 254,},
{69, 98, 173, 128, 159, 210, 145, 254, 218, 254,},
{70, 100, 174, 128, 160, 212, 146, 254, 219, 254,},
{71, 101, 176, 129, 160, 213, 146, 254, 220, 254,},
{72, 102, 177, 130, 160, 214, 147, 254, 220, 254,},
{73, 103, 178, 130, 161, 215, 148, 255, 221, 255,},
{74, 104, 179, 131, 161, 216, 148, 255, 222, 255,},
{75, 105, 181, 131, 161, 217, 149, 255, 223, 255,},
{76, 107, 182, 132, 162, 217, 150, 255, 224, 255,},
{77, 108, 183, 132, 162, 218, 150, 255, 224, 255,},
{78, 109, 184, 133, 163, 219, 151, 255, 225, 255,},
{79, 110, 185, 133, 163, 220, 152, 255, 226, 255,},
{80, 111, 187, 134, 163, 221, 153, 255, 227, 255,},
{81, 112, 188, 134, 164, 222, 153, 255, 227, 255,},
{82, 113, 189, 135, 164, 223, 154, 255, 228, 255,},
{83, 115, 190, 135, 164, 223, 155, 255, 229, 255,},
{84, 116, 191, 136, 165, 224, 155, 255, 229, 255,},
{85, 117, 192, 136, 165, 225, 156, 255, 230, 255,},
{86, 118, 193, 137, 165, 226, 157, 255, 231, 255,},
{87, 119, 194, 137, 166, 226, 157, 255, 231, 255,},
{88, 120, 195, 138, 166, 227, 158, 255, 232, 255,},
{89, 121, 196, 139, 167, 228, 159, 255, 232, 255,},
{90, 122, 197, 139, 167, 229, 159, 255, 233, 255,},
{91, 123, 198, 140, 167, 229, 160, 255, 234, 255,},
{92, 124, 199, 140, 168, 230, 161, 255, 234, 255,},
{93, 125, 200, 141, 168, 231, 162, 255, 235, 255,},
{94, 127, 201, 141, 168, 231, 162, 255, 235, 255,},
{95, 128, 202, 142, 169, 232, 163, 255, 236, 255,},
{96, 129, 203, 142, 169, 232, 164, 255, 236, 255,},
{97, 130, 204, 143, 170, 233, 164, 255, 237, 255,},
{98, 131, 205, 143, 170, 234, 165, 255, 237, 255,},
{99, 132, 206, 144, 170, 234, 166, 255, 238, 255,},
{100, 133, 207, 144, 171, 235, 166, 255, 238, 255,},
{101, 134, 208, 145, 171, 235, 167, 255, 239, 255,},
{102, 135, 209, 146, 171, 236, 168, 255, 239, 255,},
{103, 136, 209, 146, 172, 236, 168, 255, 240, 255,},
{104, 137, 210, 147, 172, 237, 169, 255, 240, 255,},
{105, 138, 211, 147, 173, 237, 170, 255, 240, 255,},
{106, 139, 212, 148, 173, 238, 170, 255, 241, 255,},
{107, 140, 213, 148, 173, 238, 171, 255, 241, 255,},
{108, 141, 213, 149, 174, 239, 172, 255, 242, 255,},
{109, 142, 214, 149, 174, 239, 172, 255, 242, 255,},
{110, 143, 215, 150, 175, 240, 173, 255, 242, 255,},
{111, 144, 216, 151, 175, 240, 174, 255, 243, 255,},
{112, 145, 217, 151, 175, 240, 174, 255, 243, 255,},
{113, 146, 217, 152, 176, 241, 175, 255, 244, 255,},
{114, 147, 218, 152, 176, 241, 176, 255, 244, 255,},
{115, 148, 219, 153, 176, 242, 177, 255, 244, 255,},
{116, 149, 219, 153, 177, 242, 177, 255, 245, 255,},
{117, 150, 220, 154, 177, 242, 178, 255, 245, 255,},
{118, 151, 221, 155, 178, 243, 179, 255, 245, 255,},
{119, 152, 222, 155, 178, 243, 179, 255, 245, 255,},
{120, 153, 222, 156, 178, 244, 180, 255, 246, 255,},
{121, 154, 223, 156, 179, 244, 181, 255, 246, 255,},
{122, 155, 224, 157, 179, 244, 181, 255, 246, 255,},
{123, 156, 224, 157, 180, 245, 182, 255, 247, 255,},
{124, 157, 225, 158, 180, 245, 183, 255, 247, 255,},
{125, 158, 225, 159, 180, 245, 183, 255, 247, 255,},
{126, 159, 226, 159, 181, 246, 184, 255, 247, 255,},
{127, 160, 227, 160, 181, 246, 185, 255, 248, 255,},
{128, 161, 227, 160, 182, 246, 185, 255, 248, 255,},
{129, 162, 228, 161, 182, 246, 186, 255, 248, 255,},
{130, 163, 228, 161, 182, 247, 187, 255, 248, 255,},
{131, 164, 229, 162, 183, 247, 187, 255, 249, 255,},
{132, 165, 230, 163, 183, 247, 188, 255, 249, 255,},
{133, 166, 230, 163, 184, 248, 189, 255, 249, 255,},
{134, 166, 231, 164, 184, 248, 189, 255, 249, 255,},
{135, 167, 231, 164, 184, 248, 190, 255, 250, 255,},
{136, 168, 232, 165, 185, 248, 191, 255, 250, 255,},
{137, 169, 232, 166, 185, 248, 191, 255, 250, 255,},
{138, 170, 233, 166, 186, 249, 192, 255, 250, 255,},
{139, 171, 233, 167, 186, 249, 192, 255, 250, 255,},
{140, 172, 234, 167, 187, 249, 193, 255, 251, 255,},
{141, 173, 234, 168, 187, 249, 194, 255, 251, 255,},
{142, 174, 235, 169, 187, 250, 194, 255, 251, 255,},
{143, 175, 235, 169, 188, 250, 195, 255, 251, 255,},
{144, 176, 236, 170, 188, 250, 196, 255, 251, 255,},
{145, 177, 236, 170, 189, 250, 196, 255, 251, 255,},
{146, 177, 237, 171, 189, 250, 197, 255, 252, 255,},
{147, 178, 237, 172, 189, 251, 198, 255, 252, 255,},
{148, 179, 238, 172, 190, 251, 198, 255, 252, 255,},
{149, 180, 238, 173, 190, 251, 199, 255, 252, 255,},
{150, 181, 238, 173, 191, 251, 200, 255, 252, 255,},
{151, 182, 239, 174, 191, 251, 200, 255, 252, 255,},
{152, 183, 239, 175, 192, 251, 201, 255, 252, 255,},
{153, 184, 240, 175, 192, 252, 202, 255, 252, 255,},
{154, 184, 240, 176, 193, 252, 202, 255, 253, 255,},
{155, 185, 240, 177, 193, 252, 203, 255, 253, 255,},
{156, 186, 241, 177, 193, 252, 203, 255, 253, 255,},
{157, 187, 241, 178, 194, 252, 204, 255, 253, 255,},
{158, 188, 242, 178, 194, 252, 205, 255, 253, 255,},
{159, 189, 242, 179, 195, 252, 205, 255, 253, 255,},
{160, 190, 242, 180, 195, 253, 206, 255, 253, 255,},
{161, 190, 243, 180, 196, 253, 207, 255, 253, 255,},
{162, 191, 243, 181, 196, 253, 207, 255, 254, 255,},
{163, 192, 243, 182, 197, 253, 208, 255, 254, 255,},
{164, 193, 244, 182, 197, 253, 209, 255, 254, 255,},
{165, 194, 244, 183, 197, 253, 209, 255, 254, 255,},
{166, 195, 244, 184, 198, 253, 210, 255, 254, 255,},
{167, 196, 245, 184, 198, 253, 210, 255, 254, 255,},
{168, 196, 245, 185, 199, 253, 211, 255, 254, 255,},
{169, 197, 245, 186, 199, 254, 212, 255, 254, 255,},
{170, 198, 246, 186, 200, 254, 212, 255, 254, 255,},
{171, 199, 246, 187, 200, 254, 213, 255, 254, 255,},
{172, 200, 246, 188, 201, 254, 214, 255, 254, 255,},
{173, 200, 246, 188, 201, 254, 214, 255, 254, 255,},
{174, 201, 247, 189, 202, 254, 215, 255, 254, 255,},
{175, 202, 247, 189, 202, 254, 215, 255, 255, 255,},
{176, 203, 247, 190, 203, 254, 216, 255, 255, 255,},
{177, 204, 248, 191, 203, 254, 217, 255, 255, 255,},
{178, 204, 248, 191, 204, 254, 217, 255, 255, 255,},
{179, 205, 248, 192, 204, 254, 218, 255, 255, 255,},
{180, 206, 248, 193, 204, 254, 218, 255, 255, 255,},
{181, 207, 249, 194, 205, 255, 219, 255, 255, 255,},
{182, 208, 249, 194, 205, 255, 220, 255, 255, 255,},
{183, 208, 249, 195, 206, 255, 220, 255, 255, 255,},
{184, 209, 249, 196, 206, 255, 221, 255, 255, 255,},
{185, 210, 250, 196, 207, 255, 221, 255, 255, 255,},
{186, 211, 250, 197, 207, 255, 222, 255, 255, 255,},
{187, 211, 250, 198, 208, 255, 223, 255, 255, 255,},
{188, 212, 250, 198, 208, 255, 223, 255, 255, 255,},
{189, 213, 250, 199, 209, 255, 224, 255, 255, 255,},
{190, 214, 251, 200, 209, 255, 224, 255, 255, 255,},
{191, 215, 251, 200, 210, 255, 225, 255, 255, 255,},
{192, 215, 251, 201, 211, 255, 225, 255, 255, 255,},
{193, 216, 251, 202, 211, 255, 226, 255, 255, 255,},
{194, 217, 251, 203, 212, 255, 227, 255, 255, 255,},
{195, 218, 252, 203, 212, 255, 227, 255, 255, 255,},
{196, 218, 252, 204, 213, 255, 228, 255, 255, 255,},
{197, 219, 252, 205, 213, 255, 228, 255, 255, 255,},
{198, 220, 252, 205, 214, 255, 229, 255, 255, 255,},
{199, 221, 252, 206, 214, 255, 229, 255, 255, 255,},
{200, 221, 252, 207, 215, 255, 230, 255, 255, 255,},
{201, 222, 252, 208, 215, 255, 231, 255, 255, 255,},
{202, 223, 253, 208, 216, 255, 231, 255, 255, 255,},
{203, 223, 253, 209, 216, 255, 232, 255, 255, 255,},
{204, 224, 253, 210, 217, 255, 232, 255, 255, 255,},
{205, 225, 253, 211, 218, 255, 233, 255, 255, 255,},
{206, 226, 253, 211, 218, 255, 233, 255, 255, 255,},
{207, 226, 253, 212, 219, 255, 234, 255, 255, 255,},
{208, 227, 253, 213, 219, 255, 234, 255, 255, 255,},
{209, 228, 254, 214, 220, 255, 235, 255, 255, 255,},
{210, 228, 254, 214, 220, 255, 236, 255, 255, 255,},
{211, 229, 254, 215, 221, 255, 236, 255, 255, 255,},
{212, 230, 254, 216, 222, 255, 237, 255, 255, 255,},
{213, 230, 254, 217, 222, 255, 237, 255, 255, 255,},
{214, 231, 254, 217, 223, 255, 238, 255, 255, 255,},
{215, 232, 254, 218, 223, 255, 238, 255, 255, 255,},
{216, 233, 254, 219, 224, 255, 239, 255, 255, 255,},
{217, 233, 254, 220, 225, 255, 239, 255, 255, 255,},
{218, 234, 255, 220, 225, 255, 240, 255, 255, 255,},
{219, 235, 255, 221, 226, 255, 240, 255, 255, 255,},
{220, 235, 255, 222, 226, 255, 241, 255, 255, 255,},
{221, 236, 255, 223, 227, 255, 241, 255, 255, 255,},
{222, 237, 255, 224, 228, 255, 242, 255, 255, 255,},
{223, 237, 255, 224, 228, 255, 242, 255, 255, 255,},
{224, 238, 255, 225, 229, 255, 243, 255, 255, 255,},
{225, 238, 255, 226, 230, 255, 243, 255, 255, 255,},
{226, 239, 255, 227, 230, 255, 244, 255, 255, 255,},
{227, 240, 255, 228, 231, 255, 244, 255, 255, 255,},
{228, 240, 255, 228, 232, 255, 245, 255, 255, 255,},
{229, 241, 255, 229, 232, 255, 245, 255, 255, 255,},
{230, 242, 255, 230, 233, 255, 246, 255, 255, 255,},
{231, 242, 255, 231, 234, 255, 246, 255, 255, 255,},
{232, 243, 255, 232, 234, 255, 247, 255, 255, 255,},
{233, 243, 255, 233, 235, 255, 247, 255, 255, 255,},
{234, 244, 255, 233, 236, 255, 247, 255, 255, 255,},
{235, 245, 255, 234, 236, 255, 248, 255, 255, 255,},
{236, 245, 255, 235, 237, 255, 248, 255, 255, 255,},
{237, 246, 255, 236, 238, 255, 249, 255, 255, 255,},
{238, 247, 255, 237, 239, 255, 249, 255, 255, 255,},
{239, 247, 255, 238, 239, 255, 250, 255, 255, 255,},
{240, 248, 255, 239, 240, 255, 250, 255, 255, 255,},
{241, 248, 255, 240, 241, 255, 251, 255, 255, 255,},
{242, 249, 255, 241, 242, 255, 251, 255, 255, 255,},
{243, 249, 255, 241, 243, 255, 251, 255, 255, 255,},
{244, 250, 255, 242, 243, 255, 252, 255, 255, 255,},
{245, 251, 255, 243, 244, 255, 252, 255, 255, 255,},
{246, 251, 255, 244, 245, 255, 253, 255, 255, 255,},
{247, 252, 255, 245, 246, 255, 253, 255, 255, 255,},
{248, 252, 255, 246, 247, 255, 253, 255, 255, 255,},
{249, 253, 255, 247, 248, 255, 254, 255, 255, 255,},
{250, 253, 255, 248, 249, 255, 254, 255, 255, 255,},
{251, 254, 255, 249, 250, 255, 254, 255, 255, 255,},
{252, 254, 255, 251, 251, 255, 255, 255, 255, 255,},
{253, 255, 255, 252, 252, 255, 255, 255, 255, 255,},
{254, 255, 255, 253, 253, 255, 255, 255, 255, 255,},
{255, 255, 255, 254, 254, 255, 255, 255, 255, 255,},
};
#else
Modeling default coef probs with distribution Replaces the default tables for single coefficient magnitudes with those obtained from an appropriate distribution. The EOB node is left unchanged. The model is represeted as a 256-size codebook where the index corresponds to the probability of the Zero or the One node. Two variations are implemented corresponding to whether the Zero node or the One-node is used as the peg. The main advantage is that the default prob tables will become considerably smaller and manageable. Besides there is substantially less risk of over-fitting for a training set. Various distributions are tried and the one that gives the best results is the family of Generalized Gaussian distributions with shape parameter 0.75. The results are within about 0.2% of fully trained tables for the Zero peg variant, and within 0.1% of the One peg variant. The forward updates are optionally (controlled by a macro) model-based, i.e. restricted to only convey probabilities from the codebook. Backward updates can also be optionally (controlled by another macro) model-based, but is turned off by default. Currently model-based forward updates work about the same as unconstrained updates, but there is a drop in performance with backward-updates being model based. The model based approach also allows the probabilities for the key frames to be adjusted from the defaults based on the base_qindex of the frame. Currently the adjustment function is a placeholder that adjusts the prob of EOB and Zero node from the nominal one at higher quality (lower qindex) or lower quality (higher qindex) ends of the range. The rest of the probabilities are then derived based on the model from the adjusted prob of zero. Change-Id: Iae050f3cbcc6d8b3f204e8dc395ae47b3b2192c9
2013-03-13 19:03:17 +01:00
const vp9_prob vp9_modelcoefprobs_gg75p1[COEFPROB_MODELS][ENTROPY_NODES - 1] = {
// Probs generated with a Generalized Gaussian (with shape parameter 0.75)
Modeling default coef probs with distribution Replaces the default tables for single coefficient magnitudes with those obtained from an appropriate distribution. The EOB node is left unchanged. The model is represeted as a 256-size codebook where the index corresponds to the probability of the Zero or the One node. Two variations are implemented corresponding to whether the Zero node or the One-node is used as the peg. The main advantage is that the default prob tables will become considerably smaller and manageable. Besides there is substantially less risk of over-fitting for a training set. Various distributions are tried and the one that gives the best results is the family of Generalized Gaussian distributions with shape parameter 0.75. The results are within about 0.2% of fully trained tables for the Zero peg variant, and within 0.1% of the One peg variant. The forward updates are optionally (controlled by a macro) model-based, i.e. restricted to only convey probabilities from the codebook. Backward updates can also be optionally (controlled by another macro) model-based, but is turned off by default. Currently model-based forward updates work about the same as unconstrained updates, but there is a drop in performance with backward-updates being model based. The model based approach also allows the probabilities for the key frames to be adjusted from the defaults based on the base_qindex of the frame. Currently the adjustment function is a placeholder that adjusts the prob of EOB and Zero node from the nominal one at higher quality (lower qindex) or lower quality (higher qindex) ends of the range. The rest of the probabilities are then derived based on the model from the adjusted prob of zero. Change-Id: Iae050f3cbcc6d8b3f204e8dc395ae47b3b2192c9
2013-03-13 19:03:17 +01:00
// source model with varying quantizer step size for a uniform quantizer
{1, 1, 3, 86, 129, 6, 87, 21, 90, 26,},
{2, 3, 9, 87, 130, 16, 89, 55, 96, 65,},
{3, 5, 14, 88, 130, 26, 90, 82, 101, 95,},
{4, 7, 20, 89, 131, 35, 92, 105, 105, 119,},
{5, 9, 25, 90, 132, 44, 93, 124, 110, 138,},
{6, 11, 30, 91, 132, 52, 95, 141, 114, 155,},
{7, 13, 35, 92, 133, 60, 96, 155, 118, 168,},
{8, 15, 40, 93, 134, 68, 97, 167, 123, 180,},
{9, 17, 44, 94, 134, 75, 99, 178, 127, 190,},
{10, 19, 49, 94, 135, 82, 100, 187, 131, 199,},
{11, 21, 54, 95, 135, 88, 101, 195, 135, 206,},
{13, 23, 58, 96, 136, 95, 102, 202, 138, 213,},
{14, 25, 63, 97, 137, 101, 104, 209, 142, 218,},
{15, 27, 67, 98, 137, 107, 105, 214, 146, 223,},
{16, 29, 71, 98, 138, 113, 106, 219, 150, 227,},
{17, 31, 76, 99, 138, 118, 107, 223, 153, 231,},
{18, 33, 80, 100, 139, 123, 109, 227, 157, 234,},
{20, 35, 84, 101, 140, 128, 110, 231, 160, 237,},
{21, 37, 88, 102, 140, 133, 111, 234, 164, 239,},
{22, 39, 92, 102, 141, 138, 112, 236, 167, 241,},
{23, 41, 95, 103, 141, 143, 114, 238, 170, 243,},
{25, 43, 99, 104, 142, 147, 115, 241, 173, 245,},
{26, 45, 103, 105, 142, 151, 116, 242, 176, 246,},
{27, 47, 107, 105, 143, 156, 117, 244, 180, 248,},
{28, 49, 110, 106, 144, 159, 119, 245, 182, 249,},
{30, 51, 114, 107, 144, 163, 120, 247, 185, 250,},
{31, 53, 117, 108, 145, 167, 121, 248, 188, 250,},
{32, 55, 121, 109, 145, 171, 123, 249, 191, 251,},
{34, 57, 124, 110, 146, 174, 124, 250, 194, 252,},
{35, 59, 127, 110, 147, 177, 125, 250, 196, 252,},
{36, 61, 130, 111, 147, 181, 127, 251, 199, 253,},
{38, 63, 134, 112, 148, 184, 128, 252, 201, 253,},
{39, 65, 137, 113, 148, 187, 129, 252, 204, 254,},
{40, 67, 140, 114, 149, 190, 130, 253, 206, 254,},
{42, 69, 143, 115, 150, 192, 132, 253, 208, 254,},
{43, 71, 146, 115, 150, 195, 133, 254, 210, 254,},
{44, 73, 149, 116, 151, 198, 134, 254, 212, 255,},
{46, 75, 152, 117, 151, 200, 136, 254, 214, 255,},
{47, 77, 154, 118, 152, 203, 137, 254, 216, 255,},
{49, 79, 157, 119, 153, 205, 139, 255, 218, 255,},
{50, 81, 160, 120, 153, 207, 140, 255, 220, 255,},
{51, 83, 163, 121, 154, 210, 141, 255, 222, 255,},
{53, 85, 165, 122, 154, 212, 143, 255, 223, 255,},
{54, 87, 168, 123, 155, 214, 144, 255, 225, 255,},
{56, 89, 170, 123, 156, 216, 145, 255, 227, 255,},
{57, 91, 173, 124, 156, 218, 147, 255, 228, 255,},
{59, 93, 175, 125, 157, 219, 148, 255, 230, 255,},
{60, 95, 178, 126, 158, 221, 150, 255, 231, 255,},
{62, 97, 180, 127, 158, 223, 151, 255, 232, 255,},
{63, 99, 182, 128, 159, 224, 152, 255, 234, 255,},
{65, 101, 184, 129, 160, 226, 154, 255, 235, 255,},
{66, 103, 187, 130, 160, 227, 155, 255, 236, 255,},
{68, 105, 189, 131, 161, 229, 157, 255, 237, 255,},
{69, 107, 191, 132, 162, 230, 158, 255, 238, 255,},
{71, 109, 193, 133, 163, 232, 159, 255, 239, 255,},
{73, 111, 195, 134, 163, 233, 161, 255, 240, 255,},
{74, 113, 197, 135, 164, 234, 162, 255, 241, 255,},
{76, 115, 199, 136, 165, 235, 164, 255, 242, 255,},
{77, 117, 201, 137, 165, 236, 165, 255, 243, 255,},
{79, 119, 203, 138, 166, 237, 167, 255, 244, 255,},
{81, 121, 205, 139, 167, 238, 168, 255, 244, 255,},
{82, 123, 206, 141, 168, 239, 170, 255, 245, 255,},
{84, 125, 208, 142, 168, 240, 171, 255, 246, 255,},
{86, 127, 210, 143, 169, 241, 173, 255, 246, 255,},
{87, 129, 212, 144, 170, 242, 174, 255, 247, 255,},
{89, 131, 213, 145, 171, 243, 176, 255, 248, 255,},
{91, 133, 215, 146, 171, 244, 177, 255, 248, 255,},
{93, 135, 216, 147, 172, 244, 179, 255, 249, 255,},
{94, 137, 218, 148, 173, 245, 180, 255, 249, 255,},
{96, 139, 220, 150, 174, 246, 181, 255, 250, 255,},
{98, 141, 221, 151, 175, 246, 183, 255, 250, 255,},
{100, 143, 222, 152, 175, 247, 184, 255, 251, 255,},
{101, 145, 224, 153, 176, 248, 186, 255, 251, 255,},
{103, 147, 225, 154, 177, 248, 187, 255, 251, 255,},
{105, 149, 226, 156, 178, 249, 189, 255, 252, 255,},
{107, 151, 228, 157, 179, 249, 190, 255, 252, 255,},
{109, 153, 229, 158, 180, 250, 192, 255, 252, 255,},
{111, 155, 230, 159, 181, 250, 193, 255, 253, 255,},
{112, 157, 231, 161, 181, 251, 195, 255, 253, 255,},
{114, 159, 233, 162, 182, 251, 196, 255, 253, 255,},
{116, 161, 234, 163, 183, 251, 198, 255, 253, 255,},
{118, 163, 235, 165, 184, 252, 199, 255, 254, 255,},
{120, 165, 236, 166, 185, 252, 201, 255, 254, 255,},
{122, 167, 237, 167, 186, 252, 202, 255, 254, 255,},
{124, 169, 238, 169, 187, 253, 204, 255, 254, 255,},
{126, 171, 239, 170, 188, 253, 205, 255, 254, 255,},
{128, 173, 240, 172, 189, 253, 207, 255, 255, 255,},
{130, 175, 241, 173, 190, 253, 208, 255, 255, 255,},
{132, 177, 242, 175, 191, 254, 210, 255, 255, 255,},
{134, 179, 242, 176, 192, 254, 211, 255, 255, 255,},
{137, 181, 243, 177, 193, 254, 213, 255, 255, 255,},
{139, 183, 244, 179, 194, 254, 214, 255, 255, 255,},
{141, 185, 245, 181, 195, 254, 216, 255, 255, 255,},
{143, 187, 245, 182, 196, 255, 217, 255, 255, 255,},
{145, 189, 246, 184, 197, 255, 218, 255, 255, 255,},
{147, 191, 247, 185, 199, 255, 220, 255, 255, 255,},
{150, 193, 247, 187, 200, 255, 221, 255, 255, 255,},
{152, 195, 248, 188, 201, 255, 223, 255, 255, 255,},
{154, 197, 249, 190, 202, 255, 224, 255, 255, 255,},
{157, 199, 249, 192, 203, 255, 225, 255, 255, 255,},
{159, 201, 250, 193, 205, 255, 227, 255, 255, 255,},
{162, 203, 250, 195, 206, 255, 228, 255, 255, 255,},
{164, 205, 251, 197, 207, 255, 229, 255, 255, 255,},
{166, 207, 251, 199, 208, 255, 231, 255, 255, 255,},
{169, 209, 252, 201, 210, 255, 232, 255, 255, 255,},
{172, 211, 252, 202, 211, 255, 233, 255, 255, 255,},
{174, 213, 252, 204, 212, 255, 235, 255, 255, 255,},
{177, 215, 253, 206, 214, 255, 236, 255, 255, 255,},
{179, 217, 253, 208, 215, 255, 237, 255, 255, 255,},
{182, 219, 254, 210, 217, 255, 238, 255, 255, 255,},
{185, 221, 254, 212, 218, 255, 240, 255, 255, 255,},
{188, 223, 254, 214, 220, 255, 241, 255, 255, 255,},
{191, 225, 254, 216, 221, 255, 242, 255, 255, 255,},
{194, 227, 255, 218, 223, 255, 243, 255, 255, 255,},
{197, 229, 255, 220, 225, 255, 244, 255, 255, 255,},
{200, 231, 255, 222, 226, 255, 245, 255, 255, 255,},
{203, 233, 255, 224, 228, 255, 247, 255, 255, 255,},
{206, 235, 255, 227, 230, 255, 248, 255, 255, 255,},
{210, 237, 255, 229, 232, 255, 249, 255, 255, 255,},
{213, 239, 255, 231, 234, 255, 250, 255, 255, 255,},
{217, 241, 255, 234, 236, 255, 251, 255, 255, 255,},
{221, 243, 255, 236, 238, 255, 252, 255, 255, 255,},
{225, 245, 255, 239, 240, 255, 252, 255, 255, 255,},
{229, 247, 255, 241, 242, 255, 253, 255, 255, 255,},
{233, 249, 255, 244, 245, 255, 254, 255, 255, 255,},
{238, 251, 255, 247, 247, 255, 255, 255, 255, 255,},
{244, 253, 255, 250, 250, 255, 255, 255, 255, 255,},
{251, 255, 255, 254, 254, 255, 255, 255, 255, 255,},
};
const vp9_prob vp9_modelcoefprobs_gg625p1[COEFPROB_MODELS][ENTROPY_NODES - 1] = {
// Probs generated with a Generalized Gaussian (with shape parameter 0.625)
// source model with varying quantizer step size for a uniform quantizer
{1, 1, 3, 87, 129, 6, 87, 20, 91, 24,},
{2, 3, 8, 88, 130, 15, 90, 50, 97, 56,},
{3, 5, 14, 90, 131, 24, 91, 74, 102, 81,},
{4, 7, 19, 91, 132, 33, 93, 93, 106, 101,},
{5, 9, 24, 92, 133, 40, 94, 110, 110, 118,},
{6, 11, 29, 93, 134, 48, 96, 125, 113, 132,},
{7, 13, 33, 93, 134, 55, 97, 138, 117, 145,},
{8, 15, 38, 94, 135, 62, 98, 149, 120, 155,},
{10, 17, 42, 95, 136, 68, 99, 159, 124, 165,},
{11, 19, 47, 96, 136, 74, 100, 168, 127, 174,},
{12, 21, 51, 97, 137, 80, 102, 176, 130, 181,},
{13, 23, 55, 98, 137, 86, 103, 184, 133, 188,},
{14, 25, 59, 98, 138, 91, 104, 190, 136, 194,},
{16, 27, 64, 99, 139, 97, 105, 196, 139, 200,},
{17, 29, 68, 100, 139, 102, 106, 201, 142, 205,},
{18, 31, 71, 101, 140, 107, 107, 206, 145, 209,},
{20, 33, 75, 102, 140, 112, 108, 211, 148, 213,},
{21, 35, 79, 102, 141, 116, 109, 215, 150, 217,},
{22, 37, 83, 103, 141, 121, 110, 218, 153, 220,},
{24, 39, 87, 104, 142, 125, 112, 222, 156, 224,},
{25, 41, 90, 105, 143, 129, 113, 225, 159, 226,},
{26, 43, 94, 105, 143, 133, 114, 227, 161, 229,},
{28, 45, 97, 106, 144, 137, 115, 230, 164, 231,},
{29, 47, 101, 107, 144, 141, 116, 232, 166, 233,},
{31, 49, 104, 108, 145, 145, 117, 234, 169, 235,},
{32, 51, 107, 108, 145, 149, 118, 236, 171, 237,},
{33, 53, 111, 109, 146, 152, 119, 238, 174, 239,},
{35, 55, 114, 110, 146, 156, 120, 240, 176, 240,},
{36, 57, 117, 111, 147, 159, 121, 241, 178, 242,},
{38, 59, 120, 112, 148, 162, 122, 242, 181, 243,},
{39, 61, 123, 112, 148, 165, 124, 244, 183, 244,},
{41, 63, 126, 113, 149, 168, 125, 245, 185, 245,},
{42, 65, 129, 114, 149, 171, 126, 246, 187, 246,},
{44, 67, 132, 115, 150, 174, 127, 247, 189, 247,},
{45, 69, 135, 116, 150, 177, 128, 248, 192, 248,},
{47, 71, 138, 116, 151, 180, 129, 248, 194, 249,},
{48, 73, 141, 117, 152, 183, 130, 249, 196, 249,},
{50, 75, 143, 118, 152, 185, 131, 250, 198, 250,},
{51, 77, 146, 119, 153, 188, 132, 250, 200, 250,},
{53, 79, 149, 120, 153, 190, 134, 251, 201, 251,},
{55, 81, 151, 120, 154, 192, 135, 251, 203, 251,},
{56, 83, 154, 121, 155, 195, 136, 252, 205, 252,},
{58, 85, 157, 122, 155, 197, 137, 252, 207, 252,},
{59, 87, 159, 123, 156, 199, 138, 253, 209, 253,},
{61, 89, 162, 124, 156, 201, 139, 253, 210, 253,},
{63, 91, 164, 125, 157, 203, 140, 253, 212, 253,},
{64, 93, 166, 126, 158, 205, 142, 254, 214, 253,},
{66, 95, 169, 126, 158, 207, 143, 254, 215, 254,},
{68, 97, 171, 127, 159, 209, 144, 254, 217, 254,},
{69, 99, 173, 128, 159, 211, 145, 254, 218, 254,},
{71, 101, 176, 129, 160, 213, 146, 254, 220, 254,},
{73, 103, 178, 130, 161, 214, 148, 255, 221, 255,},
{75, 105, 180, 131, 161, 216, 149, 255, 223, 255,},
{76, 107, 182, 132, 162, 218, 150, 255, 224, 255,},
{78, 109, 184, 133, 163, 219, 151, 255, 225, 255,},
{80, 111, 186, 134, 163, 221, 152, 255, 226, 255,},
{82, 113, 188, 135, 164, 222, 154, 255, 228, 255,},
{83, 115, 190, 136, 165, 224, 155, 255, 229, 255,},
{85, 117, 192, 137, 165, 225, 156, 255, 230, 255,},
{87, 119, 194, 137, 166, 226, 157, 255, 231, 255,},
{89, 121, 196, 138, 167, 228, 159, 255, 232, 255,},
{91, 123, 198, 139, 167, 229, 160, 255, 233, 255,},
{93, 125, 200, 140, 168, 230, 161, 255, 234, 255,},
{94, 127, 202, 141, 169, 231, 163, 255, 235, 255,},
{96, 129, 203, 142, 169, 233, 164, 255, 236, 255,},
{98, 131, 205, 143, 170, 234, 165, 255, 237, 255,},
{100, 133, 207, 145, 171, 235, 166, 255, 238, 255,},
{102, 135, 209, 146, 171, 236, 168, 255, 239, 255,},
{104, 137, 210, 147, 172, 237, 169, 255, 240, 255,},
{106, 139, 212, 148, 173, 238, 170, 255, 241, 255,},
{108, 141, 213, 149, 174, 239, 172, 255, 242, 255,},
{110, 143, 215, 150, 174, 240, 173, 255, 242, 255,},
{112, 145, 216, 151, 175, 240, 174, 255, 243, 255,},
{114, 147, 218, 152, 176, 241, 176, 255, 244, 255,},
{116, 149, 219, 153, 177, 242, 177, 255, 244, 255,},
{118, 151, 221, 154, 178, 243, 178, 255, 245, 255,},
{120, 153, 222, 156, 178, 244, 180, 255, 246, 255,},
{122, 155, 223, 157, 179, 244, 181, 255, 246, 255,},
{124, 157, 225, 158, 180, 245, 183, 255, 247, 255,},
{126, 159, 226, 159, 181, 246, 184, 255, 247, 255,},
{128, 161, 227, 160, 182, 246, 185, 255, 248, 255,},
{130, 163, 229, 162, 183, 247, 187, 255, 248, 255,},
{132, 165, 230, 163, 183, 247, 188, 255, 249, 255,},
{135, 167, 231, 164, 184, 248, 190, 255, 249, 255,},
{137, 169, 232, 165, 185, 248, 191, 255, 250, 255,},
{139, 171, 233, 167, 186, 249, 192, 255, 250, 255,},
{141, 173, 234, 168, 187, 249, 194, 255, 251, 255,},
{143, 175, 235, 169, 188, 250, 195, 255, 251, 255,},
{146, 177, 236, 171, 189, 250, 197, 255, 251, 255,},
{148, 179, 237, 172, 190, 251, 198, 255, 252, 255,},
{150, 181, 238, 173, 191, 251, 200, 255, 252, 255,},
{152, 183, 239, 175, 192, 251, 201, 255, 252, 255,},
{155, 185, 240, 176, 193, 252, 203, 255, 253, 255,},
{157, 187, 241, 178, 194, 252, 204, 255, 253, 255,},
{159, 189, 242, 179, 195, 252, 206, 255, 253, 255,},
{162, 191, 243, 181, 196, 253, 207, 255, 253, 255,},
{164, 193, 244, 182, 197, 253, 209, 255, 254, 255,},
{166, 195, 244, 184, 198, 253, 210, 255, 254, 255,},
{169, 197, 245, 185, 199, 254, 212, 255, 254, 255,},
{171, 199, 246, 187, 200, 254, 213, 255, 254, 255,},
{174, 201, 247, 189, 201, 254, 215, 255, 254, 255,},
{176, 203, 247, 190, 203, 254, 216, 255, 255, 255,},
{179, 205, 248, 192, 204, 254, 218, 255, 255, 255,},
{181, 207, 249, 194, 205, 255, 219, 255, 255, 255,},
{184, 209, 249, 195, 206, 255, 221, 255, 255, 255,},
{186, 211, 250, 197, 208, 255, 222, 255, 255, 255,},
{189, 213, 250, 199, 209, 255, 224, 255, 255, 255,},
{192, 215, 251, 201, 210, 255, 225, 255, 255, 255,},
{194, 217, 251, 203, 212, 255, 227, 255, 255, 255,},
{197, 219, 252, 205, 213, 255, 228, 255, 255, 255,},
{200, 221, 252, 207, 215, 255, 230, 255, 255, 255,},
{202, 223, 253, 209, 216, 255, 231, 255, 255, 255,},
{205, 225, 253, 211, 218, 255, 233, 255, 255, 255,},
{208, 227, 253, 213, 219, 255, 234, 255, 255, 255,},
{211, 229, 254, 215, 221, 255, 236, 255, 255, 255,},
{214, 231, 254, 217, 223, 255, 238, 255, 255, 255,},
{217, 233, 254, 219, 224, 255, 239, 255, 255, 255,},
{220, 235, 255, 222, 226, 255, 241, 255, 255, 255,},
{223, 237, 255, 224, 228, 255, 242, 255, 255, 255,},
{226, 239, 255, 227, 230, 255, 244, 255, 255, 255,},
{229, 241, 255, 229, 232, 255, 245, 255, 255, 255,},
{232, 243, 255, 232, 234, 255, 247, 255, 255, 255,},
{235, 245, 255, 235, 237, 255, 248, 255, 255, 255,},
{239, 247, 255, 238, 239, 255, 250, 255, 255, 255,},
{242, 249, 255, 241, 242, 255, 251, 255, 255, 255,},
{246, 251, 255, 244, 245, 255, 253, 255, 255, 255,},
{250, 253, 255, 248, 248, 255, 254, 255, 255, 255,},
{254, 255, 255, 253, 253, 255, 255, 255, 255, 255,},
};
#endif
Modeling default coef probs with distribution Replaces the default tables for single coefficient magnitudes with those obtained from an appropriate distribution. The EOB node is left unchanged. The model is represeted as a 256-size codebook where the index corresponds to the probability of the Zero or the One node. Two variations are implemented corresponding to whether the Zero node or the One-node is used as the peg. The main advantage is that the default prob tables will become considerably smaller and manageable. Besides there is substantially less risk of over-fitting for a training set. Various distributions are tried and the one that gives the best results is the family of Generalized Gaussian distributions with shape parameter 0.75. The results are within about 0.2% of fully trained tables for the Zero peg variant, and within 0.1% of the One peg variant. The forward updates are optionally (controlled by a macro) model-based, i.e. restricted to only convey probabilities from the codebook. Backward updates can also be optionally (controlled by another macro) model-based, but is turned off by default. Currently model-based forward updates work about the same as unconstrained updates, but there is a drop in performance with backward-updates being model based. The model based approach also allows the probabilities for the key frames to be adjusted from the defaults based on the base_qindex of the frame. Currently the adjustment function is a placeholder that adjusts the prob of EOB and Zero node from the nominal one at higher quality (lower qindex) or lower quality (higher qindex) ends of the range. The rest of the probabilities are then derived based on the model from the adjusted prob of zero. Change-Id: Iae050f3cbcc6d8b3f204e8dc395ae47b3b2192c9
2013-03-13 19:03:17 +01:00
void vp9_get_model_distribution(vp9_prob p, vp9_prob *tree_probs,
int b, int r) {
const vp9_prob (*model)[ENTROPY_NODES - 1];
#if UNCONSTRAINED_NODES == 2
if (r != INTRA_FRAME && b == PLANE_TYPE_UV)
model = vp9_modelcoefprobs_gg75;
else if (r == INTRA_FRAME && b == PLANE_TYPE_UV)
model = vp9_modelcoefprobs_gg75;
else if (r != INTRA_FRAME && b == PLANE_TYPE_Y_WITH_DC)
model = vp9_modelcoefprobs_gg75;
else
model = vp9_modelcoefprobs_gg75;
#else
if (r != INTRA_FRAME && b == PLANE_TYPE_UV)
model = vp9_modelcoefprobs_gg75p1;
else if (r == INTRA_FRAME && b == PLANE_TYPE_UV)
model = vp9_modelcoefprobs_gg75p1;
else if (r != INTRA_FRAME && b == PLANE_TYPE_Y_WITH_DC)
model = vp9_modelcoefprobs_gg75p1;
else
model = vp9_modelcoefprobs_gg625p1;
Modeling default coef probs with distribution Replaces the default tables for single coefficient magnitudes with those obtained from an appropriate distribution. The EOB node is left unchanged. The model is represeted as a 256-size codebook where the index corresponds to the probability of the Zero or the One node. Two variations are implemented corresponding to whether the Zero node or the One-node is used as the peg. The main advantage is that the default prob tables will become considerably smaller and manageable. Besides there is substantially less risk of over-fitting for a training set. Various distributions are tried and the one that gives the best results is the family of Generalized Gaussian distributions with shape parameter 0.75. The results are within about 0.2% of fully trained tables for the Zero peg variant, and within 0.1% of the One peg variant. The forward updates are optionally (controlled by a macro) model-based, i.e. restricted to only convey probabilities from the codebook. Backward updates can also be optionally (controlled by another macro) model-based, but is turned off by default. Currently model-based forward updates work about the same as unconstrained updates, but there is a drop in performance with backward-updates being model based. The model based approach also allows the probabilities for the key frames to be adjusted from the defaults based on the base_qindex of the frame. Currently the adjustment function is a placeholder that adjusts the prob of EOB and Zero node from the nominal one at higher quality (lower qindex) or lower quality (higher qindex) ends of the range. The rest of the probabilities are then derived based on the model from the adjusted prob of zero. Change-Id: Iae050f3cbcc6d8b3f204e8dc395ae47b3b2192c9
2013-03-13 19:03:17 +01:00
#endif
if (p & 1) { // odd
vpx_memcpy(tree_probs + UNCONSTRAINED_NODES,
model[(p - 1) / 2] + UNCONSTRAINED_NODES - 1,
(ENTROPY_NODES - UNCONSTRAINED_NODES) * sizeof(vp9_prob));
} else {
// interpolate
int i;
for (i = UNCONSTRAINED_NODES; i < ENTROPY_NODES; ++i)
tree_probs[i] = (model[p / 2 - 1][i - 1] + model[p / 2][i - 1]) >> 1;
}
Modeling default coef probs with distribution Replaces the default tables for single coefficient magnitudes with those obtained from an appropriate distribution. The EOB node is left unchanged. The model is represeted as a 256-size codebook where the index corresponds to the probability of the Zero or the One node. Two variations are implemented corresponding to whether the Zero node or the One-node is used as the peg. The main advantage is that the default prob tables will become considerably smaller and manageable. Besides there is substantially less risk of over-fitting for a training set. Various distributions are tried and the one that gives the best results is the family of Generalized Gaussian distributions with shape parameter 0.75. The results are within about 0.2% of fully trained tables for the Zero peg variant, and within 0.1% of the One peg variant. The forward updates are optionally (controlled by a macro) model-based, i.e. restricted to only convey probabilities from the codebook. Backward updates can also be optionally (controlled by another macro) model-based, but is turned off by default. Currently model-based forward updates work about the same as unconstrained updates, but there is a drop in performance with backward-updates being model based. The model based approach also allows the probabilities for the key frames to be adjusted from the defaults based on the base_qindex of the frame. Currently the adjustment function is a placeholder that adjusts the prob of EOB and Zero node from the nominal one at higher quality (lower qindex) or lower quality (higher qindex) ends of the range. The rest of the probabilities are then derived based on the model from the adjusted prob of zero. Change-Id: Iae050f3cbcc6d8b3f204e8dc395ae47b3b2192c9
2013-03-13 19:03:17 +01:00
}
#endif
32x32 transform for superblocks. This adds Debargha's DCT/DWT hybrid and a regular 32x32 DCT, and adds code all over the place to wrap that in the bitstream/encoder/decoder/RD. Some implementation notes (these probably need careful review): - token range is extended by 1 bit, since the value range out of this transform is [-16384,16383]. - the coefficients coming out of the FDCT are manually scaled back by 1 bit, or else they won't fit in int16_t (they are 17 bits). Because of this, the RD error scoring does not right-shift the MSE score by two (unlike for 4x4/8x8/16x16). - to compensate for this loss in precision, the quantizer is halved also. This is currently a little hacky. - FDCT and IDCT is double-only right now. Needs a fixed-point impl. - There are no default probabilities for the 32x32 transform yet; I'm simply using the 16x16 luma ones. A future commit will add newly generated probabilities for all transforms. - No ADST version. I don't think we'll add one for this level; if an ADST is desired, transform-size selection can scale back to 16x16 or lower, and use an ADST at that level. Additional notes specific to Debargha's DWT/DCT hybrid: - coefficient scale is different for the top/left 16x16 (DCT-over-DWT) block than for the rest (DWT pixel differences) of the block. Therefore, RD error scoring isn't easily scalable between coefficient and pixel domain. Thus, unfortunately, we need to compute the RD distortion in the pixel domain until we figure out how to scale these appropriately. Change-Id: I00386f20f35d7fabb19aba94c8162f8aee64ef2b
2012-12-07 23:45:05 +01:00
static vp9_tree_index cat1[2], cat2[4], cat3[6], cat4[8], cat5[10], cat6[28];
2010-05-18 17:58:33 +02:00
static void init_bit_tree(vp9_tree_index *p, int n) {
int i = 0;
2010-05-18 17:58:33 +02:00
while (++i < n) {
p[0] = p[1] = i << 1;
p += 2;
}
2010-05-18 17:58:33 +02:00
p[0] = p[1] = 0;
2010-05-18 17:58:33 +02:00
}
static void init_bit_trees() {
init_bit_tree(cat1, 1);
init_bit_tree(cat2, 2);
init_bit_tree(cat3, 3);
init_bit_tree(cat4, 4);
init_bit_tree(cat5, 5);
32x32 transform for superblocks. This adds Debargha's DCT/DWT hybrid and a regular 32x32 DCT, and adds code all over the place to wrap that in the bitstream/encoder/decoder/RD. Some implementation notes (these probably need careful review): - token range is extended by 1 bit, since the value range out of this transform is [-16384,16383]. - the coefficients coming out of the FDCT are manually scaled back by 1 bit, or else they won't fit in int16_t (they are 17 bits). Because of this, the RD error scoring does not right-shift the MSE score by two (unlike for 4x4/8x8/16x16). - to compensate for this loss in precision, the quantizer is halved also. This is currently a little hacky. - FDCT and IDCT is double-only right now. Needs a fixed-point impl. - There are no default probabilities for the 32x32 transform yet; I'm simply using the 16x16 luma ones. A future commit will add newly generated probabilities for all transforms. - No ADST version. I don't think we'll add one for this level; if an ADST is desired, transform-size selection can scale back to 16x16 or lower, and use an ADST at that level. Additional notes specific to Debargha's DWT/DCT hybrid: - coefficient scale is different for the top/left 16x16 (DCT-over-DWT) block than for the rest (DWT pixel differences) of the block. Therefore, RD error scoring isn't easily scalable between coefficient and pixel domain. Thus, unfortunately, we need to compute the RD distortion in the pixel domain until we figure out how to scale these appropriately. Change-Id: I00386f20f35d7fabb19aba94c8162f8aee64ef2b
2012-12-07 23:45:05 +01:00
init_bit_tree(cat6, 14);
2010-05-18 17:58:33 +02:00
}
vp9_extra_bit vp9_extra_bits[12] = {
{ 0, 0, 0, 0},
{ 0, 0, 0, 1},
{ 0, 0, 0, 2},
{ 0, 0, 0, 3},
{ 0, 0, 0, 4},
{ cat1, Pcat1, 1, 5},
{ cat2, Pcat2, 2, 7},
{ cat3, Pcat3, 3, 11},
{ cat4, Pcat4, 4, 19},
{ cat5, Pcat5, 5, 35},
32x32 transform for superblocks. This adds Debargha's DCT/DWT hybrid and a regular 32x32 DCT, and adds code all over the place to wrap that in the bitstream/encoder/decoder/RD. Some implementation notes (these probably need careful review): - token range is extended by 1 bit, since the value range out of this transform is [-16384,16383]. - the coefficients coming out of the FDCT are manually scaled back by 1 bit, or else they won't fit in int16_t (they are 17 bits). Because of this, the RD error scoring does not right-shift the MSE score by two (unlike for 4x4/8x8/16x16). - to compensate for this loss in precision, the quantizer is halved also. This is currently a little hacky. - FDCT and IDCT is double-only right now. Needs a fixed-point impl. - There are no default probabilities for the 32x32 transform yet; I'm simply using the 16x16 luma ones. A future commit will add newly generated probabilities for all transforms. - No ADST version. I don't think we'll add one for this level; if an ADST is desired, transform-size selection can scale back to 16x16 or lower, and use an ADST at that level. Additional notes specific to Debargha's DWT/DCT hybrid: - coefficient scale is different for the top/left 16x16 (DCT-over-DWT) block than for the rest (DWT pixel differences) of the block. Therefore, RD error scoring isn't easily scalable between coefficient and pixel domain. Thus, unfortunately, we need to compute the RD distortion in the pixel domain until we figure out how to scale these appropriately. Change-Id: I00386f20f35d7fabb19aba94c8162f8aee64ef2b
2012-12-07 23:45:05 +01:00
{ cat6, Pcat6, 14, 67},
{ 0, 0, 0, 0}
2010-05-18 17:58:33 +02:00
};
#include "vp9/common/vp9_default_coef_probs.h"
2010-05-18 17:58:33 +02:00
// This function updates and then returns n AC coefficient context
// This is currently a placeholder function to allow experimentation
// using various context models based on the energy earlier tokens
// within the current block.
//
// For now it just returns the previously used context.
#define MAX_NEIGHBORS 2
int vp9_get_coef_context(const int *scan, const int *neighbors,
int nb_pad, uint8_t *token_cache, int c, int l) {
int eob = l;
assert(nb_pad == MAX_NEIGHBORS);
if (c == eob) {
return 0;
} else {
int ctx;
assert(neighbors[MAX_NEIGHBORS * c + 0] >= 0);
if (neighbors[MAX_NEIGHBORS * c + 1] >= 0) {
ctx = (1 + token_cache[scan[neighbors[MAX_NEIGHBORS * c + 0]]] +
token_cache[scan[neighbors[MAX_NEIGHBORS * c + 1]]]) >> 1;
} else {
ctx = token_cache[scan[neighbors[MAX_NEIGHBORS * c + 0]]];
}
return vp9_pt_energy_class[ctx];
}
};
void vp9_default_coef_probs(VP9_COMMON *pc) {
Modeling default coef probs with distribution Replaces the default tables for single coefficient magnitudes with those obtained from an appropriate distribution. The EOB node is left unchanged. The model is represeted as a 256-size codebook where the index corresponds to the probability of the Zero or the One node. Two variations are implemented corresponding to whether the Zero node or the One-node is used as the peg. The main advantage is that the default prob tables will become considerably smaller and manageable. Besides there is substantially less risk of over-fitting for a training set. Various distributions are tried and the one that gives the best results is the family of Generalized Gaussian distributions with shape parameter 0.75. The results are within about 0.2% of fully trained tables for the Zero peg variant, and within 0.1% of the One peg variant. The forward updates are optionally (controlled by a macro) model-based, i.e. restricted to only convey probabilities from the codebook. Backward updates can also be optionally (controlled by another macro) model-based, but is turned off by default. Currently model-based forward updates work about the same as unconstrained updates, but there is a drop in performance with backward-updates being model based. The model based approach also allows the probabilities for the key frames to be adjusted from the defaults based on the base_qindex of the frame. Currently the adjustment function is a placeholder that adjusts the prob of EOB and Zero node from the nominal one at higher quality (lower qindex) or lower quality (higher qindex) ends of the range. The rest of the probabilities are then derived based on the model from the adjusted prob of zero. Change-Id: Iae050f3cbcc6d8b3f204e8dc395ae47b3b2192c9
2013-03-13 19:03:17 +01:00
#if CONFIG_MODELCOEFPROB
int b, r, c, p;
#endif
#if CONFIG_MODELCOEFPROB
for (b = 0; b < BLOCK_TYPES; ++b)
for (r = 0; r < REF_TYPES; ++r)
for (c = 0; c < COEF_BANDS; ++c)
for (p = 0; p < PREV_COEF_CONTEXTS; ++p) {
int t;
for (t = 0; t < UNCONSTRAINED_NODES; t++)
pc->fc.coef_probs_4x4[b][r][c][p][t] =
default_coef_probs_4x4[b][r][c][p][t];
vp9_get_model_distribution(
default_coef_probs_4x4[b][r][c][p][UNCONSTRAINED_NODES - 1],
pc->fc.coef_probs_4x4[b][r][c][p], b, r);
for (t = 0; t < UNCONSTRAINED_NODES; t++)
pc->fc.coef_probs_8x8[b][r][c][p][t] =
default_coef_probs_8x8[b][r][c][p][t];
vp9_get_model_distribution(
default_coef_probs_8x8[b][r][c][p][UNCONSTRAINED_NODES - 1],
pc->fc.coef_probs_8x8[b][r][c][p], b, r);
for (t = 0; t < UNCONSTRAINED_NODES; t++)
pc->fc.coef_probs_16x16[b][r][c][p][t] =
default_coef_probs_16x16[b][r][c][p][t];
vp9_get_model_distribution(
default_coef_probs_16x16[b][r][c][p][UNCONSTRAINED_NODES - 1],
pc->fc.coef_probs_16x16[b][r][c][p], b, r);
for (t = 0; t < UNCONSTRAINED_NODES; t++)
pc->fc.coef_probs_32x32[b][r][c][p][t] =
default_coef_probs_32x32[b][r][c][p][t];
vp9_get_model_distribution(
default_coef_probs_32x32[b][r][c][p][UNCONSTRAINED_NODES - 1],
pc->fc.coef_probs_32x32[b][r][c][p], b, r);
}
#else
vpx_memcpy(pc->fc.coef_probs_4x4, default_coef_probs_4x4,
sizeof(pc->fc.coef_probs_4x4));
vpx_memcpy(pc->fc.coef_probs_8x8, default_coef_probs_8x8,
sizeof(pc->fc.coef_probs_8x8));
vpx_memcpy(pc->fc.coef_probs_16x16, default_coef_probs_16x16,
sizeof(pc->fc.coef_probs_16x16));
vpx_memcpy(pc->fc.coef_probs_32x32, default_coef_probs_32x32,
sizeof(pc->fc.coef_probs_32x32));
Modeling default coef probs with distribution Replaces the default tables for single coefficient magnitudes with those obtained from an appropriate distribution. The EOB node is left unchanged. The model is represeted as a 256-size codebook where the index corresponds to the probability of the Zero or the One node. Two variations are implemented corresponding to whether the Zero node or the One-node is used as the peg. The main advantage is that the default prob tables will become considerably smaller and manageable. Besides there is substantially less risk of over-fitting for a training set. Various distributions are tried and the one that gives the best results is the family of Generalized Gaussian distributions with shape parameter 0.75. The results are within about 0.2% of fully trained tables for the Zero peg variant, and within 0.1% of the One peg variant. The forward updates are optionally (controlled by a macro) model-based, i.e. restricted to only convey probabilities from the codebook. Backward updates can also be optionally (controlled by another macro) model-based, but is turned off by default. Currently model-based forward updates work about the same as unconstrained updates, but there is a drop in performance with backward-updates being model based. The model based approach also allows the probabilities for the key frames to be adjusted from the defaults based on the base_qindex of the frame. Currently the adjustment function is a placeholder that adjusts the prob of EOB and Zero node from the nominal one at higher quality (lower qindex) or lower quality (higher qindex) ends of the range. The rest of the probabilities are then derived based on the model from the adjusted prob of zero. Change-Id: Iae050f3cbcc6d8b3f204e8dc395ae47b3b2192c9
2013-03-13 19:03:17 +01:00
#endif
#if CONFIG_CODE_ZEROGROUP
vpx_memcpy(pc->fc.zpc_probs_4x4, default_zpc_probs_4x4,
sizeof(pc->fc.zpc_probs_4x4));
vpx_memcpy(pc->fc.zpc_probs_8x8, default_zpc_probs_8x8,
sizeof(pc->fc.zpc_probs_8x8));
vpx_memcpy(pc->fc.zpc_probs_16x16, default_zpc_probs_16x16,
sizeof(pc->fc.zpc_probs_16x16));
vpx_memcpy(pc->fc.zpc_probs_32x32, default_zpc_probs_32x32,
sizeof(pc->fc.zpc_probs_32x32));
#endif
}
2010-05-18 17:58:33 +02:00
// Neighborhood 5-tuples for various scans and blocksizes,
// in {top, left, topleft, topright, bottomleft} order
// for each position in raster scan order.
// -1 indicates the neighbor does not exist.
DECLARE_ALIGNED(16, int,
vp9_default_zig_zag1d_4x4_neighbors[16 * MAX_NEIGHBORS]);
DECLARE_ALIGNED(16, int,
vp9_col_scan_4x4_neighbors[16 * MAX_NEIGHBORS]);
DECLARE_ALIGNED(16, int,
vp9_row_scan_4x4_neighbors[16 * MAX_NEIGHBORS]);
DECLARE_ALIGNED(16, int,
vp9_col_scan_8x8_neighbors[64 * MAX_NEIGHBORS]);
DECLARE_ALIGNED(16, int,
vp9_row_scan_8x8_neighbors[64 * MAX_NEIGHBORS]);
DECLARE_ALIGNED(16, int,
vp9_default_zig_zag1d_8x8_neighbors[64 * MAX_NEIGHBORS]);
DECLARE_ALIGNED(16, int,
vp9_col_scan_16x16_neighbors[256 * MAX_NEIGHBORS]);
DECLARE_ALIGNED(16, int,
vp9_row_scan_16x16_neighbors[256 * MAX_NEIGHBORS]);
DECLARE_ALIGNED(16, int,
vp9_default_zig_zag1d_16x16_neighbors[256 * MAX_NEIGHBORS]);
DECLARE_ALIGNED(16, int,
vp9_default_zig_zag1d_32x32_neighbors[1024 * MAX_NEIGHBORS]);
static int find_in_scan(const int *scan, int l, int idx) {
int n, l2 = l * l;
for (n = 0; n < l2; n++) {
int rc = scan[n];
if (rc == idx)
return n;
}
assert(0);
return -1;
}
static void init_scan_neighbors(const int *scan, int l, int *neighbors,
int max_neighbors) {
int l2 = l * l;
int n, i, j;
for (n = 0; n < l2; n++) {
int rc = scan[n];
assert(max_neighbors == MAX_NEIGHBORS);
i = rc / l;
j = rc % l;
if (i > 0 && j > 0) {
// col/row scan is used for adst/dct, and generally means that
// energy decreases to zero much faster in the dimension in
// which ADST is used compared to the direction in which DCT
// is used. Likewise, we find much higher correlation between
// coefficients within the direction in which DCT is used.
// Therefore, if we use ADST/DCT, prefer the DCT neighbor coeff
// as a context. If ADST or DCT is used in both directions, we
// use the combination of the two as a context.
int a = find_in_scan(scan, l, (i - 1) * l + j);
int b = find_in_scan(scan, l, i * l + j - 1);
if (scan == vp9_col_scan_4x4 || scan == vp9_col_scan_8x8 ||
scan == vp9_col_scan_16x16) {
neighbors[max_neighbors * n + 0] = a;
neighbors[max_neighbors * n + 1] = -1;
} else if (scan == vp9_row_scan_4x4 || scan == vp9_row_scan_8x8 ||
scan == vp9_row_scan_16x16) {
neighbors[max_neighbors * n + 0] = b;
neighbors[max_neighbors * n + 1] = -1;
} else {
neighbors[max_neighbors * n + 0] = a;
neighbors[max_neighbors * n + 1] = b;
}
} else if (i > 0) {
neighbors[max_neighbors * n + 0] = find_in_scan(scan, l, (i - 1) * l + j);
neighbors[max_neighbors * n + 1] = -1;
} else if (j > 0) {
neighbors[max_neighbors * n + 0] =
find_in_scan(scan, l, i * l + j - 1);
neighbors[max_neighbors * n + 1] = -1;
} else {
assert(n == 0);
// dc predictor doesn't use previous tokens
neighbors[max_neighbors * n + 0] = -1;
}
assert(neighbors[max_neighbors * n + 0] < n);
}
}
void vp9_init_neighbors() {
init_scan_neighbors(vp9_default_zig_zag1d_4x4, 4,
vp9_default_zig_zag1d_4x4_neighbors, MAX_NEIGHBORS);
init_scan_neighbors(vp9_row_scan_4x4, 4,
vp9_row_scan_4x4_neighbors, MAX_NEIGHBORS);
init_scan_neighbors(vp9_col_scan_4x4, 4,
vp9_col_scan_4x4_neighbors, MAX_NEIGHBORS);
init_scan_neighbors(vp9_default_zig_zag1d_8x8, 8,
vp9_default_zig_zag1d_8x8_neighbors, MAX_NEIGHBORS);
init_scan_neighbors(vp9_row_scan_8x8, 8,
vp9_row_scan_8x8_neighbors, MAX_NEIGHBORS);
init_scan_neighbors(vp9_col_scan_8x8, 8,
vp9_col_scan_8x8_neighbors, MAX_NEIGHBORS);
init_scan_neighbors(vp9_default_zig_zag1d_16x16, 16,
vp9_default_zig_zag1d_16x16_neighbors, MAX_NEIGHBORS);
init_scan_neighbors(vp9_row_scan_16x16, 16,
vp9_row_scan_16x16_neighbors, MAX_NEIGHBORS);
init_scan_neighbors(vp9_col_scan_16x16, 16,
vp9_col_scan_16x16_neighbors, MAX_NEIGHBORS);
init_scan_neighbors(vp9_default_zig_zag1d_32x32, 32,
vp9_default_zig_zag1d_32x32_neighbors, MAX_NEIGHBORS);
}
const int *vp9_get_coef_neighbors_handle(const int *scan, int *pad) {
if (scan == vp9_default_zig_zag1d_4x4) {
*pad = MAX_NEIGHBORS;
return vp9_default_zig_zag1d_4x4_neighbors;
} else if (scan == vp9_row_scan_4x4) {
*pad = MAX_NEIGHBORS;
return vp9_row_scan_4x4_neighbors;
} else if (scan == vp9_col_scan_4x4) {
*pad = MAX_NEIGHBORS;
return vp9_col_scan_4x4_neighbors;
} else if (scan == vp9_default_zig_zag1d_8x8) {
*pad = MAX_NEIGHBORS;
return vp9_default_zig_zag1d_8x8_neighbors;
} else if (scan == vp9_row_scan_8x8) {
*pad = 2;
return vp9_row_scan_8x8_neighbors;
} else if (scan == vp9_col_scan_8x8) {
*pad = 2;
return vp9_col_scan_8x8_neighbors;
} else if (scan == vp9_default_zig_zag1d_16x16) {
*pad = MAX_NEIGHBORS;
return vp9_default_zig_zag1d_16x16_neighbors;
} else if (scan == vp9_row_scan_16x16) {
*pad = 2;
return vp9_row_scan_16x16_neighbors;
} else if (scan == vp9_col_scan_16x16) {
*pad = 2;
return vp9_col_scan_16x16_neighbors;
} else if (scan == vp9_default_zig_zag1d_32x32) {
*pad = MAX_NEIGHBORS;
return vp9_default_zig_zag1d_32x32_neighbors;
} else {
assert(0);
return NULL;
}
}
void vp9_coef_tree_initialize() {
vp9_init_neighbors();
init_bit_trees();
vp9_tokens_from_tree(vp9_coef_encodings, vp9_coef_tree);
}
// #define COEF_COUNT_TESTING
#define COEF_COUNT_SAT 24
#define COEF_MAX_UPDATE_FACTOR 112
#define COEF_COUNT_SAT_KEY 24
#define COEF_MAX_UPDATE_FACTOR_KEY 112
#define COEF_COUNT_SAT_AFTER_KEY 24
#define COEF_MAX_UPDATE_FACTOR_AFTER_KEY 128
static void adapt_coef_probs(vp9_coeff_probs *dst_coef_probs,
vp9_coeff_probs *pre_coef_probs,
int qindex, vp9_coeff_count *coef_counts,
unsigned int (*eob_branch_count)[REF_TYPES]
[COEF_BANDS]
[PREV_COEF_CONTEXTS],
int count_sat, int update_factor) {
int t, i, j, k, l, count;
unsigned int branch_ct[ENTROPY_NODES][2];
vp9_prob coef_probs[ENTROPY_NODES];
int factor;
Modeling default coef probs with distribution Replaces the default tables for single coefficient magnitudes with those obtained from an appropriate distribution. The EOB node is left unchanged. The model is represeted as a 256-size codebook where the index corresponds to the probability of the Zero or the One node. Two variations are implemented corresponding to whether the Zero node or the One-node is used as the peg. The main advantage is that the default prob tables will become considerably smaller and manageable. Besides there is substantially less risk of over-fitting for a training set. Various distributions are tried and the one that gives the best results is the family of Generalized Gaussian distributions with shape parameter 0.75. The results are within about 0.2% of fully trained tables for the Zero peg variant, and within 0.1% of the One peg variant. The forward updates are optionally (controlled by a macro) model-based, i.e. restricted to only convey probabilities from the codebook. Backward updates can also be optionally (controlled by another macro) model-based, but is turned off by default. Currently model-based forward updates work about the same as unconstrained updates, but there is a drop in performance with backward-updates being model based. The model based approach also allows the probabilities for the key frames to be adjusted from the defaults based on the base_qindex of the frame. Currently the adjustment function is a placeholder that adjusts the prob of EOB and Zero node from the nominal one at higher quality (lower qindex) or lower quality (higher qindex) ends of the range. The rest of the probabilities are then derived based on the model from the adjusted prob of zero. Change-Id: Iae050f3cbcc6d8b3f204e8dc395ae47b3b2192c9
2013-03-13 19:03:17 +01:00
#if CONFIG_MODELCOEFPROB && MODEL_BASED_ADAPT
int entropy_nodes_adapt = UNCONSTRAINED_ADAPT_NODES;
#else
int entropy_nodes_adapt = ENTROPY_NODES;
#endif
for (i = 0; i < BLOCK_TYPES; ++i)
for (j = 0; j < REF_TYPES; ++j)
for (k = 0; k < COEF_BANDS; ++k)
for (l = 0; l < PREV_COEF_CONTEXTS; ++l) {
if (l >= 3 && k == 0)
continue;
vp9_tree_probs_from_distribution(vp9_coef_tree,
coef_probs, branch_ct,
coef_counts[i][j][k][l], 0);
branch_ct[0][1] = eob_branch_count[i][j][k][l] - branch_ct[0][0];
coef_probs[0] = get_binary_prob(branch_ct[0][0], branch_ct[0][1]);
Modeling default coef probs with distribution Replaces the default tables for single coefficient magnitudes with those obtained from an appropriate distribution. The EOB node is left unchanged. The model is represeted as a 256-size codebook where the index corresponds to the probability of the Zero or the One node. Two variations are implemented corresponding to whether the Zero node or the One-node is used as the peg. The main advantage is that the default prob tables will become considerably smaller and manageable. Besides there is substantially less risk of over-fitting for a training set. Various distributions are tried and the one that gives the best results is the family of Generalized Gaussian distributions with shape parameter 0.75. The results are within about 0.2% of fully trained tables for the Zero peg variant, and within 0.1% of the One peg variant. The forward updates are optionally (controlled by a macro) model-based, i.e. restricted to only convey probabilities from the codebook. Backward updates can also be optionally (controlled by another macro) model-based, but is turned off by default. Currently model-based forward updates work about the same as unconstrained updates, but there is a drop in performance with backward-updates being model based. The model based approach also allows the probabilities for the key frames to be adjusted from the defaults based on the base_qindex of the frame. Currently the adjustment function is a placeholder that adjusts the prob of EOB and Zero node from the nominal one at higher quality (lower qindex) or lower quality (higher qindex) ends of the range. The rest of the probabilities are then derived based on the model from the adjusted prob of zero. Change-Id: Iae050f3cbcc6d8b3f204e8dc395ae47b3b2192c9
2013-03-13 19:03:17 +01:00
for (t = 0; t < entropy_nodes_adapt; ++t) {
count = branch_ct[t][0] + branch_ct[t][1];
count = count > count_sat ? count_sat : count;
factor = (update_factor * count / count_sat);
dst_coef_probs[i][j][k][l][t] =
weighted_prob(pre_coef_probs[i][j][k][l][t],
coef_probs[t], factor);
Modeling default coef probs with distribution Replaces the default tables for single coefficient magnitudes with those obtained from an appropriate distribution. The EOB node is left unchanged. The model is represeted as a 256-size codebook where the index corresponds to the probability of the Zero or the One node. Two variations are implemented corresponding to whether the Zero node or the One-node is used as the peg. The main advantage is that the default prob tables will become considerably smaller and manageable. Besides there is substantially less risk of over-fitting for a training set. Various distributions are tried and the one that gives the best results is the family of Generalized Gaussian distributions with shape parameter 0.75. The results are within about 0.2% of fully trained tables for the Zero peg variant, and within 0.1% of the One peg variant. The forward updates are optionally (controlled by a macro) model-based, i.e. restricted to only convey probabilities from the codebook. Backward updates can also be optionally (controlled by another macro) model-based, but is turned off by default. Currently model-based forward updates work about the same as unconstrained updates, but there is a drop in performance with backward-updates being model based. The model based approach also allows the probabilities for the key frames to be adjusted from the defaults based on the base_qindex of the frame. Currently the adjustment function is a placeholder that adjusts the prob of EOB and Zero node from the nominal one at higher quality (lower qindex) or lower quality (higher qindex) ends of the range. The rest of the probabilities are then derived based on the model from the adjusted prob of zero. Change-Id: Iae050f3cbcc6d8b3f204e8dc395ae47b3b2192c9
2013-03-13 19:03:17 +01:00
#if CONFIG_MODELCOEFPROB && MODEL_BASED_ADAPT
if (t == UNCONSTRAINED_NODES - 1)
vp9_get_model_distribution(
dst_coef_probs[i][j][k][l][UNCONSTRAINED_NODES - 1],
dst_coef_probs[i][j][k][l], i, j);
#endif
}
}
}
void vp9_adapt_coef_probs(VP9_COMMON *cm) {
#ifndef DISABLE_COEF_ADAPT
int count_sat;
int update_factor; /* denominator 256 */
if (cm->frame_type == KEY_FRAME) {
update_factor = COEF_MAX_UPDATE_FACTOR_KEY;
count_sat = COEF_COUNT_SAT_KEY;
} else if (cm->last_frame_type == KEY_FRAME) {
update_factor = COEF_MAX_UPDATE_FACTOR_AFTER_KEY; /* adapt quickly */
count_sat = COEF_COUNT_SAT_AFTER_KEY;
} else {
update_factor = COEF_MAX_UPDATE_FACTOR;
count_sat = COEF_COUNT_SAT;
}
adapt_coef_probs(cm->fc.coef_probs_4x4, cm->fc.pre_coef_probs_4x4,
cm->base_qindex, cm->fc.coef_counts_4x4,
cm->fc.eob_branch_counts[TX_4X4],
count_sat, update_factor);
adapt_coef_probs(cm->fc.coef_probs_8x8, cm->fc.pre_coef_probs_8x8,
cm->base_qindex, cm->fc.coef_counts_8x8,
cm->fc.eob_branch_counts[TX_8X8],
count_sat, update_factor);
adapt_coef_probs(cm->fc.coef_probs_16x16, cm->fc.pre_coef_probs_16x16,
cm->base_qindex, cm->fc.coef_counts_16x16,
cm->fc.eob_branch_counts[TX_16X16],
count_sat, update_factor);
adapt_coef_probs(cm->fc.coef_probs_32x32, cm->fc.pre_coef_probs_32x32,
cm->base_qindex, cm->fc.coef_counts_32x32,
cm->fc.eob_branch_counts[TX_32X32],
count_sat, update_factor);
#endif
}
#if CONFIG_CODE_ZEROGROUP
OrientationType vp9_get_orientation(int rc, TX_SIZE tx_size) {
int i = rc >> (tx_size + 2);
int j = rc & ((4 << tx_size) - 1);
if (i > 2 * j)
return VERTICAL;
else if (j > 2 * i)
return HORIZONTAL;
else
return DIAGONAL;
/*
if (i == 0 && j == 0) return DIAGONAL;
while (i > 1 || j > 1) {
i >>= 1;
j >>= 1;
}
if (i == 0 && j == 1)
return HORIZONTAL; // horizontal
else if (i == 1 && j == 1)
return DIAGONAL; // diagonal
else if (i == 1 && j == 0)
return VERTICAL; // vertical
assert(0);
*/
}
int vp9_use_eoo(int c, int seg_eob, const int *scan,
TX_SIZE tx_size, int *is_last_zero, int *is_eoo) {
// NOTE: returning 0 from this function will turn off eoo symbols
// For instance we can experiment with turning eoo off for smaller blocks
// and/or lower bands
int o = vp9_get_orientation(scan[c], tx_size);
int band = get_coef_band(scan, tx_size, c);
int use_eoo = (!is_last_zero[o] &&
!is_eoo[o] &&
band <= ZPC_EOO_BAND_UPPER &&
band >= ZPC_EOO_BAND_LOWER &&
get_zpc_used(tx_size) &&
seg_eob - c > (ZPC_USEEOO_THRESH << tx_size) &&
is_eoo[0] + is_eoo[1] + is_eoo[2] < 2);
return use_eoo;
}
int vp9_is_eoo(int c, int eob, const int *scan, TX_SIZE tx_size,
const int16_t *qcoeff_ptr, int *last_nz_pos) {
int rc = scan[c];
int o = vp9_get_orientation(rc, tx_size);
int eoo = c > last_nz_pos[o];
return eoo;
}
static void adapt_zpc_probs_common(VP9_COMMON *cm,
TX_SIZE tx_size,
int count_sat,
int update_factor) {
int r, b, p, n;
int count, factor;
vp9_zpc_probs *zpc_probs;
vp9_zpc_probs *pre_zpc_probs;
vp9_zpc_count *zpc_counts;
if (!get_zpc_used(tx_size)) return;
if (tx_size == TX_32X32) {
zpc_probs = &cm->fc.zpc_probs_32x32;
pre_zpc_probs = &cm->fc.pre_zpc_probs_32x32;
zpc_counts = &cm->fc.zpc_counts_32x32;
} else if (tx_size == TX_16X16) {
zpc_probs = &cm->fc.zpc_probs_16x16;
pre_zpc_probs = &cm->fc.pre_zpc_probs_16x16;
zpc_counts = &cm->fc.zpc_counts_16x16;
} else if (tx_size == TX_8X8) {
zpc_probs = &cm->fc.zpc_probs_8x8;
pre_zpc_probs = &cm->fc.pre_zpc_probs_8x8;
zpc_counts = &cm->fc.zpc_counts_8x8;
} else {
zpc_probs = &cm->fc.zpc_probs_4x4;
pre_zpc_probs = &cm->fc.pre_zpc_probs_4x4;
zpc_counts = &cm->fc.zpc_counts_4x4;
}
for (r = 0; r < REF_TYPES; ++r) {
for (b = 0; b < ZPC_BANDS; ++b) {
for (p = 0; p < ZPC_PTOKS; ++p) {
for (n = 0; n < ZPC_NODES; ++n) {
vp9_prob prob = get_binary_prob((*zpc_counts)[r][b][p][n][0],
(*zpc_counts)[r][b][p][n][1]);
count = (*zpc_counts)[r][b][p][n][0] + (*zpc_counts)[r][b][p][n][1];
count = count > count_sat ? count_sat : count;
factor = (update_factor * count / count_sat);
(*zpc_probs)[r][b][p][n] = weighted_prob(
(*pre_zpc_probs)[r][b][p][n], prob, factor);
}
}
}
}
}
// #define ZPC_COUNT_TESTING
void vp9_adapt_zpc_probs(VP9_COMMON *cm) {
int count_sat;
int update_factor; /* denominator 256 */
if (cm->frame_type == KEY_FRAME) {
update_factor = COEF_MAX_UPDATE_FACTOR_KEY;
count_sat = COEF_COUNT_SAT_KEY;
} else if (cm->last_frame_type == KEY_FRAME) {
update_factor = COEF_MAX_UPDATE_FACTOR_AFTER_KEY; /* adapt quickly */
count_sat = COEF_COUNT_SAT_AFTER_KEY;
} else {
update_factor = COEF_MAX_UPDATE_FACTOR;
count_sat = COEF_COUNT_SAT;
}
adapt_zpc_probs_common(cm, TX_4X4, count_sat, update_factor);
adapt_zpc_probs_common(cm, TX_8X8, count_sat, update_factor);
adapt_zpc_probs_common(cm, TX_16X16, count_sat, update_factor);
adapt_zpc_probs_common(cm, TX_32X32, count_sat, update_factor);
}
#endif // CONFIG_CODE_ZEROGROUP