2 * H.26L/H.264/AVC/JVT/14496-10/... encoder/decoder
3 * Copyright (c) 2003 Michael Niedermayer <michaelni@gmx.at>
5 * This file is part of FFmpeg.
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.
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.
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
24 * H.264 / AVC / MPEG4 part10 prediction functions.
25 * @author Michael Niedermayer <michaelni@gmx.at>
29 #include "mpegvideo.h"
33 static void pred4x4_vertical_c(uint8_t *src, const uint8_t *topright, int stride){
34 const uint32_t a= ((uint32_t*)(src-stride))[0];
35 ((uint32_t*)(src+0*stride))[0]= a;
36 ((uint32_t*)(src+1*stride))[0]= a;
37 ((uint32_t*)(src+2*stride))[0]= a;
38 ((uint32_t*)(src+3*stride))[0]= a;
41 static void pred4x4_horizontal_c(uint8_t *src, const uint8_t *topright, int stride){
42 ((uint32_t*)(src+0*stride))[0]= src[-1+0*stride]*0x01010101;
43 ((uint32_t*)(src+1*stride))[0]= src[-1+1*stride]*0x01010101;
44 ((uint32_t*)(src+2*stride))[0]= src[-1+2*stride]*0x01010101;
45 ((uint32_t*)(src+3*stride))[0]= src[-1+3*stride]*0x01010101;
48 static void pred4x4_dc_c(uint8_t *src, const uint8_t *topright, int stride){
49 const int dc= ( src[-stride] + src[1-stride] + src[2-stride] + src[3-stride]
50 + src[-1+0*stride] + src[-1+1*stride] + src[-1+2*stride] + src[-1+3*stride] + 4) >>3;
52 ((uint32_t*)(src+0*stride))[0]=
53 ((uint32_t*)(src+1*stride))[0]=
54 ((uint32_t*)(src+2*stride))[0]=
55 ((uint32_t*)(src+3*stride))[0]= dc* 0x01010101;
58 static void pred4x4_left_dc_c(uint8_t *src, const uint8_t *topright, int stride){
59 const int dc= ( src[-1+0*stride] + src[-1+1*stride] + src[-1+2*stride] + src[-1+3*stride] + 2) >>2;
61 ((uint32_t*)(src+0*stride))[0]=
62 ((uint32_t*)(src+1*stride))[0]=
63 ((uint32_t*)(src+2*stride))[0]=
64 ((uint32_t*)(src+3*stride))[0]= dc* 0x01010101;
67 static void pred4x4_top_dc_c(uint8_t *src, const uint8_t *topright, int stride){
68 const int dc= ( src[-stride] + src[1-stride] + src[2-stride] + src[3-stride] + 2) >>2;
70 ((uint32_t*)(src+0*stride))[0]=
71 ((uint32_t*)(src+1*stride))[0]=
72 ((uint32_t*)(src+2*stride))[0]=
73 ((uint32_t*)(src+3*stride))[0]= dc* 0x01010101;
76 static void pred4x4_128_dc_c(uint8_t *src, const uint8_t *topright, int stride){
77 ((uint32_t*)(src+0*stride))[0]=
78 ((uint32_t*)(src+1*stride))[0]=
79 ((uint32_t*)(src+2*stride))[0]=
80 ((uint32_t*)(src+3*stride))[0]= 128U*0x01010101U;
83 static void pred4x4_127_dc_c(uint8_t *src, const uint8_t *topright, int stride){
84 ((uint32_t*)(src+0*stride))[0]=
85 ((uint32_t*)(src+1*stride))[0]=
86 ((uint32_t*)(src+2*stride))[0]=
87 ((uint32_t*)(src+3*stride))[0]= 127U*0x01010101U;
90 static void pred4x4_129_dc_c(uint8_t *src, const uint8_t *topright, int stride){
91 ((uint32_t*)(src+0*stride))[0]=
92 ((uint32_t*)(src+1*stride))[0]=
93 ((uint32_t*)(src+2*stride))[0]=
94 ((uint32_t*)(src+3*stride))[0]= 129U*0x01010101U;
98 #define LOAD_TOP_RIGHT_EDGE\
99 const int av_unused t4= topright[0];\
100 const int av_unused t5= topright[1];\
101 const int av_unused t6= topright[2];\
102 const int av_unused t7= topright[3];\
104 #define LOAD_DOWN_LEFT_EDGE\
105 const int av_unused l4= src[-1+4*stride];\
106 const int av_unused l5= src[-1+5*stride];\
107 const int av_unused l6= src[-1+6*stride];\
108 const int av_unused l7= src[-1+7*stride];\
110 #define LOAD_LEFT_EDGE\
111 const int av_unused l0= src[-1+0*stride];\
112 const int av_unused l1= src[-1+1*stride];\
113 const int av_unused l2= src[-1+2*stride];\
114 const int av_unused l3= src[-1+3*stride];\
116 #define LOAD_TOP_EDGE\
117 const int av_unused t0= src[ 0-1*stride];\
118 const int av_unused t1= src[ 1-1*stride];\
119 const int av_unused t2= src[ 2-1*stride];\
120 const int av_unused t3= src[ 3-1*stride];\
122 static void pred4x4_vertical_vp8_c(uint8_t *src, const uint8_t *topright, int stride){
123 const int lt= src[-1-1*stride];
126 uint32_t v = PACK_4U8((lt + 2*t0 + t1 + 2) >> 2,
127 (t0 + 2*t1 + t2 + 2) >> 2,
128 (t1 + 2*t2 + t3 + 2) >> 2,
129 (t2 + 2*t3 + t4 + 2) >> 2);
131 AV_WN32A(src+0*stride, v);
132 AV_WN32A(src+1*stride, v);
133 AV_WN32A(src+2*stride, v);
134 AV_WN32A(src+3*stride, v);
137 static void pred4x4_horizontal_vp8_c(uint8_t *src, const uint8_t *topright, int stride){
138 const int lt= src[-1-1*stride];
141 AV_WN32A(src+0*stride, ((lt + 2*l0 + l1 + 2) >> 2)*0x01010101);
142 AV_WN32A(src+1*stride, ((l0 + 2*l1 + l2 + 2) >> 2)*0x01010101);
143 AV_WN32A(src+2*stride, ((l1 + 2*l2 + l3 + 2) >> 2)*0x01010101);
144 AV_WN32A(src+3*stride, ((l2 + 2*l3 + l3 + 2) >> 2)*0x01010101);
147 static void pred4x4_down_right_c(uint8_t *src, const uint8_t *topright, int stride){
148 const int lt= src[-1-1*stride];
152 src[0+3*stride]=(l3 + 2*l2 + l1 + 2)>>2;
154 src[1+3*stride]=(l2 + 2*l1 + l0 + 2)>>2;
157 src[2+3*stride]=(l1 + 2*l0 + lt + 2)>>2;
161 src[3+3*stride]=(l0 + 2*lt + t0 + 2)>>2;
164 src[3+2*stride]=(lt + 2*t0 + t1 + 2)>>2;
166 src[3+1*stride]=(t0 + 2*t1 + t2 + 2)>>2;
167 src[3+0*stride]=(t1 + 2*t2 + t3 + 2)>>2;
170 static void pred4x4_down_left_c(uint8_t *src, const uint8_t *topright, int stride){
175 src[0+0*stride]=(t0 + t2 + 2*t1 + 2)>>2;
177 src[0+1*stride]=(t1 + t3 + 2*t2 + 2)>>2;
180 src[0+2*stride]=(t2 + t4 + 2*t3 + 2)>>2;
184 src[0+3*stride]=(t3 + t5 + 2*t4 + 2)>>2;
187 src[1+3*stride]=(t4 + t6 + 2*t5 + 2)>>2;
189 src[2+3*stride]=(t5 + t7 + 2*t6 + 2)>>2;
190 src[3+3*stride]=(t6 + 3*t7 + 2)>>2;
193 static void pred4x4_down_left_svq3_c(uint8_t *src, const uint8_t *topright, int stride){
196 const av_unused int unu0= t0;
197 const av_unused int unu1= l0;
199 src[0+0*stride]=(l1 + t1)>>1;
201 src[0+1*stride]=(l2 + t2)>>1;
214 src[3+3*stride]=(l3 + t3)>>1;
217 static void pred4x4_down_left_rv40_c(uint8_t *src, const uint8_t *topright, int stride){
223 src[0+0*stride]=(t0 + t2 + 2*t1 + 2 + l0 + l2 + 2*l1 + 2)>>3;
225 src[0+1*stride]=(t1 + t3 + 2*t2 + 2 + l1 + l3 + 2*l2 + 2)>>3;
228 src[0+2*stride]=(t2 + t4 + 2*t3 + 2 + l2 + l4 + 2*l3 + 2)>>3;
232 src[0+3*stride]=(t3 + t5 + 2*t4 + 2 + l3 + l5 + 2*l4 + 2)>>3;
235 src[1+3*stride]=(t4 + t6 + 2*t5 + 2 + l4 + l6 + 2*l5 + 2)>>3;
237 src[2+3*stride]=(t5 + t7 + 2*t6 + 2 + l5 + l7 + 2*l6 + 2)>>3;
238 src[3+3*stride]=(t6 + t7 + 1 + l6 + l7 + 1)>>2;
241 static void pred4x4_down_left_rv40_nodown_c(uint8_t *src, const uint8_t *topright, int stride){
246 src[0+0*stride]=(t0 + t2 + 2*t1 + 2 + l0 + l2 + 2*l1 + 2)>>3;
248 src[0+1*stride]=(t1 + t3 + 2*t2 + 2 + l1 + l3 + 2*l2 + 2)>>3;
251 src[0+2*stride]=(t2 + t4 + 2*t3 + 2 + l2 + 3*l3 + 2)>>3;
255 src[0+3*stride]=(t3 + t5 + 2*t4 + 2 + l3*4 + 2)>>3;
258 src[1+3*stride]=(t4 + t6 + 2*t5 + 2 + l3*4 + 2)>>3;
260 src[2+3*stride]=(t5 + t7 + 2*t6 + 2 + l3*4 + 2)>>3;
261 src[3+3*stride]=(t6 + t7 + 1 + 2*l3 + 1)>>2;
264 static void pred4x4_vertical_right_c(uint8_t *src, const uint8_t *topright, int stride){
265 const int lt= src[-1-1*stride];
270 src[1+2*stride]=(lt + t0 + 1)>>1;
272 src[2+2*stride]=(t0 + t1 + 1)>>1;
274 src[3+2*stride]=(t1 + t2 + 1)>>1;
275 src[3+0*stride]=(t2 + t3 + 1)>>1;
277 src[1+3*stride]=(l0 + 2*lt + t0 + 2)>>2;
279 src[2+3*stride]=(lt + 2*t0 + t1 + 2)>>2;
281 src[3+3*stride]=(t0 + 2*t1 + t2 + 2)>>2;
282 src[3+1*stride]=(t1 + 2*t2 + t3 + 2)>>2;
283 src[0+2*stride]=(lt + 2*l0 + l1 + 2)>>2;
284 src[0+3*stride]=(l0 + 2*l1 + l2 + 2)>>2;
287 static void pred4x4_vertical_left_c(uint8_t *src, const uint8_t *topright, int stride){
291 src[0+0*stride]=(t0 + t1 + 1)>>1;
293 src[0+2*stride]=(t1 + t2 + 1)>>1;
295 src[1+2*stride]=(t2 + t3 + 1)>>1;
297 src[2+2*stride]=(t3 + t4+ 1)>>1;
298 src[3+2*stride]=(t4 + t5+ 1)>>1;
299 src[0+1*stride]=(t0 + 2*t1 + t2 + 2)>>2;
301 src[0+3*stride]=(t1 + 2*t2 + t3 + 2)>>2;
303 src[1+3*stride]=(t2 + 2*t3 + t4 + 2)>>2;
305 src[2+3*stride]=(t3 + 2*t4 + t5 + 2)>>2;
306 src[3+3*stride]=(t4 + 2*t5 + t6 + 2)>>2;
309 static void pred4x4_vertical_left_rv40(uint8_t *src, const uint8_t *topright, int stride,
310 const int l0, const int l1, const int l2, const int l3, const int l4){
314 src[0+0*stride]=(2*t0 + 2*t1 + l1 + 2*l2 + l3 + 4)>>3;
316 src[0+2*stride]=(t1 + t2 + 1)>>1;
318 src[1+2*stride]=(t2 + t3 + 1)>>1;
320 src[2+2*stride]=(t3 + t4+ 1)>>1;
321 src[3+2*stride]=(t4 + t5+ 1)>>1;
322 src[0+1*stride]=(t0 + 2*t1 + t2 + l2 + 2*l3 + l4 + 4)>>3;
324 src[0+3*stride]=(t1 + 2*t2 + t3 + 2)>>2;
326 src[1+3*stride]=(t2 + 2*t3 + t4 + 2)>>2;
328 src[2+3*stride]=(t3 + 2*t4 + t5 + 2)>>2;
329 src[3+3*stride]=(t4 + 2*t5 + t6 + 2)>>2;
332 static void pred4x4_vertical_left_rv40_c(uint8_t *src, const uint8_t *topright, int stride){
336 pred4x4_vertical_left_rv40(src, topright, stride, l0, l1, l2, l3, l4);
339 static void pred4x4_vertical_left_rv40_nodown_c(uint8_t *src, const uint8_t *topright, int stride){
342 pred4x4_vertical_left_rv40(src, topright, stride, l0, l1, l2, l3, l3);
345 static void pred4x4_vertical_left_vp8_c(uint8_t *src, const uint8_t *topright, int stride){
349 src[0+0*stride]=(t0 + t1 + 1)>>1;
351 src[0+2*stride]=(t1 + t2 + 1)>>1;
353 src[1+2*stride]=(t2 + t3 + 1)>>1;
355 src[2+2*stride]=(t3 + t4 + 1)>>1;
356 src[0+1*stride]=(t0 + 2*t1 + t2 + 2)>>2;
358 src[0+3*stride]=(t1 + 2*t2 + t3 + 2)>>2;
360 src[1+3*stride]=(t2 + 2*t3 + t4 + 2)>>2;
362 src[2+3*stride]=(t3 + 2*t4 + t5 + 2)>>2;
363 src[3+2*stride]=(t4 + 2*t5 + t6 + 2)>>2;
364 src[3+3*stride]=(t5 + 2*t6 + t7 + 2)>>2;
367 static void pred4x4_horizontal_up_c(uint8_t *src, const uint8_t *topright, int stride){
370 src[0+0*stride]=(l0 + l1 + 1)>>1;
371 src[1+0*stride]=(l0 + 2*l1 + l2 + 2)>>2;
373 src[0+1*stride]=(l1 + l2 + 1)>>1;
375 src[1+1*stride]=(l1 + 2*l2 + l3 + 2)>>2;
377 src[0+2*stride]=(l2 + l3 + 1)>>1;
379 src[1+2*stride]=(l2 + 2*l3 + l3 + 2)>>2;
388 static void pred4x4_horizontal_up_rv40_c(uint8_t *src, const uint8_t *topright, int stride){
394 src[0+0*stride]=(t1 + 2*t2 + t3 + 2*l0 + 2*l1 + 4)>>3;
395 src[1+0*stride]=(t2 + 2*t3 + t4 + l0 + 2*l1 + l2 + 4)>>3;
397 src[0+1*stride]=(t3 + 2*t4 + t5 + 2*l1 + 2*l2 + 4)>>3;
399 src[1+1*stride]=(t4 + 2*t5 + t6 + l1 + 2*l2 + l3 + 4)>>3;
401 src[0+2*stride]=(t5 + 2*t6 + t7 + 2*l2 + 2*l3 + 4)>>3;
403 src[1+2*stride]=(t6 + 3*t7 + l2 + 3*l3 + 4)>>3;
405 src[1+3*stride]=(l3 + 2*l4 + l5 + 2)>>2;
407 src[2+2*stride]=(t6 + t7 + l3 + l4 + 2)>>2;
408 src[2+3*stride]=(l4 + l5 + 1)>>1;
409 src[3+3*stride]=(l4 + 2*l5 + l6 + 2)>>2;
412 static void pred4x4_horizontal_up_rv40_nodown_c(uint8_t *src, const uint8_t *topright, int stride){
417 src[0+0*stride]=(t1 + 2*t2 + t3 + 2*l0 + 2*l1 + 4)>>3;
418 src[1+0*stride]=(t2 + 2*t3 + t4 + l0 + 2*l1 + l2 + 4)>>3;
420 src[0+1*stride]=(t3 + 2*t4 + t5 + 2*l1 + 2*l2 + 4)>>3;
422 src[1+1*stride]=(t4 + 2*t5 + t6 + l1 + 2*l2 + l3 + 4)>>3;
424 src[0+2*stride]=(t5 + 2*t6 + t7 + 2*l2 + 2*l3 + 4)>>3;
426 src[1+2*stride]=(t6 + 3*t7 + l2 + 3*l3 + 4)>>3;
430 src[2+2*stride]=(t6 + t7 + 2*l3 + 2)>>2;
435 static void pred4x4_horizontal_down_c(uint8_t *src, const uint8_t *topright, int stride){
436 const int lt= src[-1-1*stride];
441 src[2+1*stride]=(lt + l0 + 1)>>1;
443 src[3+1*stride]=(l0 + 2*lt + t0 + 2)>>2;
444 src[2+0*stride]=(lt + 2*t0 + t1 + 2)>>2;
445 src[3+0*stride]=(t0 + 2*t1 + t2 + 2)>>2;
447 src[2+2*stride]=(l0 + l1 + 1)>>1;
449 src[3+2*stride]=(lt + 2*l0 + l1 + 2)>>2;
451 src[2+3*stride]=(l1 + l2+ 1)>>1;
453 src[3+3*stride]=(l0 + 2*l1 + l2 + 2)>>2;
454 src[0+3*stride]=(l2 + l3 + 1)>>1;
455 src[1+3*stride]=(l1 + 2*l2 + l3 + 2)>>2;
458 static void pred4x4_tm_vp8_c(uint8_t *src, const uint8_t *topright, int stride){
459 uint8_t *cm = ff_cropTbl + MAX_NEG_CROP - src[-1-stride];
460 uint8_t *top = src-stride;
463 for (y = 0; y < 4; y++) {
464 uint8_t *cm_in = cm + src[-1];
465 src[0] = cm_in[top[0]];
466 src[1] = cm_in[top[1]];
467 src[2] = cm_in[top[2]];
468 src[3] = cm_in[top[3]];
473 static void pred16x16_vertical_c(uint8_t *src, int stride){
475 const uint32_t a= ((uint32_t*)(src-stride))[0];
476 const uint32_t b= ((uint32_t*)(src-stride))[1];
477 const uint32_t c= ((uint32_t*)(src-stride))[2];
478 const uint32_t d= ((uint32_t*)(src-stride))[3];
481 ((uint32_t*)(src+i*stride))[0]= a;
482 ((uint32_t*)(src+i*stride))[1]= b;
483 ((uint32_t*)(src+i*stride))[2]= c;
484 ((uint32_t*)(src+i*stride))[3]= d;
488 static void pred16x16_horizontal_c(uint8_t *src, int stride){
492 ((uint32_t*)(src+i*stride))[0]=
493 ((uint32_t*)(src+i*stride))[1]=
494 ((uint32_t*)(src+i*stride))[2]=
495 ((uint32_t*)(src+i*stride))[3]= src[-1+i*stride]*0x01010101;
499 static void pred16x16_dc_c(uint8_t *src, int stride){
503 dc+= src[-1+i*stride];
510 dc= 0x01010101*((dc + 16)>>5);
513 ((uint32_t*)(src+i*stride))[0]=
514 ((uint32_t*)(src+i*stride))[1]=
515 ((uint32_t*)(src+i*stride))[2]=
516 ((uint32_t*)(src+i*stride))[3]= dc;
520 static void pred16x16_left_dc_c(uint8_t *src, int stride){
524 dc+= src[-1+i*stride];
527 dc= 0x01010101*((dc + 8)>>4);
530 ((uint32_t*)(src+i*stride))[0]=
531 ((uint32_t*)(src+i*stride))[1]=
532 ((uint32_t*)(src+i*stride))[2]=
533 ((uint32_t*)(src+i*stride))[3]= dc;
537 static void pred16x16_top_dc_c(uint8_t *src, int stride){
543 dc= 0x01010101*((dc + 8)>>4);
546 ((uint32_t*)(src+i*stride))[0]=
547 ((uint32_t*)(src+i*stride))[1]=
548 ((uint32_t*)(src+i*stride))[2]=
549 ((uint32_t*)(src+i*stride))[3]= dc;
553 static void pred16x16_128_dc_c(uint8_t *src, int stride){
557 ((uint32_t*)(src+i*stride))[0]=
558 ((uint32_t*)(src+i*stride))[1]=
559 ((uint32_t*)(src+i*stride))[2]=
560 ((uint32_t*)(src+i*stride))[3]= 0x01010101U*128U;
564 static void pred16x16_127_dc_c(uint8_t *src, int stride){
568 ((uint32_t*)(src+i*stride))[0]=
569 ((uint32_t*)(src+i*stride))[1]=
570 ((uint32_t*)(src+i*stride))[2]=
571 ((uint32_t*)(src+i*stride))[3]= 0x01010101U*127U;
575 static void pred16x16_129_dc_c(uint8_t *src, int stride){
579 ((uint32_t*)(src+i*stride))[0]=
580 ((uint32_t*)(src+i*stride))[1]=
581 ((uint32_t*)(src+i*stride))[2]=
582 ((uint32_t*)(src+i*stride))[3]= 0x01010101U*129U;
586 static inline void pred16x16_plane_compat_c(uint8_t *src, int stride, const int svq3, const int rv40){
589 uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
590 const uint8_t * const src0 = src+7-stride;
591 const uint8_t *src1 = src+8*stride-1;
592 const uint8_t *src2 = src1-2*stride; // == src+6*stride-1;
593 int H = src0[1] - src0[-1];
594 int V = src1[0] - src2[ 0];
595 for(k=2; k<=8; ++k) {
596 src1 += stride; src2 -= stride;
597 H += k*(src0[k] - src0[-k]);
598 V += k*(src1[0] - src2[ 0]);
601 H = ( 5*(H/4) ) / 16;
602 V = ( 5*(V/4) ) / 16;
604 /* required for 100% accuracy */
607 H = ( H + (H>>2) ) >> 4;
608 V = ( V + (V>>2) ) >> 4;
614 a = 16*(src1[0] + src2[16] + 1) - 7*(V+H);
615 for(j=16; j>0; --j) {
618 for(i=-16; i<0; i+=4) {
619 src[16+i] = cm[ (b ) >> 5 ];
620 src[17+i] = cm[ (b+ H) >> 5 ];
621 src[18+i] = cm[ (b+2*H) >> 5 ];
622 src[19+i] = cm[ (b+3*H) >> 5 ];
629 static void pred16x16_plane_c(uint8_t *src, int stride){
630 pred16x16_plane_compat_c(src, stride, 0, 0);
633 static void pred16x16_plane_svq3_c(uint8_t *src, int stride){
634 pred16x16_plane_compat_c(src, stride, 1, 0);
637 static void pred16x16_plane_rv40_c(uint8_t *src, int stride){
638 pred16x16_plane_compat_c(src, stride, 0, 1);
641 static void pred16x16_tm_vp8_c(uint8_t *src, int stride){
642 uint8_t *cm = ff_cropTbl + MAX_NEG_CROP - src[-1-stride];
643 uint8_t *top = src-stride;
646 for (y = 0; y < 16; y++) {
647 uint8_t *cm_in = cm + src[-1];
648 src[0] = cm_in[top[0]];
649 src[1] = cm_in[top[1]];
650 src[2] = cm_in[top[2]];
651 src[3] = cm_in[top[3]];
652 src[4] = cm_in[top[4]];
653 src[5] = cm_in[top[5]];
654 src[6] = cm_in[top[6]];
655 src[7] = cm_in[top[7]];
656 src[8] = cm_in[top[8]];
657 src[9] = cm_in[top[9]];
658 src[10] = cm_in[top[10]];
659 src[11] = cm_in[top[11]];
660 src[12] = cm_in[top[12]];
661 src[13] = cm_in[top[13]];
662 src[14] = cm_in[top[14]];
663 src[15] = cm_in[top[15]];
668 static void pred8x8_vertical_c(uint8_t *src, int stride){
670 const uint32_t a= ((uint32_t*)(src-stride))[0];
671 const uint32_t b= ((uint32_t*)(src-stride))[1];
674 ((uint32_t*)(src+i*stride))[0]= a;
675 ((uint32_t*)(src+i*stride))[1]= b;
679 static void pred8x8_horizontal_c(uint8_t *src, int stride){
683 ((uint32_t*)(src+i*stride))[0]=
684 ((uint32_t*)(src+i*stride))[1]= src[-1+i*stride]*0x01010101;
688 static void pred8x8_128_dc_c(uint8_t *src, int stride){
692 ((uint32_t*)(src+i*stride))[0]=
693 ((uint32_t*)(src+i*stride))[1]= 0x01010101U*128U;
697 static void pred8x8_127_dc_c(uint8_t *src, int stride){
701 ((uint32_t*)(src+i*stride))[0]=
702 ((uint32_t*)(src+i*stride))[1]= 0x01010101U*127U;
705 static void pred8x8_129_dc_c(uint8_t *src, int stride){
709 ((uint32_t*)(src+i*stride))[0]=
710 ((uint32_t*)(src+i*stride))[1]= 0x01010101U*129U;
714 static void pred8x8_left_dc_c(uint8_t *src, int stride){
720 dc0+= src[-1+i*stride];
721 dc2+= src[-1+(i+4)*stride];
723 dc0= 0x01010101*((dc0 + 2)>>2);
724 dc2= 0x01010101*((dc2 + 2)>>2);
727 ((uint32_t*)(src+i*stride))[0]=
728 ((uint32_t*)(src+i*stride))[1]= dc0;
731 ((uint32_t*)(src+i*stride))[0]=
732 ((uint32_t*)(src+i*stride))[1]= dc2;
736 static void pred8x8_left_dc_rv40_c(uint8_t *src, int stride){
742 dc0+= src[-1+i*stride];
743 dc0= 0x01010101*((dc0 + 4)>>3);
746 ((uint32_t*)(src+i*stride))[0]=
747 ((uint32_t*)(src+i*stride))[1]= dc0;
751 static void pred8x8_top_dc_c(uint8_t *src, int stride){
758 dc1+= src[4+i-stride];
760 dc0= 0x01010101*((dc0 + 2)>>2);
761 dc1= 0x01010101*((dc1 + 2)>>2);
764 ((uint32_t*)(src+i*stride))[0]= dc0;
765 ((uint32_t*)(src+i*stride))[1]= dc1;
768 ((uint32_t*)(src+i*stride))[0]= dc0;
769 ((uint32_t*)(src+i*stride))[1]= dc1;
773 static void pred8x8_top_dc_rv40_c(uint8_t *src, int stride){
780 dc0= 0x01010101*((dc0 + 4)>>3);
783 ((uint32_t*)(src+i*stride))[0]=
784 ((uint32_t*)(src+i*stride))[1]= dc0;
789 static void pred8x8_dc_c(uint8_t *src, int stride){
791 int dc0, dc1, dc2, dc3;
795 dc0+= src[-1+i*stride] + src[i-stride];
796 dc1+= src[4+i-stride];
797 dc2+= src[-1+(i+4)*stride];
799 dc3= 0x01010101*((dc1 + dc2 + 4)>>3);
800 dc0= 0x01010101*((dc0 + 4)>>3);
801 dc1= 0x01010101*((dc1 + 2)>>2);
802 dc2= 0x01010101*((dc2 + 2)>>2);
805 ((uint32_t*)(src+i*stride))[0]= dc0;
806 ((uint32_t*)(src+i*stride))[1]= dc1;
809 ((uint32_t*)(src+i*stride))[0]= dc2;
810 ((uint32_t*)(src+i*stride))[1]= dc3;
814 //the following 4 function should not be optimized!
815 static void pred8x8_mad_cow_dc_l0t(uint8_t *src, int stride){
816 pred8x8_top_dc_c(src, stride);
817 pred4x4_dc_c(src, NULL, stride);
820 static void pred8x8_mad_cow_dc_0lt(uint8_t *src, int stride){
821 pred8x8_dc_c(src, stride);
822 pred4x4_top_dc_c(src, NULL, stride);
825 static void pred8x8_mad_cow_dc_l00(uint8_t *src, int stride){
826 pred8x8_left_dc_c(src, stride);
827 pred4x4_128_dc_c(src + 4*stride , NULL, stride);
828 pred4x4_128_dc_c(src + 4*stride + 4, NULL, stride);
831 static void pred8x8_mad_cow_dc_0l0(uint8_t *src, int stride){
832 pred8x8_left_dc_c(src, stride);
833 pred4x4_128_dc_c(src , NULL, stride);
834 pred4x4_128_dc_c(src + 4, NULL, stride);
837 static void pred8x8_dc_rv40_c(uint8_t *src, int stride){
842 dc0+= src[-1+i*stride] + src[i-stride];
843 dc0+= src[4+i-stride];
844 dc0+= src[-1+(i+4)*stride];
846 dc0= 0x01010101*((dc0 + 8)>>4);
849 ((uint32_t*)(src+i*stride))[0]= dc0;
850 ((uint32_t*)(src+i*stride))[1]= dc0;
853 ((uint32_t*)(src+i*stride))[0]= dc0;
854 ((uint32_t*)(src+i*stride))[1]= dc0;
858 static void pred8x8_plane_c(uint8_t *src, int stride){
861 uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
862 const uint8_t * const src0 = src+3-stride;
863 const uint8_t *src1 = src+4*stride-1;
864 const uint8_t *src2 = src1-2*stride; // == src+2*stride-1;
865 int H = src0[1] - src0[-1];
866 int V = src1[0] - src2[ 0];
867 for(k=2; k<=4; ++k) {
868 src1 += stride; src2 -= stride;
869 H += k*(src0[k] - src0[-k]);
870 V += k*(src1[0] - src2[ 0]);
872 H = ( 17*H+16 ) >> 5;
873 V = ( 17*V+16 ) >> 5;
875 a = 16*(src1[0] + src2[8]+1) - 3*(V+H);
879 src[0] = cm[ (b ) >> 5 ];
880 src[1] = cm[ (b+ H) >> 5 ];
881 src[2] = cm[ (b+2*H) >> 5 ];
882 src[3] = cm[ (b+3*H) >> 5 ];
883 src[4] = cm[ (b+4*H) >> 5 ];
884 src[5] = cm[ (b+5*H) >> 5 ];
885 src[6] = cm[ (b+6*H) >> 5 ];
886 src[7] = cm[ (b+7*H) >> 5 ];
891 static void pred8x8_tm_vp8_c(uint8_t *src, int stride){
892 uint8_t *cm = ff_cropTbl + MAX_NEG_CROP - src[-1-stride];
893 uint8_t *top = src-stride;
896 for (y = 0; y < 8; y++) {
897 uint8_t *cm_in = cm + src[-1];
898 src[0] = cm_in[top[0]];
899 src[1] = cm_in[top[1]];
900 src[2] = cm_in[top[2]];
901 src[3] = cm_in[top[3]];
902 src[4] = cm_in[top[4]];
903 src[5] = cm_in[top[5]];
904 src[6] = cm_in[top[6]];
905 src[7] = cm_in[top[7]];
910 #define SRC(x,y) src[(x)+(y)*stride]
912 const int l##y = (SRC(-1,y-1) + 2*SRC(-1,y) + SRC(-1,y+1) + 2) >> 2;
913 #define PREDICT_8x8_LOAD_LEFT \
914 const int l0 = ((has_topleft ? SRC(-1,-1) : SRC(-1,0)) \
915 + 2*SRC(-1,0) + SRC(-1,1) + 2) >> 2; \
916 PL(1) PL(2) PL(3) PL(4) PL(5) PL(6) \
917 const int l7 av_unused = (SRC(-1,6) + 3*SRC(-1,7) + 2) >> 2
920 const int t##x = (SRC(x-1,-1) + 2*SRC(x,-1) + SRC(x+1,-1) + 2) >> 2;
921 #define PREDICT_8x8_LOAD_TOP \
922 const int t0 = ((has_topleft ? SRC(-1,-1) : SRC(0,-1)) \
923 + 2*SRC(0,-1) + SRC(1,-1) + 2) >> 2; \
924 PT(1) PT(2) PT(3) PT(4) PT(5) PT(6) \
925 const int t7 av_unused = ((has_topright ? SRC(8,-1) : SRC(7,-1)) \
926 + 2*SRC(7,-1) + SRC(6,-1) + 2) >> 2
929 t##x = (SRC(x-1,-1) + 2*SRC(x,-1) + SRC(x+1,-1) + 2) >> 2;
930 #define PREDICT_8x8_LOAD_TOPRIGHT \
931 int t8, t9, t10, t11, t12, t13, t14, t15; \
933 PTR(8) PTR(9) PTR(10) PTR(11) PTR(12) PTR(13) PTR(14) \
934 t15 = (SRC(14,-1) + 3*SRC(15,-1) + 2) >> 2; \
935 } else t8=t9=t10=t11=t12=t13=t14=t15= SRC(7,-1);
937 #define PREDICT_8x8_LOAD_TOPLEFT \
938 const int lt = (SRC(-1,0) + 2*SRC(-1,-1) + SRC(0,-1) + 2) >> 2
940 #define PREDICT_8x8_DC(v) \
942 for( y = 0; y < 8; y++ ) { \
943 ((uint32_t*)src)[0] = \
944 ((uint32_t*)src)[1] = v; \
948 static void pred8x8l_128_dc_c(uint8_t *src, int has_topleft, int has_topright, int stride)
950 PREDICT_8x8_DC(0x80808080);
952 static void pred8x8l_left_dc_c(uint8_t *src, int has_topleft, int has_topright, int stride)
954 PREDICT_8x8_LOAD_LEFT;
955 const uint32_t dc = ((l0+l1+l2+l3+l4+l5+l6+l7+4) >> 3) * 0x01010101;
958 static void pred8x8l_top_dc_c(uint8_t *src, int has_topleft, int has_topright, int stride)
960 PREDICT_8x8_LOAD_TOP;
961 const uint32_t dc = ((t0+t1+t2+t3+t4+t5+t6+t7+4) >> 3) * 0x01010101;
964 static void pred8x8l_dc_c(uint8_t *src, int has_topleft, int has_topright, int stride)
966 PREDICT_8x8_LOAD_LEFT;
967 PREDICT_8x8_LOAD_TOP;
968 const uint32_t dc = ((l0+l1+l2+l3+l4+l5+l6+l7
969 +t0+t1+t2+t3+t4+t5+t6+t7+8) >> 4) * 0x01010101;
972 static void pred8x8l_horizontal_c(uint8_t *src, int has_topleft, int has_topright, int stride)
974 PREDICT_8x8_LOAD_LEFT;
975 #define ROW(y) ((uint32_t*)(src+y*stride))[0] =\
976 ((uint32_t*)(src+y*stride))[1] = 0x01010101 * l##y
977 ROW(0); ROW(1); ROW(2); ROW(3); ROW(4); ROW(5); ROW(6); ROW(7);
980 static void pred8x8l_vertical_c(uint8_t *src, int has_topleft, int has_topright, int stride)
983 PREDICT_8x8_LOAD_TOP;
992 for( y = 1; y < 8; y++ )
993 *(uint64_t*)(src+y*stride) = *(uint64_t*)src;
995 static void pred8x8l_down_left_c(uint8_t *src, int has_topleft, int has_topright, int stride)
997 PREDICT_8x8_LOAD_TOP;
998 PREDICT_8x8_LOAD_TOPRIGHT;
999 SRC(0,0)= (t0 + 2*t1 + t2 + 2) >> 2;
1000 SRC(0,1)=SRC(1,0)= (t1 + 2*t2 + t3 + 2) >> 2;
1001 SRC(0,2)=SRC(1,1)=SRC(2,0)= (t2 + 2*t3 + t4 + 2) >> 2;
1002 SRC(0,3)=SRC(1,2)=SRC(2,1)=SRC(3,0)= (t3 + 2*t4 + t5 + 2) >> 2;
1003 SRC(0,4)=SRC(1,3)=SRC(2,2)=SRC(3,1)=SRC(4,0)= (t4 + 2*t5 + t6 + 2) >> 2;
1004 SRC(0,5)=SRC(1,4)=SRC(2,3)=SRC(3,2)=SRC(4,1)=SRC(5,0)= (t5 + 2*t6 + t7 + 2) >> 2;
1005 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;
1006 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;
1007 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;
1008 SRC(2,7)=SRC(3,6)=SRC(4,5)=SRC(5,4)=SRC(6,3)=SRC(7,2)= (t9 + 2*t10 + t11 + 2) >> 2;
1009 SRC(3,7)=SRC(4,6)=SRC(5,5)=SRC(6,4)=SRC(7,3)= (t10 + 2*t11 + t12 + 2) >> 2;
1010 SRC(4,7)=SRC(5,6)=SRC(6,5)=SRC(7,4)= (t11 + 2*t12 + t13 + 2) >> 2;
1011 SRC(5,7)=SRC(6,6)=SRC(7,5)= (t12 + 2*t13 + t14 + 2) >> 2;
1012 SRC(6,7)=SRC(7,6)= (t13 + 2*t14 + t15 + 2) >> 2;
1013 SRC(7,7)= (t14 + 3*t15 + 2) >> 2;
1015 static void pred8x8l_down_right_c(uint8_t *src, int has_topleft, int has_topright, int stride)
1017 PREDICT_8x8_LOAD_TOP;
1018 PREDICT_8x8_LOAD_LEFT;
1019 PREDICT_8x8_LOAD_TOPLEFT;
1020 SRC(0,7)= (l7 + 2*l6 + l5 + 2) >> 2;
1021 SRC(0,6)=SRC(1,7)= (l6 + 2*l5 + l4 + 2) >> 2;
1022 SRC(0,5)=SRC(1,6)=SRC(2,7)= (l5 + 2*l4 + l3 + 2) >> 2;
1023 SRC(0,4)=SRC(1,5)=SRC(2,6)=SRC(3,7)= (l4 + 2*l3 + l2 + 2) >> 2;
1024 SRC(0,3)=SRC(1,4)=SRC(2,5)=SRC(3,6)=SRC(4,7)= (l3 + 2*l2 + l1 + 2) >> 2;
1025 SRC(0,2)=SRC(1,3)=SRC(2,4)=SRC(3,5)=SRC(4,6)=SRC(5,7)= (l2 + 2*l1 + l0 + 2) >> 2;
1026 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;
1027 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;
1028 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;
1029 SRC(2,0)=SRC(3,1)=SRC(4,2)=SRC(5,3)=SRC(6,4)=SRC(7,5)= (t0 + 2*t1 + t2 + 2) >> 2;
1030 SRC(3,0)=SRC(4,1)=SRC(5,2)=SRC(6,3)=SRC(7,4)= (t1 + 2*t2 + t3 + 2) >> 2;
1031 SRC(4,0)=SRC(5,1)=SRC(6,2)=SRC(7,3)= (t2 + 2*t3 + t4 + 2) >> 2;
1032 SRC(5,0)=SRC(6,1)=SRC(7,2)= (t3 + 2*t4 + t5 + 2) >> 2;
1033 SRC(6,0)=SRC(7,1)= (t4 + 2*t5 + t6 + 2) >> 2;
1034 SRC(7,0)= (t5 + 2*t6 + t7 + 2) >> 2;
1037 static void pred8x8l_vertical_right_c(uint8_t *src, int has_topleft, int has_topright, int stride)
1039 PREDICT_8x8_LOAD_TOP;
1040 PREDICT_8x8_LOAD_LEFT;
1041 PREDICT_8x8_LOAD_TOPLEFT;
1042 SRC(0,6)= (l5 + 2*l4 + l3 + 2) >> 2;
1043 SRC(0,7)= (l6 + 2*l5 + l4 + 2) >> 2;
1044 SRC(0,4)=SRC(1,6)= (l3 + 2*l2 + l1 + 2) >> 2;
1045 SRC(0,5)=SRC(1,7)= (l4 + 2*l3 + l2 + 2) >> 2;
1046 SRC(0,2)=SRC(1,4)=SRC(2,6)= (l1 + 2*l0 + lt + 2) >> 2;
1047 SRC(0,3)=SRC(1,5)=SRC(2,7)= (l2 + 2*l1 + l0 + 2) >> 2;
1048 SRC(0,1)=SRC(1,3)=SRC(2,5)=SRC(3,7)= (l0 + 2*lt + t0 + 2) >> 2;
1049 SRC(0,0)=SRC(1,2)=SRC(2,4)=SRC(3,6)= (lt + t0 + 1) >> 1;
1050 SRC(1,1)=SRC(2,3)=SRC(3,5)=SRC(4,7)= (lt + 2*t0 + t1 + 2) >> 2;
1051 SRC(1,0)=SRC(2,2)=SRC(3,4)=SRC(4,6)= (t0 + t1 + 1) >> 1;
1052 SRC(2,1)=SRC(3,3)=SRC(4,5)=SRC(5,7)= (t0 + 2*t1 + t2 + 2) >> 2;
1053 SRC(2,0)=SRC(3,2)=SRC(4,4)=SRC(5,6)= (t1 + t2 + 1) >> 1;
1054 SRC(3,1)=SRC(4,3)=SRC(5,5)=SRC(6,7)= (t1 + 2*t2 + t3 + 2) >> 2;
1055 SRC(3,0)=SRC(4,2)=SRC(5,4)=SRC(6,6)= (t2 + t3 + 1) >> 1;
1056 SRC(4,1)=SRC(5,3)=SRC(6,5)=SRC(7,7)= (t2 + 2*t3 + t4 + 2) >> 2;
1057 SRC(4,0)=SRC(5,2)=SRC(6,4)=SRC(7,6)= (t3 + t4 + 1) >> 1;
1058 SRC(5,1)=SRC(6,3)=SRC(7,5)= (t3 + 2*t4 + t5 + 2) >> 2;
1059 SRC(5,0)=SRC(6,2)=SRC(7,4)= (t4 + t5 + 1) >> 1;
1060 SRC(6,1)=SRC(7,3)= (t4 + 2*t5 + t6 + 2) >> 2;
1061 SRC(6,0)=SRC(7,2)= (t5 + t6 + 1) >> 1;
1062 SRC(7,1)= (t5 + 2*t6 + t7 + 2) >> 2;
1063 SRC(7,0)= (t6 + t7 + 1) >> 1;
1065 static void pred8x8l_horizontal_down_c(uint8_t *src, int has_topleft, int has_topright, int stride)
1067 PREDICT_8x8_LOAD_TOP;
1068 PREDICT_8x8_LOAD_LEFT;
1069 PREDICT_8x8_LOAD_TOPLEFT;
1070 SRC(0,7)= (l6 + l7 + 1) >> 1;
1071 SRC(1,7)= (l5 + 2*l6 + l7 + 2) >> 2;
1072 SRC(0,6)=SRC(2,7)= (l5 + l6 + 1) >> 1;
1073 SRC(1,6)=SRC(3,7)= (l4 + 2*l5 + l6 + 2) >> 2;
1074 SRC(0,5)=SRC(2,6)=SRC(4,7)= (l4 + l5 + 1) >> 1;
1075 SRC(1,5)=SRC(3,6)=SRC(5,7)= (l3 + 2*l4 + l5 + 2) >> 2;
1076 SRC(0,4)=SRC(2,5)=SRC(4,6)=SRC(6,7)= (l3 + l4 + 1) >> 1;
1077 SRC(1,4)=SRC(3,5)=SRC(5,6)=SRC(7,7)= (l2 + 2*l3 + l4 + 2) >> 2;
1078 SRC(0,3)=SRC(2,4)=SRC(4,5)=SRC(6,6)= (l2 + l3 + 1) >> 1;
1079 SRC(1,3)=SRC(3,4)=SRC(5,5)=SRC(7,6)= (l1 + 2*l2 + l3 + 2) >> 2;
1080 SRC(0,2)=SRC(2,3)=SRC(4,4)=SRC(6,5)= (l1 + l2 + 1) >> 1;
1081 SRC(1,2)=SRC(3,3)=SRC(5,4)=SRC(7,5)= (l0 + 2*l1 + l2 + 2) >> 2;
1082 SRC(0,1)=SRC(2,2)=SRC(4,3)=SRC(6,4)= (l0 + l1 + 1) >> 1;
1083 SRC(1,1)=SRC(3,2)=SRC(5,3)=SRC(7,4)= (lt + 2*l0 + l1 + 2) >> 2;
1084 SRC(0,0)=SRC(2,1)=SRC(4,2)=SRC(6,3)= (lt + l0 + 1) >> 1;
1085 SRC(1,0)=SRC(3,1)=SRC(5,2)=SRC(7,3)= (l0 + 2*lt + t0 + 2) >> 2;
1086 SRC(2,0)=SRC(4,1)=SRC(6,2)= (t1 + 2*t0 + lt + 2) >> 2;
1087 SRC(3,0)=SRC(5,1)=SRC(7,2)= (t2 + 2*t1 + t0 + 2) >> 2;
1088 SRC(4,0)=SRC(6,1)= (t3 + 2*t2 + t1 + 2) >> 2;
1089 SRC(5,0)=SRC(7,1)= (t4 + 2*t3 + t2 + 2) >> 2;
1090 SRC(6,0)= (t5 + 2*t4 + t3 + 2) >> 2;
1091 SRC(7,0)= (t6 + 2*t5 + t4 + 2) >> 2;
1093 static void pred8x8l_vertical_left_c(uint8_t *src, int has_topleft, int has_topright, int stride)
1095 PREDICT_8x8_LOAD_TOP;
1096 PREDICT_8x8_LOAD_TOPRIGHT;
1097 SRC(0,0)= (t0 + t1 + 1) >> 1;
1098 SRC(0,1)= (t0 + 2*t1 + t2 + 2) >> 2;
1099 SRC(0,2)=SRC(1,0)= (t1 + t2 + 1) >> 1;
1100 SRC(0,3)=SRC(1,1)= (t1 + 2*t2 + t3 + 2) >> 2;
1101 SRC(0,4)=SRC(1,2)=SRC(2,0)= (t2 + t3 + 1) >> 1;
1102 SRC(0,5)=SRC(1,3)=SRC(2,1)= (t2 + 2*t3 + t4 + 2) >> 2;
1103 SRC(0,6)=SRC(1,4)=SRC(2,2)=SRC(3,0)= (t3 + t4 + 1) >> 1;
1104 SRC(0,7)=SRC(1,5)=SRC(2,3)=SRC(3,1)= (t3 + 2*t4 + t5 + 2) >> 2;
1105 SRC(1,6)=SRC(2,4)=SRC(3,2)=SRC(4,0)= (t4 + t5 + 1) >> 1;
1106 SRC(1,7)=SRC(2,5)=SRC(3,3)=SRC(4,1)= (t4 + 2*t5 + t6 + 2) >> 2;
1107 SRC(2,6)=SRC(3,4)=SRC(4,2)=SRC(5,0)= (t5 + t6 + 1) >> 1;
1108 SRC(2,7)=SRC(3,5)=SRC(4,3)=SRC(5,1)= (t5 + 2*t6 + t7 + 2) >> 2;
1109 SRC(3,6)=SRC(4,4)=SRC(5,2)=SRC(6,0)= (t6 + t7 + 1) >> 1;
1110 SRC(3,7)=SRC(4,5)=SRC(5,3)=SRC(6,1)= (t6 + 2*t7 + t8 + 2) >> 2;
1111 SRC(4,6)=SRC(5,4)=SRC(6,2)=SRC(7,0)= (t7 + t8 + 1) >> 1;
1112 SRC(4,7)=SRC(5,5)=SRC(6,3)=SRC(7,1)= (t7 + 2*t8 + t9 + 2) >> 2;
1113 SRC(5,6)=SRC(6,4)=SRC(7,2)= (t8 + t9 + 1) >> 1;
1114 SRC(5,7)=SRC(6,5)=SRC(7,3)= (t8 + 2*t9 + t10 + 2) >> 2;
1115 SRC(6,6)=SRC(7,4)= (t9 + t10 + 1) >> 1;
1116 SRC(6,7)=SRC(7,5)= (t9 + 2*t10 + t11 + 2) >> 2;
1117 SRC(7,6)= (t10 + t11 + 1) >> 1;
1118 SRC(7,7)= (t10 + 2*t11 + t12 + 2) >> 2;
1120 static void pred8x8l_horizontal_up_c(uint8_t *src, int has_topleft, int has_topright, int stride)
1122 PREDICT_8x8_LOAD_LEFT;
1123 SRC(0,0)= (l0 + l1 + 1) >> 1;
1124 SRC(1,0)= (l0 + 2*l1 + l2 + 2) >> 2;
1125 SRC(0,1)=SRC(2,0)= (l1 + l2 + 1) >> 1;
1126 SRC(1,1)=SRC(3,0)= (l1 + 2*l2 + l3 + 2) >> 2;
1127 SRC(0,2)=SRC(2,1)=SRC(4,0)= (l2 + l3 + 1) >> 1;
1128 SRC(1,2)=SRC(3,1)=SRC(5,0)= (l2 + 2*l3 + l4 + 2) >> 2;
1129 SRC(0,3)=SRC(2,2)=SRC(4,1)=SRC(6,0)= (l3 + l4 + 1) >> 1;
1130 SRC(1,3)=SRC(3,2)=SRC(5,1)=SRC(7,0)= (l3 + 2*l4 + l5 + 2) >> 2;
1131 SRC(0,4)=SRC(2,3)=SRC(4,2)=SRC(6,1)= (l4 + l5 + 1) >> 1;
1132 SRC(1,4)=SRC(3,3)=SRC(5,2)=SRC(7,1)= (l4 + 2*l5 + l6 + 2) >> 2;
1133 SRC(0,5)=SRC(2,4)=SRC(4,3)=SRC(6,2)= (l5 + l6 + 1) >> 1;
1134 SRC(1,5)=SRC(3,4)=SRC(5,3)=SRC(7,2)= (l5 + 2*l6 + l7 + 2) >> 2;
1135 SRC(0,6)=SRC(2,5)=SRC(4,4)=SRC(6,3)= (l6 + l7 + 1) >> 1;
1136 SRC(1,6)=SRC(3,5)=SRC(5,4)=SRC(7,3)= (l6 + 3*l7 + 2) >> 2;
1137 SRC(0,7)=SRC(1,7)=SRC(2,6)=SRC(2,7)=SRC(3,6)=
1138 SRC(3,7)=SRC(4,5)=SRC(4,6)=SRC(4,7)=SRC(5,5)=
1139 SRC(5,6)=SRC(5,7)=SRC(6,4)=SRC(6,5)=SRC(6,6)=
1140 SRC(6,7)=SRC(7,4)=SRC(7,5)=SRC(7,6)=SRC(7,7)= l7;
1142 #undef PREDICT_8x8_LOAD_LEFT
1143 #undef PREDICT_8x8_LOAD_TOP
1144 #undef PREDICT_8x8_LOAD_TOPLEFT
1145 #undef PREDICT_8x8_LOAD_TOPRIGHT
1146 #undef PREDICT_8x8_DC
1152 static void pred4x4_vertical_add_c(uint8_t *pix, const DCTELEM *block, int stride){
1157 pix[1*stride]= v += block[0];
1158 pix[2*stride]= v += block[4];
1159 pix[3*stride]= v += block[8];
1160 pix[4*stride]= v + block[12];
1166 static void pred4x4_horizontal_add_c(uint8_t *pix, const DCTELEM *block, int stride){
1169 uint8_t v = pix[-1];
1170 pix[0]= v += block[0];
1171 pix[1]= v += block[1];
1172 pix[2]= v += block[2];
1173 pix[3]= v + block[3];
1179 static void pred8x8l_vertical_add_c(uint8_t *pix, const DCTELEM *block, int stride){
1184 pix[1*stride]= v += block[0];
1185 pix[2*stride]= v += block[8];
1186 pix[3*stride]= v += block[16];
1187 pix[4*stride]= v += block[24];
1188 pix[5*stride]= v += block[32];
1189 pix[6*stride]= v += block[40];
1190 pix[7*stride]= v += block[48];
1191 pix[8*stride]= v + block[56];
1197 static void pred8x8l_horizontal_add_c(uint8_t *pix, const DCTELEM *block, int stride){
1200 uint8_t v = pix[-1];
1201 pix[0]= v += block[0];
1202 pix[1]= v += block[1];
1203 pix[2]= v += block[2];
1204 pix[3]= v += block[3];
1205 pix[4]= v += block[4];
1206 pix[5]= v += block[5];
1207 pix[6]= v += block[6];
1208 pix[7]= v + block[7];
1214 static void pred16x16_vertical_add_c(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){
1217 pred4x4_vertical_add_c(pix + block_offset[i], block + i*16, stride);
1220 static void pred16x16_horizontal_add_c(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){
1223 pred4x4_horizontal_add_c(pix + block_offset[i], block + i*16, stride);
1226 static void pred8x8_vertical_add_c(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){
1229 pred4x4_vertical_add_c(pix + block_offset[i], block + i*16, stride);
1232 static void pred8x8_horizontal_add_c(uint8_t *pix, const int *block_offset, const DCTELEM *block, int stride){
1235 pred4x4_horizontal_add_c(pix + block_offset[i], block + i*16, stride);
1240 * Set the intra prediction function pointers.
1242 void ff_h264_pred_init(H264PredContext *h, int codec_id){
1243 // MpegEncContext * const s = &h->s;
1245 if(codec_id != CODEC_ID_RV40){
1246 if(codec_id == CODEC_ID_VP8) {
1247 h->pred4x4[VERT_PRED ]= pred4x4_vertical_vp8_c;
1248 h->pred4x4[HOR_PRED ]= pred4x4_horizontal_vp8_c;
1250 h->pred4x4[VERT_PRED ]= pred4x4_vertical_c;
1251 h->pred4x4[HOR_PRED ]= pred4x4_horizontal_c;
1253 h->pred4x4[DC_PRED ]= pred4x4_dc_c;
1254 if(codec_id == CODEC_ID_SVQ3)
1255 h->pred4x4[DIAG_DOWN_LEFT_PRED ]= pred4x4_down_left_svq3_c;
1257 h->pred4x4[DIAG_DOWN_LEFT_PRED ]= pred4x4_down_left_c;
1258 h->pred4x4[DIAG_DOWN_RIGHT_PRED]= pred4x4_down_right_c;
1259 h->pred4x4[VERT_RIGHT_PRED ]= pred4x4_vertical_right_c;
1260 h->pred4x4[HOR_DOWN_PRED ]= pred4x4_horizontal_down_c;
1261 if (codec_id == CODEC_ID_VP8) {
1262 h->pred4x4[VERT_LEFT_PRED ]= pred4x4_vertical_left_vp8_c;
1264 h->pred4x4[VERT_LEFT_PRED ]= pred4x4_vertical_left_c;
1265 h->pred4x4[HOR_UP_PRED ]= pred4x4_horizontal_up_c;
1266 if(codec_id != CODEC_ID_VP8) {
1267 h->pred4x4[LEFT_DC_PRED ]= pred4x4_left_dc_c;
1268 h->pred4x4[TOP_DC_PRED ]= pred4x4_top_dc_c;
1269 h->pred4x4[DC_128_PRED ]= pred4x4_128_dc_c;
1271 h->pred4x4[TM_VP8_PRED ]= pred4x4_tm_vp8_c;
1272 h->pred4x4[DC_127_PRED ]= pred4x4_127_dc_c;
1273 h->pred4x4[DC_129_PRED ]= pred4x4_129_dc_c;
1274 h->pred4x4[VERT_VP8_PRED ]= pred4x4_vertical_c;
1275 h->pred4x4[HOR_VP8_PRED ]= pred4x4_horizontal_c;
1278 h->pred4x4[VERT_PRED ]= pred4x4_vertical_c;
1279 h->pred4x4[HOR_PRED ]= pred4x4_horizontal_c;
1280 h->pred4x4[DC_PRED ]= pred4x4_dc_c;
1281 h->pred4x4[DIAG_DOWN_LEFT_PRED ]= pred4x4_down_left_rv40_c;
1282 h->pred4x4[DIAG_DOWN_RIGHT_PRED]= pred4x4_down_right_c;
1283 h->pred4x4[VERT_RIGHT_PRED ]= pred4x4_vertical_right_c;
1284 h->pred4x4[HOR_DOWN_PRED ]= pred4x4_horizontal_down_c;
1285 h->pred4x4[VERT_LEFT_PRED ]= pred4x4_vertical_left_rv40_c;
1286 h->pred4x4[HOR_UP_PRED ]= pred4x4_horizontal_up_rv40_c;
1287 h->pred4x4[LEFT_DC_PRED ]= pred4x4_left_dc_c;
1288 h->pred4x4[TOP_DC_PRED ]= pred4x4_top_dc_c;
1289 h->pred4x4[DC_128_PRED ]= pred4x4_128_dc_c;
1290 h->pred4x4[DIAG_DOWN_LEFT_PRED_RV40_NODOWN]= pred4x4_down_left_rv40_nodown_c;
1291 h->pred4x4[HOR_UP_PRED_RV40_NODOWN]= pred4x4_horizontal_up_rv40_nodown_c;
1292 h->pred4x4[VERT_LEFT_PRED_RV40_NODOWN]= pred4x4_vertical_left_rv40_nodown_c;
1295 h->pred8x8l[VERT_PRED ]= pred8x8l_vertical_c;
1296 h->pred8x8l[HOR_PRED ]= pred8x8l_horizontal_c;
1297 h->pred8x8l[DC_PRED ]= pred8x8l_dc_c;
1298 h->pred8x8l[DIAG_DOWN_LEFT_PRED ]= pred8x8l_down_left_c;
1299 h->pred8x8l[DIAG_DOWN_RIGHT_PRED]= pred8x8l_down_right_c;
1300 h->pred8x8l[VERT_RIGHT_PRED ]= pred8x8l_vertical_right_c;
1301 h->pred8x8l[HOR_DOWN_PRED ]= pred8x8l_horizontal_down_c;
1302 h->pred8x8l[VERT_LEFT_PRED ]= pred8x8l_vertical_left_c;
1303 h->pred8x8l[HOR_UP_PRED ]= pred8x8l_horizontal_up_c;
1304 h->pred8x8l[LEFT_DC_PRED ]= pred8x8l_left_dc_c;
1305 h->pred8x8l[TOP_DC_PRED ]= pred8x8l_top_dc_c;
1306 h->pred8x8l[DC_128_PRED ]= pred8x8l_128_dc_c;
1308 h->pred8x8[VERT_PRED8x8 ]= pred8x8_vertical_c;
1309 h->pred8x8[HOR_PRED8x8 ]= pred8x8_horizontal_c;
1310 if (codec_id != CODEC_ID_VP8) {
1311 h->pred8x8[PLANE_PRED8x8]= pred8x8_plane_c;
1313 h->pred8x8[PLANE_PRED8x8]= pred8x8_tm_vp8_c;
1314 if(codec_id != CODEC_ID_RV40 && codec_id != CODEC_ID_VP8){
1315 h->pred8x8[DC_PRED8x8 ]= pred8x8_dc_c;
1316 h->pred8x8[LEFT_DC_PRED8x8]= pred8x8_left_dc_c;
1317 h->pred8x8[TOP_DC_PRED8x8 ]= pred8x8_top_dc_c;
1318 h->pred8x8[ALZHEIMER_DC_L0T_PRED8x8 ]= pred8x8_mad_cow_dc_l0t;
1319 h->pred8x8[ALZHEIMER_DC_0LT_PRED8x8 ]= pred8x8_mad_cow_dc_0lt;
1320 h->pred8x8[ALZHEIMER_DC_L00_PRED8x8 ]= pred8x8_mad_cow_dc_l00;
1321 h->pred8x8[ALZHEIMER_DC_0L0_PRED8x8 ]= pred8x8_mad_cow_dc_0l0;
1323 h->pred8x8[DC_PRED8x8 ]= pred8x8_dc_rv40_c;
1324 h->pred8x8[LEFT_DC_PRED8x8]= pred8x8_left_dc_rv40_c;
1325 h->pred8x8[TOP_DC_PRED8x8 ]= pred8x8_top_dc_rv40_c;
1326 if (codec_id == CODEC_ID_VP8) {
1327 h->pred8x8[DC_127_PRED8x8]= pred8x8_127_dc_c;
1328 h->pred8x8[DC_129_PRED8x8]= pred8x8_129_dc_c;
1331 h->pred8x8[DC_128_PRED8x8 ]= pred8x8_128_dc_c;
1333 h->pred16x16[DC_PRED8x8 ]= pred16x16_dc_c;
1334 h->pred16x16[VERT_PRED8x8 ]= pred16x16_vertical_c;
1335 h->pred16x16[HOR_PRED8x8 ]= pred16x16_horizontal_c;
1338 h->pred16x16[PLANE_PRED8x8 ]= pred16x16_plane_svq3_c;
1341 h->pred16x16[PLANE_PRED8x8 ]= pred16x16_plane_rv40_c;
1344 h->pred16x16[PLANE_PRED8x8 ]= pred16x16_tm_vp8_c;
1345 h->pred16x16[DC_127_PRED8x8]= pred16x16_127_dc_c;
1346 h->pred16x16[DC_129_PRED8x8]= pred16x16_129_dc_c;
1349 h->pred16x16[PLANE_PRED8x8 ]= pred16x16_plane_c;
1352 h->pred16x16[LEFT_DC_PRED8x8]= pred16x16_left_dc_c;
1353 h->pred16x16[TOP_DC_PRED8x8 ]= pred16x16_top_dc_c;
1354 h->pred16x16[DC_128_PRED8x8 ]= pred16x16_128_dc_c;
1356 //special lossless h/v prediction for h264
1357 h->pred4x4_add [VERT_PRED ]= pred4x4_vertical_add_c;
1358 h->pred4x4_add [ HOR_PRED ]= pred4x4_horizontal_add_c;
1359 h->pred8x8l_add [VERT_PRED ]= pred8x8l_vertical_add_c;
1360 h->pred8x8l_add [ HOR_PRED ]= pred8x8l_horizontal_add_c;
1361 h->pred8x8_add [VERT_PRED8x8]= pred8x8_vertical_add_c;
1362 h->pred8x8_add [ HOR_PRED8x8]= pred8x8_horizontal_add_c;
1363 h->pred16x16_add[VERT_PRED8x8]= pred16x16_vertical_add_c;
1364 h->pred16x16_add[ HOR_PRED8x8]= pred16x16_horizontal_add_c;
1366 if (ARCH_ARM) ff_h264_pred_init_arm(h, codec_id);
1367 if (HAVE_MMX) ff_h264_pred_init_x86(h, codec_id);