libavcodec/h264idct.c

   1 /*
   2  * H.264 IDCT
   3  * Copyright (c) 2004 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 h264-idct.c
  24  * H.264 IDCT.
  25  * @author Michael Niedermayer <michaelni@gmx.at>
  26  */
  27
  28 #include "dsputil.h"
  29
  30 static av_always_inline void idct_internal(uint8_t *dst, DCTELEM *block, int stride, int block_stride, int shift, int add){
  31     int i;
  32     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
  33
  34     block[0] += 1<<(shift-1);
  35
  36     for(i=0; i<4; i++){
  37         const int z0=  block[0 + block_stride*i]     +  block[2 + block_stride*i];
  38         const int z1=  block[0 + block_stride*i]     -  block[2 + block_stride*i];
  39         const int z2= (block[1 + block_stride*i]>>1) -  block[3 + block_stride*i];
  40         const int z3=  block[1 + block_stride*i]     + (block[3 + block_stride*i]>>1);
  41
  42         block[0 + block_stride*i]= z0 + z3;
  43         block[1 + block_stride*i]= z1 + z2;
  44         block[2 + block_stride*i]= z1 - z2;
  45         block[3 + block_stride*i]= z0 - z3;
  46     }
  47
  48     for(i=0; i<4; i++){
  49         const int z0=  block[i + block_stride*0]     +  block[i + block_stride*2];
  50         const int z1=  block[i + block_stride*0]     -  block[i + block_stride*2];
  51         const int z2= (block[i + block_stride*1]>>1) -  block[i + block_stride*3];
  52         const int z3=  block[i + block_stride*1]     + (block[i + block_stride*3]>>1);
  53
  54         dst[i + 0*stride]= cm[ add*dst[i + 0*stride] + ((z0 + z3) >> shift) ];
  55         dst[i + 1*stride]= cm[ add*dst[i + 1*stride] + ((z1 + z2) >> shift) ];
  56         dst[i + 2*stride]= cm[ add*dst[i + 2*stride] + ((z1 - z2) >> shift) ];
  57         dst[i + 3*stride]= cm[ add*dst[i + 3*stride] + ((z0 - z3) >> shift) ];
  58     }
  59 }
  60
  61 void ff_h264_idct_add_c(uint8_t *dst, DCTELEM *block, int stride){
  62     idct_internal(dst, block, stride, 4, 6, 1);
  63 }
  64
  65 void ff_h264_lowres_idct_add_c(uint8_t *dst, int stride, DCTELEM *block){
  66     idct_internal(dst, block, stride, 8, 3, 1);
  67 }
  68
  69 void ff_h264_lowres_idct_put_c(uint8_t *dst, int stride, DCTELEM *block){
  70     idct_internal(dst, block, stride, 8, 3, 0);
  71 }
  72
  73 void ff_h264_idct8_add_c(uint8_t *dst, DCTELEM *block, int stride){
  74     int i;
  75     DCTELEM (*src)[8] = (DCTELEM(*)[8])block;
  76     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
  77
  78     block[0] += 32;
  79
  80     for( i = 0; i < 8; i++ )
  81     {
  82         const int a0 =  src[i][0] + src[i][4];
  83         const int a2 =  src[i][0] - src[i][4];
  84         const int a4 = (src[i][2]>>1) - src[i][6];
  85         const int a6 = (src[i][6]>>1) + src[i][2];
  86
  87         const int b0 = a0 + a6;
  88         const int b2 = a2 + a4;
  89         const int b4 = a2 - a4;
  90         const int b6 = a0 - a6;
  91
  92         const int a1 = -src[i][3] + src[i][5] - src[i][7] - (src[i][7]>>1);
  93         const int a3 =  src[i][1] + src[i][7] - src[i][3] - (src[i][3]>>1);
  94         const int a5 = -src[i][1] + src[i][7] + src[i][5] + (src[i][5]>>1);
  95         const int a7 =  src[i][3] + src[i][5] + src[i][1] + (src[i][1]>>1);
  96
  97         const int b1 = (a7>>2) + a1;
  98         const int b3 =  a3 + (a5>>2);
  99         const int b5 = (a3>>2) - a5;
 100         const int b7 =  a7 - (a1>>2);
 101
 102         src[i][0] = b0 + b7;
 103         src[i][7] = b0 - b7;
 104         src[i][1] = b2 + b5;
 105         src[i][6] = b2 - b5;
 106         src[i][2] = b4 + b3;
 107         src[i][5] = b4 - b3;
 108         src[i][3] = b6 + b1;
 109         src[i][4] = b6 - b1;
 110     }
 111     for( i = 0; i < 8; i++ )
 112     {
 113         const int a0 =  src[0][i] + src[4][i];
 114         const int a2 =  src[0][i] - src[4][i];
 115         const int a4 = (src[2][i]>>1) - src[6][i];
 116         const int a6 = (src[6][i]>>1) + src[2][i];
 117
 118         const int b0 = a0 + a6;
 119         const int b2 = a2 + a4;
 120         const int b4 = a2 - a4;
 121         const int b6 = a0 - a6;
 122
 123         const int a1 = -src[3][i] + src[5][i] - src[7][i] - (src[7][i]>>1);
 124         const int a3 =  src[1][i] + src[7][i] - src[3][i] - (src[3][i]>>1);
 125         const int a5 = -src[1][i] + src[7][i] + src[5][i] + (src[5][i]>>1);
 126         const int a7 =  src[3][i] + src[5][i] + src[1][i] + (src[1][i]>>1);
 127
 128         const int b1 = (a7>>2) + a1;
 129         const int b3 =  a3 + (a5>>2);
 130         const int b5 = (a3>>2) - a5;
 131         const int b7 =  a7 - (a1>>2);
 132
 133         dst[i + 0*stride] = cm[ dst[i + 0*stride] + ((b0 + b7) >> 6) ];
 134         dst[i + 1*stride] = cm[ dst[i + 1*stride] + ((b2 + b5) >> 6) ];
 135         dst[i + 2*stride] = cm[ dst[i + 2*stride] + ((b4 + b3) >> 6) ];
 136         dst[i + 3*stride] = cm[ dst[i + 3*stride] + ((b6 + b1) >> 6) ];
 137         dst[i + 4*stride] = cm[ dst[i + 4*stride] + ((b6 - b1) >> 6) ];
 138         dst[i + 5*stride] = cm[ dst[i + 5*stride] + ((b4 - b3) >> 6) ];
 139         dst[i + 6*stride] = cm[ dst[i + 6*stride] + ((b2 - b5) >> 6) ];
 140         dst[i + 7*stride] = cm[ dst[i + 7*stride] + ((b0 - b7) >> 6) ];
 141     }
 142 }
 143
 144 // assumes all AC coefs are 0
 145 void ff_h264_idct_dc_add_c(uint8_t *dst, DCTELEM *block, int stride){
 146     int i, j;
 147     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
 148     int dc = (block[0] + 32) >> 6;
 149     for( j = 0; j < 4; j++ )
 150     {
 151         for( i = 0; i < 4; i++ )
 152             dst[i] = cm[ dst[i] + dc ];
 153         dst += stride;
 154     }
 155 }
 156
 157 void ff_h264_idct8_dc_add_c(uint8_t *dst, DCTELEM *block, int stride){
 158     int i, j;
 159     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
 160     int dc = (block[0] + 32) >> 6;
 161     for( j = 0; j < 8; j++ )
 162     {
 163         for( i = 0; i < 8; i++ )
 164             dst[i] = cm[ dst[i] + dc ];
 165         dst += stride;
 166     }
 167 }