falkTX
/
FFmpeg
mirror of https://github.com/falkTX/FFmpeg.git
1
0
Fork 0
Code Issues 0 Releases 338 Wiki Activity
You can not select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
10865 Commits
32 Branches
766MB
Tree: c92a30bb06
Branches Tags
${ item.name }
Create branch ${ searchTerm }
from 'c92a30bb06'
${ noResults }
FFmpeg/libavcodec/h264pred.c

1060 lines
35KB
Raw Blame History 

  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_notop_c(uint8_t *src, uint8_t *topright, int stride){

  202.     LOAD_LEFT_EDGE

  203.     LOAD_DOWN_LEFT_EDGE

  204. 

  205.     src[0+0*stride]=(l0 + l2 + 2*l1 + 2)>>2;

  206.     src[1+0*stride]=

  207.     src[0+1*stride]=(l1 + l3 + 2*l2 + 2)>>2;

  208.     src[2+0*stride]=

  209.     src[1+1*stride]=

  210.     src[0+2*stride]=(l2 + l4 + 2*l3 + 2)>>2;

  211.     src[3+0*stride]=

  212.     src[2+1*stride]=

  213.     src[1+2*stride]=

  214.     src[0+3*stride]=(l3 + l5 + 2*l4 + 2)>>2;

  215.     src[3+1*stride]=

  216.     src[2+2*stride]=

  217.     src[1+3*stride]=(l4 + l6 + 2*l5 + 2)>>2;

  218.     src[3+2*stride]=

  219.     src[2+3*stride]=(l5 + l7 + 2*l6 + 2)>>2;

  220.     src[3+3*stride]=(l6 + l7 + 1)>>1;

  221. }

  222. 

  223. static void pred4x4_down_left_rv40_nodown_c(uint8_t *src, uint8_t *topright, int stride){

  224.     LOAD_TOP_EDGE

  225.     LOAD_TOP_RIGHT_EDGE

  226.     LOAD_LEFT_EDGE

  227. 

  228.     src[0+0*stride]=(t0 + t2 + 2*t1 + 2 + l0 + l2 + 2*l1 + 2)>>3;

  229.     src[1+0*stride]=

  230.     src[0+1*stride]=(t1 + t3 + 2*t2 + 2 + l1 + l3 + 2*l2 + 2)>>3;

  231.     src[2+0*stride]=

  232.     src[1+1*stride]=

  233.     src[0+2*stride]=(t2 + t4 + 2*t3 + 2 + l2 + 3*l3 + 2)>>3;

  234.     src[3+0*stride]=

  235.     src[2+1*stride]=

  236.     src[1+2*stride]=

  237.     src[0+3*stride]=(t3 + t5 + 2*t4 + 2 + l3*4 + 2)>>3;

  238.     src[3+1*stride]=

  239.     src[2+2*stride]=

  240.     src[1+3*stride]=(t4 + t6 + 2*t5 + 2 + l3*4 + 2)>>3;

  241.     src[3+2*stride]=

  242.     src[2+3*stride]=(t5 + t7 + 2*t6 + 2 + l3*4 + 2)>>3;

  243.     src[3+3*stride]=(t6 + t7 + 1 + 2*l3 + 1)>>2;

  244. }

  245. 

  246. static void pred4x4_vertical_right_c(uint8_t *src, uint8_t *topright, int stride){

  247.     const int lt= src[-1-1*stride];

  248.     LOAD_TOP_EDGE

  249.     LOAD_LEFT_EDGE

  250. 

  251.     src[0+0*stride]=

  252.     src[1+2*stride]=(lt + t0 + 1)>>1;

  253.     src[1+0*stride]=

  254.     src[2+2*stride]=(t0 + t1 + 1)>>1;

  255.     src[2+0*stride]=

  256.     src[3+2*stride]=(t1 + t2 + 1)>>1;

  257.     src[3+0*stride]=(t2 + t3 + 1)>>1;

  258.     src[0+1*stride]=

  259.     src[1+3*stride]=(l0 + 2*lt + t0 + 2)>>2;

  260.     src[1+1*stride]=

  261.     src[2+3*stride]=(lt + 2*t0 + t1 + 2)>>2;

  262.     src[2+1*stride]=

  263.     src[3+3*stride]=(t0 + 2*t1 + t2 + 2)>>2;

  264.     src[3+1*stride]=(t1 + 2*t2 + t3 + 2)>>2;

  265.     src[0+2*stride]=(lt + 2*l0 + l1 + 2)>>2;

  266.     src[0+3*stride]=(l0 + 2*l1 + l2 + 2)>>2;

  267. }

  268. 

  269. static void pred4x4_vertical_left_c(uint8_t *src, uint8_t *topright, int stride){

  270.     LOAD_TOP_EDGE

  271.     LOAD_TOP_RIGHT_EDGE

  272. 

  273.     src[0+0*stride]=(t0 + t1 + 1)>>1;

  274.     src[1+0*stride]=

  275.     src[0+2*stride]=(t1 + t2 + 1)>>1;

  276.     src[2+0*stride]=

  277.     src[1+2*stride]=(t2 + t3 + 1)>>1;

  278.     src[3+0*stride]=

  279.     src[2+2*stride]=(t3 + t4+ 1)>>1;

  280.     src[3+2*stride]=(t4 + t5+ 1)>>1;

  281.     src[0+1*stride]=(t0 + 2*t1 + t2 + 2)>>2;

  282.     src[1+1*stride]=

  283.     src[0+3*stride]=(t1 + 2*t2 + t3 + 2)>>2;

  284.     src[2+1*stride]=

  285.     src[1+3*stride]=(t2 + 2*t3 + t4 + 2)>>2;

  286.     src[3+1*stride]=

  287.     src[2+3*stride]=(t3 + 2*t4 + t5 + 2)>>2;

  288.     src[3+3*stride]=(t4 + 2*t5 + t6 + 2)>>2;

  289. }

  290. 

  291. static void pred4x4_vertical_left_rv40_c(uint8_t *src, uint8_t *topright, int stride){

  292.     LOAD_TOP_EDGE

  293.     LOAD_TOP_RIGHT_EDGE

  294.     LOAD_LEFT_EDGE

  295. 

  296.     src[0+0*stride]=(2*t0 + 2*t1 + l1 + 2*l2 + l3 + 4)>>3;

  297.     src[1+0*stride]=

  298.     src[0+2*stride]=(t1 + t2 + 1)>>1;

  299.     src[2+0*stride]=

  300.     src[1+2*stride]=(t2 + t3 + 1)>>1;

  301.     src[3+0*stride]=

  302.     src[2+2*stride]=(t3 + t4+ 1)>>1;

  303.     src[3+2*stride]=(t4 + t5+ 1)>>1;

  304.     src[0+1*stride]=(t0 + 2*t1 + t2 + l2 + 2*l3 + l3 + 4)>>3;

  305.     src[1+1*stride]=

  306.     src[0+3*stride]=(t1 + 2*t2 + t3 + 2)>>2;

  307.     src[2+1*stride]=

  308.     src[1+3*stride]=(t2 + 2*t3 + t4 + 2)>>2;

  309.     src[3+1*stride]=

  310.     src[2+3*stride]=(t3 + 2*t4 + t5 + 2)>>2;

  311.     src[3+3*stride]=(t4 + 2*t5 + t6 + 2)>>2;

  312. }

  313. 

  314. static void pred4x4_horizontal_up_c(uint8_t *src, uint8_t *topright, int stride){

  315.     LOAD_LEFT_EDGE

  316. 

  317.     src[0+0*stride]=(l0 + l1 + 1)>>1;

  318.     src[1+0*stride]=(l0 + 2*l1 + l2 + 2)>>2;

  319.     src[2+0*stride]=

  320.     src[0+1*stride]=(l1 + l2 + 1)>>1;

  321.     src[3+0*stride]=

  322.     src[1+1*stride]=(l1 + 2*l2 + l3 + 2)>>2;

  323.     src[2+1*stride]=

  324.     src[0+2*stride]=(l2 + l3 + 1)>>1;

  325.     src[3+1*stride]=

  326.     src[1+2*stride]=(l2 + 2*l3 + l3 + 2)>>2;

  327.     src[3+2*stride]=

  328.     src[1+3*stride]=

  329.     src[0+3*stride]=

  330.     src[2+2*stride]=

  331.     src[2+3*stride]=

  332.     src[3+3*stride]=l3;

  333. }

  334. 

  335. static void pred4x4_horizontal_up_rv40_c(uint8_t *src, uint8_t *topright, int stride){

  336.     LOAD_LEFT_EDGE

  337.     LOAD_DOWN_LEFT_EDGE

  338.     LOAD_TOP_EDGE

  339.     LOAD_TOP_RIGHT_EDGE

  340. 

  341.     src[0+0*stride]=(t1 + 2*t2 + t3 + 2*l0 + 2*l1 + 4)>>3;

  342.     src[1+0*stride]=(t2 + 2*t3 + t4 + l0 + 2*l1 + l2 + 4)>>3;

  343.     src[2+0*stride]=

  344.     src[0+1*stride]=(t3 + 2*t4 + t5 + 2*l1 + 2*l2 + 4)>>3;

  345.     src[3+0*stride]=

  346.     src[1+1*stride]=(t4 + 2*t5 + t6 + l1 + 2*l2 + l3 + 4)>>3;

  347.     src[2+1*stride]=

  348.     src[0+2*stride]=(t5 + 2*t6 + t7 + 2*l2 + 2*l3 + 4)>>3;

  349.     src[3+1*stride]=

  350.     src[1+2*stride]=(t6 + 3*t7 + l2 + 3*l3 + 4)>>3;

  351.     src[3+2*stride]=

  352.     src[1+3*stride]=(l3 + 2*l4 + l5 + 2)>>2;

  353.     src[0+3*stride]=

  354.     src[2+2*stride]=(t6 + t7 + l3 + l4 + 2)>>2;

  355.     src[2+3*stride]=(l4 + l5 + 1)>>1;

  356.     src[3+3*stride]=(l4 + 2*l5 + l6 + 2)>>2;

  357. }

  358. 

  359. static void pred4x4_horizontal_up_rv40_nodown_c(uint8_t *src, uint8_t *topright, int stride){

  360.     LOAD_LEFT_EDGE

  361.     LOAD_TOP_EDGE

  362.     LOAD_TOP_RIGHT_EDGE

  363. 

  364.     src[0+0*stride]=(t1 + 2*t2 + t3 + 2*l0 + 2*l1 + 4)>>3;

  365.     src[1+0*stride]=(t2 + 2*t3 + t4 + l0 + 2*l1 + l2 + 4)>>3;

  366.     src[2+0*stride]=

  367.     src[0+1*stride]=(t3 + 2*t4 + t5 + 2*l1 + 2*l2 + 4)>>3;

  368.     src[3+0*stride]=

  369.     src[1+1*stride]=(t4 + 2*t5 + t6 + l1 + 2*l2 + l3 + 4)>>3;

  370.     src[2+1*stride]=

  371.     src[0+2*stride]=(t5 + 2*t6 + t7 + 2*l2 + 2*l3 + 4)>>3;

  372.     src[3+1*stride]=

  373.     src[1+2*stride]=(t6 + 3*t7 + l2 + 3*l3 + 4)>>3;

  374.     src[3+2*stride]=

  375.     src[1+3*stride]=l3;

  376.     src[0+3*stride]=

  377.     src[2+2*stride]=(t6 + t7 + 2*l3 + 2)>>2;

  378.     src[2+3*stride]=

  379.     src[3+3*stride]=l3;

  380. }

  381. 

  382. static void pred4x4_horizontal_down_c(uint8_t *src, uint8_t *topright, int stride){

  383.     const int lt= src[-1-1*stride];

  384.     LOAD_TOP_EDGE

  385.     LOAD_LEFT_EDGE

  386. 

  387.     src[0+0*stride]=

  388.     src[2+1*stride]=(lt + l0 + 1)>>1;

  389.     src[1+0*stride]=

  390.     src[3+1*stride]=(l0 + 2*lt + t0 + 2)>>2;

  391.     src[2+0*stride]=(lt + 2*t0 + t1 + 2)>>2;

  392.     src[3+0*stride]=(t0 + 2*t1 + t2 + 2)>>2;

  393.     src[0+1*stride]=

  394.     src[2+2*stride]=(l0 + l1 + 1)>>1;

  395.     src[1+1*stride]=

  396.     src[3+2*stride]=(lt + 2*l0 + l1 + 2)>>2;

  397.     src[0+2*stride]=

  398.     src[2+3*stride]=(l1 + l2+ 1)>>1;

  399.     src[1+2*stride]=

  400.     src[3+3*stride]=(l0 + 2*l1 + l2 + 2)>>2;

  401.     src[0+3*stride]=(l2 + l3 + 1)>>1;

  402.     src[1+3*stride]=(l1 + 2*l2 + l3 + 2)>>2;

  403. }

  404. 

  405. static void pred16x16_vertical_c(uint8_t *src, int stride){

  406.     int i;

  407.     const uint32_t a= ((uint32_t*)(src-stride))[0];

  408.     const uint32_t b= ((uint32_t*)(src-stride))[1];

  409.     const uint32_t c= ((uint32_t*)(src-stride))[2];

  410.     const uint32_t d= ((uint32_t*)(src-stride))[3];

  411. 

  412.     for(i=0; i<16; i++){

  413.         ((uint32_t*)(src+i*stride))[0]= a;

  414.         ((uint32_t*)(src+i*stride))[1]= b;

  415.         ((uint32_t*)(src+i*stride))[2]= c;

  416.         ((uint32_t*)(src+i*stride))[3]= d;

  417.     }

  418. }

  419. 

  420. static void pred16x16_horizontal_c(uint8_t *src, int stride){

  421.     int i;

  422. 

  423.     for(i=0; i<16; i++){

  424.         ((uint32_t*)(src+i*stride))[0]=

  425.         ((uint32_t*)(src+i*stride))[1]=

  426.         ((uint32_t*)(src+i*stride))[2]=

  427.         ((uint32_t*)(src+i*stride))[3]= src[-1+i*stride]*0x01010101;

  428.     }

  429. }

  430. 

  431. static void pred16x16_dc_c(uint8_t *src, int stride){

  432.     int i, dc=0;

  433. 

  434.     for(i=0;i<16; i++){

  435.         dc+= src[-1+i*stride];

  436.     }

  437. 

  438.     for(i=0;i<16; i++){

  439.         dc+= src[i-stride];

  440.     }

  441. 

  442.     dc= 0x01010101*((dc + 16)>>5);

  443. 

  444.     for(i=0; i<16; i++){

  445.         ((uint32_t*)(src+i*stride))[0]=

  446.         ((uint32_t*)(src+i*stride))[1]=

  447.         ((uint32_t*)(src+i*stride))[2]=

  448.         ((uint32_t*)(src+i*stride))[3]= dc;

  449.     }

  450. }

  451. 

  452. static void pred16x16_left_dc_c(uint8_t *src, int stride){

  453.     int i, dc=0;

  454. 

  455.     for(i=0;i<16; i++){

  456.         dc+= src[-1+i*stride];

  457.     }

  458. 

  459.     dc= 0x01010101*((dc + 8)>>4);

  460. 

  461.     for(i=0; i<16; i++){

  462.         ((uint32_t*)(src+i*stride))[0]=

  463.         ((uint32_t*)(src+i*stride))[1]=

  464.         ((uint32_t*)(src+i*stride))[2]=

  465.         ((uint32_t*)(src+i*stride))[3]= dc;

  466.     }

  467. }

  468. 

  469. static void pred16x16_top_dc_c(uint8_t *src, int stride){

  470.     int i, dc=0;

  471. 

  472.     for(i=0;i<16; i++){

  473.         dc+= src[i-stride];

  474.     }

  475.     dc= 0x01010101*((dc + 8)>>4);

  476. 

  477.     for(i=0; i<16; i++){

  478.         ((uint32_t*)(src+i*stride))[0]=

  479.         ((uint32_t*)(src+i*stride))[1]=

  480.         ((uint32_t*)(src+i*stride))[2]=

  481.         ((uint32_t*)(src+i*stride))[3]= dc;

  482.     }

  483. }

  484. 

  485. static void pred16x16_128_dc_c(uint8_t *src, int stride){

  486.     int i;

  487. 

  488.     for(i=0; i<16; i++){

  489.         ((uint32_t*)(src+i*stride))[0]=

  490.         ((uint32_t*)(src+i*stride))[1]=

  491.         ((uint32_t*)(src+i*stride))[2]=

  492.         ((uint32_t*)(src+i*stride))[3]= 0x01010101U*128U;

  493.     }

  494. }

  495. 

  496. static inline void pred16x16_plane_compat_c(uint8_t *src, int stride, const int svq3, const int rv40){

  497.   int i, j, k;

  498.   int a;

  499.   uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;

  500.   const uint8_t * const src0 = src+7-stride;

  501.   const uint8_t *src1 = src+8*stride-1;

  502.   const uint8_t *src2 = src1-2*stride;      // == src+6*stride-1;

  503.   int H = src0[1] - src0[-1];

  504.   int V = src1[0] - src2[ 0];

  505.   for(k=2; k<=8; ++k) {

  506.     src1 += stride; src2 -= stride;

  507.     H += k*(src0[k] - src0[-k]);

  508.     V += k*(src1[0] - src2[ 0]);

  509.   }

  510.   if(svq3){

  511.     H = ( 5*(H/4) ) / 16;

  512.     V = ( 5*(V/4) ) / 16;

  513. 

  514.     /* required for 100% accuracy */

  515.     i = H; H = V; V = i;

  516.   }else if(rv40){

  517.     H = ( H + (H>>2) ) >> 4;

  518.     V = ( V + (V>>2) ) >> 4;

  519.   }else{

  520.     H = ( 5*H+32 ) >> 6;

  521.     V = ( 5*V+32 ) >> 6;

  522.   }

  523. 

  524.   a = 16*(src1[0] + src2[16] + 1) - 7*(V+H);

  525.   for(j=16; j>0; --j) {

  526.     int b = a;

  527.     a += V;

  528.     for(i=-16; i<0; i+=4) {

  529.       src[16+i] = cm[ (b    ) >> 5 ];

  530.       src[17+i] = cm[ (b+  H) >> 5 ];

  531.       src[18+i] = cm[ (b+2*H) >> 5 ];

  532.       src[19+i] = cm[ (b+3*H) >> 5 ];

  533.       b += 4*H;

  534.     }

  535.     src += stride;

  536.   }

  537. }

  538. 

  539. static void pred16x16_plane_c(uint8_t *src, int stride){

  540.     pred16x16_plane_compat_c(src, stride, 0, 0);

  541. }

  542. 

  543. static void pred16x16_plane_svq3_c(uint8_t *src, int stride){

  544.     pred16x16_plane_compat_c(src, stride, 1, 0);

  545. }

  546. 

  547. static void pred16x16_plane_rv40_c(uint8_t *src, int stride){

  548.     pred16x16_plane_compat_c(src, stride, 0, 1);

  549. }

  550. 

  551. static void pred8x8_vertical_c(uint8_t *src, int stride){

  552.     int i;

  553.     const uint32_t a= ((uint32_t*)(src-stride))[0];

  554.     const uint32_t b= ((uint32_t*)(src-stride))[1];

  555. 

  556.     for(i=0; i<8; i++){

  557.         ((uint32_t*)(src+i*stride))[0]= a;

  558.         ((uint32_t*)(src+i*stride))[1]= b;

  559.     }

  560. }

  561. 

  562. static void pred8x8_horizontal_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]= src[-1+i*stride]*0x01010101;

  568.     }

  569. }

  570. 

  571. static void pred8x8_128_dc_c(uint8_t *src, int stride){

  572.     int i;

  573. 

  574.     for(i=0; i<8; i++){

  575.         ((uint32_t*)(src+i*stride))[0]=

  576.         ((uint32_t*)(src+i*stride))[1]= 0x01010101U*128U;

  577.     }

  578. }

  579. 

  580. static void pred8x8_left_dc_c(uint8_t *src, int stride){

  581.     int i;

  582.     int dc0, dc2;

  583. 

  584.     dc0=dc2=0;

  585.     for(i=0;i<4; i++){

  586.         dc0+= src[-1+i*stride];

  587.         dc2+= src[-1+(i+4)*stride];

  588.     }

  589.     dc0= 0x01010101*((dc0 + 2)>>2);

  590.     dc2= 0x01010101*((dc2 + 2)>>2);

  591. 

  592.     for(i=0; i<4; i++){

  593.         ((uint32_t*)(src+i*stride))[0]=

  594.         ((uint32_t*)(src+i*stride))[1]= dc0;

  595.     }

  596.     for(i=4; i<8; i++){

  597.         ((uint32_t*)(src+i*stride))[0]=

  598.         ((uint32_t*)(src+i*stride))[1]= dc2;

  599.     }

  600. }

  601. 

  602. static void pred8x8_left_dc_rv40_c(uint8_t *src, int stride){

  603.     int i;

  604.     int dc0;

  605. 

  606.     dc0=0;

  607.     for(i=0;i<8; i++)

  608.         dc0+= src[-1+i*stride];

  609.     dc0= 0x01010101*((dc0 + 4)>>3);

  610. 

  611.     for(i=0; i<8; i++){

  612.         ((uint32_t*)(src+i*stride))[0]=

  613.         ((uint32_t*)(src+i*stride))[1]= dc0;

  614.     }

  615. }

  616. 

  617. static void pred8x8_top_dc_c(uint8_t *src, int stride){

  618.     int i;

  619.     int dc0, dc1;

  620. 

  621.     dc0=dc1=0;

  622.     for(i=0;i<4; i++){

  623.         dc0+= src[i-stride];

  624.         dc1+= src[4+i-stride];

  625.     }

  626.     dc0= 0x01010101*((dc0 + 2)>>2);

  627.     dc1= 0x01010101*((dc1 + 2)>>2);

  628. 

  629.     for(i=0; i<4; i++){

  630.         ((uint32_t*)(src+i*stride))[0]= dc0;

  631.         ((uint32_t*)(src+i*stride))[1]= dc1;

  632.     }

  633.     for(i=4; i<8; i++){

  634.         ((uint32_t*)(src+i*stride))[0]= dc0;

  635.         ((uint32_t*)(src+i*stride))[1]= dc1;

  636.     }

  637. }

  638. 

  639. static void pred8x8_top_dc_rv40_c(uint8_t *src, int stride){

  640.     int i;

  641.     int dc0;

  642. 

  643.     dc0=0;

  644.     for(i=0;i<8; i++)

  645.         dc0+= src[i-stride];

  646.     dc0= 0x01010101*((dc0 + 4)>>3);

  647. 

  648.     for(i=0; i<8; i++){

  649.         ((uint32_t*)(src+i*stride))[0]=

  650.         ((uint32_t*)(src+i*stride))[1]= dc0;

  651.     }

  652. }

  653. 

  654. 

  655. static void pred8x8_dc_c(uint8_t *src, int stride){

  656.     int i;

  657.     int dc0, dc1, dc2, dc3;

  658. 

  659.     dc0=dc1=dc2=0;

  660.     for(i=0;i<4; i++){

  661.         dc0+= src[-1+i*stride] + src[i-stride];

  662.         dc1+= src[4+i-stride];

  663.         dc2+= src[-1+(i+4)*stride];

  664.     }

  665.     dc3= 0x01010101*((dc1 + dc2 + 4)>>3);

  666.     dc0= 0x01010101*((dc0 + 4)>>3);

  667.     dc1= 0x01010101*((dc1 + 2)>>2);

  668.     dc2= 0x01010101*((dc2 + 2)>>2);

  669. 

  670.     for(i=0; i<4; i++){

  671.         ((uint32_t*)(src+i*stride))[0]= dc0;

  672.         ((uint32_t*)(src+i*stride))[1]= dc1;

  673.     }

  674.     for(i=4; i<8; i++){

  675.         ((uint32_t*)(src+i*stride))[0]= dc2;

  676.         ((uint32_t*)(src+i*stride))[1]= dc3;

  677.     }

  678. }

  679. 

  680. static void pred8x8_dc_rv40_c(uint8_t *src, int stride){

  681.     int i;

  682.     int dc0=0;

  683. 

  684.     for(i=0;i<4; i++){

  685.         dc0+= src[-1+i*stride] + src[i-stride];

  686.         dc0+= src[4+i-stride];

  687.         dc0+= src[-1+(i+4)*stride];

  688.     }

  689.     dc0= 0x01010101*((dc0 + 8)>>4);

  690. 

  691.     for(i=0; i<4; i++){

  692.         ((uint32_t*)(src+i*stride))[0]= dc0;

  693.         ((uint32_t*)(src+i*stride))[1]= dc0;

  694.     }

  695.     for(i=4; i<8; i++){

  696.         ((uint32_t*)(src+i*stride))[0]= dc0;

  697.         ((uint32_t*)(src+i*stride))[1]= dc0;

  698.     }

  699. }

  700. 

  701. static void pred8x8_plane_c(uint8_t *src, int stride){

  702.   int j, k;

  703.   int a;

  704.   uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;

  705.   const uint8_t * const src0 = src+3-stride;

  706.   const uint8_t *src1 = src+4*stride-1;

  707.   const uint8_t *src2 = src1-2*stride;      // == src+2*stride-1;

  708.   int H = src0[1] - src0[-1];

  709.   int V = src1[0] - src2[ 0];

  710.   for(k=2; k<=4; ++k) {

  711.     src1 += stride; src2 -= stride;

  712.     H += k*(src0[k] - src0[-k]);

  713.     V += k*(src1[0] - src2[ 0]);

  714.   }

  715.   H = ( 17*H+16 ) >> 5;

  716.   V = ( 17*V+16 ) >> 5;

  717. 

  718.   a = 16*(src1[0] + src2[8]+1) - 3*(V+H);

  719.   for(j=8; j>0; --j) {

  720.     int b = a;

  721.     a += V;

  722.     src[0] = cm[ (b    ) >> 5 ];

  723.     src[1] = cm[ (b+  H) >> 5 ];

  724.     src[2] = cm[ (b+2*H) >> 5 ];

  725.     src[3] = cm[ (b+3*H) >> 5 ];

  726.     src[4] = cm[ (b+4*H) >> 5 ];

  727.     src[5] = cm[ (b+5*H) >> 5 ];

  728.     src[6] = cm[ (b+6*H) >> 5 ];

  729.     src[7] = cm[ (b+7*H) >> 5 ];

  730.     src += stride;

  731.   }

  732. }

  733. 

  734. #define SRC(x,y) src[(x)+(y)*stride]

  735. #define PL(y) \

  736.     const int l##y = (SRC(-1,y-1) + 2*SRC(-1,y) + SRC(-1,y+1) + 2) >> 2;

  737. #define PREDICT_8x8_LOAD_LEFT \

  738.     const int l0 = ((has_topleft ? SRC(-1,-1) : SRC(-1,0)) \

  739.                      + 2*SRC(-1,0) + SRC(-1,1) + 2) >> 2; \

  740.     PL(1) PL(2) PL(3) PL(4) PL(5) PL(6) \

  741.     const int l7 av_unused = (SRC(-1,6) + 3*SRC(-1,7) + 2) >> 2

  742. 

  743. #define PT(x) \

  744.     const int t##x = (SRC(x-1,-1) + 2*SRC(x,-1) + SRC(x+1,-1) + 2) >> 2;

  745. #define PREDICT_8x8_LOAD_TOP \

  746.     const int t0 = ((has_topleft ? SRC(-1,-1) : SRC(0,-1)) \

  747.                      + 2*SRC(0,-1) + SRC(1,-1) + 2) >> 2; \

  748.     PT(1) PT(2) PT(3) PT(4) PT(5) PT(6) \

  749.     const int t7 av_unused = ((has_topright ? SRC(8,-1) : SRC(7,-1)) \

  750.                      + 2*SRC(7,-1) + SRC(6,-1) + 2) >> 2

  751. 

  752. #define PTR(x) \

  753.     t##x = (SRC(x-1,-1) + 2*SRC(x,-1) + SRC(x+1,-1) + 2) >> 2;

  754. #define PREDICT_8x8_LOAD_TOPRIGHT \

  755.     int t8, t9, t10, t11, t12, t13, t14, t15; \

  756.     if(has_topright) { \

  757.         PTR(8) PTR(9) PTR(10) PTR(11) PTR(12) PTR(13) PTR(14) \

  758.         t15 = (SRC(14,-1) + 3*SRC(15,-1) + 2) >> 2; \

  759.     } else t8=t9=t10=t11=t12=t13=t14=t15= SRC(7,-1);

  760. 

  761. #define PREDICT_8x8_LOAD_TOPLEFT \

  762.     const int lt = (SRC(-1,0) + 2*SRC(-1,-1) + SRC(0,-1) + 2) >> 2

  763. 

  764. #define PREDICT_8x8_DC(v) \

  765.     int y; \

  766.     for( y = 0; y < 8; y++ ) { \

  767.         ((uint32_t*)src)[0] = \

  768.         ((uint32_t*)src)[1] = v; \

  769.         src += stride; \

  770.     }

  771. 

  772. static void pred8x8l_128_dc_c(uint8_t *src, int has_topleft, int has_topright, int stride)

  773. {

  774.     PREDICT_8x8_DC(0x80808080);

  775. }

  776. static void pred8x8l_left_dc_c(uint8_t *src, int has_topleft, int has_topright, int stride)

  777. {

  778.     PREDICT_8x8_LOAD_LEFT;

  779.     const uint32_t dc = ((l0+l1+l2+l3+l4+l5+l6+l7+4) >> 3) * 0x01010101;

  780.     PREDICT_8x8_DC(dc);

  781. }

  782. static void pred8x8l_top_dc_c(uint8_t *src, int has_topleft, int has_topright, int stride)

  783. {

  784.     PREDICT_8x8_LOAD_TOP;

  785.     const uint32_t dc = ((t0+t1+t2+t3+t4+t5+t6+t7+4) >> 3) * 0x01010101;

  786.     PREDICT_8x8_DC(dc);

  787. }

  788. static void pred8x8l_dc_c(uint8_t *src, int has_topleft, int has_topright, int stride)

  789. {

  790.     PREDICT_8x8_LOAD_LEFT;

  791.     PREDICT_8x8_LOAD_TOP;

  792.     const uint32_t dc = ((l0+l1+l2+l3+l4+l5+l6+l7

  793.                          +t0+t1+t2+t3+t4+t5+t6+t7+8) >> 4) * 0x01010101;

  794.     PREDICT_8x8_DC(dc);

  795. }

  796. static void pred8x8l_horizontal_c(uint8_t *src, int has_topleft, int has_topright, int stride)

  797. {

  798.     PREDICT_8x8_LOAD_LEFT;

  799. #define ROW(y) ((uint32_t*)(src+y*stride))[0] =\

  800.                ((uint32_t*)(src+y*stride))[1] = 0x01010101 * l##y

  801.     ROW(0); ROW(1); ROW(2); ROW(3); ROW(4); ROW(5); ROW(6); ROW(7);

  802. #undef ROW

  803. }

  804. static void pred8x8l_vertical_c(uint8_t *src, int has_topleft, int has_topright, int stride)

  805. {

  806.     int y;

  807.     PREDICT_8x8_LOAD_TOP;

  808.     src[0] = t0;

  809.     src[1] = t1;

  810.     src[2] = t2;

  811.     src[3] = t3;

  812.     src[4] = t4;

  813.     src[5] = t5;

  814.     src[6] = t6;

  815.     src[7] = t7;

  816.     for( y = 1; y < 8; y++ )

  817.         *(uint64_t*)(src+y*stride) = *(uint64_t*)src;

  818. }

  819. static void pred8x8l_down_left_c(uint8_t *src, int has_topleft, int has_topright, int stride)

  820. {

  821.     PREDICT_8x8_LOAD_TOP;

  822.     PREDICT_8x8_LOAD_TOPRIGHT;

  823.     SRC(0,0)= (t0 + 2*t1 + t2 + 2) >> 2;

  824.     SRC(0,1)=SRC(1,0)= (t1 + 2*t2 + t3 + 2) >> 2;

  825.     SRC(0,2)=SRC(1,1)=SRC(2,0)= (t2 + 2*t3 + t4 + 2) >> 2;

  826.     SRC(0,3)=SRC(1,2)=SRC(2,1)=SRC(3,0)= (t3 + 2*t4 + t5 + 2) >> 2;

  827.     SRC(0,4)=SRC(1,3)=SRC(2,2)=SRC(3,1)=SRC(4,0)= (t4 + 2*t5 + t6 + 2) >> 2;

  828.     SRC(0,5)=SRC(1,4)=SRC(2,3)=SRC(3,2)=SRC(4,1)=SRC(5,0)= (t5 + 2*t6 + t7 + 2) >> 2;

  829.     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;

  830.     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;

  831.     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;

  832.     SRC(2,7)=SRC(3,6)=SRC(4,5)=SRC(5,4)=SRC(6,3)=SRC(7,2)= (t9 + 2*t10 + t11 + 2) >> 2;

  833.     SRC(3,7)=SRC(4,6)=SRC(5,5)=SRC(6,4)=SRC(7,3)= (t10 + 2*t11 + t12 + 2) >> 2;

  834.     SRC(4,7)=SRC(5,6)=SRC(6,5)=SRC(7,4)= (t11 + 2*t12 + t13 + 2) >> 2;

  835.     SRC(5,7)=SRC(6,6)=SRC(7,5)= (t12 + 2*t13 + t14 + 2) >> 2;

  836.     SRC(6,7)=SRC(7,6)= (t13 + 2*t14 + t15 + 2) >> 2;

  837.     SRC(7,7)= (t14 + 3*t15 + 2) >> 2;

  838. }

  839. static void pred8x8l_down_right_c(uint8_t *src, int has_topleft, int has_topright, int stride)

  840. {

  841.     PREDICT_8x8_LOAD_TOP;

  842.     PREDICT_8x8_LOAD_LEFT;

  843.     PREDICT_8x8_LOAD_TOPLEFT;

  844.     SRC(0,7)= (l7 + 2*l6 + l5 + 2) >> 2;

  845.     SRC(0,6)=SRC(1,7)= (l6 + 2*l5 + l4 + 2) >> 2;

  846.     SRC(0,5)=SRC(1,6)=SRC(2,7)= (l5 + 2*l4 + l3 + 2) >> 2;

  847.     SRC(0,4)=SRC(1,5)=SRC(2,6)=SRC(3,7)= (l4 + 2*l3 + l2 + 2) >> 2;

  848.     SRC(0,3)=SRC(1,4)=SRC(2,5)=SRC(3,6)=SRC(4,7)= (l3 + 2*l2 + l1 + 2) >> 2;

  849.     SRC(0,2)=SRC(1,3)=SRC(2,4)=SRC(3,5)=SRC(4,6)=SRC(5,7)= (l2 + 2*l1 + l0 + 2) >> 2;

  850.     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;

  851.     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;

  852.     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;

  853.     SRC(2,0)=SRC(3,1)=SRC(4,2)=SRC(5,3)=SRC(6,4)=SRC(7,5)= (t0 + 2*t1 + t2 + 2) >> 2;

  854.     SRC(3,0)=SRC(4,1)=SRC(5,2)=SRC(6,3)=SRC(7,4)= (t1 + 2*t2 + t3 + 2) >> 2;

  855.     SRC(4,0)=SRC(5,1)=SRC(6,2)=SRC(7,3)= (t2 + 2*t3 + t4 + 2) >> 2;

  856.     SRC(5,0)=SRC(6,1)=SRC(7,2)= (t3 + 2*t4 + t5 + 2) >> 2;

  857.     SRC(6,0)=SRC(7,1)= (t4 + 2*t5 + t6 + 2) >> 2;

  858.     SRC(7,0)= (t5 + 2*t6 + t7 + 2) >> 2;

  859. 

  860. }

  861. static void pred8x8l_vertical_right_c(uint8_t *src, int has_topleft, int has_topright, int stride)

  862. {

  863.     PREDICT_8x8_LOAD_TOP;

  864.     PREDICT_8x8_LOAD_LEFT;

  865.     PREDICT_8x8_LOAD_TOPLEFT;

  866.     SRC(0,6)= (l5 + 2*l4 + l3 + 2) >> 2;

  867.     SRC(0,7)= (l6 + 2*l5 + l4 + 2) >> 2;

  868.     SRC(0,4)=SRC(1,6)= (l3 + 2*l2 + l1 + 2) >> 2;

  869.     SRC(0,5)=SRC(1,7)= (l4 + 2*l3 + l2 + 2) >> 2;

  870.     SRC(0,2)=SRC(1,4)=SRC(2,6)= (l1 + 2*l0 + lt + 2) >> 2;

  871.     SRC(0,3)=SRC(1,5)=SRC(2,7)= (l2 + 2*l1 + l0 + 2) >> 2;

  872.     SRC(0,1)=SRC(1,3)=SRC(2,5)=SRC(3,7)= (l0 + 2*lt + t0 + 2) >> 2;

  873.     SRC(0,0)=SRC(1,2)=SRC(2,4)=SRC(3,6)= (lt + t0 + 1) >> 1;

  874.     SRC(1,1)=SRC(2,3)=SRC(3,5)=SRC(4,7)= (lt + 2*t0 + t1 + 2) >> 2;

  875.     SRC(1,0)=SRC(2,2)=SRC(3,4)=SRC(4,6)= (t0 + t1 + 1) >> 1;

  876.     SRC(2,1)=SRC(3,3)=SRC(4,5)=SRC(5,7)= (t0 + 2*t1 + t2 + 2) >> 2;

  877.     SRC(2,0)=SRC(3,2)=SRC(4,4)=SRC(5,6)= (t1 + t2 + 1) >> 1;

  878.     SRC(3,1)=SRC(4,3)=SRC(5,5)=SRC(6,7)= (t1 + 2*t2 + t3 + 2) >> 2;

  879.     SRC(3,0)=SRC(4,2)=SRC(5,4)=SRC(6,6)= (t2 + t3 + 1) >> 1;

  880.     SRC(4,1)=SRC(5,3)=SRC(6,5)=SRC(7,7)= (t2 + 2*t3 + t4 + 2) >> 2;

  881.     SRC(4,0)=SRC(5,2)=SRC(6,4)=SRC(7,6)= (t3 + t4 + 1) >> 1;

  882.     SRC(5,1)=SRC(6,3)=SRC(7,5)= (t3 + 2*t4 + t5 + 2) >> 2;

  883.     SRC(5,0)=SRC(6,2)=SRC(7,4)= (t4 + t5 + 1) >> 1;

  884.     SRC(6,1)=SRC(7,3)= (t4 + 2*t5 + t6 + 2) >> 2;

  885.     SRC(6,0)=SRC(7,2)= (t5 + t6 + 1) >> 1;

  886.     SRC(7,1)= (t5 + 2*t6 + t7 + 2) >> 2;

  887.     SRC(7,0)= (t6 + t7 + 1) >> 1;

  888. }

  889. static void pred8x8l_horizontal_down_c(uint8_t *src, int has_topleft, int has_topright, int stride)

  890. {

  891.     PREDICT_8x8_LOAD_TOP;

  892.     PREDICT_8x8_LOAD_LEFT;

  893.     PREDICT_8x8_LOAD_TOPLEFT;

  894.     SRC(0,7)= (l6 + l7 + 1) >> 1;

  895.     SRC(1,7)= (l5 + 2*l6 + l7 + 2) >> 2;

  896.     SRC(0,6)=SRC(2,7)= (l5 + l6 + 1) >> 1;

  897.     SRC(1,6)=SRC(3,7)= (l4 + 2*l5 + l6 + 2) >> 2;

  898.     SRC(0,5)=SRC(2,6)=SRC(4,7)= (l4 + l5 + 1) >> 1;

  899.     SRC(1,5)=SRC(3,6)=SRC(5,7)= (l3 + 2*l4 + l5 + 2) >> 2;

  900.     SRC(0,4)=SRC(2,5)=SRC(4,6)=SRC(6,7)= (l3 + l4 + 1) >> 1;

  901.     SRC(1,4)=SRC(3,5)=SRC(5,6)=SRC(7,7)= (l2 + 2*l3 + l4 + 2) >> 2;

  902.     SRC(0,3)=SRC(2,4)=SRC(4,5)=SRC(6,6)= (l2 + l3 + 1) >> 1;

  903.     SRC(1,3)=SRC(3,4)=SRC(5,5)=SRC(7,6)= (l1 + 2*l2 + l3 + 2) >> 2;

  904.     SRC(0,2)=SRC(2,3)=SRC(4,4)=SRC(6,5)= (l1 + l2 + 1) >> 1;

  905.     SRC(1,2)=SRC(3,3)=SRC(5,4)=SRC(7,5)= (l0 + 2*l1 + l2 + 2) >> 2;

  906.     SRC(0,1)=SRC(2,2)=SRC(4,3)=SRC(6,4)= (l0 + l1 + 1) >> 1;

  907.     SRC(1,1)=SRC(3,2)=SRC(5,3)=SRC(7,4)= (lt + 2*l0 + l1 + 2) >> 2;

  908.     SRC(0,0)=SRC(2,1)=SRC(4,2)=SRC(6,3)= (lt + l0 + 1) >> 1;

  909.     SRC(1,0)=SRC(3,1)=SRC(5,2)=SRC(7,3)= (l0 + 2*lt + t0 + 2) >> 2;

  910.     SRC(2,0)=SRC(4,1)=SRC(6,2)= (t1 + 2*t0 + lt + 2) >> 2;

  911.     SRC(3,0)=SRC(5,1)=SRC(7,2)= (t2 + 2*t1 + t0 + 2) >> 2;

  912.     SRC(4,0)=SRC(6,1)= (t3 + 2*t2 + t1 + 2) >> 2;

  913.     SRC(5,0)=SRC(7,1)= (t4 + 2*t3 + t2 + 2) >> 2;

  914.     SRC(6,0)= (t5 + 2*t4 + t3 + 2) >> 2;

  915.     SRC(7,0)= (t6 + 2*t5 + t4 + 2) >> 2;

  916. }

  917. static void pred8x8l_vertical_left_c(uint8_t *src, int has_topleft, int has_topright, int stride)

  918. {

  919.     PREDICT_8x8_LOAD_TOP;

  920.     PREDICT_8x8_LOAD_TOPRIGHT;

  921.     SRC(0,0)= (t0 + t1 + 1) >> 1;

  922.     SRC(0,1)= (t0 + 2*t1 + t2 + 2) >> 2;

  923.     SRC(0,2)=SRC(1,0)= (t1 + t2 + 1) >> 1;

  924.     SRC(0,3)=SRC(1,1)= (t1 + 2*t2 + t3 + 2) >> 2;

  925.     SRC(0,4)=SRC(1,2)=SRC(2,0)= (t2 + t3 + 1) >> 1;

  926.     SRC(0,5)=SRC(1,3)=SRC(2,1)= (t2 + 2*t3 + t4 + 2) >> 2;

  927.     SRC(0,6)=SRC(1,4)=SRC(2,2)=SRC(3,0)= (t3 + t4 + 1) >> 1;

  928.     SRC(0,7)=SRC(1,5)=SRC(2,3)=SRC(3,1)= (t3 + 2*t4 + t5 + 2) >> 2;

  929.     SRC(1,6)=SRC(2,4)=SRC(3,2)=SRC(4,0)= (t4 + t5 + 1) >> 1;

  930.     SRC(1,7)=SRC(2,5)=SRC(3,3)=SRC(4,1)= (t4 + 2*t5 + t6 + 2) >> 2;

  931.     SRC(2,6)=SRC(3,4)=SRC(4,2)=SRC(5,0)= (t5 + t6 + 1) >> 1;

  932.     SRC(2,7)=SRC(3,5)=SRC(4,3)=SRC(5,1)= (t5 + 2*t6 + t7 + 2) >> 2;

  933.     SRC(3,6)=SRC(4,4)=SRC(5,2)=SRC(6,0)= (t6 + t7 + 1) >> 1;

  934.     SRC(3,7)=SRC(4,5)=SRC(5,3)=SRC(6,1)= (t6 + 2*t7 + t8 + 2) >> 2;

  935.     SRC(4,6)=SRC(5,4)=SRC(6,2)=SRC(7,0)= (t7 + t8 + 1) >> 1;

  936.     SRC(4,7)=SRC(5,5)=SRC(6,3)=SRC(7,1)= (t7 + 2*t8 + t9 + 2) >> 2;

  937.     SRC(5,6)=SRC(6,4)=SRC(7,2)= (t8 + t9 + 1) >> 1;

  938.     SRC(5,7)=SRC(6,5)=SRC(7,3)= (t8 + 2*t9 + t10 + 2) >> 2;

  939.     SRC(6,6)=SRC(7,4)= (t9 + t10 + 1) >> 1;

  940.     SRC(6,7)=SRC(7,5)= (t9 + 2*t10 + t11 + 2) >> 2;

  941.     SRC(7,6)= (t10 + t11 + 1) >> 1;

  942.     SRC(7,7)= (t10 + 2*t11 + t12 + 2) >> 2;

  943. }

  944. static void pred8x8l_horizontal_up_c(uint8_t *src, int has_topleft, int has_topright, int stride)

  945. {

  946.     PREDICT_8x8_LOAD_LEFT;

  947.     SRC(0,0)= (l0 + l1 + 1) >> 1;

  948.     SRC(1,0)= (l0 + 2*l1 + l2 + 2) >> 2;

  949.     SRC(0,1)=SRC(2,0)= (l1 + l2 + 1) >> 1;

  950.     SRC(1,1)=SRC(3,0)= (l1 + 2*l2 + l3 + 2) >> 2;

  951.     SRC(0,2)=SRC(2,1)=SRC(4,0)= (l2 + l3 + 1) >> 1;

  952.     SRC(1,2)=SRC(3,1)=SRC(5,0)= (l2 + 2*l3 + l4 + 2) >> 2;

  953.     SRC(0,3)=SRC(2,2)=SRC(4,1)=SRC(6,0)= (l3 + l4 + 1) >> 1;

  954.     SRC(1,3)=SRC(3,2)=SRC(5,1)=SRC(7,0)= (l3 + 2*l4 + l5 + 2) >> 2;

  955.     SRC(0,4)=SRC(2,3)=SRC(4,2)=SRC(6,1)= (l4 + l5 + 1) >> 1;

  956.     SRC(1,4)=SRC(3,3)=SRC(5,2)=SRC(7,1)= (l4 + 2*l5 + l6 + 2) >> 2;

  957.     SRC(0,5)=SRC(2,4)=SRC(4,3)=SRC(6,2)= (l5 + l6 + 1) >> 1;

  958.     SRC(1,5)=SRC(3,4)=SRC(5,3)=SRC(7,2)= (l5 + 2*l6 + l7 + 2) >> 2;

  959.     SRC(0,6)=SRC(2,5)=SRC(4,4)=SRC(6,3)= (l6 + l7 + 1) >> 1;

  960.     SRC(1,6)=SRC(3,5)=SRC(5,4)=SRC(7,3)= (l6 + 3*l7 + 2) >> 2;

  961.     SRC(0,7)=SRC(1,7)=SRC(2,6)=SRC(2,7)=SRC(3,6)=

  962.     SRC(3,7)=SRC(4,5)=SRC(4,6)=SRC(4,7)=SRC(5,5)=

  963.     SRC(5,6)=SRC(5,7)=SRC(6,4)=SRC(6,5)=SRC(6,6)=

  964.     SRC(6,7)=SRC(7,4)=SRC(7,5)=SRC(7,6)=SRC(7,7)= l7;

  965. }

  966. #undef PREDICT_8x8_LOAD_LEFT

  967. #undef PREDICT_8x8_LOAD_TOP

  968. #undef PREDICT_8x8_LOAD_TOPLEFT

  969. #undef PREDICT_8x8_LOAD_TOPRIGHT

  970. #undef PREDICT_8x8_DC

  971. #undef PTR

  972. #undef PT

  973. #undef PL

  974. #undef SRC

  975. 

  976. /**

  977.  * Sets the intra prediction function pointers.

  978.  */

  979. void ff_h264_pred_init(H264PredContext *h, int codec_id){

  980. //    MpegEncContext * const s = &h->s;

  981. 

  982.     if(codec_id != CODEC_ID_RV40){

  983.         h->pred4x4[VERT_PRED           ]= pred4x4_vertical_c;

  984.         h->pred4x4[HOR_PRED            ]= pred4x4_horizontal_c;

  985.         h->pred4x4[DC_PRED             ]= pred4x4_dc_c;

  986.         if(codec_id == CODEC_ID_SVQ3)

  987.             h->pred4x4[DIAG_DOWN_LEFT_PRED ]= pred4x4_down_left_svq3_c;

  988.         else

  989.             h->pred4x4[DIAG_DOWN_LEFT_PRED ]= pred4x4_down_left_c;

  990.         h->pred4x4[DIAG_DOWN_RIGHT_PRED]= pred4x4_down_right_c;

  991.         h->pred4x4[VERT_RIGHT_PRED     ]= pred4x4_vertical_right_c;

  992.         h->pred4x4[HOR_DOWN_PRED       ]= pred4x4_horizontal_down_c;

  993.         h->pred4x4[VERT_LEFT_PRED      ]= pred4x4_vertical_left_c;

  994.         h->pred4x4[HOR_UP_PRED         ]= pred4x4_horizontal_up_c;

  995.         h->pred4x4[LEFT_DC_PRED        ]= pred4x4_left_dc_c;

  996.         h->pred4x4[TOP_DC_PRED         ]= pred4x4_top_dc_c;

  997.         h->pred4x4[DC_128_PRED         ]= pred4x4_128_dc_c;

  998.     }else{

  999.         h->pred4x4[VERT_PRED           ]= pred4x4_vertical_c;

  1000.         h->pred4x4[HOR_PRED            ]= pred4x4_horizontal_c;

  1001.         h->pred4x4[DC_PRED             ]= pred4x4_dc_c;

  1002.         h->pred4x4[DIAG_DOWN_LEFT_PRED ]= pred4x4_down_left_rv40_c;

  1003.         h->pred4x4[DIAG_DOWN_RIGHT_PRED]= pred4x4_down_right_c;

  1004.         h->pred4x4[VERT_RIGHT_PRED     ]= pred4x4_vertical_right_c;

  1005.         h->pred4x4[HOR_DOWN_PRED       ]= pred4x4_horizontal_down_c;

  1006.         h->pred4x4[VERT_LEFT_PRED      ]= pred4x4_vertical_left_rv40_c;

  1007.         h->pred4x4[HOR_UP_PRED         ]= pred4x4_horizontal_up_rv40_c;

  1008.         h->pred4x4[LEFT_DC_PRED        ]= pred4x4_left_dc_c;

  1009.         h->pred4x4[TOP_DC_PRED         ]= pred4x4_top_dc_c;

  1010.         h->pred4x4[DC_128_PRED         ]= pred4x4_128_dc_c;

  1011.         h->pred4x4[DIAG_DOWN_LEFT_PRED_RV40_NODOWN]= pred4x4_down_left_rv40_nodown_c;

  1012.         h->pred4x4[HOR_UP_PRED_RV40_NODOWN]= pred4x4_horizontal_up_rv40_nodown_c;

  1013.     }

  1014. 

  1015.     h->pred8x8l[VERT_PRED           ]= pred8x8l_vertical_c;

  1016.     h->pred8x8l[HOR_PRED            ]= pred8x8l_horizontal_c;

  1017.     h->pred8x8l[DC_PRED             ]= pred8x8l_dc_c;

  1018.     h->pred8x8l[DIAG_DOWN_LEFT_PRED ]= pred8x8l_down_left_c;

  1019.     h->pred8x8l[DIAG_DOWN_RIGHT_PRED]= pred8x8l_down_right_c;

  1020.     h->pred8x8l[VERT_RIGHT_PRED     ]= pred8x8l_vertical_right_c;

  1021.     h->pred8x8l[HOR_DOWN_PRED       ]= pred8x8l_horizontal_down_c;

  1022.     h->pred8x8l[VERT_LEFT_PRED      ]= pred8x8l_vertical_left_c;

  1023.     h->pred8x8l[HOR_UP_PRED         ]= pred8x8l_horizontal_up_c;

  1024.     h->pred8x8l[LEFT_DC_PRED        ]= pred8x8l_left_dc_c;

  1025.     h->pred8x8l[TOP_DC_PRED         ]= pred8x8l_top_dc_c;

  1026.     h->pred8x8l[DC_128_PRED         ]= pred8x8l_128_dc_c;

  1027. 

  1028.     h->pred8x8[VERT_PRED8x8   ]= pred8x8_vertical_c;

  1029.     h->pred8x8[HOR_PRED8x8    ]= pred8x8_horizontal_c;

  1030.     h->pred8x8[PLANE_PRED8x8  ]= pred8x8_plane_c;

  1031.     if(codec_id != CODEC_ID_RV40){

  1032.         h->pred8x8[DC_PRED8x8     ]= pred8x8_dc_c;

  1033.         h->pred8x8[LEFT_DC_PRED8x8]= pred8x8_left_dc_c;

  1034.         h->pred8x8[TOP_DC_PRED8x8 ]= pred8x8_top_dc_c;

  1035.     }else{

  1036.         h->pred8x8[DC_PRED8x8     ]= pred8x8_dc_rv40_c;

  1037.         h->pred8x8[LEFT_DC_PRED8x8]= pred8x8_left_dc_rv40_c;

  1038.         h->pred8x8[TOP_DC_PRED8x8 ]= pred8x8_top_dc_rv40_c;

  1039.     }

  1040.     h->pred8x8[DC_128_PRED8x8 ]= pred8x8_128_dc_c;

  1041. 

  1042.     h->pred16x16[DC_PRED8x8     ]= pred16x16_dc_c;

  1043.     h->pred16x16[VERT_PRED8x8   ]= pred16x16_vertical_c;

  1044.     h->pred16x16[HOR_PRED8x8    ]= pred16x16_horizontal_c;

  1045.     h->pred16x16[PLANE_PRED8x8  ]= pred16x16_plane_c;

  1046.     switch(codec_id){

  1047.     case CODEC_ID_SVQ3:

  1048.        h->pred16x16[PLANE_PRED8x8  ]= pred16x16_plane_svq3_c;

  1049.        break;

  1050.     case CODEC_ID_RV40:

  1051.        h->pred16x16[PLANE_PRED8x8  ]= pred16x16_plane_rv40_c;

  1052.        break;

  1053.     default:

  1054.        h->pred16x16[PLANE_PRED8x8  ]= pred16x16_plane_c;

  1055.     }

  1056.     h->pred16x16[LEFT_DC_PRED8x8]= pred16x16_left_dc_c;

  1057.     h->pred16x16[TOP_DC_PRED8x8 ]= pred16x16_top_dc_c;

  1058.     h->pred16x16[DC_128_PRED8x8 ]= pred16x16_128_dc_c;

  1059. }