• Main Page
  • Related Pages
  • Modules
  • Data Structures
  • Files
  • Examples
  • File List
  • Globals

libavcodec/h264idct_template.c

Go to the documentation of this file.
00001 /*
00002  * H.264 IDCT
00003  * Copyright (c) 2004-2011 Michael Niedermayer <michaelni@gmx.at>
00004  *
00005  * This file is part of Libav.
00006  *
00007  * Libav is free software; you can redistribute it and/or
00008  * modify it under the terms of the GNU Lesser General Public
00009  * License as published by the Free Software Foundation; either
00010  * version 2.1 of the License, or (at your option) any later version.
00011  *
00012  * Libav is distributed in the hope that it will be useful,
00013  * but WITHOUT ANY WARRANTY; without even the implied warranty of
00014  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
00015  * Lesser General Public License for more details.
00016  *
00017  * You should have received a copy of the GNU Lesser General Public
00018  * License along with Libav; if not, write to the Free Software
00019  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
00020  */
00021 
00028 #include "bit_depth_template.c"
00029 
00030 #ifndef AVCODEC_H264IDCT_INTERNAL_H
00031 #define AVCODEC_H264IDCT_INTERNAL_H
00032 //FIXME this table is a duplicate from h264data.h, and will be removed once the tables from, h264 have been split
00033 static const uint8_t scan8[16*3]={
00034  4+ 1*8, 5+ 1*8, 4+ 2*8, 5+ 2*8,
00035  6+ 1*8, 7+ 1*8, 6+ 2*8, 7+ 2*8,
00036  4+ 3*8, 5+ 3*8, 4+ 4*8, 5+ 4*8,
00037  6+ 3*8, 7+ 3*8, 6+ 4*8, 7+ 4*8,
00038  4+ 6*8, 5+ 6*8, 4+ 7*8, 5+ 7*8,
00039  6+ 6*8, 7+ 6*8, 6+ 7*8, 7+ 7*8,
00040  4+ 8*8, 5+ 8*8, 4+ 9*8, 5+ 9*8,
00041  6+ 8*8, 7+ 8*8, 6+ 9*8, 7+ 9*8,
00042  4+11*8, 5+11*8, 4+12*8, 5+12*8,
00043  6+11*8, 7+11*8, 6+12*8, 7+12*8,
00044  4+13*8, 5+13*8, 4+14*8, 5+14*8,
00045  6+13*8, 7+13*8, 6+14*8, 7+14*8
00046 };
00047 #endif
00048 
00049 void FUNCC(ff_h264_idct_add)(uint8_t *_dst, DCTELEM *_block, int stride)
00050 {
00051     int i;
00052     pixel *dst = (pixel*)_dst;
00053     dctcoef *block = (dctcoef*)_block;
00054     stride /= sizeof(pixel);
00055 
00056     block[0] += 1 << 5;
00057 
00058     for(i=0; i<4; i++){
00059         const int z0=  block[i + 4*0]     +  block[i + 4*2];
00060         const int z1=  block[i + 4*0]     -  block[i + 4*2];
00061         const int z2= (block[i + 4*1]>>1) -  block[i + 4*3];
00062         const int z3=  block[i + 4*1]     + (block[i + 4*3]>>1);
00063 
00064         block[i + 4*0]= z0 + z3;
00065         block[i + 4*1]= z1 + z2;
00066         block[i + 4*2]= z1 - z2;
00067         block[i + 4*3]= z0 - z3;
00068     }
00069 
00070     for(i=0; i<4; i++){
00071         const int z0=  block[0 + 4*i]     +  block[2 + 4*i];
00072         const int z1=  block[0 + 4*i]     -  block[2 + 4*i];
00073         const int z2= (block[1 + 4*i]>>1) -  block[3 + 4*i];
00074         const int z3=  block[1 + 4*i]     + (block[3 + 4*i]>>1);
00075 
00076         dst[i + 0*stride]= av_clip_pixel(dst[i + 0*stride] + ((z0 + z3) >> 6));
00077         dst[i + 1*stride]= av_clip_pixel(dst[i + 1*stride] + ((z1 + z2) >> 6));
00078         dst[i + 2*stride]= av_clip_pixel(dst[i + 2*stride] + ((z1 - z2) >> 6));
00079         dst[i + 3*stride]= av_clip_pixel(dst[i + 3*stride] + ((z0 - z3) >> 6));
00080     }
00081 }
00082 
00083 void FUNCC(ff_h264_idct8_add)(uint8_t *_dst, DCTELEM *_block, int stride){
00084     int i;
00085     pixel *dst = (pixel*)_dst;
00086     dctcoef *block = (dctcoef*)_block;
00087     stride /= sizeof(pixel);
00088 
00089     block[0] += 32;
00090 
00091     for( i = 0; i < 8; i++ )
00092     {
00093         const int a0 =  block[i+0*8] + block[i+4*8];
00094         const int a2 =  block[i+0*8] - block[i+4*8];
00095         const int a4 = (block[i+2*8]>>1) - block[i+6*8];
00096         const int a6 = (block[i+6*8]>>1) + block[i+2*8];
00097 
00098         const int b0 = a0 + a6;
00099         const int b2 = a2 + a4;
00100         const int b4 = a2 - a4;
00101         const int b6 = a0 - a6;
00102 
00103         const int a1 = -block[i+3*8] + block[i+5*8] - block[i+7*8] - (block[i+7*8]>>1);
00104         const int a3 =  block[i+1*8] + block[i+7*8] - block[i+3*8] - (block[i+3*8]>>1);
00105         const int a5 = -block[i+1*8] + block[i+7*8] + block[i+5*8] + (block[i+5*8]>>1);
00106         const int a7 =  block[i+3*8] + block[i+5*8] + block[i+1*8] + (block[i+1*8]>>1);
00107 
00108         const int b1 = (a7>>2) + a1;
00109         const int b3 =  a3 + (a5>>2);
00110         const int b5 = (a3>>2) - a5;
00111         const int b7 =  a7 - (a1>>2);
00112 
00113         block[i+0*8] = b0 + b7;
00114         block[i+7*8] = b0 - b7;
00115         block[i+1*8] = b2 + b5;
00116         block[i+6*8] = b2 - b5;
00117         block[i+2*8] = b4 + b3;
00118         block[i+5*8] = b4 - b3;
00119         block[i+3*8] = b6 + b1;
00120         block[i+4*8] = b6 - b1;
00121     }
00122     for( i = 0; i < 8; i++ )
00123     {
00124         const int a0 =  block[0+i*8] + block[4+i*8];
00125         const int a2 =  block[0+i*8] - block[4+i*8];
00126         const int a4 = (block[2+i*8]>>1) - block[6+i*8];
00127         const int a6 = (block[6+i*8]>>1) + block[2+i*8];
00128 
00129         const int b0 = a0 + a6;
00130         const int b2 = a2 + a4;
00131         const int b4 = a2 - a4;
00132         const int b6 = a0 - a6;
00133 
00134         const int a1 = -block[3+i*8] + block[5+i*8] - block[7+i*8] - (block[7+i*8]>>1);
00135         const int a3 =  block[1+i*8] + block[7+i*8] - block[3+i*8] - (block[3+i*8]>>1);
00136         const int a5 = -block[1+i*8] + block[7+i*8] + block[5+i*8] + (block[5+i*8]>>1);
00137         const int a7 =  block[3+i*8] + block[5+i*8] + block[1+i*8] + (block[1+i*8]>>1);
00138 
00139         const int b1 = (a7>>2) + a1;
00140         const int b3 =  a3 + (a5>>2);
00141         const int b5 = (a3>>2) - a5;
00142         const int b7 =  a7 - (a1>>2);
00143 
00144         dst[i + 0*stride] = av_clip_pixel( dst[i + 0*stride] + ((b0 + b7) >> 6) );
00145         dst[i + 1*stride] = av_clip_pixel( dst[i + 1*stride] + ((b2 + b5) >> 6) );
00146         dst[i + 2*stride] = av_clip_pixel( dst[i + 2*stride] + ((b4 + b3) >> 6) );
00147         dst[i + 3*stride] = av_clip_pixel( dst[i + 3*stride] + ((b6 + b1) >> 6) );
00148         dst[i + 4*stride] = av_clip_pixel( dst[i + 4*stride] + ((b6 - b1) >> 6) );
00149         dst[i + 5*stride] = av_clip_pixel( dst[i + 5*stride] + ((b4 - b3) >> 6) );
00150         dst[i + 6*stride] = av_clip_pixel( dst[i + 6*stride] + ((b2 - b5) >> 6) );
00151         dst[i + 7*stride] = av_clip_pixel( dst[i + 7*stride] + ((b0 - b7) >> 6) );
00152     }
00153 }
00154 
00155 // assumes all AC coefs are 0
00156 void FUNCC(ff_h264_idct_dc_add)(uint8_t *_dst, DCTELEM *block, int stride){
00157     int i, j;
00158     int dc = (((dctcoef*)block)[0] + 32) >> 6;
00159     pixel *dst = (pixel*)_dst;
00160     stride /= sizeof(pixel);
00161     for( j = 0; j < 4; j++ )
00162     {
00163         for( i = 0; i < 4; i++ )
00164             dst[i] = av_clip_pixel( dst[i] + dc );
00165         dst += stride;
00166     }
00167 }
00168 
00169 void FUNCC(ff_h264_idct8_dc_add)(uint8_t *_dst, DCTELEM *block, int stride){
00170     int i, j;
00171     int dc = (((dctcoef*)block)[0] + 32) >> 6;
00172     pixel *dst = (pixel*)_dst;
00173     stride /= sizeof(pixel);
00174     for( j = 0; j < 8; j++ )
00175     {
00176         for( i = 0; i < 8; i++ )
00177             dst[i] = av_clip_pixel( dst[i] + dc );
00178         dst += stride;
00179     }
00180 }
00181 
00182 void FUNCC(ff_h264_idct_add16)(uint8_t *dst, const int *block_offset, DCTELEM *block, int stride, const uint8_t nnzc[15*8]){
00183     int i;
00184     for(i=0; i<16; i++){
00185         int nnz = nnzc[ scan8[i] ];
00186         if(nnz){
00187             if(nnz==1 && ((dctcoef*)block)[i*16]) FUNCC(ff_h264_idct_dc_add)(dst + block_offset[i], block + i*16*sizeof(pixel), stride);
00188             else                                  FUNCC(ff_h264_idct_add   )(dst + block_offset[i], block + i*16*sizeof(pixel), stride);
00189         }
00190     }
00191 }
00192 
00193 void FUNCC(ff_h264_idct_add16intra)(uint8_t *dst, const int *block_offset, DCTELEM *block, int stride, const uint8_t nnzc[15*8]){
00194     int i;
00195     for(i=0; i<16; i++){
00196         if(nnzc[ scan8[i] ])             FUNCC(ff_h264_idct_add   )(dst + block_offset[i], block + i*16*sizeof(pixel), stride);
00197         else if(((dctcoef*)block)[i*16]) FUNCC(ff_h264_idct_dc_add)(dst + block_offset[i], block + i*16*sizeof(pixel), stride);
00198     }
00199 }
00200 
00201 void FUNCC(ff_h264_idct8_add4)(uint8_t *dst, const int *block_offset, DCTELEM *block, int stride, const uint8_t nnzc[15*8]){
00202     int i;
00203     for(i=0; i<16; i+=4){
00204         int nnz = nnzc[ scan8[i] ];
00205         if(nnz){
00206             if(nnz==1 && ((dctcoef*)block)[i*16]) FUNCC(ff_h264_idct8_dc_add)(dst + block_offset[i], block + i*16*sizeof(pixel), stride);
00207             else                                  FUNCC(ff_h264_idct8_add   )(dst + block_offset[i], block + i*16*sizeof(pixel), stride);
00208         }
00209     }
00210 }
00211 
00212 void FUNCC(ff_h264_idct_add8)(uint8_t **dest, const int *block_offset, DCTELEM *block, int stride, const uint8_t nnzc[15*8]){
00213     int i, j;
00214     for(j=1; j<3; j++){
00215         for(i=j*16; i<j*16+4; i++){
00216             if(nnzc[ scan8[i] ])
00217                 FUNCC(ff_h264_idct_add   )(dest[j-1] + block_offset[i], block + i*16*sizeof(pixel), stride);
00218             else if(((dctcoef*)block)[i*16])
00219                 FUNCC(ff_h264_idct_dc_add)(dest[j-1] + block_offset[i], block + i*16*sizeof(pixel), stride);
00220         }
00221     }
00222 }
00223 
00224 void FUNCC(ff_h264_idct_add8_422)(uint8_t **dest, const int *block_offset, DCTELEM *block, int stride, const uint8_t nnzc[15*8]){
00225     int i, j;
00226 
00227     for(j=1; j<3; j++){
00228         for(i=j*16; i<j*16+4; i++){
00229             if(nnzc[ scan8[i] ])
00230                 FUNCC(ff_h264_idct_add   )(dest[j-1] + block_offset[i], block + i*16*sizeof(pixel), stride);
00231             else if(((dctcoef*)block)[i*16])
00232                 FUNCC(ff_h264_idct_dc_add)(dest[j-1] + block_offset[i], block + i*16*sizeof(pixel), stride);
00233         }
00234     }
00235 
00236     for(j=1; j<3; j++){
00237         for(i=j*16+4; i<j*16+8; i++){
00238             if(nnzc[ scan8[i+4] ])
00239                 FUNCC(ff_h264_idct_add   )(dest[j-1] + block_offset[i+4], block + i*16*sizeof(pixel), stride);
00240             else if(((dctcoef*)block)[i*16])
00241                 FUNCC(ff_h264_idct_dc_add)(dest[j-1] + block_offset[i+4], block + i*16*sizeof(pixel), stride);
00242         }
00243     }
00244 }
00245 
00250 void FUNCC(ff_h264_luma_dc_dequant_idct)(DCTELEM *_output, DCTELEM *_input, int qmul){
00251 #define stride 16
00252     int i;
00253     int temp[16];
00254     static const uint8_t x_offset[4]={0, 2*stride, 8*stride, 10*stride};
00255     dctcoef *input = (dctcoef*)_input;
00256     dctcoef *output = (dctcoef*)_output;
00257 
00258     for(i=0; i<4; i++){
00259         const int z0= input[4*i+0] + input[4*i+1];
00260         const int z1= input[4*i+0] - input[4*i+1];
00261         const int z2= input[4*i+2] - input[4*i+3];
00262         const int z3= input[4*i+2] + input[4*i+3];
00263 
00264         temp[4*i+0]= z0+z3;
00265         temp[4*i+1]= z0-z3;
00266         temp[4*i+2]= z1-z2;
00267         temp[4*i+3]= z1+z2;
00268     }
00269 
00270     for(i=0; i<4; i++){
00271         const int offset= x_offset[i];
00272         const int z0= temp[4*0+i] + temp[4*2+i];
00273         const int z1= temp[4*0+i] - temp[4*2+i];
00274         const int z2= temp[4*1+i] - temp[4*3+i];
00275         const int z3= temp[4*1+i] + temp[4*3+i];
00276 
00277         output[stride* 0+offset]= ((((z0 + z3)*qmul + 128 ) >> 8));
00278         output[stride* 1+offset]= ((((z1 + z2)*qmul + 128 ) >> 8));
00279         output[stride* 4+offset]= ((((z1 - z2)*qmul + 128 ) >> 8));
00280         output[stride* 5+offset]= ((((z0 - z3)*qmul + 128 ) >> 8));
00281     }
00282 #undef stride
00283 }
00284 
00285 void FUNCC(ff_h264_chroma422_dc_dequant_idct)(DCTELEM *_block, int qmul){
00286     const int stride= 16*2;
00287     const int xStride= 16;
00288     int i;
00289     int temp[8];
00290     static const uint8_t x_offset[2]={0, 16};
00291     dctcoef *block = (dctcoef*)_block;
00292 
00293     for(i=0; i<4; i++){
00294         temp[2*i+0] = block[stride*i + xStride*0] + block[stride*i + xStride*1];
00295         temp[2*i+1] = block[stride*i + xStride*0] - block[stride*i + xStride*1];
00296     }
00297 
00298     for(i=0; i<2; i++){
00299         const int offset= x_offset[i];
00300         const int z0= temp[2*0+i] + temp[2*2+i];
00301         const int z1= temp[2*0+i] - temp[2*2+i];
00302         const int z2= temp[2*1+i] - temp[2*3+i];
00303         const int z3= temp[2*1+i] + temp[2*3+i];
00304 
00305         block[stride*0+offset]= ((z0 + z3)*qmul + 128) >> 8;
00306         block[stride*1+offset]= ((z1 + z2)*qmul + 128) >> 8;
00307         block[stride*2+offset]= ((z1 - z2)*qmul + 128) >> 8;
00308         block[stride*3+offset]= ((z0 - z3)*qmul + 128) >> 8;
00309     }
00310 }
00311 
00312 void FUNCC(ff_h264_chroma_dc_dequant_idct)(DCTELEM *_block, int qmul){
00313     const int stride= 16*2;
00314     const int xStride= 16;
00315     int a,b,c,d,e;
00316     dctcoef *block = (dctcoef*)_block;
00317 
00318     a= block[stride*0 + xStride*0];
00319     b= block[stride*0 + xStride*1];
00320     c= block[stride*1 + xStride*0];
00321     d= block[stride*1 + xStride*1];
00322 
00323     e= a-b;
00324     a= a+b;
00325     b= c-d;
00326     c= c+d;
00327 
00328     block[stride*0 + xStride*0]= ((a+c)*qmul) >> 7;
00329     block[stride*0 + xStride*1]= ((e+b)*qmul) >> 7;
00330     block[stride*1 + xStride*0]= ((a-c)*qmul) >> 7;
00331     block[stride*1 + xStride*1]= ((e-b)*qmul) >> 7;
00332 }
Generated on Thu Jul 11 2013 15:38:19 for Libav by doxygen 1.7.1