1 /*
2 * H.26L/H.264/AVC/JVT/14496-10/... encoder/decoder
3 * Copyright (c) 2003 Michael Niedermayer <michaelni@gmx.at>
4 *
5 * This file is part of FFmpeg.
6 *
7 * FFmpeg is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
11 *
12 * FFmpeg is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * Lesser General Public License for more details.
16 *
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with FFmpeg; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20 */
21
22 /**
23 * @file
24 * H.264 / AVC / MPEG-4 part10 prediction functions.
25 * @author Michael Niedermayer <michaelni@gmx.at>
26 */
27
28 #include "config.h"
29 #include "libavutil/attributes.h"
30 #include "libavutil/avassert.h"
31 #include "libavutil/intreadwrite.h"
32 #include "codec_id.h"
33 #include "h264pred.h"
34 #include "mathops.h"
35
36 #define BIT_DEPTH 8
37 #include "h264pred_template.c"
38 #undef BIT_DEPTH
39
40 #define BIT_DEPTH 9
41 #include "h264pred_template.c"
42 #undef BIT_DEPTH
43
44 #define BIT_DEPTH 10
45 #include "h264pred_template.c"
46 #undef BIT_DEPTH
47
48 #define BIT_DEPTH 12
49 #include "h264pred_template.c"
50 #undef BIT_DEPTH
51
52 #define BIT_DEPTH 14
53 #include "h264pred_template.c"
54 #undef BIT_DEPTH
55
pred4x4_127_dc_c(uint8_t * src,const uint8_t * topright,ptrdiff_t _stride)56 static void pred4x4_127_dc_c(uint8_t *src, const uint8_t *topright,
57 ptrdiff_t _stride)
58 {
59 int stride = _stride;
60 const uint32_t a = 0x7F7F7F7FU;
61
62 AV_WN32A(src + 0 * stride, a);
63 AV_WN32A(src + 1 * stride, a);
64 AV_WN32A(src + 2 * stride, a);
65 AV_WN32A(src + 3 * stride, a);
66 }
67
pred4x4_129_dc_c(uint8_t * src,const uint8_t * topright,ptrdiff_t _stride)68 static void pred4x4_129_dc_c(uint8_t *src, const uint8_t *topright,
69 ptrdiff_t _stride)
70 {
71 int stride = _stride;
72 const uint32_t a = 0x81818181U;
73
74 AV_WN32A(src + 0 * stride, a);
75 AV_WN32A(src + 1 * stride, a);
76 AV_WN32A(src + 2 * stride, a);
77 AV_WN32A(src + 3 * stride, a);
78 }
79
pred4x4_vertical_vp8_c(uint8_t * src,const uint8_t * topright,ptrdiff_t stride)80 static void pred4x4_vertical_vp8_c(uint8_t *src, const uint8_t *topright,
81 ptrdiff_t stride)
82 {
83 const unsigned lt = src[-1-1*stride];
84 LOAD_TOP_EDGE
85 LOAD_TOP_RIGHT_EDGE
86 uint32_t v = PACK_4U8((lt + 2*t0 + t1 + 2) >> 2,
87 (t0 + 2*t1 + t2 + 2) >> 2,
88 (t1 + 2*t2 + t3 + 2) >> 2,
89 (t2 + 2*t3 + t4 + 2) >> 2);
90
91 AV_WN32A(src+0*stride, v);
92 AV_WN32A(src+1*stride, v);
93 AV_WN32A(src+2*stride, v);
94 AV_WN32A(src+3*stride, v);
95 }
96
pred4x4_horizontal_vp8_c(uint8_t * src,const uint8_t * topright,ptrdiff_t stride)97 static void pred4x4_horizontal_vp8_c(uint8_t *src, const uint8_t *topright,
98 ptrdiff_t stride)
99 {
100 const unsigned lt = src[-1-1*stride];
101 LOAD_LEFT_EDGE
102
103 AV_WN32A(src+0*stride, ((lt + 2*l0 + l1 + 2) >> 2)*0x01010101);
104 AV_WN32A(src+1*stride, ((l0 + 2*l1 + l2 + 2) >> 2)*0x01010101);
105 AV_WN32A(src+2*stride, ((l1 + 2*l2 + l3 + 2) >> 2)*0x01010101);
106 AV_WN32A(src+3*stride, ((l2 + 2*l3 + l3 + 2) >> 2)*0x01010101);
107 }
108
pred4x4_down_left_svq3_c(uint8_t * src,const uint8_t * topright,ptrdiff_t stride)109 static void pred4x4_down_left_svq3_c(uint8_t *src, const uint8_t *topright,
110 ptrdiff_t stride)
111 {
112 LOAD_TOP_EDGE
113 LOAD_LEFT_EDGE
114
115 src[0+0*stride]=(l1 + t1)>>1;
116 src[1+0*stride]=
117 src[0+1*stride]=(l2 + t2)>>1;
118 src[2+0*stride]=
119 src[1+1*stride]=
120 src[0+2*stride]=
121 src[3+0*stride]=
122 src[2+1*stride]=
123 src[1+2*stride]=
124 src[0+3*stride]=
125 src[3+1*stride]=
126 src[2+2*stride]=
127 src[1+3*stride]=
128 src[3+2*stride]=
129 src[2+3*stride]=
130 src[3+3*stride]=(l3 + t3)>>1;
131 }
132
pred4x4_down_left_rv40_c(uint8_t * src,const uint8_t * topright,ptrdiff_t stride)133 static void pred4x4_down_left_rv40_c(uint8_t *src, const uint8_t *topright,
134 ptrdiff_t stride)
135 {
136 LOAD_TOP_EDGE
137 LOAD_TOP_RIGHT_EDGE
138 LOAD_LEFT_EDGE
139 LOAD_DOWN_LEFT_EDGE
140
141 src[0+0*stride]=(t0 + t2 + 2*t1 + 2 + l0 + l2 + 2*l1 + 2)>>3;
142 src[1+0*stride]=
143 src[0+1*stride]=(t1 + t3 + 2*t2 + 2 + l1 + l3 + 2*l2 + 2)>>3;
144 src[2+0*stride]=
145 src[1+1*stride]=
146 src[0+2*stride]=(t2 + t4 + 2*t3 + 2 + l2 + l4 + 2*l3 + 2)>>3;
147 src[3+0*stride]=
148 src[2+1*stride]=
149 src[1+2*stride]=
150 src[0+3*stride]=(t3 + t5 + 2*t4 + 2 + l3 + l5 + 2*l4 + 2)>>3;
151 src[3+1*stride]=
152 src[2+2*stride]=
153 src[1+3*stride]=(t4 + t6 + 2*t5 + 2 + l4 + l6 + 2*l5 + 2)>>3;
154 src[3+2*stride]=
155 src[2+3*stride]=(t5 + t7 + 2*t6 + 2 + l5 + l7 + 2*l6 + 2)>>3;
156 src[3+3*stride]=(t6 + t7 + 1 + l6 + l7 + 1)>>2;
157 }
158
pred4x4_down_left_rv40_nodown_c(uint8_t * src,const uint8_t * topright,ptrdiff_t stride)159 static void pred4x4_down_left_rv40_nodown_c(uint8_t *src,
160 const uint8_t *topright,
161 ptrdiff_t stride)
162 {
163 LOAD_TOP_EDGE
164 LOAD_TOP_RIGHT_EDGE
165 LOAD_LEFT_EDGE
166
167 src[0+0*stride]=(t0 + t2 + 2*t1 + 2 + l0 + l2 + 2*l1 + 2)>>3;
168 src[1+0*stride]=
169 src[0+1*stride]=(t1 + t3 + 2*t2 + 2 + l1 + l3 + 2*l2 + 2)>>3;
170 src[2+0*stride]=
171 src[1+1*stride]=
172 src[0+2*stride]=(t2 + t4 + 2*t3 + 2 + l2 + 3*l3 + 2)>>3;
173 src[3+0*stride]=
174 src[2+1*stride]=
175 src[1+2*stride]=
176 src[0+3*stride]=(t3 + t5 + 2*t4 + 2 + l3*4 + 2)>>3;
177 src[3+1*stride]=
178 src[2+2*stride]=
179 src[1+3*stride]=(t4 + t6 + 2*t5 + 2 + l3*4 + 2)>>3;
180 src[3+2*stride]=
181 src[2+3*stride]=(t5 + t7 + 2*t6 + 2 + l3*4 + 2)>>3;
182 src[3+3*stride]=(t6 + t7 + 1 + 2*l3 + 1)>>2;
183 }
184
pred4x4_vertical_left_rv40(uint8_t * src,const uint8_t * topright,ptrdiff_t stride,const int l0,const int l1,const int l2,const int l3,const int l4)185 static void pred4x4_vertical_left_rv40(uint8_t *src, const uint8_t *topright,
186 ptrdiff_t stride,
187 const int l0, const int l1, const int l2,
188 const int l3, const int l4)
189 {
190 LOAD_TOP_EDGE
191 LOAD_TOP_RIGHT_EDGE
192
193 src[0+0*stride]=(2*t0 + 2*t1 + l1 + 2*l2 + l3 + 4)>>3;
194 src[1+0*stride]=
195 src[0+2*stride]=(t1 + t2 + 1)>>1;
196 src[2+0*stride]=
197 src[1+2*stride]=(t2 + t3 + 1)>>1;
198 src[3+0*stride]=
199 src[2+2*stride]=(t3 + t4+ 1)>>1;
200 src[3+2*stride]=(t4 + t5+ 1)>>1;
201 src[0+1*stride]=(t0 + 2*t1 + t2 + l2 + 2*l3 + l4 + 4)>>3;
202 src[1+1*stride]=
203 src[0+3*stride]=(t1 + 2*t2 + t3 + 2)>>2;
204 src[2+1*stride]=
205 src[1+3*stride]=(t2 + 2*t3 + t4 + 2)>>2;
206 src[3+1*stride]=
207 src[2+3*stride]=(t3 + 2*t4 + t5 + 2)>>2;
208 src[3+3*stride]=(t4 + 2*t5 + t6 + 2)>>2;
209 }
210
pred4x4_vertical_left_rv40_c(uint8_t * src,const uint8_t * topright,ptrdiff_t stride)211 static void pred4x4_vertical_left_rv40_c(uint8_t *src, const uint8_t *topright,
212 ptrdiff_t stride)
213 {
214 LOAD_LEFT_EDGE
215 LOAD_DOWN_LEFT_EDGE
216
217 pred4x4_vertical_left_rv40(src, topright, stride, l0, l1, l2, l3, l4);
218 }
219
pred4x4_vertical_left_rv40_nodown_c(uint8_t * src,const uint8_t * topright,ptrdiff_t stride)220 static void pred4x4_vertical_left_rv40_nodown_c(uint8_t *src,
221 const uint8_t *topright,
222 ptrdiff_t stride)
223 {
224 LOAD_LEFT_EDGE
225
226 pred4x4_vertical_left_rv40(src, topright, stride, l0, l1, l2, l3, l3);
227 }
228
pred4x4_vertical_left_vp8_c(uint8_t * src,const uint8_t * topright,ptrdiff_t stride)229 static void pred4x4_vertical_left_vp8_c(uint8_t *src, const uint8_t *topright,
230 ptrdiff_t stride)
231 {
232 LOAD_TOP_EDGE
233 LOAD_TOP_RIGHT_EDGE
234
235 src[0+0*stride]=(t0 + t1 + 1)>>1;
236 src[1+0*stride]=
237 src[0+2*stride]=(t1 + t2 + 1)>>1;
238 src[2+0*stride]=
239 src[1+2*stride]=(t2 + t3 + 1)>>1;
240 src[3+0*stride]=
241 src[2+2*stride]=(t3 + t4 + 1)>>1;
242 src[0+1*stride]=(t0 + 2*t1 + t2 + 2)>>2;
243 src[1+1*stride]=
244 src[0+3*stride]=(t1 + 2*t2 + t3 + 2)>>2;
245 src[2+1*stride]=
246 src[1+3*stride]=(t2 + 2*t3 + t4 + 2)>>2;
247 src[3+1*stride]=
248 src[2+3*stride]=(t3 + 2*t4 + t5 + 2)>>2;
249 src[3+2*stride]=(t4 + 2*t5 + t6 + 2)>>2;
250 src[3+3*stride]=(t5 + 2*t6 + t7 + 2)>>2;
251 }
252
pred4x4_horizontal_up_rv40_c(uint8_t * src,const uint8_t * topright,ptrdiff_t stride)253 static void pred4x4_horizontal_up_rv40_c(uint8_t *src, const uint8_t *topright,
254 ptrdiff_t stride)
255 {
256 LOAD_LEFT_EDGE
257 LOAD_DOWN_LEFT_EDGE
258 LOAD_TOP_EDGE
259 LOAD_TOP_RIGHT_EDGE
260
261 src[0+0*stride]=(t1 + 2*t2 + t3 + 2*l0 + 2*l1 + 4)>>3;
262 src[1+0*stride]=(t2 + 2*t3 + t4 + l0 + 2*l1 + l2 + 4)>>3;
263 src[2+0*stride]=
264 src[0+1*stride]=(t3 + 2*t4 + t5 + 2*l1 + 2*l2 + 4)>>3;
265 src[3+0*stride]=
266 src[1+1*stride]=(t4 + 2*t5 + t6 + l1 + 2*l2 + l3 + 4)>>3;
267 src[2+1*stride]=
268 src[0+2*stride]=(t5 + 2*t6 + t7 + 2*l2 + 2*l3 + 4)>>3;
269 src[3+1*stride]=
270 src[1+2*stride]=(t6 + 3*t7 + l2 + 3*l3 + 4)>>3;
271 src[3+2*stride]=
272 src[1+3*stride]=(l3 + 2*l4 + l5 + 2)>>2;
273 src[0+3*stride]=
274 src[2+2*stride]=(t6 + t7 + l3 + l4 + 2)>>2;
275 src[2+3*stride]=(l4 + l5 + 1)>>1;
276 src[3+3*stride]=(l4 + 2*l5 + l6 + 2)>>2;
277 }
278
pred4x4_horizontal_up_rv40_nodown_c(uint8_t * src,const uint8_t * topright,ptrdiff_t stride)279 static void pred4x4_horizontal_up_rv40_nodown_c(uint8_t *src,
280 const uint8_t *topright,
281 ptrdiff_t stride)
282 {
283 LOAD_LEFT_EDGE
284 LOAD_TOP_EDGE
285 LOAD_TOP_RIGHT_EDGE
286
287 src[0+0*stride]=(t1 + 2*t2 + t3 + 2*l0 + 2*l1 + 4)>>3;
288 src[1+0*stride]=(t2 + 2*t3 + t4 + l0 + 2*l1 + l2 + 4)>>3;
289 src[2+0*stride]=
290 src[0+1*stride]=(t3 + 2*t4 + t5 + 2*l1 + 2*l2 + 4)>>3;
291 src[3+0*stride]=
292 src[1+1*stride]=(t4 + 2*t5 + t6 + l1 + 2*l2 + l3 + 4)>>3;
293 src[2+1*stride]=
294 src[0+2*stride]=(t5 + 2*t6 + t7 + 2*l2 + 2*l3 + 4)>>3;
295 src[3+1*stride]=
296 src[1+2*stride]=(t6 + 3*t7 + l2 + 3*l3 + 4)>>3;
297 src[3+2*stride]=
298 src[1+3*stride]=l3;
299 src[0+3*stride]=
300 src[2+2*stride]=(t6 + t7 + 2*l3 + 2)>>2;
301 src[2+3*stride]=
302 src[3+3*stride]=l3;
303 }
304
pred4x4_tm_vp8_c(uint8_t * src,const uint8_t * topright,ptrdiff_t stride)305 static void pred4x4_tm_vp8_c(uint8_t *src, const uint8_t *topright,
306 ptrdiff_t stride)
307 {
308 const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP - src[-1-stride];
309 uint8_t *top = src-stride;
310 int y;
311
312 for (y = 0; y < 4; y++) {
313 const uint8_t *cm_in = cm + src[-1];
314 src[0] = cm_in[top[0]];
315 src[1] = cm_in[top[1]];
316 src[2] = cm_in[top[2]];
317 src[3] = cm_in[top[3]];
318 src += stride;
319 }
320 }
321
pred16x16_plane_svq3_c(uint8_t * src,ptrdiff_t stride)322 static void pred16x16_plane_svq3_c(uint8_t *src, ptrdiff_t stride)
323 {
324 pred16x16_plane_compat_8_c(src, stride, 1, 0);
325 }
326
pred16x16_plane_rv40_c(uint8_t * src,ptrdiff_t stride)327 static void pred16x16_plane_rv40_c(uint8_t *src, ptrdiff_t stride)
328 {
329 pred16x16_plane_compat_8_c(src, stride, 0, 1);
330 }
331
pred16x16_tm_vp8_c(uint8_t * src,ptrdiff_t stride)332 static void pred16x16_tm_vp8_c(uint8_t *src, ptrdiff_t stride)
333 {
334 const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP - src[-1-stride];
335 uint8_t *top = src-stride;
336 int y;
337
338 for (y = 0; y < 16; y++) {
339 const uint8_t *cm_in = cm + src[-1];
340 src[0] = cm_in[top[0]];
341 src[1] = cm_in[top[1]];
342 src[2] = cm_in[top[2]];
343 src[3] = cm_in[top[3]];
344 src[4] = cm_in[top[4]];
345 src[5] = cm_in[top[5]];
346 src[6] = cm_in[top[6]];
347 src[7] = cm_in[top[7]];
348 src[8] = cm_in[top[8]];
349 src[9] = cm_in[top[9]];
350 src[10] = cm_in[top[10]];
351 src[11] = cm_in[top[11]];
352 src[12] = cm_in[top[12]];
353 src[13] = cm_in[top[13]];
354 src[14] = cm_in[top[14]];
355 src[15] = cm_in[top[15]];
356 src += stride;
357 }
358 }
359
pred8x8_left_dc_rv40_c(uint8_t * src,ptrdiff_t stride)360 static void pred8x8_left_dc_rv40_c(uint8_t *src, ptrdiff_t stride)
361 {
362 int i;
363 unsigned dc0;
364
365 dc0=0;
366 for(i=0;i<8; i++)
367 dc0+= src[-1+i*stride];
368 dc0= 0x01010101*((dc0 + 4)>>3);
369
370 for(i=0; i<8; i++){
371 ((uint32_t*)(src+i*stride))[0]=
372 ((uint32_t*)(src+i*stride))[1]= dc0;
373 }
374 }
375
pred8x8_top_dc_rv40_c(uint8_t * src,ptrdiff_t stride)376 static void pred8x8_top_dc_rv40_c(uint8_t *src, ptrdiff_t stride)
377 {
378 int i;
379 unsigned dc0;
380
381 dc0=0;
382 for(i=0;i<8; i++)
383 dc0+= src[i-stride];
384 dc0= 0x01010101*((dc0 + 4)>>3);
385
386 for(i=0; i<8; i++){
387 ((uint32_t*)(src+i*stride))[0]=
388 ((uint32_t*)(src+i*stride))[1]= dc0;
389 }
390 }
391
pred8x8_dc_rv40_c(uint8_t * src,ptrdiff_t stride)392 static void pred8x8_dc_rv40_c(uint8_t *src, ptrdiff_t stride)
393 {
394 int i;
395 unsigned dc0 = 0;
396
397 for(i=0;i<4; i++){
398 dc0+= src[-1+i*stride] + src[i-stride];
399 dc0+= src[4+i-stride];
400 dc0+= src[-1+(i+4)*stride];
401 }
402 dc0= 0x01010101*((dc0 + 8)>>4);
403
404 for(i=0; i<4; i++){
405 ((uint32_t*)(src+i*stride))[0]= dc0;
406 ((uint32_t*)(src+i*stride))[1]= dc0;
407 }
408 for(i=4; i<8; i++){
409 ((uint32_t*)(src+i*stride))[0]= dc0;
410 ((uint32_t*)(src+i*stride))[1]= dc0;
411 }
412 }
413
pred8x8_tm_vp8_c(uint8_t * src,ptrdiff_t stride)414 static void pred8x8_tm_vp8_c(uint8_t *src, ptrdiff_t stride)
415 {
416 const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP - src[-1-stride];
417 uint8_t *top = src-stride;
418 int y;
419
420 for (y = 0; y < 8; y++) {
421 const uint8_t *cm_in = cm + src[-1];
422 src[0] = cm_in[top[0]];
423 src[1] = cm_in[top[1]];
424 src[2] = cm_in[top[2]];
425 src[3] = cm_in[top[3]];
426 src[4] = cm_in[top[4]];
427 src[5] = cm_in[top[5]];
428 src[6] = cm_in[top[6]];
429 src[7] = cm_in[top[7]];
430 src += stride;
431 }
432 }
433
434 /**
435 * Set the intra prediction function pointers.
436 */
ff_h264_pred_init(H264PredContext * h,int codec_id,const int bit_depth,int chroma_format_idc)437 av_cold void ff_h264_pred_init(H264PredContext *h, int codec_id,
438 const int bit_depth,
439 int chroma_format_idc)
440 {
441 #undef FUNC
442 #undef FUNCC
443 #define FUNC(a, depth) a ## _ ## depth
444 #define FUNCC(a, depth) a ## _ ## depth ## _c
445 #define FUNCD(a) a ## _c
446
447 #define H264_PRED(depth) \
448 h->pred4x4[VERT_PRED ] = FUNCC(pred4x4_vertical, depth);\
449 h->pred4x4[HOR_PRED ] = FUNCC(pred4x4_horizontal, depth);\
450 h->pred4x4[DC_PRED ] = FUNCC(pred4x4_dc, depth);\
451 h->pred4x4[DIAG_DOWN_LEFT_PRED ] = FUNCC(pred4x4_down_left, depth);\
452 h->pred4x4[DIAG_DOWN_RIGHT_PRED] = FUNCC(pred4x4_down_right, depth);\
453 h->pred4x4[VERT_RIGHT_PRED ] = FUNCC(pred4x4_vertical_right, depth);\
454 h->pred4x4[HOR_DOWN_PRED ] = FUNCC(pred4x4_horizontal_down, depth);\
455 h->pred4x4[VERT_LEFT_PRED ] = FUNCC(pred4x4_vertical_left, depth);\
456 h->pred4x4[HOR_UP_PRED ] = FUNCC(pred4x4_horizontal_up, depth);\
457 h->pred4x4[LEFT_DC_PRED ] = FUNCC(pred4x4_left_dc, depth);\
458 h->pred4x4[TOP_DC_PRED ] = FUNCC(pred4x4_top_dc, depth);\
459 if (depth > 8 || codec_id != AV_CODEC_ID_VP8)\
460 h->pred4x4[DC_128_PRED ] = FUNCC(pred4x4_128_dc, depth);\
461 \
462 h->pred8x8l[VERT_PRED ]= FUNCC(pred8x8l_vertical , depth);\
463 h->pred8x8l[HOR_PRED ]= FUNCC(pred8x8l_horizontal , depth);\
464 h->pred8x8l[DC_PRED ]= FUNCC(pred8x8l_dc , depth);\
465 h->pred8x8l[DIAG_DOWN_LEFT_PRED ]= FUNCC(pred8x8l_down_left , depth);\
466 h->pred8x8l[DIAG_DOWN_RIGHT_PRED]= FUNCC(pred8x8l_down_right , depth);\
467 h->pred8x8l[VERT_RIGHT_PRED ]= FUNCC(pred8x8l_vertical_right , depth);\
468 h->pred8x8l[HOR_DOWN_PRED ]= FUNCC(pred8x8l_horizontal_down , depth);\
469 h->pred8x8l[VERT_LEFT_PRED ]= FUNCC(pred8x8l_vertical_left , depth);\
470 h->pred8x8l[HOR_UP_PRED ]= FUNCC(pred8x8l_horizontal_up , depth);\
471 h->pred8x8l[LEFT_DC_PRED ]= FUNCC(pred8x8l_left_dc , depth);\
472 h->pred8x8l[TOP_DC_PRED ]= FUNCC(pred8x8l_top_dc , depth);\
473 h->pred8x8l[DC_128_PRED ]= FUNCC(pred8x8l_128_dc , depth);\
474 \
475 if (chroma_format_idc <= 1) {\
476 h->pred8x8[VERT_PRED8x8 ]= FUNCC(pred8x8_vertical , depth);\
477 h->pred8x8[HOR_PRED8x8 ]= FUNCC(pred8x8_horizontal , depth);\
478 h->pred8x8[PLANE_PRED8x8] = FUNCC(pred8x8_plane, depth);\
479 } else {\
480 h->pred8x8[VERT_PRED8x8 ]= FUNCC(pred8x16_vertical , depth);\
481 h->pred8x8[HOR_PRED8x8 ]= FUNCC(pred8x16_horizontal , depth);\
482 h->pred8x8[PLANE_PRED8x8] = FUNCC(pred8x16_plane, depth);\
483 }\
484 if (depth > 8 || (codec_id != AV_CODEC_ID_RV40 && \
485 codec_id != AV_CODEC_ID_VP7 && \
486 codec_id != AV_CODEC_ID_VP8)) { \
487 if (chroma_format_idc <= 1) {\
488 h->pred8x8[DC_PRED8x8 ]= FUNCC(pred8x8_dc , depth);\
489 h->pred8x8[LEFT_DC_PRED8x8]= FUNCC(pred8x8_left_dc , depth);\
490 h->pred8x8[TOP_DC_PRED8x8 ]= FUNCC(pred8x8_top_dc , depth);\
491 h->pred8x8[ALZHEIMER_DC_L0T_PRED8x8 ]= FUNC(pred8x8_mad_cow_dc_l0t, depth);\
492 h->pred8x8[ALZHEIMER_DC_0LT_PRED8x8 ]= FUNC(pred8x8_mad_cow_dc_0lt, depth);\
493 h->pred8x8[ALZHEIMER_DC_L00_PRED8x8 ]= FUNC(pred8x8_mad_cow_dc_l00, depth);\
494 h->pred8x8[ALZHEIMER_DC_0L0_PRED8x8 ]= FUNC(pred8x8_mad_cow_dc_0l0, depth);\
495 } else {\
496 h->pred8x8[DC_PRED8x8 ]= FUNCC(pred8x16_dc , depth);\
497 h->pred8x8[LEFT_DC_PRED8x8]= FUNCC(pred8x16_left_dc , depth);\
498 h->pred8x8[TOP_DC_PRED8x8 ]= FUNCC(pred8x16_top_dc , depth);\
499 h->pred8x8[ALZHEIMER_DC_L0T_PRED8x8 ]= FUNC(pred8x16_mad_cow_dc_l0t, depth);\
500 h->pred8x8[ALZHEIMER_DC_0LT_PRED8x8 ]= FUNC(pred8x16_mad_cow_dc_0lt, depth);\
501 h->pred8x8[ALZHEIMER_DC_L00_PRED8x8 ]= FUNC(pred8x16_mad_cow_dc_l00, depth);\
502 h->pred8x8[ALZHEIMER_DC_0L0_PRED8x8 ]= FUNC(pred8x16_mad_cow_dc_0l0, depth);\
503 }\
504 }else{\
505 h->pred8x8[DC_PRED8x8 ]= FUNCD(pred8x8_dc_rv40);\
506 h->pred8x8[LEFT_DC_PRED8x8]= FUNCD(pred8x8_left_dc_rv40);\
507 h->pred8x8[TOP_DC_PRED8x8 ]= FUNCD(pred8x8_top_dc_rv40);\
508 }\
509 if (chroma_format_idc <= 1) {\
510 h->pred8x8[DC_128_PRED8x8 ]= FUNCC(pred8x8_128_dc , depth);\
511 } else {\
512 h->pred8x8[DC_128_PRED8x8 ]= FUNCC(pred8x16_128_dc , depth);\
513 }\
514 \
515 h->pred16x16[DC_PRED8x8 ]= FUNCC(pred16x16_dc , depth);\
516 h->pred16x16[VERT_PRED8x8 ]= FUNCC(pred16x16_vertical , depth);\
517 h->pred16x16[HOR_PRED8x8 ]= FUNCC(pred16x16_horizontal , depth);\
518 h->pred16x16[PLANE_PRED8x8 ]= FUNCC(pred16x16_plane , depth);\
519 h->pred16x16[LEFT_DC_PRED8x8]= FUNCC(pred16x16_left_dc , depth);\
520 h->pred16x16[TOP_DC_PRED8x8 ]= FUNCC(pred16x16_top_dc , depth);\
521 h->pred16x16[DC_128_PRED8x8 ]= FUNCC(pred16x16_128_dc , depth);\
522 \
523 /* special lossless h/v prediction for H.264 */ \
524 h->pred4x4_add [VERT_PRED ]= FUNCC(pred4x4_vertical_add , depth);\
525 h->pred4x4_add [ HOR_PRED ]= FUNCC(pred4x4_horizontal_add , depth);\
526 h->pred8x8l_add [VERT_PRED ]= FUNCC(pred8x8l_vertical_add , depth);\
527 h->pred8x8l_add [ HOR_PRED ]= FUNCC(pred8x8l_horizontal_add , depth);\
528 h->pred8x8l_filter_add [VERT_PRED ]= FUNCC(pred8x8l_vertical_filter_add , depth);\
529 h->pred8x8l_filter_add [ HOR_PRED ]= FUNCC(pred8x8l_horizontal_filter_add , depth);\
530 if (chroma_format_idc <= 1) {\
531 h->pred8x8_add[VERT_PRED8x8] = FUNCC(pred8x8_vertical_add, depth);\
532 h->pred8x8_add[ HOR_PRED8x8] = FUNCC(pred8x8_horizontal_add, depth);\
533 } else {\
534 h->pred8x8_add [VERT_PRED8x8]= FUNCC(pred8x16_vertical_add , depth);\
535 h->pred8x8_add [ HOR_PRED8x8]= FUNCC(pred8x16_horizontal_add , depth);\
536 }\
537 h->pred16x16_add[VERT_PRED8x8]= FUNCC(pred16x16_vertical_add , depth);\
538 h->pred16x16_add[ HOR_PRED8x8]= FUNCC(pred16x16_horizontal_add , depth);\
539
540 switch (bit_depth) {
541 case 9:
542 H264_PRED(9)
543 break;
544 case 10:
545 H264_PRED(10)
546 break;
547 case 12:
548 H264_PRED(12)
549 break;
550 case 14:
551 H264_PRED(14)
552 break;
553 default:
554 av_assert0(bit_depth<=8);
555 H264_PRED(8)
556 switch (codec_id) {
557 case AV_CODEC_ID_SVQ3:
558 h->pred4x4[DIAG_DOWN_LEFT_PRED] = FUNCD(pred4x4_down_left_svq3);
559 h->pred16x16[PLANE_PRED8x8 ] = FUNCD(pred16x16_plane_svq3);
560 break;
561 case AV_CODEC_ID_RV40:
562 h->pred4x4[DIAG_DOWN_LEFT_PRED] = FUNCD(pred4x4_down_left_rv40);
563 h->pred4x4[VERT_LEFT_PRED ] = FUNCD(pred4x4_vertical_left_rv40);
564 h->pred4x4[HOR_UP_PRED ] = FUNCD(pred4x4_horizontal_up_rv40);
565 h->pred4x4[DIAG_DOWN_LEFT_PRED_RV40_NODOWN] = FUNCD(pred4x4_down_left_rv40_nodown);
566 h->pred4x4[HOR_UP_PRED_RV40_NODOWN] = FUNCD(pred4x4_horizontal_up_rv40_nodown);
567 h->pred4x4[VERT_LEFT_PRED_RV40_NODOWN] = FUNCD(pred4x4_vertical_left_rv40_nodown);
568 h->pred16x16[PLANE_PRED8x8 ] = FUNCD(pred16x16_plane_rv40);
569 break;
570 case AV_CODEC_ID_VP7:
571 case AV_CODEC_ID_VP8:
572 h->pred4x4[VERT_PRED ] = FUNCD(pred4x4_vertical_vp8);
573 h->pred4x4[HOR_PRED ] = FUNCD(pred4x4_horizontal_vp8);
574 h->pred4x4[VERT_LEFT_PRED ] = FUNCD(pred4x4_vertical_left_vp8);
575 h->pred4x4[TM_VP8_PRED ] = FUNCD(pred4x4_tm_vp8);
576 h->pred4x4[VERT_VP8_PRED ] = FUNCC(pred4x4_vertical, 8);
577 h->pred4x4[DC_127_PRED ] = FUNCD(pred4x4_127_dc);
578 h->pred4x4[DC_129_PRED ] = FUNCD(pred4x4_129_dc);
579 h->pred4x4[HOR_VP8_PRED ] = FUNCC(pred4x4_horizontal, 8);
580 h->pred8x8[PLANE_PRED8x8 ] = FUNCD(pred8x8_tm_vp8);
581 h->pred8x8[DC_127_PRED8x8 ] = FUNCC(pred8x8_127_dc, 8);
582 h->pred8x8[DC_129_PRED8x8 ] = FUNCC(pred8x8_129_dc, 8);
583 h->pred16x16[PLANE_PRED8x8 ] = FUNCD(pred16x16_tm_vp8);
584 h->pred16x16[DC_127_PRED8x8] = FUNCC(pred16x16_127_dc, 8);
585 h->pred16x16[DC_129_PRED8x8] = FUNCC(pred16x16_129_dc, 8);
586 break;
587 }
588 break;
589 }
590
591 #if ARCH_AARCH64
592 ff_h264_pred_init_aarch64(h, codec_id, bit_depth, chroma_format_idc);
593 #elif ARCH_ARM
594 ff_h264_pred_init_arm(h, codec_id, bit_depth, chroma_format_idc);
595 #elif ARCH_X86
596 ff_h264_pred_init_x86(h, codec_id, bit_depth, chroma_format_idc);
597 #elif ARCH_MIPS
598 ff_h264_pred_init_mips(h, codec_id, bit_depth, chroma_format_idc);
599 #elif ARCH_LOONGARCH
600 ff_h264_pred_init_loongarch(h, codec_id, bit_depth, chroma_format_idc);
601 #endif
602 }
603