Line data Source code
1 : /*
2 : * Copyright (c) 2016, Alliance for Open Media. All rights reserved
3 : *
4 : * This source code is subject to the terms of the BSD 2 Clause License and
5 : * the Alliance for Open Media Patent License 1.0. If the BSD 2 Clause License
6 : * was not distributed with this source code in the LICENSE file, you can
7 : * obtain it at www.aomedia.org/license/software. If the Alliance for Open
8 : * Media Patent License 1.0 was not distributed with this source code in the
9 : * PATENTS file, you can obtain it at www.aomedia.org/license/patent.
10 : */
11 :
12 : /*!\file
13 : * \brief Describes film grain parameters and film grain synthesis
14 : *
15 : */
16 :
17 : #include <stdio.h>
18 : #include <string.h>
19 : #include <stdlib.h>
20 : #include "EbDefinitions.h"
21 : #include "grainSynthesis.h"
22 :
23 : // Samples with Gaussian distribution in the range of [-2048, 2047] (12 bits)
24 : // with zero mean and standard deviation of about 512.
25 : // should be divided by 4 for 10-bit range and 16 for 8-bit range.
26 : static const int32_t gaussian_sequence[2048] = {
27 : 56, 568, -180, 172, 124, -84, 172, -64, -900, 24, 820,
28 : 224, 1248, 996, 272, -8, -916, -388, -732, -104, -188, 800,
29 : 112, -652, -320, -376, 140, -252, 492, -168, 44, -788, 588,
30 : -584, 500, -228, 12, 680, 272, -476, 972, -100, 652, 368,
31 : 432, -196, -720, -192, 1000, -332, 652, -136, -552, -604, -4,
32 : 192, -220, -136, 1000, -52, 372, -96, -624, 124, -24, 396,
33 : 540, -12, -104, 640, 464, 244, -208, -84, 368, -528, -740,
34 : 248, -968, -848, 608, 376, -60, -292, -40, -156, 252, -292,
35 : 248, 224, -280, 400, -244, 244, -60, 76, -80, 212, 532,
36 : 340, 128, -36, 824, -352, -60, -264, -96, -612, 416, -704,
37 : 220, -204, 640, -160, 1220, -408, 900, 336, 20, -336, -96,
38 : -792, 304, 48, -28, -1232, -1172, -448, 104, -292, -520, 244,
39 : 60, -948, 0, -708, 268, 108, 356, -548, 488, -344, -136,
40 : 488, -196, -224, 656, -236, -1128, 60, 4, 140, 276, -676,
41 : -376, 168, -108, 464, 8, 564, 64, 240, 308, -300, -400,
42 : -456, -136, 56, 120, -408, -116, 436, 504, -232, 328, 844,
43 : -164, -84, 784, -168, 232, -224, 348, -376, 128, 568, 96,
44 : -1244, -288, 276, 848, 832, -360, 656, 464, -384, -332, -356,
45 : 728, -388, 160, -192, 468, 296, 224, 140, -776, -100, 280,
46 : 4, 196, 44, -36, -648, 932, 16, 1428, 28, 528, 808,
47 : 772, 20, 268, 88, -332, -284, 124, -384, -448, 208, -228,
48 : -1044, -328, 660, 380, -148, -300, 588, 240, 540, 28, 136,
49 : -88, -436, 256, 296, -1000, 1400, 0, -48, 1056, -136, 264,
50 : -528, -1108, 632, -484, -592, -344, 796, 124, -668, -768, 388,
51 : 1296, -232, -188, -200, -288, -4, 308, 100, -168, 256, -500,
52 : 204, -508, 648, -136, 372, -272, -120, -1004, -552, -548, -384,
53 : 548, -296, 428, -108, -8, -912, -324, -224, -88, -112, -220,
54 : -100, 996, -796, 548, 360, -216, 180, 428, -200, -212, 148,
55 : 96, 148, 284, 216, -412, -320, 120, -300, -384, -604, -572,
56 : -332, -8, -180, -176, 696, 116, -88, 628, 76, 44, -516,
57 : 240, -208, -40, 100, -592, 344, -308, -452, -228, 20, 916,
58 : -1752, -136, -340, -804, 140, 40, 512, 340, 248, 184, -492,
59 : 896, -156, 932, -628, 328, -688, -448, -616, -752, -100, 560,
60 : -1020, 180, -800, -64, 76, 576, 1068, 396, 660, 552, -108,
61 : -28, 320, -628, 312, -92, -92, -472, 268, 16, 560, 516,
62 : -672, -52, 492, -100, 260, 384, 284, 292, 304, -148, 88,
63 : -152, 1012, 1064, -228, 164, -376, -684, 592, -392, 156, 196,
64 : -524, -64, -884, 160, -176, 636, 648, 404, -396, -436, 864,
65 : 424, -728, 988, -604, 904, -592, 296, -224, 536, -176, -920,
66 : 436, -48, 1176, -884, 416, -776, -824, -884, 524, -548, -564,
67 : -68, -164, -96, 692, 364, -692, -1012, -68, 260, -480, 876,
68 : -1116, 452, -332, -352, 892, -1088, 1220, -676, 12, -292, 244,
69 : 496, 372, -32, 280, 200, 112, -440, -96, 24, -644, -184,
70 : 56, -432, 224, -980, 272, -260, 144, -436, 420, 356, 364,
71 : -528, 76, 172, -744, -368, 404, -752, -416, 684, -688, 72,
72 : 540, 416, 92, 444, 480, -72, -1416, 164, -1172, -68, 24,
73 : 424, 264, 1040, 128, -912, -524, -356, 64, 876, -12, 4,
74 : -88, 532, 272, -524, 320, 276, -508, 940, 24, -400, -120,
75 : 756, 60, 236, -412, 100, 376, -484, 400, -100, -740, -108,
76 : -260, 328, -268, 224, -200, -416, 184, -604, -564, -20, 296,
77 : 60, 892, -888, 60, 164, 68, -760, 216, -296, 904, -336,
78 : -28, 404, -356, -568, -208, -1480, -512, 296, 328, -360, -164,
79 : -1560, -776, 1156, -428, 164, -504, -112, 120, -216, -148, -264,
80 : 308, 32, 64, -72, 72, 116, 176, -64, -272, 460, -536,
81 : -784, -280, 348, 108, -752, -132, 524, -540, -776, 116, -296,
82 : -1196, -288, -560, 1040, -472, 116, -848, -1116, 116, 636, 696,
83 : 284, -176, 1016, 204, -864, -648, -248, 356, 972, -584, -204,
84 : 264, 880, 528, -24, -184, 116, 448, -144, 828, 524, 212,
85 : -212, 52, 12, 200, 268, -488, -404, -880, 824, -672, -40,
86 : 908, -248, 500, 716, -576, 492, -576, 16, 720, -108, 384,
87 : 124, 344, 280, 576, -500, 252, 104, -308, 196, -188, -8,
88 : 1268, 296, 1032, -1196, 436, 316, 372, -432, -200, -660, 704,
89 : -224, 596, -132, 268, 32, -452, 884, 104, -1008, 424, -1348,
90 : -280, 4, -1168, 368, 476, 696, 300, -8, 24, 180, -592,
91 : -196, 388, 304, 500, 724, -160, 244, -84, 272, -256, -420,
92 : 320, 208, -144, -156, 156, 364, 452, 28, 540, 316, 220,
93 : -644, -248, 464, 72, 360, 32, -388, 496, -680, -48, 208,
94 : -116, -408, 60, -604, -392, 548, -840, 784, -460, 656, -544,
95 : -388, -264, 908, -800, -628, -612, -568, 572, -220, 164, 288,
96 : -16, -308, 308, -112, -636, -760, 280, -668, 432, 364, 240,
97 : -196, 604, 340, 384, 196, 592, -44, -500, 432, -580, -132,
98 : 636, -76, 392, 4, -412, 540, 508, 328, -356, -36, 16,
99 : -220, -64, -248, -60, 24, -192, 368, 1040, 92, -24, -1044,
100 : -32, 40, 104, 148, 192, -136, -520, 56, -816, -224, 732,
101 : 392, 356, 212, -80, -424, -1008, -324, 588, -1496, 576, 460,
102 : -816, -848, 56, -580, -92, -1372, -112, -496, 200, 364, 52,
103 : -140, 48, -48, -60, 84, 72, 40, 132, -356, -268, -104,
104 : -284, -404, 732, -520, 164, -304, -540, 120, 328, -76, -460,
105 : 756, 388, 588, 236, -436, -72, -176, -404, -316, -148, 716,
106 : -604, 404, -72, -88, -888, -68, 944, 88, -220, -344, 960,
107 : 472, 460, -232, 704, 120, 832, -228, 692, -508, 132, -476,
108 : 844, -748, -364, -44, 1116, -1104, -1056, 76, 428, 552, -692,
109 : 60, 356, 96, -384, -188, -612, -576, 736, 508, 892, 352,
110 : -1132, 504, -24, -352, 324, 332, -600, -312, 292, 508, -144,
111 : -8, 484, 48, 284, -260, -240, 256, -100, -292, -204, -44,
112 : 472, -204, 908, -188, -1000, -256, 92, 1164, -392, 564, 356,
113 : 652, -28, -884, 256, 484, -192, 760, -176, 376, -524, -452,
114 : -436, 860, -736, 212, 124, 504, -476, 468, 76, -472, 552,
115 : -692, -944, -620, 740, -240, 400, 132, 20, 192, -196, 264,
116 : -668, -1012, -60, 296, -316, -828, 76, -156, 284, -768, -448,
117 : -832, 148, 248, 652, 616, 1236, 288, -328, -400, -124, 588,
118 : 220, 520, -696, 1032, 768, -740, -92, -272, 296, 448, -464,
119 : 412, -200, 392, 440, -200, 264, -152, -260, 320, 1032, 216,
120 : 320, -8, -64, 156, -1016, 1084, 1172, 536, 484, -432, 132,
121 : 372, -52, -256, 84, 116, -352, 48, 116, 304, -384, 412,
122 : 924, -300, 528, 628, 180, 648, 44, -980, -220, 1320, 48,
123 : 332, 748, 524, -268, -720, 540, -276, 564, -344, -208, -196,
124 : 436, 896, 88, -392, 132, 80, -964, -288, 568, 56, -48,
125 : -456, 888, 8, 552, -156, -292, 948, 288, 128, -716, -292,
126 : 1192, -152, 876, 352, -600, -260, -812, -468, -28, -120, -32,
127 : -44, 1284, 496, 192, 464, 312, -76, -516, -380, -456, -1012,
128 : -48, 308, -156, 36, 492, -156, -808, 188, 1652, 68, -120,
129 : -116, 316, 160, -140, 352, 808, -416, 592, 316, -480, 56,
130 : 528, -204, -568, 372, -232, 752, -344, 744, -4, 324, -416,
131 : -600, 768, 268, -248, -88, -132, -420, -432, 80, -288, 404,
132 : -316, -1216, -588, 520, -108, 92, -320, 368, -480, -216, -92,
133 : 1688, -300, 180, 1020, -176, 820, -68, -228, -260, 436, -904,
134 : 20, 40, -508, 440, -736, 312, 332, 204, 760, -372, 728,
135 : 96, -20, -632, -520, -560, 336, 1076, -64, -532, 776, 584,
136 : 192, 396, -728, -520, 276, -188, 80, -52, -612, -252, -48,
137 : 648, 212, -688, 228, -52, -260, 428, -412, -272, -404, 180,
138 : 816, -796, 48, 152, 484, -88, -216, 988, 696, 188, -528,
139 : 648, -116, -180, 316, 476, 12, -564, 96, 476, -252, -364,
140 : -376, -392, 556, -256, -576, 260, -352, 120, -16, -136, -260,
141 : -492, 72, 556, 660, 580, 616, 772, 436, 424, -32, -324,
142 : -1268, 416, -324, -80, 920, 160, 228, 724, 32, -516, 64,
143 : 384, 68, -128, 136, 240, 248, -204, -68, 252, -932, -120,
144 : -480, -628, -84, 192, 852, -404, -288, -132, 204, 100, 168,
145 : -68, -196, -868, 460, 1080, 380, -80, 244, 0, 484, -888,
146 : 64, 184, 352, 600, 460, 164, 604, -196, 320, -64, 588,
147 : -184, 228, 12, 372, 48, -848, -344, 224, 208, -200, 484,
148 : 128, -20, 272, -468, -840, 384, 256, -720, -520, -464, -580,
149 : 112, -120, 644, -356, -208, -608, -528, 704, 560, -424, 392,
150 : 828, 40, 84, 200, -152, 0, -144, 584, 280, -120, 80,
151 : -556, -972, -196, -472, 724, 80, 168, -32, 88, 160, -688,
152 : 0, 160, 356, 372, -776, 740, -128, 676, -248, -480, 4,
153 : -364, 96, 544, 232, -1032, 956, 236, 356, 20, -40, 300,
154 : 24, -676, -596, 132, 1120, -104, 532, -1096, 568, 648, 444,
155 : 508, 380, 188, -376, -604, 1488, 424, 24, 756, -220, -192,
156 : 716, 120, 920, 688, 168, 44, -460, 568, 284, 1144, 1160,
157 : 600, 424, 888, 656, -356, -320, 220, 316, -176, -724, -188,
158 : -816, -628, -348, -228, -380, 1012, -452, -660, 736, 928, 404,
159 : -696, -72, -268, -892, 128, 184, -344, -780, 360, 336, 400,
160 : 344, 428, 548, -112, 136, -228, -216, -820, -516, 340, 92,
161 : -136, 116, -300, 376, -244, 100, -316, -520, -284, -12, 824,
162 : 164, -548, -180, -128, 116, -924, -828, 268, -368, -580, 620,
163 : 192, 160, 0, -1676, 1068, 424, -56, -360, 468, -156, 720,
164 : 288, -528, 556, -364, 548, -148, 504, 316, 152, -648, -620,
165 : -684, -24, -376, -384, -108, -920, -1032, 768, 180, -264, -508,
166 : -1268, -260, -60, 300, -240, 988, 724, -376, -576, -212, -736,
167 : 556, 192, 1092, -620, -880, 376, -56, -4, -216, -32, 836,
168 : 268, 396, 1332, 864, -600, 100, 56, -412, -92, 356, 180,
169 : 884, -468, -436, 292, -388, -804, -704, -840, 368, -348, 140,
170 : -724, 1536, 940, 372, 112, -372, 436, -480, 1136, 296, -32,
171 : -228, 132, -48, -220, 868, -1016, -60, -1044, -464, 328, 916,
172 : 244, 12, -736, -296, 360, 468, -376, -108, -92, 788, 368,
173 : -56, 544, 400, -672, -420, 728, 16, 320, 44, -284, -380,
174 : -796, 488, 132, 204, -596, -372, 88, -152, -908, -636, -572,
175 : -624, -116, -692, -200, -56, 276, -88, 484, -324, 948, 864,
176 : 1000, -456, -184, -276, 292, -296, 156, 676, 320, 160, 908,
177 : -84, -1236, -288, -116, 260, -372, -644, 732, -756, -96, 84,
178 : 344, -520, 348, -688, 240, -84, 216, -1044, -136, -676, -396,
179 : -1500, 960, -40, 176, 168, 1516, 420, -504, -344, -364, -360,
180 : 1216, -940, -380, -212, 252, -660, -708, 484, -444, -152, 928,
181 : -120, 1112, 476, -260, 560, -148, -344, 108, -196, 228, -288,
182 : 504, 560, -328, -88, 288, -1008, 460, -228, 468, -836, -196,
183 : 76, 388, 232, 412, -1168, -716, -644, 756, -172, -356, -504,
184 : 116, 432, 528, 48, 476, -168, -608, 448, 160, -532, -272,
185 : 28, -676, -12, 828, 980, 456, 520, 104, -104, 256, -344,
186 : -4, -28, -368, -52, -524, -572, -556, -200, 768, 1124, -208,
187 : -512, 176, 232, 248, -148, -888, 604, -600, -304, 804, -156,
188 : -212, 488, -192, -804, -256, 368, -360, -916, -328, 228, -240,
189 : -448, -472, 856, -556, -364, 572, -12, -156, -368, -340, 432,
190 : 252, -752, -152, 288, 268, -580, -848, -592, 108, -76, 244,
191 : 312, -716, 592, -80, 436, 360, 4, -248, 160, 516, 584,
192 : 732, 44, -468, -280, -292, -156, -588, 28, 308, 912, 24,
193 : 124, 156, 180, -252, 944, -924, -772, -520, -428, -624, 300,
194 : -212, -1144, 32, -724, 800, -1128, -212, -1288, -848, 180, -416,
195 : 440, 192, -576, -792, -76, -1080, 80, -532, -352, -132, 380,
196 : -820, 148, 1112, 128, 164, 456, 700, -924, 144, -668, -384,
197 : 648, -832, 508, 552, -52, -100, -656, 208, -568, 748, -88,
198 : 680, 232, 300, 192, -408, -1012, -152, -252, -268, 272, -876,
199 : -664, -648, -332, -136, 16, 12, 1152, -28, 332, -536, 320,
200 : -672, -460, -316, 532, -260, 228, -40, 1052, -816, 180, 88,
201 : -496, -556, -672, -368, 428, 92, 356, 404, -408, 252, 196,
202 : -176, -556, 792, 268, 32, 372, 40, 96, -332, 328, 120,
203 : 372, -900, -40, 472, -264, -592, 952, 128, 656, 112, 664,
204 : -232, 420, 4, -344, -464, 556, 244, -416, -32, 252, 0,
205 : -412, 188, -696, 508, -476, 324, -1096, 656, -312, 560, 264,
206 : -136, 304, 160, -64, -580, 248, 336, -720, 560, -348, -288,
207 : -276, -196, -500, 852, -544, -236, -1128, -992, -776, 116, 56,
208 : 52, 860, 884, 212, -12, 168, 1020, 512, -552, 924, -148,
209 : 716, 188, 164, -340, -520, -184, 880, -152, -680, -208, -1156,
210 : -300, -528, -472, 364, 100, -744, -1056, -32, 540, 280, 144,
211 : -676, -32, -232, -280, -224, 96, 568, -76, 172, 148, 148,
212 : 104, 32, -296, -32, 788, -80, 32, -16, 280, 288, 944,
213 : 428, -484
214 : };
215 :
216 : static const int32_t gauss_bits = 11;
217 :
218 : static int32_t luma_subblock_size_y = 32;
219 : static int32_t luma_subblock_size_x = 32;
220 :
221 : static int32_t chroma_subblock_size_y = 16;
222 : static int32_t chroma_subblock_size_x = 16;
223 :
224 : static const int32_t min_luma_legal_range = 16;
225 : static const int32_t max_luma_legal_range = 235;
226 :
227 : static const int32_t min_chroma_legal_range = 16;
228 : static const int32_t max_chroma_legal_range = 240;
229 :
230 : static int32_t scaling_lut_y[256];
231 : static int32_t scaling_lut_cb[256];
232 : static int32_t scaling_lut_cr[256];
233 :
234 : static int32_t grain_center;
235 : static int32_t grain_min;
236 : static int32_t grain_max;
237 :
238 : static uint16_t random_register = 0; // random number generator register
239 :
240 : //----------------------------------------------------------------------
241 : // todo: aomlib memory functions (to be replaced by Eb functions)
242 : /*
243 : #define ADDRESS_STORAGE_SIZE sizeof(size_t)
244 : #define DEFAULT_ALIGNMENT (2 * sizeof(void *))
245 : #define AOM_MAX_ALLOCABLE_MEMORY 8589934592 // 8 GB
246 : //returns an addr aligned to the byte boundary specified by align
247 : #define align_addr(addr, align) \
248 : (void *)(((size_t)(addr) + ((align)-1)) & ~(size_t)((align)-1))
249 :
250 : // Returns 0 in case of overflow of nmemb * size.
251 : static int32_t check_size_argument_overflow(uint64_t nmemb, uint64_t size) {
252 : const uint64_t total_size = nmemb * size;
253 : if (nmemb == 0) return 1;
254 : if (size > AOM_MAX_ALLOCABLE_MEMORY / nmemb) return 0;
255 : if (total_size != (size_t)total_size) return 0;
256 : return 1;
257 : }
258 :
259 : static size_t GetAlignedMallocSize(size_t size, size_t align) {
260 : return size + align - 1 + ADDRESS_STORAGE_SIZE;
261 : }
262 :
263 : static size_t *GetMallocAddressLocation(void *const mem) {
264 : return ((size_t *)mem) - 1;
265 : }
266 :
267 : static void SetActualMallocAddress(void *const mem,
268 : const void *const malloc_addr) {
269 : size_t *const malloc_addr_location = GetMallocAddressLocation(mem);
270 : *malloc_addr_location = (size_t)malloc_addr;
271 : }
272 :
273 : static void *GetActualMallocAddress(void *const mem) {
274 : const size_t *const malloc_addr_location = GetMallocAddressLocation(mem);
275 : return (void *)(*malloc_addr_location);
276 : }
277 :
278 : void *eb_aom_memalign(size_t align, size_t size) {
279 : void *x = NULL;
280 : const size_t aligned_size = GetAlignedMallocSize(size, align);
281 : #if defined(AOM_MAX_ALLOCABLE_MEMORY)
282 : if (!check_size_argument_overflow(1, aligned_size)) return NULL;
283 : #endif
284 : void *const addr = malloc(aligned_size);
285 : if (addr) {
286 : x = align_addr((uint8_t *)addr + ADDRESS_STORAGE_SIZE, align);
287 : SetActualMallocAddress(x, addr);
288 : }
289 : return x;
290 : }
291 :
292 : void *eb_aom_malloc(size_t size) { return eb_aom_memalign(DEFAULT_ALIGNMENT, size); }
293 :
294 : void eb_aom_free(void *memblk) {
295 : if (memblk) {
296 : void *addr = GetActualMallocAddress(memblk);
297 : free(addr);
298 : }
299 : }
300 : */
301 : //--------------------------------------------------------------------
302 :
303 0 : static void init_arrays(aom_film_grain_t *params, int32_t luma_stride,
304 : int32_t chroma_stride, int32_t ***pred_pos_luma_p,
305 : int32_t ***pred_pos_chroma_p, int32_t **luma_grain_block,
306 : int32_t **cb_grain_block, int32_t **cr_grain_block,
307 : int32_t **y_line_buf, int32_t **cb_line_buf, int32_t **cr_line_buf,
308 : int32_t **y_col_buf, int32_t **cb_col_buf, int32_t **cr_col_buf,
309 : int32_t luma_grain_samples, int32_t chroma_grain_samples,
310 : int32_t chroma_subsamp_y, int32_t chroma_subsamp_x) {
311 0 : memset(scaling_lut_y, 0, sizeof(*scaling_lut_y) * 256);
312 0 : memset(scaling_lut_cb, 0, sizeof(*scaling_lut_cb) * 256);
313 0 : memset(scaling_lut_cr, 0, sizeof(*scaling_lut_cr) * 256);
314 :
315 0 : int32_t num_pos_luma = 2 * params->ar_coeff_lag * (params->ar_coeff_lag + 1);
316 0 : int32_t num_pos_chroma = num_pos_luma;
317 0 : if (params->num_y_points > 0) ++num_pos_chroma;
318 :
319 : int32_t **pred_pos_luma;
320 : int32_t **pred_pos_chroma;
321 :
322 0 : pred_pos_luma = (int32_t**)malloc(sizeof(*pred_pos_luma) * num_pos_luma);
323 : ASSERT(pred_pos_luma != NULL);
324 0 : for (int32_t row = 0; row < num_pos_luma; row++) {
325 0 : pred_pos_luma[row] = (int32_t *)malloc(sizeof(**pred_pos_luma) * 3);
326 : ASSERT(pred_pos_luma[row]);
327 : }
328 :
329 : pred_pos_chroma =
330 0 : (int32_t **)malloc(sizeof(*pred_pos_chroma) * num_pos_chroma);
331 : ASSERT(pred_pos_chroma != NULL);
332 0 : for (int32_t row = 0; row < num_pos_chroma; row++) {
333 0 : pred_pos_chroma[row] = (int32_t *)malloc(sizeof(**pred_pos_chroma) * 3);
334 : ASSERT(pred_pos_chroma[row]);
335 : }
336 :
337 0 : int32_t pos_ar_index = 0;
338 :
339 0 : for (int32_t row = -params->ar_coeff_lag; row < 0; row++) {
340 0 : for (int32_t col = -params->ar_coeff_lag; col < params->ar_coeff_lag + 1;
341 0 : col++) {
342 0 : pred_pos_luma[pos_ar_index][0] = row;
343 0 : pred_pos_luma[pos_ar_index][1] = col;
344 0 : pred_pos_luma[pos_ar_index][2] = 0;
345 :
346 0 : pred_pos_chroma[pos_ar_index][0] = row;
347 0 : pred_pos_chroma[pos_ar_index][1] = col;
348 0 : pred_pos_chroma[pos_ar_index][2] = 0;
349 0 : ++pos_ar_index;
350 : }
351 : }
352 :
353 0 : for (int32_t col = -params->ar_coeff_lag; col < 0; col++) {
354 0 : pred_pos_luma[pos_ar_index][0] = 0;
355 0 : pred_pos_luma[pos_ar_index][1] = col;
356 0 : pred_pos_luma[pos_ar_index][2] = 0;
357 :
358 0 : pred_pos_chroma[pos_ar_index][0] = 0;
359 0 : pred_pos_chroma[pos_ar_index][1] = col;
360 0 : pred_pos_chroma[pos_ar_index][2] = 0;
361 :
362 0 : ++pos_ar_index;
363 : }
364 :
365 0 : if (params->num_y_points > 0) {
366 0 : pred_pos_chroma[pos_ar_index][0] = 0;
367 0 : pred_pos_chroma[pos_ar_index][1] = 0;
368 0 : pred_pos_chroma[pos_ar_index][2] = 1;
369 : }
370 :
371 0 : *pred_pos_luma_p = pred_pos_luma;
372 0 : *pred_pos_chroma_p = pred_pos_chroma;
373 :
374 0 : *y_line_buf = (int32_t *)malloc(sizeof(**y_line_buf) * luma_stride * 2);
375 0 : *cb_line_buf = (int32_t *)malloc(sizeof(**cb_line_buf) * chroma_stride *
376 0 : (2 >> chroma_subsamp_y));
377 0 : *cr_line_buf = (int32_t *)malloc(sizeof(**cr_line_buf) * chroma_stride *
378 0 : (2 >> chroma_subsamp_y));
379 :
380 0 : *y_col_buf =
381 0 : (int32_t *)malloc(sizeof(**y_col_buf) * (luma_subblock_size_y + 2) * 2);
382 0 : *cb_col_buf =
383 0 : (int32_t *)malloc(sizeof(**cb_col_buf) *
384 0 : (chroma_subblock_size_y + (2 >> chroma_subsamp_y)) *
385 0 : (2 >> chroma_subsamp_x));
386 0 : *cr_col_buf =
387 0 : (int32_t *)malloc(sizeof(**cr_col_buf) *
388 0 : (chroma_subblock_size_y + (2 >> chroma_subsamp_y)) *
389 0 : (2 >> chroma_subsamp_x));
390 :
391 0 : *luma_grain_block =
392 0 : (int32_t *)malloc(sizeof(**luma_grain_block) * luma_grain_samples);
393 0 : *cb_grain_block =
394 0 : (int32_t *)malloc(sizeof(**cb_grain_block) * chroma_grain_samples);
395 0 : *cr_grain_block =
396 0 : (int32_t *)malloc(sizeof(**cr_grain_block) * chroma_grain_samples);
397 0 : }
398 :
399 0 : static void dealloc_arrays(aom_film_grain_t *params, int32_t ***pred_pos_luma,
400 : int32_t ***pred_pos_chroma, int32_t **luma_grain_block,
401 : int32_t **cb_grain_block, int32_t **cr_grain_block,
402 : int32_t **y_line_buf, int32_t **cb_line_buf,
403 : int32_t **cr_line_buf, int32_t **y_col_buf, int32_t **cb_col_buf,
404 : int32_t **cr_col_buf) {
405 0 : int32_t num_pos_luma = 2 * params->ar_coeff_lag * (params->ar_coeff_lag + 1);
406 0 : int32_t num_pos_chroma = num_pos_luma + 1;
407 :
408 0 : for (int32_t row = 0; row < num_pos_luma; row++)
409 0 : free((*pred_pos_luma)[row]);
410 0 : free(*pred_pos_luma);
411 :
412 0 : for (int32_t row = 0; row < num_pos_chroma; row++)
413 0 : free((*pred_pos_chroma)[row]);
414 0 : free((*pred_pos_chroma));
415 :
416 0 : free(*y_line_buf);
417 :
418 0 : free(*cb_line_buf);
419 :
420 0 : free(*cr_line_buf);
421 :
422 0 : free(*y_col_buf);
423 :
424 0 : free(*cb_col_buf);
425 :
426 0 : free(*cr_col_buf);
427 :
428 0 : free(*luma_grain_block);
429 :
430 0 : free(*cb_grain_block);
431 :
432 0 : free(*cr_grain_block);
433 0 : }
434 :
435 : // get a number between 0 and 2^bits - 1
436 0 : static INLINE int32_t get_random_number(int32_t bits) {
437 : uint16_t bit;
438 0 : bit = ((random_register >> 0) ^ (random_register >> 1) ^
439 0 : (random_register >> 3) ^ (random_register >> 12)) &
440 : 1;
441 0 : random_register = (random_register >> 1) | (bit << 15);
442 0 : return (random_register >> (16 - bits)) & ((1 << bits) - 1);
443 : }
444 :
445 0 : static void init_random_generator(int32_t luma_line, uint16_t seed) {
446 : // same for the picture
447 :
448 0 : uint16_t msb = (seed >> 8) & 255;
449 0 : uint16_t lsb = seed & 255;
450 :
451 0 : random_register = (msb << 8) + lsb;
452 :
453 : // changes for each row
454 0 : int32_t luma_num = luma_line >> 5;
455 :
456 0 : random_register ^= ((luma_num * 37 + 178) & 255) << 8;
457 0 : random_register ^= ((luma_num * 173 + 105) & 255);
458 0 : }
459 :
460 0 : static void generate_luma_grain_block(
461 : aom_film_grain_t *params, int32_t **pred_pos_luma, int32_t *luma_grain_block,
462 : int32_t luma_block_size_y, int32_t luma_block_size_x, int32_t luma_grain_stride,
463 : int32_t left_pad, int32_t top_pad, int32_t right_pad, int32_t bottom_pad) {
464 0 : if (params->num_y_points == 0) return;
465 :
466 0 : int32_t bit_depth = params->bit_depth;
467 0 : int32_t gauss_sec_shift = 12 - bit_depth + params->grain_scale_shift;
468 :
469 0 : int32_t num_pos_luma = 2 * params->ar_coeff_lag * (params->ar_coeff_lag + 1);
470 0 : int32_t rounding_offset = (1 << (params->ar_coeff_shift - 1));
471 :
472 0 : for (int32_t i = 0; i < luma_block_size_y; i++)
473 0 : for (int32_t j = 0; j < luma_block_size_x; j++)
474 0 : luma_grain_block[i * luma_grain_stride + j] =
475 0 : (gaussian_sequence[get_random_number(gauss_bits)] +
476 0 : ((1 << gauss_sec_shift) >> 1)) >>
477 : gauss_sec_shift;
478 :
479 0 : for (int32_t i = top_pad; i < luma_block_size_y - bottom_pad; i++)
480 0 : for (int32_t j = left_pad; j < luma_block_size_x - right_pad; j++) {
481 0 : int32_t wsum = 0;
482 0 : for (int32_t pos = 0; pos < num_pos_luma; pos++) {
483 0 : wsum = wsum + params->ar_coeffs_y[pos] *
484 0 : luma_grain_block[(i + pred_pos_luma[pos][0]) *
485 0 : luma_grain_stride +
486 0 : j + pred_pos_luma[pos][1]];
487 : }
488 0 : luma_grain_block[i * luma_grain_stride + j] =
489 0 : clamp(luma_grain_block[i * luma_grain_stride + j] +
490 0 : ((wsum + rounding_offset) >> params->ar_coeff_shift),
491 : grain_min, grain_max);
492 : }
493 : }
494 :
495 0 : static void generate_chroma_grain_blocks(
496 : aom_film_grain_t *params,
497 : // int32_t** pred_pos_luma,
498 : int32_t **pred_pos_chroma, int32_t *luma_grain_block, int32_t *cb_grain_block,
499 : int32_t *cr_grain_block, int32_t luma_grain_stride, int32_t chroma_block_size_y,
500 : int32_t chroma_block_size_x, int32_t chroma_grain_stride, int32_t left_pad, int32_t top_pad,
501 : int32_t right_pad, int32_t bottom_pad, int32_t chroma_subsamp_y, int32_t chroma_subsamp_x) {
502 0 : int32_t bit_depth = params->bit_depth;
503 0 : int32_t gauss_sec_shift = 12 - bit_depth + params->grain_scale_shift;
504 :
505 0 : int32_t num_pos_chroma = 2 * params->ar_coeff_lag * (params->ar_coeff_lag + 1);
506 0 : if (params->num_y_points > 0) ++num_pos_chroma;
507 0 : int32_t rounding_offset = (1 << (params->ar_coeff_shift - 1));
508 :
509 0 : if (params->num_cb_points) {
510 0 : init_random_generator(7 << 5, params->random_seed);
511 :
512 0 : for (int32_t i = 0; i < chroma_block_size_y; i++)
513 0 : for (int32_t j = 0; j < chroma_block_size_x; j++)
514 0 : cb_grain_block[i * chroma_grain_stride + j] =
515 0 : (gaussian_sequence[get_random_number(gauss_bits)] +
516 0 : ((1 << gauss_sec_shift) >> 1)) >>
517 : gauss_sec_shift;
518 : }
519 0 : if (params->num_cr_points) {
520 0 : init_random_generator(11 << 5, params->random_seed);
521 :
522 0 : for (int32_t i = 0; i < chroma_block_size_y; i++)
523 0 : for (int32_t j = 0; j < chroma_block_size_x; j++)
524 0 : cr_grain_block[i * chroma_grain_stride + j] =
525 0 : (gaussian_sequence[get_random_number(gauss_bits)] +
526 0 : ((1 << gauss_sec_shift) >> 1)) >>
527 : gauss_sec_shift;
528 : }
529 :
530 0 : for (int32_t i = top_pad; i < chroma_block_size_y - bottom_pad; i++)
531 0 : for (int32_t j = left_pad; j < chroma_block_size_x - right_pad; j++) {
532 0 : int32_t wsum_cb = 0;
533 0 : int32_t wsum_cr = 0;
534 0 : for (int32_t pos = 0; pos < num_pos_chroma; pos++) {
535 0 : if (pred_pos_chroma[pos][2] == 0) {
536 0 : wsum_cb = wsum_cb + params->ar_coeffs_cb[pos] *
537 0 : cb_grain_block[(i + pred_pos_chroma[pos][0]) *
538 0 : chroma_grain_stride +
539 0 : j + pred_pos_chroma[pos][1]];
540 0 : wsum_cr = wsum_cr + params->ar_coeffs_cr[pos] *
541 0 : cr_grain_block[(i + pred_pos_chroma[pos][0]) *
542 0 : chroma_grain_stride +
543 0 : j + pred_pos_chroma[pos][1]];
544 : }
545 0 : else if (pred_pos_chroma[pos][2] == 1) {
546 0 : int32_t av_luma = 0;
547 0 : int32_t luma_coord_y = ((i - top_pad) << chroma_subsamp_y) + top_pad;
548 0 : int32_t luma_coord_x = ((j - left_pad) << chroma_subsamp_x) + left_pad;
549 :
550 0 : for (int32_t k = luma_coord_y; k < luma_coord_y + chroma_subsamp_y + 1;
551 0 : k++)
552 0 : for (int32_t l = luma_coord_x; l < luma_coord_x + chroma_subsamp_x + 1;
553 0 : l++)
554 0 : av_luma += luma_grain_block[k * luma_grain_stride + l];
555 :
556 0 : av_luma =
557 0 : (av_luma + ((1 << (chroma_subsamp_y + chroma_subsamp_x)) >> 1)) >>
558 0 : (chroma_subsamp_y + chroma_subsamp_x);
559 :
560 0 : wsum_cb = wsum_cb + params->ar_coeffs_cb[pos] * av_luma;
561 0 : wsum_cr = wsum_cr + params->ar_coeffs_cr[pos] * av_luma;
562 : }
563 : else {
564 0 : printf(
565 : "Grain synthesis: prediction between two chroma components is "
566 : "not supported!");
567 0 : exit(1);
568 : }
569 : }
570 0 : if (params->num_cb_points)
571 0 : cb_grain_block[i * chroma_grain_stride + j] =
572 0 : clamp(cb_grain_block[i * chroma_grain_stride + j] +
573 0 : ((wsum_cb + rounding_offset) >> params->ar_coeff_shift),
574 : grain_min, grain_max);
575 0 : if (params->num_cr_points)
576 0 : cr_grain_block[i * chroma_grain_stride + j] =
577 0 : clamp(cr_grain_block[i * chroma_grain_stride + j] +
578 0 : ((wsum_cr + rounding_offset) >> params->ar_coeff_shift),
579 : grain_min, grain_max);
580 : }
581 0 : }
582 :
583 0 : static void init_scaling_function(int32_t scaling_points[][2], int32_t num_points,
584 : int32_t scaling_lut[]) {
585 0 : if (num_points == 0) return;
586 :
587 0 : for (int32_t i = 0; i < scaling_points[0][0]; i++)
588 0 : scaling_lut[i] = scaling_points[0][1];
589 :
590 0 : for (int32_t point = 0; point < num_points - 1; point++) {
591 0 : int32_t delta_y = scaling_points[point + 1][1] - scaling_points[point][1];
592 0 : int32_t delta_x = scaling_points[point + 1][0] - scaling_points[point][0];
593 :
594 0 : int64_t delta = delta_y * ((65536 + (delta_x >> 1)) / delta_x);
595 :
596 0 : for (int32_t x = 0; x < delta_x; x++) {
597 0 : scaling_lut[scaling_points[point][0] + x] =
598 0 : scaling_points[point][1] + (int32_t)((x * delta + 32768) >> 16);
599 : }
600 : }
601 :
602 0 : for (int32_t i = scaling_points[num_points - 1][0]; i < 256; i++)
603 0 : scaling_lut[i] = scaling_points[num_points - 1][1];
604 : }
605 :
606 : // function that extracts samples from a LUT (and interpolates intemediate
607 : // frames for 10- and 12-bit video)
608 0 : static int32_t scale_LUT(int32_t *scaling_lut, int32_t index, int32_t bit_depth) {
609 0 : int32_t x = index >> (bit_depth - 8);
610 :
611 0 : if (!(bit_depth - 8) || x == 255)
612 0 : return scaling_lut[x];
613 : else
614 0 : return scaling_lut[x] + (((scaling_lut[x + 1] - scaling_lut[x]) *
615 0 : (index & ((1 << (bit_depth - 8)) - 1)) +
616 0 : (1 << (bit_depth - 9))) >>
617 0 : (bit_depth - 8));
618 : }
619 :
620 0 : static void add_noise_to_block(aom_film_grain_t *params, uint8_t *luma,
621 : uint8_t *cb, uint8_t *cr, int32_t luma_stride,
622 : int32_t chroma_stride, int32_t *luma_grain,
623 : int32_t *cb_grain, int32_t *cr_grain,
624 : int32_t luma_grain_stride, int32_t chroma_grain_stride,
625 : int32_t half_luma_height, int32_t half_luma_width,
626 : int32_t bit_depth, int32_t chroma_subsamp_y,
627 : int32_t chroma_subsamp_x) {
628 0 : int32_t cb_mult = params->cb_mult - 128; // fixed scale
629 0 : int32_t cb_luma_mult = params->cb_luma_mult - 128; // fixed scale
630 0 : int32_t cb_offset = params->cb_offset - 256;
631 :
632 0 : int32_t cr_mult = params->cr_mult - 128; // fixed scale
633 0 : int32_t cr_luma_mult = params->cr_luma_mult - 128; // fixed scale
634 0 : int32_t cr_offset = params->cr_offset - 256;
635 :
636 0 : int32_t rounding_offset = (1 << (params->scaling_shift - 1));
637 :
638 0 : int32_t apply_y = params->num_y_points > 0 ? 1 : 0;
639 0 : int32_t apply_cb = params->num_cb_points > 0 ? 1 : 0;
640 0 : int32_t apply_cr = params->num_cr_points > 0 ? 1 : 0;
641 :
642 0 : if (params->chroma_scaling_from_luma) {
643 0 : cb_mult = 0; // fixed scale
644 0 : cb_luma_mult = 64; // fixed scale
645 0 : cb_offset = 0;
646 :
647 0 : cr_mult = 0; // fixed scale
648 0 : cr_luma_mult = 64; // fixed scale
649 0 : cr_offset = 0;
650 : }
651 :
652 : int32_t min_luma, max_luma, min_chroma, max_chroma;
653 :
654 0 : if (params->clip_to_restricted_range) {
655 0 : min_luma = min_luma_legal_range;
656 0 : max_luma = max_luma_legal_range;
657 :
658 0 : min_chroma = min_chroma_legal_range;
659 0 : max_chroma = max_chroma_legal_range;
660 : }
661 : else {
662 0 : min_luma = min_chroma = 0;
663 0 : max_luma = max_chroma = 255;
664 : }
665 :
666 0 : for (int32_t i = 0; i < (half_luma_height << (1 - chroma_subsamp_y)); i++) {
667 0 : for (int32_t j = 0; j < (half_luma_width << (1 - chroma_subsamp_x)); j++) {
668 0 : int32_t average_luma = 0;
669 0 : if (chroma_subsamp_x) {
670 0 : average_luma = (luma[(i << chroma_subsamp_y) * luma_stride +
671 0 : (j << chroma_subsamp_x)] +
672 0 : luma[(i << chroma_subsamp_y) * luma_stride +
673 0 : (j << chroma_subsamp_x) + 1] +
674 : 1) >>
675 : 1;
676 : }
677 : else
678 0 : average_luma = luma[(i << chroma_subsamp_y) * luma_stride + j];
679 0 : if (apply_cb) {
680 0 : cb[i * chroma_stride + j] = clamp(
681 0 : cb[i * chroma_stride + j] +
682 0 : ((scale_LUT(scaling_lut_cb,
683 0 : clamp(((average_luma * cb_luma_mult +
684 0 : cb_mult * cb[i * chroma_stride + j]) >>
685 : 6) +
686 : cb_offset,
687 0 : 0, (256 << (bit_depth - 8)) - 1),
688 0 : 8) *
689 0 : cb_grain[i * chroma_grain_stride + j] +
690 0 : rounding_offset) >>
691 0 : params->scaling_shift),
692 : min_chroma, max_chroma);
693 : }
694 :
695 0 : if (apply_cr) {
696 0 : cr[i * chroma_stride + j] = clamp(
697 0 : cr[i * chroma_stride + j] +
698 0 : ((scale_LUT(scaling_lut_cr,
699 0 : clamp(((average_luma * cr_luma_mult +
700 0 : cr_mult * cr[i * chroma_stride + j]) >>
701 : 6) +
702 : cr_offset,
703 0 : 0, (256 << (bit_depth - 8)) - 1),
704 0 : 8) *
705 0 : cr_grain[i * chroma_grain_stride + j] +
706 0 : rounding_offset) >>
707 0 : params->scaling_shift),
708 : min_chroma, max_chroma);
709 : }
710 : }
711 : }
712 :
713 0 : if (apply_y) {
714 0 : for (int32_t i = 0; i < (half_luma_height << 1); i++) {
715 0 : for (int32_t j = 0; j < (half_luma_width << 1); j++) {
716 0 : luma[i * luma_stride + j] =
717 0 : clamp(luma[i * luma_stride + j] +
718 0 : ((scale_LUT(scaling_lut_y, luma[i * luma_stride + j], 8) *
719 0 : luma_grain[i * luma_grain_stride + j] +
720 0 : rounding_offset) >>
721 0 : params->scaling_shift),
722 : min_luma, max_luma);
723 : }
724 : }
725 : }
726 0 : }
727 :
728 0 : static void add_noise_to_block_hbd(
729 : aom_film_grain_t *params, uint16_t *luma, uint16_t *cb, uint16_t *cr,
730 : int32_t luma_stride, int32_t chroma_stride, int32_t *luma_grain, int32_t *cb_grain,
731 : int32_t *cr_grain, int32_t luma_grain_stride, int32_t chroma_grain_stride,
732 : int32_t half_luma_height, int32_t half_luma_width, int32_t bit_depth,
733 : int32_t chroma_subsamp_y, int32_t chroma_subsamp_x) {
734 0 : int32_t cb_mult = params->cb_mult - 128; // fixed scale
735 0 : int32_t cb_luma_mult = params->cb_luma_mult - 128; // fixed scale
736 : // offset value depends on the bit depth
737 0 : int32_t cb_offset = (params->cb_offset << (bit_depth - 8)) - (1 << bit_depth);
738 :
739 0 : int32_t cr_mult = params->cr_mult - 128; // fixed scale
740 0 : int32_t cr_luma_mult = params->cr_luma_mult - 128; // fixed scale
741 : // offset value depends on the bit depth
742 0 : int32_t cr_offset = (params->cr_offset << (bit_depth - 8)) - (1 << bit_depth);
743 :
744 0 : int32_t rounding_offset = (1 << (params->scaling_shift - 1));
745 :
746 0 : int32_t apply_y = params->num_y_points > 0 ? 1 : 0;
747 0 : int32_t apply_cb = params->num_cb_points > 0 ? 1 : 0;
748 0 : int32_t apply_cr = params->num_cr_points > 0 ? 1 : 0;
749 :
750 0 : if (params->chroma_scaling_from_luma) {
751 0 : cb_mult = 0; // fixed scale
752 0 : cb_luma_mult = 64; // fixed scale
753 0 : cb_offset = 0;
754 :
755 0 : cr_mult = 0; // fixed scale
756 0 : cr_luma_mult = 64; // fixed scale
757 0 : cr_offset = 0;
758 : }
759 :
760 : int32_t min_luma, max_luma, min_chroma, max_chroma;
761 :
762 0 : if (params->clip_to_restricted_range) {
763 0 : min_luma = min_luma_legal_range << (bit_depth - 8);
764 0 : max_luma = max_luma_legal_range << (bit_depth - 8);
765 :
766 0 : min_chroma = min_chroma_legal_range << (bit_depth - 8);
767 0 : max_chroma = max_chroma_legal_range << (bit_depth - 8);
768 : }
769 : else {
770 0 : min_luma = min_chroma = 0;
771 0 : max_luma = max_chroma = (256 << (bit_depth - 8)) - 1;
772 : }
773 :
774 0 : for (int32_t i = 0; i < (half_luma_height << (1 - chroma_subsamp_y)); i++) {
775 0 : for (int32_t j = 0; j < (half_luma_width << (1 - chroma_subsamp_x)); j++) {
776 0 : int32_t average_luma = 0;
777 0 : if (chroma_subsamp_x) {
778 0 : average_luma = (luma[(i << chroma_subsamp_y) * luma_stride +
779 0 : (j << chroma_subsamp_x)] +
780 0 : luma[(i << chroma_subsamp_y) * luma_stride +
781 0 : (j << chroma_subsamp_x) + 1] +
782 : 1) >>
783 : 1;
784 : }
785 : else
786 0 : average_luma = luma[(i << chroma_subsamp_y) * luma_stride + j];
787 0 : if (apply_cb) {
788 0 : cb[i * chroma_stride + j] = clamp(
789 0 : cb[i * chroma_stride + j] +
790 0 : ((scale_LUT(scaling_lut_cb,
791 0 : clamp(((average_luma * cb_luma_mult +
792 0 : cb_mult * cb[i * chroma_stride + j]) >>
793 : 6) +
794 : cb_offset,
795 0 : 0, (256 << (bit_depth - 8)) - 1),
796 0 : bit_depth) *
797 0 : cb_grain[i * chroma_grain_stride + j] +
798 0 : rounding_offset) >>
799 0 : params->scaling_shift),
800 : min_chroma, max_chroma);
801 : }
802 0 : if (apply_cr) {
803 0 : cr[i * chroma_stride + j] = clamp(
804 0 : cr[i * chroma_stride + j] +
805 0 : ((scale_LUT(scaling_lut_cr,
806 0 : clamp(((average_luma * cr_luma_mult +
807 0 : cr_mult * cr[i * chroma_stride + j]) >>
808 : 6) +
809 : cr_offset,
810 0 : 0, (256 << (bit_depth - 8)) - 1),
811 0 : bit_depth) *
812 0 : cr_grain[i * chroma_grain_stride + j] +
813 0 : rounding_offset) >>
814 0 : params->scaling_shift),
815 : min_chroma, max_chroma);
816 : }
817 : }
818 : }
819 :
820 0 : if (apply_y) {
821 0 : for (int32_t i = 0; i < (half_luma_height << 1); i++) {
822 0 : for (int32_t j = 0; j < (half_luma_width << 1); j++) {
823 0 : luma[i * luma_stride + j] =
824 0 : clamp(luma[i * luma_stride + j] +
825 0 : ((scale_LUT(scaling_lut_y, luma[i * luma_stride + j],
826 0 : bit_depth) *
827 0 : luma_grain[i * luma_grain_stride + j] +
828 0 : rounding_offset) >>
829 0 : params->scaling_shift),
830 : min_luma, max_luma);
831 : }
832 : }
833 : }
834 0 : }
835 :
836 0 : int32_t film_grain_params_equal(aom_film_grain_t *pars_a, aom_film_grain_t *pars_b) {
837 0 : if (pars_a->apply_grain != pars_b->apply_grain)
838 0 : return 0;
839 0 : if (pars_a->overlap_flag != pars_b->overlap_flag)
840 0 : return 0;
841 0 : if (pars_a->clip_to_restricted_range != pars_b->clip_to_restricted_range)
842 0 : return 0;
843 0 : if (pars_a->chroma_scaling_from_luma != pars_b->chroma_scaling_from_luma)
844 0 : return 0;
845 0 : if (pars_a->grain_scale_shift != pars_b->grain_scale_shift)
846 0 : return 0;
847 0 : if (pars_a->ar_coeff_shift != pars_b->ar_coeff_shift)
848 0 : return 0;
849 0 : if (pars_a->cb_mult != pars_b->cb_mult)
850 0 : return 0;
851 0 : if (pars_a->cb_luma_mult != pars_b->cb_luma_mult)
852 0 : return 0;
853 0 : if (pars_a->cb_offset != pars_b->cb_offset)
854 0 : return 0;
855 0 : if (pars_a->cr_mult != pars_b->cr_mult)
856 0 : return 0;
857 0 : if (pars_a->cr_luma_mult != pars_b->cr_luma_mult)
858 0 : return 0;
859 0 : if (pars_a->cr_offset != pars_b->cr_offset)
860 0 : return 0;
861 :
862 0 : if (pars_a->scaling_shift != pars_b->scaling_shift)
863 0 : return 0;
864 0 : if (pars_a->ar_coeff_lag != pars_b->ar_coeff_lag)
865 0 : return 0;
866 :
867 0 : if (pars_a->num_y_points != pars_b->num_y_points)
868 0 : return 0;
869 :
870 0 : if (pars_a->num_cb_points != pars_b->num_cb_points)
871 0 : return 0;
872 :
873 0 : if (pars_a->num_cr_points != pars_b->num_cr_points)
874 0 : return 0;
875 :
876 0 : if (memcmp(pars_a->scaling_points_y, pars_b->scaling_points_y, sizeof(pars_b->scaling_points_y)))
877 0 : return 0;
878 :
879 0 : if (memcmp(pars_a->scaling_points_cb, pars_b->scaling_points_cb, sizeof(pars_b->scaling_points_cb)))
880 0 : return 0;
881 :
882 0 : if (memcmp(pars_a->scaling_points_cr, pars_b->scaling_points_cr, sizeof(pars_b->scaling_points_cr)))
883 0 : return 0;
884 :
885 0 : if (memcmp(pars_a->ar_coeffs_y, pars_b->ar_coeffs_y, sizeof(pars_b->ar_coeffs_y)))
886 0 : return 0;
887 :
888 0 : if (memcmp(pars_a->ar_coeffs_cb, pars_b->ar_coeffs_cb, sizeof(pars_b->ar_coeffs_cb)))
889 0 : return 0;
890 :
891 0 : if (memcmp(pars_a->ar_coeffs_cr, pars_b->ar_coeffs_cr, sizeof(pars_b->ar_coeffs_cr)))
892 0 : return 0;
893 :
894 0 : return 1;
895 : }
896 :
897 0 : void fgn_copy_rect(uint8_t *src, int32_t src_stride, uint8_t *dst,
898 : int32_t dst_stride, int32_t width, int32_t height,
899 : int32_t use_high_bit_depth) {
900 0 : int32_t hbd_coeff = use_high_bit_depth ? 2 : 1;
901 0 : while (height) {
902 0 : memcpy(dst, src, width * sizeof(uint8_t) * hbd_coeff);
903 0 : src += src_stride * hbd_coeff;
904 0 : dst += dst_stride * hbd_coeff;
905 0 : --height;
906 : }
907 0 : return;
908 : }
909 :
910 0 : static void copy_area(int32_t *src, int32_t src_stride, int32_t *dst, int32_t dst_stride,
911 : int32_t width, int32_t height) {
912 0 : while (height) {
913 0 : memcpy(dst, src, width * sizeof(*src));
914 0 : src += src_stride;
915 0 : dst += dst_stride;
916 0 : --height;
917 : }
918 0 : return;
919 : }
920 :
921 0 : static void ver_boundary_overlap(int32_t *left_block, int32_t left_stride,
922 : int32_t *right_block, int32_t right_stride,
923 : int32_t *dst_block, int32_t dst_stride, int32_t width,
924 : int32_t height) {
925 0 : if (width == 1) {
926 0 : while (height) {
927 0 : *dst_block = clamp((*left_block * 23 + *right_block * 22 + 16) >> 5,
928 : grain_min, grain_max);
929 0 : left_block += left_stride;
930 0 : right_block += right_stride;
931 0 : dst_block += dst_stride;
932 0 : --height;
933 : }
934 0 : return;
935 : }
936 0 : else if (width == 2) {
937 0 : while (height) {
938 0 : dst_block[0] = clamp((27 * left_block[0] + 17 * right_block[0] + 16) >> 5,
939 : grain_min, grain_max);
940 0 : dst_block[1] = clamp((17 * left_block[1] + 27 * right_block[1] + 16) >> 5,
941 : grain_min, grain_max);
942 0 : left_block += left_stride;
943 0 : right_block += right_stride;
944 0 : dst_block += dst_stride;
945 0 : --height;
946 : }
947 0 : return;
948 : }
949 : }
950 :
951 0 : static void hor_boundary_overlap(int32_t *top_block, int32_t top_stride,
952 : int32_t *bottom_block, int32_t bottom_stride,
953 : int32_t *dst_block, int32_t dst_stride, int32_t width,
954 : int32_t height) {
955 0 : if (height == 1) {
956 0 : while (width) {
957 0 : *dst_block = clamp((*top_block * 23 + *bottom_block * 22 + 16) >> 5,
958 : grain_min, grain_max);
959 0 : ++top_block;
960 0 : ++bottom_block;
961 0 : ++dst_block;
962 0 : --width;
963 : }
964 0 : return;
965 : }
966 0 : else if (height == 2) {
967 0 : while (width) {
968 0 : dst_block[0] = clamp((27 * top_block[0] + 17 * bottom_block[0] + 16) >> 5,
969 : grain_min, grain_max);
970 0 : dst_block[dst_stride] = clamp((17 * top_block[top_stride] +
971 0 : 27 * bottom_block[bottom_stride] + 16) >>
972 : 5,
973 : grain_min, grain_max);
974 0 : ++top_block;
975 0 : ++bottom_block;
976 0 : ++dst_block;
977 0 : --width;
978 : }
979 0 : return;
980 : }
981 : }
982 :
983 0 : void eb_av1_add_film_grain_run(aom_film_grain_t *params, uint8_t *luma,
984 : uint8_t *cb, uint8_t *cr, int32_t height, int32_t width,
985 : int32_t luma_stride, int32_t chroma_stride,
986 : int32_t use_high_bit_depth, int32_t chroma_subsamp_y,
987 : int32_t chroma_subsamp_x) {
988 : int32_t **pred_pos_luma;
989 : int32_t **pred_pos_chroma;
990 : int32_t *luma_grain_block;
991 : int32_t *cb_grain_block;
992 : int32_t *cr_grain_block;
993 :
994 : int32_t *y_line_buf;
995 : int32_t *cb_line_buf;
996 : int32_t *cr_line_buf;
997 :
998 : int32_t *y_col_buf;
999 : int32_t *cb_col_buf;
1000 : int32_t *cr_col_buf;
1001 :
1002 0 : random_register = params->random_seed;
1003 :
1004 0 : int32_t left_pad = 3;
1005 0 : int32_t right_pad = 3; // padding to offset for AR coefficients
1006 0 : int32_t top_pad = 3;
1007 0 : int32_t bottom_pad = 0;
1008 :
1009 0 : int32_t ar_padding = 3; // maximum lag used for stabilization of AR coefficients
1010 :
1011 0 : luma_subblock_size_y = 32;
1012 0 : luma_subblock_size_x = 32;
1013 :
1014 0 : chroma_subblock_size_y = luma_subblock_size_y >> chroma_subsamp_y;
1015 0 : chroma_subblock_size_x = luma_subblock_size_x >> chroma_subsamp_x;
1016 :
1017 : // Initial padding is only needed for generation of
1018 : // film grain templates (to stabilize the AR process)
1019 : // Only a 64x64 luma and 32x32 chroma part of a template
1020 : // is used later for adding grain, padding can be discarded
1021 :
1022 0 : int32_t luma_block_size_y =
1023 0 : top_pad + 2 * ar_padding + luma_subblock_size_y * 2 + bottom_pad;
1024 0 : int32_t luma_block_size_x = left_pad + 2 * ar_padding + luma_subblock_size_x * 2 +
1025 0 : 2 * ar_padding + right_pad;
1026 :
1027 0 : int32_t chroma_block_size_y = top_pad + (2 >> chroma_subsamp_y) * ar_padding +
1028 0 : chroma_subblock_size_y * 2 + bottom_pad;
1029 0 : int32_t chroma_block_size_x = left_pad + (2 >> chroma_subsamp_x) * ar_padding +
1030 0 : chroma_subblock_size_x * 2 +
1031 0 : (2 >> chroma_subsamp_x) * ar_padding + right_pad;
1032 :
1033 0 : int32_t luma_grain_stride = luma_block_size_x;
1034 0 : int32_t chroma_grain_stride = chroma_block_size_x;
1035 :
1036 0 : int32_t overlap = params->overlap_flag;
1037 0 : int32_t bit_depth = params->bit_depth;
1038 :
1039 0 : grain_center = 128 << (bit_depth - 8);
1040 0 : grain_min = 0 - grain_center;
1041 0 : grain_max = (256 << (bit_depth - 8)) - 1 - grain_center;
1042 :
1043 0 : init_arrays(params, luma_stride, chroma_stride, &pred_pos_luma,
1044 : &pred_pos_chroma, &luma_grain_block, &cb_grain_block,
1045 : &cr_grain_block, &y_line_buf, &cb_line_buf, &cr_line_buf,
1046 : &y_col_buf, &cb_col_buf, &cr_col_buf,
1047 : luma_block_size_y * luma_block_size_x,
1048 : chroma_block_size_y * chroma_block_size_x, chroma_subsamp_y,
1049 : chroma_subsamp_x);
1050 :
1051 0 : generate_luma_grain_block(params, pred_pos_luma, luma_grain_block,
1052 : luma_block_size_y, luma_block_size_x,
1053 : luma_grain_stride, left_pad, top_pad, right_pad,
1054 : bottom_pad);
1055 :
1056 0 : generate_chroma_grain_blocks(
1057 : params,
1058 : // pred_pos_luma,
1059 : pred_pos_chroma, luma_grain_block, cb_grain_block, cr_grain_block,
1060 : luma_grain_stride, chroma_block_size_y, chroma_block_size_x,
1061 : chroma_grain_stride, left_pad, top_pad, right_pad, bottom_pad,
1062 : chroma_subsamp_y, chroma_subsamp_x);
1063 :
1064 0 : init_scaling_function(params->scaling_points_y, params->num_y_points,
1065 : scaling_lut_y);
1066 :
1067 0 : if (params->chroma_scaling_from_luma) {
1068 0 : memcpy(scaling_lut_cb, scaling_lut_y, sizeof(*scaling_lut_y) * 256);
1069 0 : memcpy(scaling_lut_cr, scaling_lut_y, sizeof(*scaling_lut_y) * 256);
1070 : }
1071 : else {
1072 0 : init_scaling_function(params->scaling_points_cb, params->num_cb_points,
1073 : scaling_lut_cb);
1074 0 : init_scaling_function(params->scaling_points_cr, params->num_cr_points,
1075 : scaling_lut_cr);
1076 : }
1077 0 : for (int32_t y = 0; y < height / 2; y += (luma_subblock_size_y >> 1)) {
1078 0 : init_random_generator(y * 2, params->random_seed);
1079 :
1080 0 : for (int32_t x = 0; x < width / 2; x += (luma_subblock_size_x >> 1)) {
1081 0 : int32_t offset_y = get_random_number(8);
1082 0 : int32_t offset_x = (offset_y >> 4) & 15;
1083 0 : offset_y &= 15;
1084 :
1085 0 : int32_t luma_offset_y = left_pad + 2 * ar_padding + (offset_y << 1);
1086 0 : int32_t luma_offset_x = top_pad + 2 * ar_padding + (offset_x << 1);
1087 :
1088 0 : int32_t chroma_offset_y = top_pad + (2 >> chroma_subsamp_y) * ar_padding +
1089 0 : offset_y * (2 >> chroma_subsamp_y);
1090 0 : int32_t chroma_offset_x = left_pad + (2 >> chroma_subsamp_x) * ar_padding +
1091 0 : offset_x * (2 >> chroma_subsamp_x);
1092 :
1093 0 : if (overlap && x) {
1094 0 : ver_boundary_overlap(
1095 : y_col_buf, 2,
1096 0 : luma_grain_block + luma_offset_y * luma_grain_stride +
1097 : luma_offset_x,
1098 : luma_grain_stride, y_col_buf, 2, 2,
1099 0 : AOMMIN(luma_subblock_size_y + 2, height - (y << 1)));
1100 :
1101 0 : ver_boundary_overlap(
1102 : cb_col_buf, 2 >> chroma_subsamp_x,
1103 0 : cb_grain_block + chroma_offset_y * chroma_grain_stride +
1104 : chroma_offset_x,
1105 : chroma_grain_stride, cb_col_buf, 2 >> chroma_subsamp_x,
1106 : 2 >> chroma_subsamp_x,
1107 0 : AOMMIN(chroma_subblock_size_y + (2 >> chroma_subsamp_y),
1108 : (height - (y << 1)) >> chroma_subsamp_y));
1109 :
1110 0 : ver_boundary_overlap(
1111 : cr_col_buf, 2 >> chroma_subsamp_x,
1112 0 : cr_grain_block + chroma_offset_y * chroma_grain_stride +
1113 : chroma_offset_x,
1114 : chroma_grain_stride, cr_col_buf, 2 >> chroma_subsamp_x,
1115 : 2 >> chroma_subsamp_x,
1116 0 : AOMMIN(chroma_subblock_size_y + (2 >> chroma_subsamp_y),
1117 : (height - (y << 1)) >> chroma_subsamp_y));
1118 :
1119 0 : int32_t i = y ? 1 : 0;
1120 :
1121 0 : if (use_high_bit_depth) {
1122 0 : add_noise_to_block_hbd(
1123 : params,
1124 0 : (uint16_t *)luma + ((y + i) << 1) * luma_stride + (x << 1),
1125 : (uint16_t *)cb +
1126 0 : ((y + i) << (1 - chroma_subsamp_y)) * chroma_stride +
1127 0 : (x << (1 - chroma_subsamp_x)),
1128 : (uint16_t *)cr +
1129 0 : ((y + i) << (1 - chroma_subsamp_y)) * chroma_stride +
1130 0 : (x << (1 - chroma_subsamp_x)),
1131 0 : luma_stride, chroma_stride, y_col_buf + i * 4,
1132 0 : cb_col_buf + i * (2 - chroma_subsamp_y) * (2 - chroma_subsamp_x),
1133 0 : cr_col_buf + i * (2 - chroma_subsamp_y) * (2 - chroma_subsamp_x),
1134 : 2, (2 - chroma_subsamp_x),
1135 0 : AOMMIN(luma_subblock_size_y >> 1, height / 2 - y) - i, 1,
1136 : bit_depth, chroma_subsamp_y, chroma_subsamp_x);
1137 : }
1138 : else {
1139 0 : add_noise_to_block(
1140 0 : params, luma + ((y + i) << 1) * luma_stride + (x << 1),
1141 0 : cb + ((y + i) << (1 - chroma_subsamp_y)) * chroma_stride +
1142 0 : (x << (1 - chroma_subsamp_x)),
1143 0 : cr + ((y + i) << (1 - chroma_subsamp_y)) * chroma_stride +
1144 0 : (x << (1 - chroma_subsamp_x)),
1145 0 : luma_stride, chroma_stride, y_col_buf + i * 4,
1146 0 : cb_col_buf + i * (2 - chroma_subsamp_y) * (2 - chroma_subsamp_x),
1147 0 : cr_col_buf + i * (2 - chroma_subsamp_y) * (2 - chroma_subsamp_x),
1148 : 2, (2 - chroma_subsamp_x),
1149 0 : AOMMIN(luma_subblock_size_y >> 1, height / 2 - y) - i, 1,
1150 : bit_depth, chroma_subsamp_y, chroma_subsamp_x);
1151 : }
1152 : }
1153 :
1154 0 : if (overlap && y) {
1155 0 : if (x) {
1156 : ASSERT(y_col_buf != NULL);
1157 0 : hor_boundary_overlap(y_line_buf + (x << 1), luma_stride, y_col_buf, 2,
1158 0 : y_line_buf + (x << 1), luma_stride, 2, 2);
1159 :
1160 0 : hor_boundary_overlap(cb_line_buf + x * (2 >> chroma_subsamp_x),
1161 : chroma_stride, cb_col_buf, 2 >> chroma_subsamp_x,
1162 0 : cb_line_buf + x * (2 >> chroma_subsamp_x),
1163 : chroma_stride, 2 >> chroma_subsamp_x,
1164 : 2 >> chroma_subsamp_y);
1165 :
1166 0 : hor_boundary_overlap(cr_line_buf + x * (2 >> chroma_subsamp_x),
1167 : chroma_stride, cr_col_buf, 2 >> chroma_subsamp_x,
1168 0 : cr_line_buf + x * (2 >> chroma_subsamp_x),
1169 : chroma_stride, 2 >> chroma_subsamp_x,
1170 : 2 >> chroma_subsamp_y);
1171 : }
1172 :
1173 0 : hor_boundary_overlap(
1174 0 : y_line_buf + ((x ? x + 1 : 0) << 1), luma_stride,
1175 0 : luma_grain_block + luma_offset_y * luma_grain_stride +
1176 0 : luma_offset_x + (x ? 2 : 0),
1177 0 : luma_grain_stride, y_line_buf + ((x ? x + 1 : 0) << 1), luma_stride,
1178 0 : AOMMIN(luma_subblock_size_x - ((x ? 1 : 0) << 1),
1179 : width - ((x ? x + 1 : 0) << 1)),
1180 : 2);
1181 :
1182 0 : hor_boundary_overlap(
1183 0 : cb_line_buf + ((x ? x + 1 : 0) << (1 - chroma_subsamp_x)),
1184 : chroma_stride,
1185 0 : cb_grain_block + chroma_offset_y * chroma_grain_stride +
1186 0 : chroma_offset_x + ((x ? 1 : 0) << (1 - chroma_subsamp_x)),
1187 : chroma_grain_stride,
1188 0 : cb_line_buf + ((x ? x + 1 : 0) << (1 - chroma_subsamp_x)),
1189 : chroma_stride,
1190 0 : AOMMIN(chroma_subblock_size_x -
1191 : ((x ? 1 : 0) << (1 - chroma_subsamp_x)),
1192 : (width - ((x ? x + 1 : 0) << 1)) >> chroma_subsamp_x),
1193 : 2 >> chroma_subsamp_y);
1194 :
1195 0 : hor_boundary_overlap(
1196 0 : cr_line_buf + ((x ? x + 1 : 0) << (1 - chroma_subsamp_x)),
1197 : chroma_stride,
1198 0 : cr_grain_block + chroma_offset_y * chroma_grain_stride +
1199 0 : chroma_offset_x + ((x ? 1 : 0) << (1 - chroma_subsamp_x)),
1200 : chroma_grain_stride,
1201 0 : cr_line_buf + ((x ? x + 1 : 0) << (1 - chroma_subsamp_x)),
1202 : chroma_stride,
1203 0 : AOMMIN(chroma_subblock_size_x -
1204 : ((x ? 1 : 0) << (1 - chroma_subsamp_x)),
1205 : (width - ((x ? x + 1 : 0) << 1)) >> chroma_subsamp_x),
1206 : 2 >> chroma_subsamp_y);
1207 :
1208 0 : if (use_high_bit_depth) {
1209 0 : add_noise_to_block_hbd(
1210 0 : params, (uint16_t *)luma + (y << 1) * luma_stride + (x << 1),
1211 0 : (uint16_t *)cb + (y << (1 - chroma_subsamp_y)) * chroma_stride +
1212 0 : (x << ((1 - chroma_subsamp_x))),
1213 0 : (uint16_t *)cr + (y << (1 - chroma_subsamp_y)) * chroma_stride +
1214 0 : (x << ((1 - chroma_subsamp_x))),
1215 0 : luma_stride, chroma_stride, y_line_buf + (x << 1),
1216 0 : cb_line_buf + (x << (1 - chroma_subsamp_x)),
1217 0 : cr_line_buf + (x << (1 - chroma_subsamp_x)), luma_stride,
1218 : chroma_stride, 1,
1219 0 : AOMMIN(luma_subblock_size_x >> 1, width / 2 - x), bit_depth,
1220 : chroma_subsamp_y, chroma_subsamp_x);
1221 : }
1222 : else {
1223 0 : add_noise_to_block(
1224 0 : params, luma + (y << 1) * luma_stride + (x << 1),
1225 0 : cb + (y << (1 - chroma_subsamp_y)) * chroma_stride +
1226 0 : (x << ((1 - chroma_subsamp_x))),
1227 0 : cr + (y << (1 - chroma_subsamp_y)) * chroma_stride +
1228 0 : (x << ((1 - chroma_subsamp_x))),
1229 0 : luma_stride, chroma_stride, y_line_buf + (x << 1),
1230 0 : cb_line_buf + (x << (1 - chroma_subsamp_x)),
1231 0 : cr_line_buf + (x << (1 - chroma_subsamp_x)), luma_stride,
1232 : chroma_stride, 1,
1233 0 : AOMMIN(luma_subblock_size_x >> 1, width / 2 - x), bit_depth,
1234 : chroma_subsamp_y, chroma_subsamp_x);
1235 : }
1236 : }
1237 :
1238 0 : int32_t i = overlap && y ? 1 : 0;
1239 0 : int32_t j = overlap && x ? 1 : 0;
1240 :
1241 0 : if (use_high_bit_depth) {
1242 0 : add_noise_to_block_hbd(
1243 : params,
1244 0 : (uint16_t *)luma + ((y + i) << 1) * luma_stride + ((x + j) << 1),
1245 : (uint16_t *)cb +
1246 0 : ((y + i) << (1 - chroma_subsamp_y)) * chroma_stride +
1247 0 : ((x + j) << (1 - chroma_subsamp_x)),
1248 : (uint16_t *)cr +
1249 0 : ((y + i) << (1 - chroma_subsamp_y)) * chroma_stride +
1250 0 : ((x + j) << (1 - chroma_subsamp_x)),
1251 : luma_stride, chroma_stride,
1252 0 : luma_grain_block + (luma_offset_y + (i << 1)) * luma_grain_stride +
1253 0 : luma_offset_x + (j << 1),
1254 : cb_grain_block +
1255 0 : (chroma_offset_y + (i << (1 - chroma_subsamp_y))) *
1256 0 : chroma_grain_stride +
1257 0 : chroma_offset_x + (j << (1 - chroma_subsamp_x)),
1258 : cr_grain_block +
1259 0 : (chroma_offset_y + (i << (1 - chroma_subsamp_y))) *
1260 0 : chroma_grain_stride +
1261 0 : chroma_offset_x + (j << (1 - chroma_subsamp_x)),
1262 : luma_grain_stride, chroma_grain_stride,
1263 0 : AOMMIN(luma_subblock_size_y >> 1, height / 2 - y) - i,
1264 0 : AOMMIN(luma_subblock_size_x >> 1, width / 2 - x) - j, bit_depth,
1265 : chroma_subsamp_y, chroma_subsamp_x);
1266 : }
1267 : else {
1268 0 : add_noise_to_block(
1269 0 : params, luma + ((y + i) << 1) * luma_stride + ((x + j) << 1),
1270 0 : cb + ((y + i) << (1 - chroma_subsamp_y)) * chroma_stride +
1271 0 : ((x + j) << (1 - chroma_subsamp_x)),
1272 0 : cr + ((y + i) << (1 - chroma_subsamp_y)) * chroma_stride +
1273 0 : ((x + j) << (1 - chroma_subsamp_x)),
1274 : luma_stride, chroma_stride,
1275 0 : luma_grain_block + (luma_offset_y + (i << 1)) * luma_grain_stride +
1276 0 : luma_offset_x + (j << 1),
1277 : cb_grain_block +
1278 0 : (chroma_offset_y + (i << (1 - chroma_subsamp_y))) *
1279 0 : chroma_grain_stride +
1280 0 : chroma_offset_x + (j << (1 - chroma_subsamp_x)),
1281 : cr_grain_block +
1282 0 : (chroma_offset_y + (i << (1 - chroma_subsamp_y))) *
1283 0 : chroma_grain_stride +
1284 0 : chroma_offset_x + (j << (1 - chroma_subsamp_x)),
1285 : luma_grain_stride, chroma_grain_stride,
1286 0 : AOMMIN(luma_subblock_size_y >> 1, height / 2 - y) - i,
1287 0 : AOMMIN(luma_subblock_size_x >> 1, width / 2 - x) - j, bit_depth,
1288 : chroma_subsamp_y, chroma_subsamp_x);
1289 : }
1290 :
1291 0 : if (overlap) {
1292 0 : if (x) {
1293 : // Copy overlapped column bufer to line buffer
1294 0 : copy_area(y_col_buf + (luma_subblock_size_y << 1), 2,
1295 0 : y_line_buf + (x << 1), luma_stride, 2, 2);
1296 :
1297 0 : copy_area(
1298 0 : cb_col_buf + (chroma_subblock_size_y << (1 - chroma_subsamp_x)),
1299 : 2 >> chroma_subsamp_x,
1300 0 : cb_line_buf + (x << (1 - chroma_subsamp_x)), chroma_stride,
1301 : 2 >> chroma_subsamp_x, 2 >> chroma_subsamp_y);
1302 :
1303 0 : copy_area(
1304 0 : cr_col_buf + (chroma_subblock_size_y << (1 - chroma_subsamp_x)),
1305 : 2 >> chroma_subsamp_x,
1306 0 : cr_line_buf + (x << (1 - chroma_subsamp_x)), chroma_stride,
1307 : 2 >> chroma_subsamp_x, 2 >> chroma_subsamp_y);
1308 : }
1309 :
1310 : // Copy grain to the line buffer for overlap with a bottom block
1311 0 : copy_area(
1312 : luma_grain_block +
1313 0 : (luma_offset_y + luma_subblock_size_y) * luma_grain_stride +
1314 0 : luma_offset_x + ((x ? 2 : 0)),
1315 0 : luma_grain_stride, y_line_buf + ((x ? x + 1 : 0) << 1), luma_stride,
1316 0 : AOMMIN(luma_subblock_size_x, width - (x << 1)) - (x ? 2 : 0), 2);
1317 :
1318 0 : copy_area(cb_grain_block +
1319 0 : (chroma_offset_y + chroma_subblock_size_y) *
1320 0 : chroma_grain_stride +
1321 0 : chroma_offset_x + (x ? 2 >> chroma_subsamp_x : 0),
1322 : chroma_grain_stride,
1323 0 : cb_line_buf + ((x ? x + 1 : 0) << (1 - chroma_subsamp_x)),
1324 : chroma_stride,
1325 0 : AOMMIN(chroma_subblock_size_x,
1326 : ((width - (x << 1)) >> chroma_subsamp_x)) -
1327 0 : (x ? 2 >> chroma_subsamp_x : 0),
1328 : 2 >> chroma_subsamp_y);
1329 :
1330 0 : copy_area(cr_grain_block +
1331 0 : (chroma_offset_y + chroma_subblock_size_y) *
1332 0 : chroma_grain_stride +
1333 0 : chroma_offset_x + (x ? 2 >> chroma_subsamp_x : 0),
1334 : chroma_grain_stride,
1335 0 : cr_line_buf + ((x ? x + 1 : 0) << (1 - chroma_subsamp_x)),
1336 : chroma_stride,
1337 0 : AOMMIN(chroma_subblock_size_x,
1338 : ((width - (x << 1)) >> chroma_subsamp_x)) -
1339 0 : (x ? 2 >> chroma_subsamp_x : 0),
1340 : 2 >> chroma_subsamp_y);
1341 :
1342 : // Copy grain to the column buffer for overlap with the next block to
1343 : // the right
1344 :
1345 0 : copy_area(luma_grain_block + luma_offset_y * luma_grain_stride +
1346 0 : luma_offset_x + luma_subblock_size_x,
1347 : luma_grain_stride, y_col_buf, 2, 2,
1348 0 : AOMMIN(luma_subblock_size_y + 2, height - (y << 1)));
1349 :
1350 0 : copy_area(cb_grain_block + chroma_offset_y * chroma_grain_stride +
1351 0 : chroma_offset_x + chroma_subblock_size_x,
1352 : chroma_grain_stride, cb_col_buf, 2 >> chroma_subsamp_x,
1353 : 2 >> chroma_subsamp_x,
1354 0 : AOMMIN(chroma_subblock_size_y + (2 >> chroma_subsamp_y),
1355 : (height - (y << 1)) >> chroma_subsamp_y));
1356 :
1357 0 : copy_area(cr_grain_block + chroma_offset_y * chroma_grain_stride +
1358 0 : chroma_offset_x + chroma_subblock_size_x,
1359 : chroma_grain_stride, cr_col_buf, 2 >> chroma_subsamp_x,
1360 : 2 >> chroma_subsamp_x,
1361 0 : AOMMIN(chroma_subblock_size_y + (2 >> chroma_subsamp_y),
1362 : (height - (y << 1)) >> chroma_subsamp_y));
1363 : }
1364 : }
1365 : }
1366 :
1367 0 : dealloc_arrays(params, &pred_pos_luma, &pred_pos_chroma, &luma_grain_block,
1368 : &cb_grain_block, &cr_grain_block, &y_line_buf, &cb_line_buf,
1369 : &cr_line_buf, &y_col_buf, &cb_col_buf, &cr_col_buf);
1370 0 : }
1371 :
1372 : /*
1373 : void av1_film_grain_write_updated(const aom_film_grain_t *pars,
1374 : int32_t monochrome,
1375 : struct AomWriteBitBuffer *wb) {
1376 : eb_aom_wb_write_literal(wb, pars->num_y_points, 4); // max 14
1377 : for (int32_t i = 0; i < pars->num_y_points; i++) {
1378 : eb_aom_wb_write_literal(wb, pars->scaling_points_y[i][0], 8);
1379 : eb_aom_wb_write_literal(wb, pars->scaling_points_y[i][1], 8);
1380 : }
1381 :
1382 : if (!monochrome)
1383 : eb_aom_wb_write_bit(wb, pars->chroma_scaling_from_luma);
1384 :
1385 : if (!(monochrome || pars->chroma_scaling_from_luma)) {
1386 : eb_aom_wb_write_literal(wb, pars->num_cb_points, 4); // max 10
1387 : for (int32_t i = 0; i < pars->num_cb_points; i++) {
1388 : eb_aom_wb_write_literal(wb, pars->scaling_points_cb[i][0], 8);
1389 : eb_aom_wb_write_literal(wb, pars->scaling_points_cb[i][1], 8);
1390 : }
1391 :
1392 : eb_aom_wb_write_literal(wb, pars->num_cr_points, 4); // max 10
1393 : for (int32_t i = 0; i < pars->num_cr_points; i++) {
1394 : eb_aom_wb_write_literal(wb, pars->scaling_points_cr[i][0], 8);
1395 : eb_aom_wb_write_literal(wb, pars->scaling_points_cr[i][1], 8);
1396 : }
1397 : }
1398 :
1399 : eb_aom_wb_write_literal(wb, pars->scaling_shift - 8, 2); // 8 + value
1400 :
1401 : // AR coefficients
1402 : // Only sent if the corresponsing scaling function has
1403 : // more than 0 points
1404 : eb_aom_wb_write_literal(wb, pars->ar_coeff_lag, 2);
1405 :
1406 : int32_t num_pos_luma = 2 * pars->ar_coeff_lag * (pars->ar_coeff_lag + 1);
1407 : int32_t num_pos_chroma = num_pos_luma;
1408 : if (pars->num_y_points > 0) ++num_pos_chroma;
1409 :
1410 : if (pars->num_y_points)
1411 : for (int32_t i = 0; i < num_pos_luma; i++)
1412 : eb_aom_wb_write_literal(wb, pars->ar_coeffs_y[i] + 128, 8);
1413 :
1414 : if (pars->num_cb_points || pars->chroma_scaling_from_luma)
1415 : for (int32_t i = 0; i < num_pos_chroma; i++)
1416 : eb_aom_wb_write_literal(wb, pars->ar_coeffs_cb[i] + 128, 8);
1417 :
1418 : if (pars->num_cr_points || pars->chroma_scaling_from_luma)
1419 : for (int32_t i = 0; i < num_pos_chroma; i++)
1420 : eb_aom_wb_write_literal(wb, pars->ar_coeffs_cr[i] + 128, 8);
1421 :
1422 : eb_aom_wb_write_literal(wb, pars->ar_coeff_shift - 6, 2); // 8 + value
1423 :
1424 : eb_aom_wb_write_literal(wb, pars->grain_scale_shift, 2);
1425 :
1426 : if (pars->num_cb_points) {
1427 : eb_aom_wb_write_literal(wb, pars->cb_mult, 8);
1428 : eb_aom_wb_write_literal(wb, pars->cb_luma_mult, 8);
1429 : eb_aom_wb_write_literal(wb, pars->cb_offset, 9);
1430 : }
1431 :
1432 : if (pars->num_cr_points) {
1433 : eb_aom_wb_write_literal(wb, pars->cr_mult, 8);
1434 : eb_aom_wb_write_literal(wb, pars->cr_luma_mult, 8);
1435 : eb_aom_wb_write_literal(wb, pars->cr_offset, 9);
1436 : }
1437 :
1438 : eb_aom_wb_write_bit(wb, pars->overlap_flag);
1439 :
1440 : eb_aom_wb_write_bit(wb, pars->clip_to_restricted_range);
1441 : }
1442 : */
1443 : /*
1444 : void av1_film_grain_read_updated(aom_film_grain_t *pars,
1445 : int32_t monochrome,
1446 : struct aom_read_bit_buffer *rb,
1447 : struct aom_internal_error_info *error) {
1448 : // Scaling functions parameters
1449 : pars->num_y_points = aom_rb_read_literal(rb, 4); // max 14
1450 : if (pars->num_y_points > 14)
1451 : aom_internal_error(error, AOM_CODEC_UNSUP_BITSTREAM,
1452 : "Number of points for film grain luma scaling function "
1453 : "exceeds the maximum value.");
1454 : for (int32_t i = 0; i < pars->num_y_points; i++) {
1455 : pars->scaling_points_y[i][0] = aom_rb_read_literal(rb, 8);
1456 : if (i && pars->scaling_points_y[i - 1][0] >= pars->scaling_points_y[i][0])
1457 : aom_internal_error(error, AOM_CODEC_UNSUP_BITSTREAM,
1458 : "First coordinateg of the scaling function points "
1459 : "shall be increasing.");
1460 : pars->scaling_points_y[i][1] = aom_rb_read_literal(rb, 8);
1461 : }
1462 :
1463 : if (!monochrome)
1464 : pars->chroma_scaling_from_luma = aom_rb_read_bit(rb);
1465 :
1466 : if (monochrome || pars->chroma_scaling_from_luma) {
1467 : pars->num_cb_points = 0;
1468 : pars->num_cr_points = 0;
1469 : } else {
1470 : pars->num_cb_points = aom_rb_read_literal(rb, 4); // max 10
1471 : if (pars->num_cb_points > 10)
1472 : aom_internal_error(error, AOM_CODEC_UNSUP_BITSTREAM,
1473 : "Number of points for film grain cb scaling function "
1474 : "exceeds the maximum value.");
1475 : for (int32_t i = 0; i < pars->num_cb_points; i++) {
1476 : pars->scaling_points_cb[i][0] = aom_rb_read_literal(rb, 8);
1477 : if (i &&
1478 : pars->scaling_points_cb[i - 1][0] >= pars->scaling_points_cb[i][0])
1479 : aom_internal_error(error, AOM_CODEC_UNSUP_BITSTREAM,
1480 : "First coordinate of the scaling function points "
1481 : "shall be increasing.");
1482 : pars->scaling_points_cb[i][1] = aom_rb_read_literal(rb, 8);
1483 : }
1484 :
1485 : pars->num_cr_points = aom_rb_read_literal(rb, 4); // max 10
1486 : if (pars->num_cr_points > 10)
1487 : aom_internal_error(error, AOM_CODEC_UNSUP_BITSTREAM,
1488 : "Number of points for film grain cr scaling function "
1489 : "exceeds the maximum value.");
1490 : for (int32_t i = 0; i < pars->num_cr_points; i++) {
1491 : pars->scaling_points_cr[i][0] = aom_rb_read_literal(rb, 8);
1492 : if (i &&
1493 : pars->scaling_points_cr[i - 1][0] >= pars->scaling_points_cr[i][0])
1494 : aom_internal_error(error, AOM_CODEC_UNSUP_BITSTREAM,
1495 : "First coordinate of the scaling function points "
1496 : "shall be increasing.");
1497 : pars->scaling_points_cr[i][1] = aom_rb_read_literal(rb, 8);
1498 : }
1499 : }
1500 :
1501 : pars->scaling_shift = aom_rb_read_literal(rb, 2) + 8; // 8 + value
1502 :
1503 : // AR coefficients
1504 : // Only sent if the corresponsing scaling function has
1505 : // more than 0 points
1506 : pars->ar_coeff_lag = aom_rb_read_literal(rb, 2);
1507 :
1508 : int32_t num_pos_luma = 2 * pars->ar_coeff_lag * (pars->ar_coeff_lag + 1);
1509 : int32_t num_pos_chroma = num_pos_luma;
1510 : if (pars->num_y_points > 0) ++num_pos_chroma;
1511 :
1512 : if (pars->num_y_points)
1513 : for (int32_t i = 0; i < num_pos_luma; i++)
1514 : pars->ar_coeffs_y[i] = aom_rb_read_literal(rb, 8) - 128;
1515 :
1516 : if (pars->num_cb_points || pars->chroma_scaling_from_luma)
1517 : for (int32_t i = 0; i < num_pos_chroma; i++)
1518 : pars->ar_coeffs_cb[i] = aom_rb_read_literal(rb, 8) - 128;
1519 :
1520 : if (pars->num_cr_points || pars->chroma_scaling_from_luma)
1521 : for (int32_t i = 0; i < num_pos_chroma; i++)
1522 : pars->ar_coeffs_cr[i] = aom_rb_read_literal(rb, 8) - 128;
1523 :
1524 : pars->ar_coeff_shift = aom_rb_read_literal(rb, 2) + 6; // 6 + value
1525 :
1526 : pars->grain_scale_shift = aom_rb_read_literal(rb, 2);
1527 :
1528 : if (pars->num_cb_points) {
1529 : pars->cb_mult = aom_rb_read_literal(rb, 8);
1530 : pars->cb_luma_mult = aom_rb_read_literal(rb, 8);
1531 : pars->cb_offset = aom_rb_read_literal(rb, 9);
1532 : }
1533 :
1534 : if (pars->num_cr_points) {
1535 : pars->cr_mult = aom_rb_read_literal(rb, 8);
1536 : pars->cr_luma_mult = aom_rb_read_literal(rb, 8);
1537 : pars->cr_offset = aom_rb_read_literal(rb, 9);
1538 : }
1539 :
1540 : pars->overlap_flag = aom_rb_read_bit(rb);
1541 :
1542 : pars->clip_to_restricted_range = aom_rb_read_bit(rb);
1543 : }
1544 : */
|