Horizontal and vertical prediction functions needed for new lossless mode
[libav.git] / libavcodec / h264pred.c
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 h264pred.c
24 * H.264 / AVC / MPEG4 part10 prediction functions.
25 * @author Michael Niedermayer <michaelni@gmx.at>
26 */
27
28 #include "avcodec.h"
29 #include "mpegvideo.h"
30 #include "h264pred.h"
31
32 static void pred4x4_vertical_c(uint8_t *src, uint8_t *topright, int stride){
33 const uint32_t a= ((uint32_t*)(src-stride))[0];
34 ((uint32_t*)(src+0*stride))[0]= a;
35 ((uint32_t*)(src+1*stride))[0]= a;
36 ((uint32_t*)(src+2*stride))[0]= a;
37 ((uint32_t*)(src+3*stride))[0]= a;
38 }
39
40 static void pred4x4_horizontal_c(uint8_t *src, uint8_t *topright, int stride){
41 ((uint32_t*)(src+0*stride))[0]= src[-1+0*stride]*0x01010101;
42 ((uint32_t*)(src+1*stride))[0]= src[-1+1*stride]*0x01010101;
43 ((uint32_t*)(src+2*stride))[0]= src[-1+2*stride]*0x01010101;
44 ((uint32_t*)(src+3*stride))[0]= src[-1+3*stride]*0x01010101;
45 }
46
47 static void pred4x4_dc_c(uint8_t *src, uint8_t *topright, int stride){
48 const int dc= ( src[-stride] + src[1-stride] + src[2-stride] + src[3-stride]
49 + src[-1+0*stride] + src[-1+1*stride] + src[-1+2*stride] + src[-1+3*stride] + 4) >>3;
50
51 ((uint32_t*)(src+0*stride))[0]=
52 ((uint32_t*)(src+1*stride))[0]=
53 ((uint32_t*)(src+2*stride))[0]=
54 ((uint32_t*)(src+3*stride))[0]= dc* 0x01010101;
55 }
56
57 static void pred4x4_left_dc_c(uint8_t *src, uint8_t *topright, int stride){
58 const int dc= ( src[-1+0*stride] + src[-1+1*stride] + src[-1+2*stride] + src[-1+3*stride] + 2) >>2;
59
60 ((uint32_t*)(src+0*stride))[0]=
61 ((uint32_t*)(src+1*stride))[0]=
62 ((uint32_t*)(src+2*stride))[0]=
63 ((uint32_t*)(src+3*stride))[0]= dc* 0x01010101;
64 }
65
66 static void pred4x4_top_dc_c(uint8_t *src, uint8_t *topright, int stride){
67 const int dc= ( src[-stride] + src[1-stride] + src[2-stride] + src[3-stride] + 2) >>2;
68
69 ((uint32_t*)(src+0*stride))[0]=
70 ((uint32_t*)(src+1*stride))[0]=
71 ((uint32_t*)(src+2*stride))[0]=
72 ((uint32_t*)(src+3*stride))[0]= dc* 0x01010101;
73 }
74
75 static void pred4x4_128_dc_c(uint8_t *src, uint8_t *topright, int stride){
76 ((uint32_t*)(src+0*stride))[0]=
77 ((uint32_t*)(src+1*stride))[0]=
78 ((uint32_t*)(src+2*stride))[0]=
79 ((uint32_t*)(src+3*stride))[0]= 128U*0x01010101U;
80 }
81
82
83 #define LOAD_TOP_RIGHT_EDGE\
84 const int av_unused t4= topright[0];\
85 const int av_unused t5= topright[1];\
86 const int av_unused t6= topright[2];\
87 const int av_unused t7= topright[3];\
88
89 #define LOAD_DOWN_LEFT_EDGE\
90 const int av_unused l4= src[-1+4*stride];\
91 const int av_unused l5= src[-1+5*stride];\
92 const int av_unused l6= src[-1+6*stride];\
93 const int av_unused l7= src[-1+7*stride];\
94
95 #define LOAD_LEFT_EDGE\
96 const int av_unused l0= src[-1+0*stride];\
97 const int av_unused l1= src[-1+1*stride];\
98 const int av_unused l2= src[-1+2*stride];\
99 const int av_unused l3= src[-1+3*stride];\
100
101 #define LOAD_TOP_EDGE\
102 const int av_unused t0= src[ 0-1*stride];\
103 const int av_unused t1= src[ 1-1*stride];\
104 const int av_unused t2= src[ 2-1*stride];\
105 const int av_unused t3= src[ 3-1*stride];\
106
107 static void pred4x4_down_right_c(uint8_t *src, uint8_t *topright, int stride){
108 const int lt= src[-1-1*stride];
109 LOAD_TOP_EDGE
110 LOAD_LEFT_EDGE
111
112 src[0+3*stride]=(l3 + 2*l2 + l1 + 2)>>2;
113 src[0+2*stride]=
114 src[1+3*stride]=(l2 + 2*l1 + l0 + 2)>>2;
115 src[0+1*stride]=
116 src[1+2*stride]=
117 src[2+3*stride]=(l1 + 2*l0 + lt + 2)>>2;
118 src[0+0*stride]=
119 src[1+1*stride]=
120 src[2+2*stride]=
121 src[3+3*stride]=(l0 + 2*lt + t0 + 2)>>2;
122 src[1+0*stride]=
123 src[2+1*stride]=
124 src[3+2*stride]=(lt + 2*t0 + t1 + 2)>>2;
125 src[2+0*stride]=
126 src[3+1*stride]=(t0 + 2*t1 + t2 + 2)>>2;
127 src[3+0*stride]=(t1 + 2*t2 + t3 + 2)>>2;
128 }
129
130 static void pred4x4_down_left_c(uint8_t *src, uint8_t *topright, int stride){
131 LOAD_TOP_EDGE
132 LOAD_TOP_RIGHT_EDGE
133 // LOAD_LEFT_EDGE
134
135 src[0+0*stride]=(t0 + t2 + 2*t1 + 2)>>2;
136 src[1+0*stride]=
137 src[0+1*stride]=(t1 + t3 + 2*t2 + 2)>>2;
138 src[2+0*stride]=
139 src[1+1*stride]=
140 src[0+2*stride]=(t2 + t4 + 2*t3 + 2)>>2;
141 src[3+0*stride]=
142 src[2+1*stride]=
143 src[1+2*stride]=
144 src[0+3*stride]=(t3 + t5 + 2*t4 + 2)>>2;
145 src[3+1*stride]=
146 src[2+2*stride]=
147 src[1+3*stride]=(t4 + t6 + 2*t5 + 2)>>2;
148 src[3+2*stride]=
149 src[2+3*stride]=(t5 + t7 + 2*t6 + 2)>>2;
150 src[3+3*stride]=(t6 + 3*t7 + 2)>>2;
151 }
152
153 static void pred4x4_down_left_svq3_c(uint8_t *src, uint8_t *topright, int stride){
154 LOAD_TOP_EDGE
155 LOAD_LEFT_EDGE
156 const av_unused int unu0= t0;
157 const av_unused int unu1= l0;
158
159 src[0+0*stride]=(l1 + t1)>>1;
160 src[1+0*stride]=
161 src[0+1*stride]=(l2 + t2)>>1;
162 src[2+0*stride]=
163 src[1+1*stride]=
164 src[0+2*stride]=
165 src[3+0*stride]=
166 src[2+1*stride]=
167 src[1+2*stride]=
168 src[0+3*stride]=
169 src[3+1*stride]=
170 src[2+2*stride]=
171 src[1+3*stride]=
172 src[3+2*stride]=
173 src[2+3*stride]=
174 src[3+3*stride]=(l3 + t3)>>1;
175 }
176
177 static void pred4x4_down_left_rv40_c(uint8_t *src, uint8_t *topright, int stride){
178 LOAD_TOP_EDGE
179 LOAD_TOP_RIGHT_EDGE
180 LOAD_LEFT_EDGE
181 LOAD_DOWN_LEFT_EDGE
182
183 src[0+0*stride]=(t0 + t2 + 2*t1 + 2 + l0 + l2 + 2*l1 + 2)>>3;
184 src[1+0*stride]=
185 src[0+1*stride]=(t1 + t3 + 2*t2 + 2 + l1 + l3 + 2*l2 + 2)>>3;
186 src[2+0*stride]=
187 src[1+1*stride]=
188 src[0+2*stride]=(t2 + t4 + 2*t3 + 2 + l2 + l4 + 2*l3 + 2)>>3;
189 src[3+0*stride]=
190 src[2+1*stride]=
191 src[1+2*stride]=
192 src[0+3*stride]=(t3 + t5 + 2*t4 + 2 + l3 + l5 + 2*l4 + 2)>>3;
193 src[3+1*stride]=
194 src[2+2*stride]=
195 src[1+3*stride]=(t4 + t6 + 2*t5 + 2 + l4 + l6 + 2*l5 + 2)>>3;
196 src[3+2*stride]=
197 src[2+3*stride]=(t5 + t7 + 2*t6 + 2 + l5 + l7 + 2*l6 + 2)>>3;
198 src[3+3*stride]=(t6 + t7 + 1 + l6 + l7 + 1)>>2;
199 }
200
201 static void pred4x4_down_left_rv40_nodown_c(uint8_t *src, uint8_t *topright, int stride){
202 LOAD_TOP_EDGE
203 LOAD_TOP_RIGHT_EDGE
204 LOAD_LEFT_EDGE
205
206 src[0+0*stride]=(t0 + t2 + 2*t1 + 2 + l0 + l2 + 2*l1 + 2)>>3;
207 src[1+0*stride]=
208 src[0+1*stride]=(t1 + t3 + 2*t2 + 2 + l1 + l3 + 2*l2 + 2)>>3;
209 src[2+0*stride]=
210 src[1+1*stride]=
211 src[0+2*stride]=(t2 + t4 + 2*t3 + 2 + l2 + 3*l3 + 2)>>3;
212 src[3+0*stride]=
213 src[2+1*stride]=
214 src[1+2*stride]=
215 src[0+3*stride]=(t3 + t5 + 2*t4 + 2 + l3*4 + 2)>>3;
216 src[3+1*stride]=
217 src[2+2*stride]=
218 src[1+3*stride]=(t4 + t6 + 2*t5 + 2 + l3*4 + 2)>>3;
219 src[3+2*stride]=
220 src[2+3*stride]=(t5 + t7 + 2*t6 + 2 + l3*4 + 2)>>3;
221 src[3+3*stride]=(t6 + t7 + 1 + 2*l3 + 1)>>2;
222 }
223
224 static void pred4x4_vertical_right_c(uint8_t *src, uint8_t *topright, int stride){
225 const int lt= src[-1-1*stride];
226 LOAD_TOP_EDGE
227 LOAD_LEFT_EDGE
228
229 src[0+0*stride]=
230 src[1+2*stride]=(lt + t0 + 1)>>1;
231 src[1+0*stride]=
232 src[2+2*stride]=(t0 + t1 + 1)>>1;
233 src[2+0*stride]=
234 src[3+2*stride]=(t1 + t2 + 1)>>1;
235 src[3+0*stride]=(t2 + t3 + 1)>>1;
236 src[0+1*stride]=
237 src[1+3*stride]=(l0 + 2*lt + t0 + 2)>>2;
238 src[1+1*stride]=
239 src[2+3*stride]=(lt + 2*t0 + t1 + 2)>>2;
240 src[2+1*stride]=
241 src[3+3*stride]=(t0 + 2*t1 + t2 + 2)>>2;
242 src[3+1*stride]=(t1 + 2*t2 + t3 + 2)>>2;
243 src[0+2*stride]=(lt + 2*l0 + l1 + 2)>>2;
244 src[0+3*stride]=(l0 + 2*l1 + l2 + 2)>>2;
245 }
246
247 static void pred4x4_vertical_left_c(uint8_t *src, uint8_t *topright, int stride){
248 LOAD_TOP_EDGE
249 LOAD_TOP_RIGHT_EDGE
250
251 src[0+0*stride]=(t0 + t1 + 1)>>1;
252 src[1+0*stride]=
253 src[0+2*stride]=(t1 + t2 + 1)>>1;
254 src[2+0*stride]=
255 src[1+2*stride]=(t2 + t3 + 1)>>1;
256 src[3+0*stride]=
257 src[2+2*stride]=(t3 + t4+ 1)>>1;
258 src[3+2*stride]=(t4 + t5+ 1)>>1;
259 src[0+1*stride]=(t0 + 2*t1 + t2 + 2)>>2;
260 src[1+1*stride]=
261 src[0+3*stride]=(t1 + 2*t2 + t3 + 2)>>2;
262 src[2+1*stride]=
263 src[1+3*stride]=(t2 + 2*t3 + t4 + 2)>>2;
264 src[3+1*stride]=
265 src[2+3*stride]=(t3 + 2*t4 + t5 + 2)>>2;
266 src[3+3*stride]=(t4 + 2*t5 + t6 + 2)>>2;
267 }
268
269 static void pred4x4_vertical_left_rv40(uint8_t *src, uint8_t *topright, int stride,
270 const int l0, const int l1, const int l2, const int l3, const int l4){
271 LOAD_TOP_EDGE
272 LOAD_TOP_RIGHT_EDGE
273
274 src[0+0*stride]=(2*t0 + 2*t1 + l1 + 2*l2 + l3 + 4)>>3;
275 src[1+0*stride]=
276 src[0+2*stride]=(t1 + t2 + 1)>>1;
277 src[2+0*stride]=
278 src[1+2*stride]=(t2 + t3 + 1)>>1;
279 src[3+0*stride]=
280 src[2+2*stride]=(t3 + t4+ 1)>>1;
281 src[3+2*stride]=(t4 + t5+ 1)>>1;
282 src[0+1*stride]=(t0 + 2*t1 + t2 + l2 + 2*l3 + l4 + 4)>>3;
283 src[1+1*stride]=
284 src[0+3*stride]=(t1 + 2*t2 + t3 + 2)>>2;
285 src[2+1*stride]=
286 src[1+3*stride]=(t2 + 2*t3 + t4 + 2)>>2;
287 src[3+1*stride]=
288 src[2+3*stride]=(t3 + 2*t4 + t5 + 2)>>2;
289 src[3+3*stride]=(t4 + 2*t5 + t6 + 2)>>2;
290 }
291
292 static void pred4x4_vertical_left_rv40_c(uint8_t *src, uint8_t *topright, int stride){
293 LOAD_LEFT_EDGE
294 LOAD_DOWN_LEFT_EDGE
295
296 pred4x4_vertical_left_rv40(src, topright, stride, l0, l1, l2, l3, l4);
297 }
298
299 static void pred4x4_vertical_left_rv40_nodown_c(uint8_t *src, uint8_t *topright, int stride){
300 LOAD_LEFT_EDGE
301
302 pred4x4_vertical_left_rv40(src, topright, stride, l0, l1, l2, l3, l3);
303 }
304
305 static void pred4x4_horizontal_up_c(uint8_t *src, uint8_t *topright, int stride){
306 LOAD_LEFT_EDGE
307
308 src[0+0*stride]=(l0 + l1 + 1)>>1;
309 src[1+0*stride]=(l0 + 2*l1 + l2 + 2)>>2;
310 src[2+0*stride]=
311 src[0+1*stride]=(l1 + l2 + 1)>>1;
312 src[3+0*stride]=
313 src[1+1*stride]=(l1 + 2*l2 + l3 + 2)>>2;
314 src[2+1*stride]=
315 src[0+2*stride]=(l2 + l3 + 1)>>1;
316 src[3+1*stride]=
317 src[1+2*stride]=(l2 + 2*l3 + l3 + 2)>>2;
318 src[3+2*stride]=
319 src[1+3*stride]=
320 src[0+3*stride]=
321 src[2+2*stride]=
322 src[2+3*stride]=
323 src[3+3*stride]=l3;
324 }
325
326 static void pred4x4_horizontal_up_rv40_c(uint8_t *src, uint8_t *topright, int stride){
327 LOAD_LEFT_EDGE
328 LOAD_DOWN_LEFT_EDGE
329 LOAD_TOP_EDGE
330 LOAD_TOP_RIGHT_EDGE
331
332 src[0+0*stride]=(t1 + 2*t2 + t3 + 2*l0 + 2*l1 + 4)>>3;
333 src[1+0*stride]=(t2 + 2*t3 + t4 + l0 + 2*l1 + l2 + 4)>>3;
334 src[2+0*stride]=
335 src[0+1*stride]=(t3 + 2*t4 + t5 + 2*l1 + 2*l2 + 4)>>3;
336 src[3+0*stride]=
337 src[1+1*stride]=(t4 + 2*t5 + t6 + l1 + 2*l2 + l3 + 4)>>3;
338 src[2+1*stride]=
339 src[0+2*stride]=(t5 + 2*t6 + t7 + 2*l2 + 2*l3 + 4)>>3;
340 src[3+1*stride]=
341 src[1+2*stride]=(t6 + 3*t7 + l2 + 3*l3 + 4)>>3;
342 src[3+2*stride]=
343 src[1+3*stride]=(l3 + 2*l4 + l5 + 2)>>2;
344 src[0+3*stride]=
345 src[2+2*stride]=(t6 + t7 + l3 + l4 + 2)>>2;
346 src[2+3*stride]=(l4 + l5 + 1)>>1;
347 src[3+3*stride]=(l4 + 2*l5 + l6 + 2)>>2;
348 }
349
350 static void pred4x4_horizontal_up_rv40_nodown_c(uint8_t *src, uint8_t *topright, int stride){
351 LOAD_LEFT_EDGE
352 LOAD_TOP_EDGE
353 LOAD_TOP_RIGHT_EDGE
354
355 src[0+0*stride]=(t1 + 2*t2 + t3 + 2*l0 + 2*l1 + 4)>>3;
356 src[1+0*stride]=(t2 + 2*t3 + t4 + l0 + 2*l1 + l2 + 4)>>3;
357 src[2+0*stride]=
358 src[0+1*stride]=(t3 + 2*t4 + t5 + 2*l1 + 2*l2 + 4)>>3;
359 src[3+0*stride]=
360 src[1+1*stride]=(t4 + 2*t5 + t6 + l1 + 2*l2 + l3 + 4)>>3;
361 src[2+1*stride]=
362 src[0+2*stride]=(t5 + 2*t6 + t7 + 2*l2 + 2*l3 + 4)>>3;
363 src[3+1*stride]=
364 src[1+2*stride]=(t6 + 3*t7 + l2 + 3*l3 + 4)>>3;
365 src[3+2*stride]=
366 src[1+3*stride]=l3;
367 src[0+3*stride]=
368 src[2+2*stride]=(t6 + t7 + 2*l3 + 2)>>2;
369 src[2+3*stride]=
370 src[3+3*stride]=l3;
371 }
372
373 static void pred4x4_horizontal_down_c(uint8_t *src, uint8_t *topright, int stride){
374 const int lt= src[-1-1*stride];
375 LOAD_TOP_EDGE
376 LOAD_LEFT_EDGE
377
378 src[0+0*stride]=
379 src[2+1*stride]=(lt + l0 + 1)>>1;
380 src[1+0*stride]=
381 src[3+1*stride]=(l0 + 2*lt + t0 + 2)>>2;
382 src[2+0*stride]=(lt + 2*t0 + t1 + 2)>>2;
383 src[3+0*stride]=(t0 + 2*t1 + t2 + 2)>>2;
384 src[0+1*stride]=
385 src[2+2*stride]=(l0 + l1 + 1)>>1;
386 src[1+1*stride]=
387 src[3+2*stride]=(lt + 2*l0 + l1 + 2)>>2;
388 src[0+2*stride]=
389 src[2+3*stride]=(l1 + l2+ 1)>>1;
390 src[1+2*stride]=
391 src[3+3*stride]=(l0 + 2*l1 + l2 + 2)>>2;
392 src[0+3*stride]=(l2 + l3 + 1)>>1;
393 src[1+3*stride]=(l1 + 2*l2 + l3 + 2)>>2;
394 }
395
396 static void pred16x16_vertical_c(uint8_t *src, int stride){
397 int i;
398 const uint32_t a= ((uint32_t*)(src-stride))[0];
399 const uint32_t b= ((uint32_t*)(src-stride))[1];
400 const uint32_t c= ((uint32_t*)(src-stride))[2];
401 const uint32_t d= ((uint32_t*)(src-stride))[3];
402
403 for(i=0; i<16; i++){
404 ((uint32_t*)(src+i*stride))[0]= a;
405 ((uint32_t*)(src+i*stride))[1]= b;
406 ((uint32_t*)(src+i*stride))[2]= c;
407 ((uint32_t*)(src+i*stride))[3]= d;
408 }
409 }
410
411 static void pred16x16_horizontal_c(uint8_t *src, int stride){
412 int i;
413
414 for(i=0; i<16; i++){
415 ((uint32_t*)(src+i*stride))[0]=
416 ((uint32_t*)(src+i*stride))[1]=
417 ((uint32_t*)(src+i*stride))[2]=
418 ((uint32_t*)(src+i*stride))[3]= src[-1+i*stride]*0x01010101;
419 }
420 }
421
422 static void pred16x16_dc_c(uint8_t *src, int stride){
423 int i, dc=0;
424
425 for(i=0;i<16; i++){
426 dc+= src[-1+i*stride];
427 }
428
429 for(i=0;i<16; i++){
430 dc+= src[i-stride];
431 }
432
433 dc= 0x01010101*((dc + 16)>>5);
434
435 for(i=0; i<16; i++){
436 ((uint32_t*)(src+i*stride))[0]=
437 ((uint32_t*)(src+i*stride))[1]=
438 ((uint32_t*)(src+i*stride))[2]=
439 ((uint32_t*)(src+i*stride))[3]= dc;
440 }
441 }
442
443 static void pred16x16_left_dc_c(uint8_t *src, int stride){
444 int i, dc=0;
445
446 for(i=0;i<16; i++){
447 dc+= src[-1+i*stride];
448 }
449
450 dc= 0x01010101*((dc + 8)>>4);
451
452 for(i=0; i<16; i++){
453 ((uint32_t*)(src+i*stride))[0]=
454 ((uint32_t*)(src+i*stride))[1]=
455 ((uint32_t*)(src+i*stride))[2]=
456 ((uint32_t*)(src+i*stride))[3]= dc;
457 }
458 }
459
460 static void pred16x16_top_dc_c(uint8_t *src, int stride){
461 int i, dc=0;
462
463 for(i=0;i<16; i++){
464 dc+= src[i-stride];
465 }
466 dc= 0x01010101*((dc + 8)>>4);
467
468 for(i=0; i<16; i++){
469 ((uint32_t*)(src+i*stride))[0]=
470 ((uint32_t*)(src+i*stride))[1]=
471 ((uint32_t*)(src+i*stride))[2]=
472 ((uint32_t*)(src+i*stride))[3]= dc;
473 }
474 }
475
476 static void pred16x16_128_dc_c(uint8_t *src, int stride){
477 int i;
478
479 for(i=0; i<16; i++){
480 ((uint32_t*)(src+i*stride))[0]=
481 ((uint32_t*)(src+i*stride))[1]=
482 ((uint32_t*)(src+i*stride))[2]=
483 ((uint32_t*)(src+i*stride))[3]= 0x01010101U*128U;
484 }
485 }
486
487 static inline void pred16x16_plane_compat_c(uint8_t *src, int stride, const int svq3, const int rv40){
488 int i, j, k;
489 int a;
490 uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
491 const uint8_t * const src0 = src+7-stride;
492 const uint8_t *src1 = src+8*stride-1;
493 const uint8_t *src2 = src1-2*stride; // == src+6*stride-1;
494 int H = src0[1] - src0[-1];
495 int V = src1[0] - src2[ 0];
496 for(k=2; k<=8; ++k) {
497 src1 += stride; src2 -= stride;
498 H += k*(src0[k] - src0[-k]);
499 V += k*(src1[0] - src2[ 0]);
500 }
501 if(svq3){
502 H = ( 5*(H/4) ) / 16;
503 V = ( 5*(V/4) ) / 16;
504
505 /* required for 100% accuracy */
506 i = H; H = V; V = i;
507 }else if(rv40){
508 H = ( H + (H>>2) ) >> 4;
509 V = ( V + (V>>2) ) >> 4;
510 }else{
511 H = ( 5*H+32 ) >> 6;
512 V = ( 5*V+32 ) >> 6;
513 }
514
515 a = 16*(src1[0] + src2[16] + 1) - 7*(V+H);
516 for(j=16; j>0; --j) {
517 int b = a;
518 a += V;
519 for(i=-16; i<0; i+=4) {
520 src[16+i] = cm[ (b ) >> 5 ];
521 src[17+i] = cm[ (b+ H) >> 5 ];
522 src[18+i] = cm[ (b+2*H) >> 5 ];
523 src[19+i] = cm[ (b+3*H) >> 5 ];
524 b += 4*H;
525 }
526 src += stride;
527 }
528 }
529
530 static void pred16x16_plane_c(uint8_t *src, int stride){
531 pred16x16_plane_compat_c(src, stride, 0, 0);
532 }
533
534 static void pred16x16_plane_svq3_c(uint8_t *src, int stride){
535 pred16x16_plane_compat_c(src, stride, 1, 0);
536 }
537
538 static void pred16x16_plane_rv40_c(uint8_t *src, int stride){
539 pred16x16_plane_compat_c(src, stride, 0, 1);
540 }
541
542 static void pred8x8_vertical_c(uint8_t *src, int stride){
543 int i;
544 const uint32_t a= ((uint32_t*)(src-stride))[0];
545 const uint32_t b= ((uint32_t*)(src-stride))[1];
546
547 for(i=0; i<8; i++){
548 ((uint32_t*)(src+i*stride))[0]= a;
549 ((uint32_t*)(src+i*stride))[1]= b;
550 }
551 }
552
553 static void pred8x8_horizontal_c(uint8_t *src, int stride){
554 int i;
555
556 for(i=0; i<8; i++){
557 ((uint32_t*)(src+i*stride))[0]=
558 ((uint32_t*)(src+i*stride))[1]= src[-1+i*stride]*0x01010101;
559 }
560 }
561
562 static void pred8x8_128_dc_c(uint8_t *src, int stride){
563 int i;
564
565 for(i=0; i<8; i++){
566 ((uint32_t*)(src+i*stride))[0]=
567 ((uint32_t*)(src+i*stride))[1]= 0x01010101U*128U;
568 }
569 }
570
571 static void pred8x8_left_dc_c(uint8_t *src, int stride){
572 int i;
573 int dc0, dc2;
574
575 dc0=dc2=0;
576 for(i=0;i<4; i++){
577 dc0+= src[-1+i*stride];
578 dc2+= src[-1+(i+4)*stride];
579 }
580 dc0= 0x01010101*((dc0 + 2)>>2);
581 dc2= 0x01010101*((dc2 + 2)>>2);
582
583 for(i=0; i<4; i++){
584 ((uint32_t*)(src+i*stride))[0]=
585 ((uint32_t*)(src+i*stride))[1]= dc0;
586 }
587 for(i=4; i<8; i++){
588 ((uint32_t*)(src+i*stride))[0]=
589 ((uint32_t*)(src+i*stride))[1]= dc2;
590 }
591 }
592
593 static void pred8x8_left_dc_rv40_c(uint8_t *src, int stride){
594 int i;
595 int dc0;
596
597 dc0=0;
598 for(i=0;i<8; i++)
599 dc0+= src[-1+i*stride];
600 dc0= 0x01010101*((dc0 + 4)>>3);
601
602 for(i=0; i<8; i++){
603 ((uint32_t*)(src+i*stride))[0]=
604 ((uint32_t*)(src+i*stride))[1]= dc0;
605 }
606 }
607
608 static void pred8x8_top_dc_c(uint8_t *src, int stride){
609 int i;
610 int dc0, dc1;
611
612 dc0=dc1=0;
613 for(i=0;i<4; i++){
614 dc0+= src[i-stride];
615 dc1+= src[4+i-stride];
616 }
617 dc0= 0x01010101*((dc0 + 2)>>2);
618 dc1= 0x01010101*((dc1 + 2)>>2);
619
620 for(i=0; i<4; i++){
621 ((uint32_t*)(src+i*stride))[0]= dc0;
622 ((uint32_t*)(src+i*stride))[1]= dc1;
623 }
624 for(i=4; i<8; i++){
625 ((uint32_t*)(src+i*stride))[0]= dc0;
626 ((uint32_t*)(src+i*stride))[1]= dc1;
627 }
628 }
629
630 static void pred8x8_top_dc_rv40_c(uint8_t *src, int stride){
631 int i;
632 int dc0;
633
634 dc0=0;
635 for(i=0;i<8; i++)
636 dc0+= src[i-stride];
637 dc0= 0x01010101*((dc0 + 4)>>3);
638
639 for(i=0; i<8; i++){
640 ((uint32_t*)(src+i*stride))[0]=
641 ((uint32_t*)(src+i*stride))[1]= dc0;
642 }
643 }
644
645
646 static void pred8x8_dc_c(uint8_t *src, int stride){
647 int i;
648 int dc0, dc1, dc2, dc3;
649
650 dc0=dc1=dc2=0;
651 for(i=0;i<4; i++){
652 dc0+= src[-1+i*stride] + src[i-stride];
653 dc1+= src[4+i-stride];
654 dc2+= src[-1+(i+4)*stride];
655 }
656 dc3= 0x01010101*((dc1 + dc2 + 4)>>3);
657 dc0= 0x01010101*((dc0 + 4)>>3);
658 dc1= 0x01010101*((dc1 + 2)>>2);
659 dc2= 0x01010101*((dc2 + 2)>>2);
660
661 for(i=0; i<4; i++){
662 ((uint32_t*)(src+i*stride))[0]= dc0;
663 ((uint32_t*)(src+i*stride))[1]= dc1;
664 }
665 for(i=4; i<8; i++){
666 ((uint32_t*)(src+i*stride))[0]= dc2;
667 ((uint32_t*)(src+i*stride))[1]= dc3;
668 }
669 }
670
671 //the following 4 function should not be optimized!
672 static void pred8x8_mad_cow_dc_l0t(uint8_t *src, int stride){
673 pred8x8_top_dc_c(src, stride);
674 pred4x4_dc_c(src, NULL, stride);
675 }
676
677 static void pred8x8_mad_cow_dc_0lt(uint8_t *src, int stride){
678 pred8x8_dc_c(src, stride);
679 pred4x4_top_dc_c(src, NULL, stride);
680 }
681
682 static void pred8x8_mad_cow_dc_l00(uint8_t *src, int stride){
683 pred8x8_left_dc_c(src, stride);
684 pred4x4_128_dc_c(src + 4*stride , NULL, stride);
685 pred4x4_128_dc_c(src + 4*stride + 4, NULL, stride);
686 }
687
688 static void pred8x8_mad_cow_dc_0l0(uint8_t *src, int stride){
689 pred8x8_left_dc_c(src, stride);
690 pred4x4_128_dc_c(src , NULL, stride);
691 pred4x4_128_dc_c(src + 4, NULL, stride);
692 }
693
694 static void pred8x8_dc_rv40_c(uint8_t *src, int stride){
695 int i;
696 int dc0=0;
697
698 for(i=0;i<4; i++){
699 dc0+= src[-1+i*stride] + src[i-stride];
700 dc0+= src[4+i-stride];
701 dc0+= src[-1+(i+4)*stride];
702 }
703 dc0= 0x01010101*((dc0 + 8)>>4);
704
705 for(i=0; i<4; i++){
706 ((uint32_t*)(src+i*stride))[0]= dc0;
707 ((uint32_t*)(src+i*stride))[1]= dc0;
708 }
709 for(i=4; i<8; i++){
710 ((uint32_t*)(src+i*stride))[0]= dc0;
711 ((uint32_t*)(src+i*stride))[1]= dc0;
712 }
713 }
714
715 static void pred8x8_plane_c(uint8_t *src, int stride){
716 int j, k;
717 int a;
718 uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
719 const uint8_t * const src0 = src+3-stride;
720 const uint8_t *src1 = src+4*stride-1;
721 const uint8_t *src2 = src1-2*stride; // == src+2*stride-1;
722 int H = src0[1] - src0[-1];
723 int V = src1[0] - src2[ 0];
724 for(k=2; k<=4; ++k) {
725 src1 += stride; src2 -= stride;
726 H += k*(src0[k] - src0[-k]);
727 V += k*(src1[0] - src2[ 0]);
728 }
729 H = ( 17*H+16 ) >> 5;
730 V = ( 17*V+16 ) >> 5;
731
732 a = 16*(src1[0] + src2[8]+1) - 3*(V+H);
733 for(j=8; j>0; --j) {
734 int b = a;
735 a += V;
736 src[0] = cm[ (b ) >> 5 ];
737 src[1] = cm[ (b+ H) >> 5 ];
738 src[2] = cm[ (b+2*H) >> 5 ];
739 src[3] = cm[ (b+3*H) >> 5 ];
740 src[4] = cm[ (b+4*H) >> 5 ];
741 src[5] = cm[ (b+5*H) >> 5 ];
742 src[6] = cm[ (b+6*H) >> 5 ];
743 src[7] = cm[ (b+7*H) >> 5 ];
744 src += stride;
745 }
746 }
747
748 #define SRC(x,y) src[(x)+(y)*stride]
749 #define PL(y) \
750 const int l##y = (SRC(-1,y-1) + 2*SRC(-1,y) + SRC(-1,y+1) + 2) >> 2;
751 #define PREDICT_8x8_LOAD_LEFT \
752 const int l0 = ((has_topleft ? SRC(-1,-1) : SRC(-1,0)) \
753 + 2*SRC(-1,0) + SRC(-1,1) + 2) >> 2; \
754 PL(1) PL(2) PL(3) PL(4) PL(5) PL(6) \
755 const int l7 av_unused = (SRC(-1,6) + 3*SRC(-1,7) + 2) >> 2
756
757 #define PT(x) \
758 const int t##x = (SRC(x-1,-1) + 2*SRC(x,-1) + SRC(x+1,-1) + 2) >> 2;
759 #define PREDICT_8x8_LOAD_TOP \
760 const int t0 = ((has_topleft ? SRC(-1,-1) : SRC(0,-1)) \
761 + 2*SRC(0,-1) + SRC(1,-1) + 2) >> 2; \
762 PT(1) PT(2) PT(3) PT(4) PT(5) PT(6) \
763 const int t7 av_unused = ((has_topright ? SRC(8,-1) : SRC(7,-1)) \
764 + 2*SRC(7,-1) + SRC(6,-1) + 2) >> 2
765
766 #define PTR(x) \
767 t##x = (SRC(x-1,-1) + 2*SRC(x,-1) + SRC(x+1,-1) + 2) >> 2;
768 #define PREDICT_8x8_LOAD_TOPRIGHT \
769 int t8, t9, t10, t11, t12, t13, t14, t15; \
770 if(has_topright) { \
771 PTR(8) PTR(9) PTR(10) PTR(11) PTR(12) PTR(13) PTR(14) \
772 t15 = (SRC(14,-1) + 3*SRC(15,-1) + 2) >> 2; \
773 } else t8=t9=t10=t11=t12=t13=t14=t15= SRC(7,-1);
774
775 #define PREDICT_8x8_LOAD_TOPLEFT \
776 const int lt = (SRC(-1,0) + 2*SRC(-1,-1) + SRC(0,-1) + 2) >> 2
777
778 #define PREDICT_8x8_DC(v) \
779 int y; \
780 for( y = 0; y < 8; y++ ) { \
781 ((uint32_t*)src)[0] = \
782 ((uint32_t*)src)[1] = v; \
783 src += stride; \
784 }
785
786 static void pred8x8l_128_dc_c(uint8_t *src, int has_topleft, int has_topright, int stride)
787 {
788 PREDICT_8x8_DC(0x80808080);
789 }
790 static void pred8x8l_left_dc_c(uint8_t *src, int has_topleft, int has_topright, int stride)
791 {
792 PREDICT_8x8_LOAD_LEFT;
793 const uint32_t dc = ((l0+l1+l2+l3+l4+l5+l6+l7+4) >> 3) * 0x01010101;
794 PREDICT_8x8_DC(dc);
795 }
796 static void pred8x8l_top_dc_c(uint8_t *src, int has_topleft, int has_topright, int stride)
797 {
798 PREDICT_8x8_LOAD_TOP;
799 const uint32_t dc = ((t0+t1+t2+t3+t4+t5+t6+t7+4) >> 3) * 0x01010101;
800 PREDICT_8x8_DC(dc);
801 }
802 static void pred8x8l_dc_c(uint8_t *src, int has_topleft, int has_topright, int stride)
803 {
804 PREDICT_8x8_LOAD_LEFT;
805 PREDICT_8x8_LOAD_TOP;
806 const uint32_t dc = ((l0+l1+l2+l3+l4+l5+l6+l7
807 +t0+t1+t2+t3+t4+t5+t6+t7+8) >> 4) * 0x01010101;
808 PREDICT_8x8_DC(dc);
809 }
810 static void pred8x8l_horizontal_c(uint8_t *src, int has_topleft, int has_topright, int stride)
811 {
812 PREDICT_8x8_LOAD_LEFT;
813 #define ROW(y) ((uint32_t*)(src+y*stride))[0] =\
814 ((uint32_t*)(src+y*stride))[1] = 0x01010101 * l##y
815 ROW(0); ROW(1); ROW(2); ROW(3); ROW(4); ROW(5); ROW(6); ROW(7);
816 #undef ROW
817 }
818 static void pred8x8l_vertical_c(uint8_t *src, int has_topleft, int has_topright, int stride)
819 {
820 int y;
821 PREDICT_8x8_LOAD_TOP;
822 src[0] = t0;
823 src[1] = t1;
824 src[2] = t2;
825 src[3] = t3;
826 src[4] = t4;
827 src[5] = t5;
828 src[6] = t6;
829 src[7] = t7;
830 for( y = 1; y < 8; y++ )
831 *(uint64_t*)(src+y*stride) = *(uint64_t*)src;
832 }
833 static void pred8x8l_down_left_c(uint8_t *src, int has_topleft, int has_topright, int stride)
834 {
835 PREDICT_8x8_LOAD_TOP;
836 PREDICT_8x8_LOAD_TOPRIGHT;
837 SRC(0,0)= (t0 + 2*t1 + t2 + 2) >> 2;
838 SRC(0,1)=SRC(1,0)= (t1 + 2*t2 + t3 + 2) >> 2;
839 SRC(0,2)=SRC(1,1)=SRC(2,0)= (t2 + 2*t3 + t4 + 2) >> 2;
840 SRC(0,3)=SRC(1,2)=SRC(2,1)=SRC(3,0)= (t3 + 2*t4 + t5 + 2) >> 2;
841 SRC(0,4)=SRC(1,3)=SRC(2,2)=SRC(3,1)=SRC(4,0)= (t4 + 2*t5 + t6 + 2) >> 2;
842 SRC(0,5)=SRC(1,4)=SRC(2,3)=SRC(3,2)=SRC(4,1)=SRC(5,0)= (t5 + 2*t6 + t7 + 2) >> 2;
843 SRC(0,6)=SRC(1,5)=SRC(2,4)=SRC(3,3)=SRC(4,2)=SRC(5,1)=SRC(6,0)= (t6 + 2*t7 + t8 + 2) >> 2;
844 SRC(0,7)=SRC(1,6)=SRC(2,5)=SRC(3,4)=SRC(4,3)=SRC(5,2)=SRC(6,1)=SRC(7,0)= (t7 + 2*t8 + t9 + 2) >> 2;
845 SRC(1,7)=SRC(2,6)=SRC(3,5)=SRC(4,4)=SRC(5,3)=SRC(6,2)=SRC(7,1)= (t8 + 2*t9 + t10 + 2) >> 2;
846 SRC(2,7)=SRC(3,6)=SRC(4,5)=SRC(5,4)=SRC(6,3)=SRC(7,2)= (t9 + 2*t10 + t11 + 2) >> 2;
847 SRC(3,7)=SRC(4,6)=SRC(5,5)=SRC(6,4)=SRC(7,3)= (t10 + 2*t11 + t12 + 2) >> 2;
848 SRC(4,7)=SRC(5,6)=SRC(6,5)=SRC(7,4)= (t11 + 2*t12 + t13 + 2) >> 2;
849 SRC(5,7)=SRC(6,6)=SRC(7,5)= (t12 + 2*t13 + t14 + 2) >> 2;
850 SRC(6,7)=SRC(7,6)= (t13 + 2*t14 + t15 + 2) >> 2;
851 SRC(7,7)= (t14 + 3*t15 + 2) >> 2;
852 }
853 static void pred8x8l_down_right_c(uint8_t *src, int has_topleft, int has_topright, int stride)
854 {
855 PREDICT_8x8_LOAD_TOP;
856 PREDICT_8x8_LOAD_LEFT;
857 PREDICT_8x8_LOAD_TOPLEFT;
858 SRC(0,7)= (l7 + 2*l6 + l5 + 2) >> 2;
859 SRC(0,6)=SRC(1,7)= (l6 + 2*l5 + l4 + 2) >> 2;
860 SRC(0,5)=SRC(1,6)=SRC(2,7)= (l5 + 2*l4 + l3 + 2) >> 2;
861 SRC(0,4)=SRC(1,5)=SRC(2,6)=SRC(3,7)= (l4 + 2*l3 + l2 + 2) >> 2;
862 SRC(0,3)=SRC(1,4)=SRC(2,5)=SRC(3,6)=SRC(4,7)= (l3 + 2*l2 + l1 + 2) >> 2;
863 SRC(0,2)=SRC(1,3)=SRC(2,4)=SRC(3,5)=SRC(4,6)=SRC(5,7)= (l2 + 2*l1 + l0 + 2) >> 2;
864 SRC(0,1)=SRC(1,2)=SRC(2,3)=SRC(3,4)=SRC(4,5)=SRC(5,6)=SRC(6,7)= (l1 + 2*l0 + lt + 2) >> 2;
865 SRC(0,0)=SRC(1,1)=SRC(2,2)=SRC(3,3)=SRC(4,4)=SRC(5,5)=SRC(6,6)=SRC(7,7)= (l0 + 2*lt + t0 + 2) >> 2;
866 SRC(1,0)=SRC(2,1)=SRC(3,2)=SRC(4,3)=SRC(5,4)=SRC(6,5)=SRC(7,6)= (lt + 2*t0 + t1 + 2) >> 2;
867 SRC(2,0)=SRC(3,1)=SRC(4,2)=SRC(5,3)=SRC(6,4)=SRC(7,5)= (t0 + 2*t1 + t2 + 2) >> 2;
868 SRC(3,0)=SRC(4,1)=SRC(5,2)=SRC(6,3)=SRC(7,4)= (t1 + 2*t2 + t3 + 2) >> 2;
869 SRC(4,0)=SRC(5,1)=SRC(6,2)=SRC(7,3)= (t2 + 2*t3 + t4 + 2) >> 2;
870 SRC(5,0)=SRC(6,1)=SRC(7,2)= (t3 + 2*t4 + t5 + 2) >> 2;
871 SRC(6,0)=SRC(7,1)= (t4 + 2*t5 + t6 + 2) >> 2;
872 SRC(7,0)= (t5 + 2*t6 + t7 + 2) >> 2;
873
874 }
875 static void pred8x8l_vertical_right_c(uint8_t *src, int has_topleft, int has_topright, int stride)
876 {
877 PREDICT_8x8_LOAD_TOP;
878 PREDICT_8x8_LOAD_LEFT;
879 PREDICT_8x8_LOAD_TOPLEFT;
880 SRC(0,6)= (l5 + 2*l4 + l3 + 2) >> 2;
881 SRC(0,7)= (l6 + 2*l5 + l4 + 2) >> 2;
882 SRC(0,4)=SRC(1,6)= (l3 + 2*l2 + l1 + 2) >> 2;
883 SRC(0,5)=SRC(1,7)= (l4 + 2*l3 + l2 + 2) >> 2;
884 SRC(0,2)=SRC(1,4)=SRC(2,6)= (l1 + 2*l0 + lt + 2) >> 2;
885 SRC(0,3)=SRC(1,5)=SRC(2,7)= (l2 + 2*l1 + l0 + 2) >> 2;
886 SRC(0,1)=SRC(1,3)=SRC(2,5)=SRC(3,7)= (l0 + 2*lt + t0 + 2) >> 2;
887 SRC(0,0)=SRC(1,2)=SRC(2,4)=SRC(3,6)= (lt + t0 + 1) >> 1;
888 SRC(1,1)=SRC(2,3)=SRC(3,5)=SRC(4,7)= (lt + 2*t0 + t1 + 2) >> 2;
889 SRC(1,0)=SRC(2,2)=SRC(3,4)=SRC(4,6)= (t0 + t1 + 1) >> 1;
890 SRC(2,1)=SRC(3,3)=SRC(4,5)=SRC(5,7)= (t0 + 2*t1 + t2 + 2) >> 2;
891 SRC(2,0)=SRC(3,2)=SRC(4,4)=SRC(5,6)= (t1 + t2 + 1) >> 1;
892 SRC(3,1)=SRC(4,3)=SRC(5,5)=SRC(6,7)= (t1 + 2*t2 + t3 + 2) >> 2;
893 SRC(3,0)=SRC(4,2)=SRC(5,4)=SRC(6,6)= (t2 + t3 + 1) >> 1;
894 SRC(4,1)=SRC(5,3)=SRC(6,5)=SRC(7,7)= (t2 + 2*t3 + t4 + 2) >> 2;
895 SRC(4,0)=SRC(5,2)=SRC(6,4)=SRC(7,6)= (t3 + t4 + 1) >> 1;
896 SRC(5,1)=SRC(6,3)=SRC(7,5)= (t3 + 2*t4 + t5 + 2) >> 2;
897 SRC(5,0)=SRC(6,2)=SRC(7,4)= (t4 + t5 + 1) >> 1;
898 SRC(6,1)=SRC(7,3)= (t4 + 2*t5 + t6 + 2) >> 2;
899 SRC(6,0)=SRC(7,2)= (t5 + t6 + 1) >> 1;
900 SRC(7,1)= (t5 + 2*t6 + t7 + 2) >> 2;
901 SRC(7,0)= (t6 + t7 + 1) >> 1;
902 }
903 static void pred8x8l_horizontal_down_c(uint8_t *src, int has_topleft, int has_topright, int stride)
904 {
905 PREDICT_8x8_LOAD_TOP;
906 PREDICT_8x8_LOAD_LEFT;
907 PREDICT_8x8_LOAD_TOPLEFT;
908 SRC(0,7)= (l6 + l7 + 1) >> 1;
909 SRC(1,7)= (l5 + 2*l6 + l7 + 2) >> 2;
910 SRC(0,6)=SRC(2,7)= (l5 + l6 + 1) >> 1;
911 SRC(1,6)=SRC(3,7)= (l4 + 2*l5 + l6 + 2) >> 2;
912 SRC(0,5)=SRC(2,6)=SRC(4,7)= (l4 + l5 + 1) >> 1;
913 SRC(1,5)=SRC(3,6)=SRC(5,7)= (l3 + 2*l4 + l5 + 2) >> 2;
914 SRC(0,4)=SRC(2,5)=SRC(4,6)=SRC(6,7)= (l3 + l4 + 1) >> 1;
915 SRC(1,4)=SRC(3,5)=SRC(5,6)=SRC(7,7)= (l2 + 2*l3 + l4 + 2) >> 2;
916 SRC(0,3)=SRC(2,4)=SRC(4,5)=SRC(6,6)= (l2 + l3 + 1) >> 1;
917 SRC(1,3)=SRC(3,4)=SRC(5,5)=SRC(7,6)= (l1 + 2*l2 + l3 + 2) >> 2;
918 SRC(0,2)=SRC(2,3)=SRC(4,4)=SRC(6,5)= (l1 + l2 + 1) >> 1;
919 SRC(1,2)=SRC(3,3)=SRC(5,4)=SRC(7,5)= (l0 + 2*l1 + l2 + 2) >> 2;
920 SRC(0,1)=SRC(2,2)=SRC(4,3)=SRC(6,4)= (l0 + l1 + 1) >> 1;
921 SRC(1,1)=SRC(3,2)=SRC(5,3)=SRC(7,4)= (lt + 2*l0 + l1 + 2) >> 2;
922 SRC(0,0)=SRC(2,1)=SRC(4,2)=SRC(6,3)= (lt + l0 + 1) >> 1;
923 SRC(1,0)=SRC(3,1)=SRC(5,2)=SRC(7,3)= (l0 + 2*lt + t0 + 2) >> 2;
924 SRC(2,0)=SRC(4,1)=SRC(6,2)= (t1 + 2*t0 + lt + 2) >> 2;
925 SRC(3,0)=SRC(5,1)=SRC(7,2)= (t2 + 2*t1 + t0 + 2) >> 2;
926 SRC(4,0)=SRC(6,1)= (t3 + 2*t2 + t1 + 2) >> 2;
927 SRC(5,0)=SRC(7,1)= (t4 + 2*t3 + t2 + 2) >> 2;
928 SRC(6,0)= (t5 + 2*t4 + t3 + 2) >> 2;
929 SRC(7,0)= (t6 + 2*t5 + t4 + 2) >> 2;
930 }
931 static void pred8x8l_vertical_left_c(uint8_t *src, int has_topleft, int has_topright, int stride)
932 {
933 PREDICT_8x8_LOAD_TOP;
934 PREDICT_8x8_LOAD_TOPRIGHT;
935 SRC(0,0)= (t0 + t1 + 1) >> 1;
936 SRC(0,1)= (t0 + 2*t1 + t2 + 2) >> 2;
937 SRC(0,2)=SRC(1,0)= (t1 + t2 + 1) >> 1;
938 SRC(0,3)=SRC(1,1)= (t1 + 2*t2 + t3 + 2) >> 2;
939 SRC(0,4)=SRC(1,2)=SRC(2,0)= (t2 + t3 + 1) >> 1;
940 SRC(0,5)=SRC(1,3)=SRC(2,1)= (t2 + 2*t3 + t4 + 2) >> 2;
941 SRC(0,6)=SRC(1,4)=SRC(2,2)=SRC(3,0)= (t3 + t4 + 1) >> 1;
942 SRC(0,7)=SRC(1,5)=SRC(2,3)=SRC(3,1)= (t3 + 2*t4 + t5 + 2) >> 2;
943 SRC(1,6)=SRC(2,4)=SRC(3,2)=SRC(4,0)= (t4 + t5 + 1) >> 1;
944 SRC(1,7)=SRC(2,5)=SRC(3,3)=SRC(4,1)= (t4 + 2*t5 + t6 + 2) >> 2;
945 SRC(2,6)=SRC(3,4)=SRC(4,2)=SRC(5,0)= (t5 + t6 + 1) >> 1;
946 SRC(2,7)=SRC(3,5)=SRC(4,3)=SRC(5,1)= (t5 + 2*t6 + t7 + 2) >> 2;
947 SRC(3,6)=SRC(4,4)=SRC(5,2)=SRC(6,0)= (t6 + t7 + 1) >> 1;
948 SRC(3,7)=SRC(4,5)=SRC(5,3)=SRC(6,1)= (t6 + 2*t7 + t8 + 2) >> 2;
949 SRC(4,6)=SRC(5,4)=SRC(6,2)=SRC(7,0)= (t7 + t8 + 1) >> 1;
950 SRC(4,7)=SRC(5,5)=SRC(6,3)=SRC(7,1)= (t7 + 2*t8 + t9 + 2) >> 2;
951 SRC(5,6)=SRC(6,4)=SRC(7,2)= (t8 + t9 + 1) >> 1;
952 SRC(5,7)=SRC(6,5)=SRC(7,3)= (t8 + 2*t9 + t10 + 2) >> 2;
953 SRC(6,6)=SRC(7,4)= (t9 + t10 + 1) >> 1;
954 SRC(6,7)=SRC(7,5)= (t9 + 2*t10 + t11 + 2) >> 2;
955 SRC(7,6)= (t10 + t11 + 1) >> 1;
956 SRC(7,7)= (t10 + 2*t11 + t12 + 2) >> 2;
957 }
958 static void pred8x8l_horizontal_up_c(uint8_t *src, int has_topleft, int has_topright, int stride)
959 {
960 PREDICT_8x8_LOAD_LEFT;
961 SRC(0,0)= (l0 + l1 + 1) >> 1;
962 SRC(1,0)= (l0 + 2*l1 + l2 + 2) >> 2;
963 SRC(0,1)=SRC(2,0)= (l1 + l2 + 1) >> 1;
964 SRC(1,1)=SRC(3,0)= (l1 + 2*l2 + l3 + 2) >> 2;
965 SRC(0,2)=SRC(2,1)=SRC(4,0)= (l2 + l3 + 1) >> 1;
966 SRC(1,2)=SRC(3,1)=SRC(5,0)= (l2 + 2*l3 + l4 + 2) >> 2;
967 SRC(0,3)=SRC(2,2)=SRC(4,1)=SRC(6,0)= (l3 + l4 + 1) >> 1;
968 SRC(1,3)=SRC(3,2)=SRC(5,1)=SRC(7,0)= (l3 + 2*l4 + l5 + 2) >> 2;
969 SRC(0,4)=SRC(2,3)=SRC(4,2)=SRC(6,1)= (l4 + l5 + 1) >> 1;
970 SRC(1,4)=SRC(3,3)=SRC(5,2)=SRC(7,1)= (l4 + 2*l5 + l6 + 2) >> 2;
971 SRC(0,5)=SRC(2,4)=SRC(4,3)=SRC(6,2)= (l5 + l6 + 1) >> 1;
972 SRC(1,5)=SRC(3,4)=SRC(5,3)=SRC(7,2)= (l5 + 2*l6 + l7 + 2) >> 2;
973 SRC(0,6)=SRC(2,5)=SRC(4,4)=SRC(6,3)= (l6 + l7 + 1) >> 1;
974 SRC(1,6)=SRC(3,5)=SRC(5,4)=SRC(7,3)= (l6 + 3*l7 + 2) >> 2;
975 SRC(0,7)=SRC(1,7)=SRC(2,6)=SRC(2,7)=SRC(3,6)=
976 SRC(3,7)=SRC(4,5)=SRC(4,6)=SRC(4,7)=SRC(5,5)=
977 SRC(5,6)=SRC(5,7)=SRC(6,4)=SRC(6,5)=SRC(6,6)=
978 SRC(6,7)=SRC(7,4)=SRC(7,5)=SRC(7,6)=SRC(7,7)= l7;
979 }
980 #undef PREDICT_8x8_LOAD_LEFT
981 #undef PREDICT_8x8_LOAD_TOP
982 #undef PREDICT_8x8_LOAD_TOPLEFT
983 #undef PREDICT_8x8_LOAD_TOPRIGHT
984 #undef PREDICT_8x8_DC
985 #undef PTR
986 #undef PT
987 #undef PL
988 #undef SRC
989
990 static void pred4x4_vertical_add_c(uint8_t *pix, const DCTELEM *block, int stride){
991 int i;
992 for(i=0; i<4; i++){
993 pix[0]= pix[0-stride] + block[0];
994 pix[1]= pix[1-stride] + block[1];
995 pix[2]= pix[2-stride] + block[2];
996 pix[3]= pix[3-stride] + block[3];
997 pix+= stride;
998 block+= 4;
999 }
1000 }
1001
1002 static void pred4x4_horizontal_add_c(uint8_t *pix, const DCTELEM *block, int stride){
1003 int i;
1004 for(i=0; i<4; i++){
1005 pix[0]= pix[-1] + block[0];
1006 pix[1]= pix[ 0] + block[1];
1007 pix[2]= pix[ 1] + block[2];
1008 pix[3]= pix[ 2] + block[3];
1009 pix+= stride;
1010 block+= 4;
1011 }
1012 }
1013
1014 static void pred8x8l_vertical_add_c(uint8_t *pix, const DCTELEM *block, int stride){
1015 int i;
1016 for(i=0; i<8; i++){
1017 pix[0]= pix[0-stride] + block[0];
1018 pix[1]= pix[1-stride] + block[1];
1019 pix[2]= pix[2-stride] + block[2];
1020 pix[3]= pix[3-stride] + block[3];
1021 pix[4]= pix[4-stride] + block[4];
1022 pix[5]= pix[5-stride] + block[5];
1023 pix[6]= pix[6-stride] + block[6];
1024 pix[7]= pix[7-stride] + block[7];
1025 pix+= stride;
1026 block+= 8;
1027 }
1028 }
1029
1030 static void pred8x8l_horizontal_add_c(uint8_t *pix, const DCTELEM *block, int stride){
1031 int i;
1032 for(i=0; i<8; i++){
1033 pix[0]= pix[-1] + block[0];
1034 pix[1]= pix[ 0] + block[1];
1035 pix[2]= pix[ 1] + block[2];
1036 pix[3]= pix[ 2] + block[3];
1037 pix[4]= pix[ 3] + block[4];
1038 pix[5]= pix[ 4] + block[5];
1039 pix[6]= pix[ 5] + block[6];
1040 pix[7]= pix[ 6] + block[7];
1041 pix+= stride;
1042 block+= 8;
1043 }
1044 }
1045
1046 static void pred16x16_vertical_add_c(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){
1047 int i;
1048 for(i=0; i<16; i++)
1049 pred4x4_vertical_add_c(pix + block_offset[i], block + i*16, stride);
1050 }
1051
1052 static void pred16x16_horizontal_add_c(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){
1053 int i;
1054 for(i=0; i<16; i++)
1055 pred4x4_horizontal_add_c(pix + block_offset[i], block + i*16, stride);
1056 }
1057
1058 static void pred8x8_vertical_add_c(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){
1059 int i;
1060 for(i=0; i<4; i++)
1061 pred4x4_vertical_add_c(pix + block_offset[i], block + i*16, stride);
1062 }
1063
1064 static void pred8x8_horizontal_add_c(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){
1065 int i;
1066 for(i=0; i<4; i++)
1067 pred4x4_horizontal_add_c(pix + block_offset[i], block + i*16, stride);
1068 }
1069
1070
1071 /**
1072 * Sets the intra prediction function pointers.
1073 */
1074 void ff_h264_pred_init(H264PredContext *h, int codec_id){
1075 // MpegEncContext * const s = &h->s;
1076
1077 if(codec_id != CODEC_ID_RV40){
1078 h->pred4x4[VERT_PRED ]= pred4x4_vertical_c;
1079 h->pred4x4[HOR_PRED ]= pred4x4_horizontal_c;
1080 h->pred4x4[DC_PRED ]= pred4x4_dc_c;
1081 if(codec_id == CODEC_ID_SVQ3)
1082 h->pred4x4[DIAG_DOWN_LEFT_PRED ]= pred4x4_down_left_svq3_c;
1083 else
1084 h->pred4x4[DIAG_DOWN_LEFT_PRED ]= pred4x4_down_left_c;
1085 h->pred4x4[DIAG_DOWN_RIGHT_PRED]= pred4x4_down_right_c;
1086 h->pred4x4[VERT_RIGHT_PRED ]= pred4x4_vertical_right_c;
1087 h->pred4x4[HOR_DOWN_PRED ]= pred4x4_horizontal_down_c;
1088 h->pred4x4[VERT_LEFT_PRED ]= pred4x4_vertical_left_c;
1089 h->pred4x4[HOR_UP_PRED ]= pred4x4_horizontal_up_c;
1090 h->pred4x4[LEFT_DC_PRED ]= pred4x4_left_dc_c;
1091 h->pred4x4[TOP_DC_PRED ]= pred4x4_top_dc_c;
1092 h->pred4x4[DC_128_PRED ]= pred4x4_128_dc_c;
1093 }else{
1094 h->pred4x4[VERT_PRED ]= pred4x4_vertical_c;
1095 h->pred4x4[HOR_PRED ]= pred4x4_horizontal_c;
1096 h->pred4x4[DC_PRED ]= pred4x4_dc_c;
1097 h->pred4x4[DIAG_DOWN_LEFT_PRED ]= pred4x4_down_left_rv40_c;
1098 h->pred4x4[DIAG_DOWN_RIGHT_PRED]= pred4x4_down_right_c;
1099 h->pred4x4[VERT_RIGHT_PRED ]= pred4x4_vertical_right_c;
1100 h->pred4x4[HOR_DOWN_PRED ]= pred4x4_horizontal_down_c;
1101 h->pred4x4[VERT_LEFT_PRED ]= pred4x4_vertical_left_rv40_c;
1102 h->pred4x4[HOR_UP_PRED ]= pred4x4_horizontal_up_rv40_c;
1103 h->pred4x4[LEFT_DC_PRED ]= pred4x4_left_dc_c;
1104 h->pred4x4[TOP_DC_PRED ]= pred4x4_top_dc_c;
1105 h->pred4x4[DC_128_PRED ]= pred4x4_128_dc_c;
1106 h->pred4x4[DIAG_DOWN_LEFT_PRED_RV40_NODOWN]= pred4x4_down_left_rv40_nodown_c;
1107 h->pred4x4[HOR_UP_PRED_RV40_NODOWN]= pred4x4_horizontal_up_rv40_nodown_c;
1108 h->pred4x4[VERT_LEFT_PRED_RV40_NODOWN]= pred4x4_vertical_left_rv40_nodown_c;
1109 }
1110
1111 h->pred8x8l[VERT_PRED ]= pred8x8l_vertical_c;
1112 h->pred8x8l[HOR_PRED ]= pred8x8l_horizontal_c;
1113 h->pred8x8l[DC_PRED ]= pred8x8l_dc_c;
1114 h->pred8x8l[DIAG_DOWN_LEFT_PRED ]= pred8x8l_down_left_c;
1115 h->pred8x8l[DIAG_DOWN_RIGHT_PRED]= pred8x8l_down_right_c;
1116 h->pred8x8l[VERT_RIGHT_PRED ]= pred8x8l_vertical_right_c;
1117 h->pred8x8l[HOR_DOWN_PRED ]= pred8x8l_horizontal_down_c;
1118 h->pred8x8l[VERT_LEFT_PRED ]= pred8x8l_vertical_left_c;
1119 h->pred8x8l[HOR_UP_PRED ]= pred8x8l_horizontal_up_c;
1120 h->pred8x8l[LEFT_DC_PRED ]= pred8x8l_left_dc_c;
1121 h->pred8x8l[TOP_DC_PRED ]= pred8x8l_top_dc_c;
1122 h->pred8x8l[DC_128_PRED ]= pred8x8l_128_dc_c;
1123
1124 h->pred8x8[VERT_PRED8x8 ]= pred8x8_vertical_c;
1125 h->pred8x8[HOR_PRED8x8 ]= pred8x8_horizontal_c;
1126 h->pred8x8[PLANE_PRED8x8 ]= pred8x8_plane_c;
1127 if(codec_id != CODEC_ID_RV40){
1128 h->pred8x8[DC_PRED8x8 ]= pred8x8_dc_c;
1129 h->pred8x8[LEFT_DC_PRED8x8]= pred8x8_left_dc_c;
1130 h->pred8x8[TOP_DC_PRED8x8 ]= pred8x8_top_dc_c;
1131 h->pred8x8[ALZHEIMER_DC_L0T_PRED8x8 ]= pred8x8_mad_cow_dc_l0t;
1132 h->pred8x8[ALZHEIMER_DC_0LT_PRED8x8 ]= pred8x8_mad_cow_dc_0lt;
1133 h->pred8x8[ALZHEIMER_DC_L00_PRED8x8 ]= pred8x8_mad_cow_dc_l00;
1134 h->pred8x8[ALZHEIMER_DC_0L0_PRED8x8 ]= pred8x8_mad_cow_dc_0l0;
1135 }else{
1136 h->pred8x8[DC_PRED8x8 ]= pred8x8_dc_rv40_c;
1137 h->pred8x8[LEFT_DC_PRED8x8]= pred8x8_left_dc_rv40_c;
1138 h->pred8x8[TOP_DC_PRED8x8 ]= pred8x8_top_dc_rv40_c;
1139 }
1140 h->pred8x8[DC_128_PRED8x8 ]= pred8x8_128_dc_c;
1141
1142 h->pred16x16[DC_PRED8x8 ]= pred16x16_dc_c;
1143 h->pred16x16[VERT_PRED8x8 ]= pred16x16_vertical_c;
1144 h->pred16x16[HOR_PRED8x8 ]= pred16x16_horizontal_c;
1145 h->pred16x16[PLANE_PRED8x8 ]= pred16x16_plane_c;
1146 switch(codec_id){
1147 case CODEC_ID_SVQ3:
1148 h->pred16x16[PLANE_PRED8x8 ]= pred16x16_plane_svq3_c;
1149 break;
1150 case CODEC_ID_RV40:
1151 h->pred16x16[PLANE_PRED8x8 ]= pred16x16_plane_rv40_c;
1152 break;
1153 default:
1154 h->pred16x16[PLANE_PRED8x8 ]= pred16x16_plane_c;
1155 }
1156 h->pred16x16[LEFT_DC_PRED8x8]= pred16x16_left_dc_c;
1157 h->pred16x16[TOP_DC_PRED8x8 ]= pred16x16_top_dc_c;
1158 h->pred16x16[DC_128_PRED8x8 ]= pred16x16_128_dc_c;
1159
1160 //special lossless h/v prediction for h264
1161 h->pred4x4_add [VERT_PRED ]= pred4x4_vertical_add_c;
1162 h->pred4x4_add [ HOR_PRED ]= pred4x4_horizontal_add_c;
1163 h->pred8x8l_add [VERT_PRED ]= pred8x8l_vertical_add_c;
1164 h->pred8x8l_add [ HOR_PRED ]= pred8x8l_horizontal_add_c;
1165 h->pred8x8_add [VERT_PRED8x8]= pred8x8_vertical_add_c;
1166 h->pred8x8_add [ HOR_PRED8x8]= pred8x8_horizontal_add_c;
1167 h->pred16x16_add[VERT_PRED8x8]= pred16x16_vertical_add_c;
1168 h->pred16x16_add[ HOR_PRED8x8]= pred16x16_horizontal_add_c;
1169 }