libavcodec/faanidct.c

   1 /*
   2  * Floating point AAN IDCT
   3  * Copyright (c) 2008 Michael Niedermayer <michaelni@gmx.at>
   4  *
   5  * This file is part of Libav.
   6  *
   7  * Libav 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  * Libav 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 Libav; if not, write to the Free Software
  19  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  20  */
  21 #include "faanidct.h"
  22 #include "libavutil/common.h"
  23
  24 /* To allow switching to double. */
  25 #define FLOAT float
  26
  27 #define B0 1.0000000000000000000000
  28 #define B1 1.3870398453221474618216 // cos(pi*1/16)sqrt(2)
  29 #define B2 1.3065629648763765278566 // cos(pi*2/16)sqrt(2)
  30 #define B3 1.1758756024193587169745 // cos(pi*3/16)sqrt(2)
  31 #define B4 1.0000000000000000000000 // cos(pi*4/16)sqrt(2)
  32 #define B5 0.7856949583871021812779 // cos(pi*5/16)sqrt(2)
  33 #define B6 0.5411961001461969843997 // cos(pi*6/16)sqrt(2)
  34 #define B7 0.2758993792829430123360 // cos(pi*7/16)sqrt(2)
  35
  36 #define A4 0.70710678118654752438 // cos(pi*4/16)
  37 #define A2 0.92387953251128675613 // cos(pi*2/16)
  38
  39 static const FLOAT prescale[64]={
  40 B0*B0/8, B0*B1/8, B0*B2/8, B0*B3/8, B0*B4/8, B0*B5/8, B0*B6/8, B0*B7/8,
  41 B1*B0/8, B1*B1/8, B1*B2/8, B1*B3/8, B1*B4/8, B1*B5/8, B1*B6/8, B1*B7/8,
  42 B2*B0/8, B2*B1/8, B2*B2/8, B2*B3/8, B2*B4/8, B2*B5/8, B2*B6/8, B2*B7/8,
  43 B3*B0/8, B3*B1/8, B3*B2/8, B3*B3/8, B3*B4/8, B3*B5/8, B3*B6/8, B3*B7/8,
  44 B4*B0/8, B4*B1/8, B4*B2/8, B4*B3/8, B4*B4/8, B4*B5/8, B4*B6/8, B4*B7/8,
  45 B5*B0/8, B5*B1/8, B5*B2/8, B5*B3/8, B5*B4/8, B5*B5/8, B5*B6/8, B5*B7/8,
  46 B6*B0/8, B6*B1/8, B6*B2/8, B6*B3/8, B6*B4/8, B6*B5/8, B6*B6/8, B6*B7/8,
  47 B7*B0/8, B7*B1/8, B7*B2/8, B7*B3/8, B7*B4/8, B7*B5/8, B7*B6/8, B7*B7/8,
  48 };
  49
  50 static inline void p8idct(int16_t data[64], FLOAT temp[64], uint8_t *dest, int stride, int x, int y, int type){
  51     int i;
  52     FLOAT av_unused tmp0;
  53     FLOAT s04, d04, s17, d17, s26, d26, s53, d53;
  54     FLOAT os07, os16, os25, os34;
  55     FLOAT od07, od16, od25, od34;
  56
  57     for(i=0; i<y*8; i+=y){
  58         s17= temp[1*x + i] + temp[7*x + i];
  59         d17= temp[1*x + i] - temp[7*x + i];
  60         s53= temp[5*x + i] + temp[3*x + i];
  61         d53= temp[5*x + i] - temp[3*x + i];
  62
  63         od07=  s17 + s53;
  64         od25= (s17 - s53)*(2*A4);
  65
  66 #if 0 //these 2 are equivalent
  67         tmp0= (d17 + d53)*(2*A2);
  68         od34=  d17*( 2*B6) - tmp0;
  69         od16=  d53*(-2*B2) + tmp0;
  70 #else
  71         od34=  d17*(2*(B6-A2)) - d53*(2*A2);
  72         od16=  d53*(2*(A2-B2)) + d17*(2*A2);
  73 #endif
  74
  75         od16 -= od07;
  76         od25 -= od16;
  77         od34 += od25;
  78
  79         s26 = temp[2*x + i] + temp[6*x + i];
  80         d26 = temp[2*x + i] - temp[6*x + i];
  81         d26*= 2*A4;
  82         d26-= s26;
  83
  84         s04= temp[0*x + i] + temp[4*x + i];
  85         d04= temp[0*x + i] - temp[4*x + i];
  86
  87         os07= s04 + s26;
  88         os34= s04 - s26;
  89         os16= d04 + d26;
  90         os25= d04 - d26;
  91
  92         if(type==0){
  93             temp[0*x + i]= os07 + od07;
  94             temp[7*x + i]= os07 - od07;
  95             temp[1*x + i]= os16 + od16;
  96             temp[6*x + i]= os16 - od16;
  97             temp[2*x + i]= os25 + od25;
  98             temp[5*x + i]= os25 - od25;
  99             temp[3*x + i]= os34 - od34;
 100             temp[4*x + i]= os34 + od34;
 101         }else if(type==1){
 102             data[0*x + i]= lrintf(os07 + od07);
 103             data[7*x + i]= lrintf(os07 - od07);
 104             data[1*x + i]= lrintf(os16 + od16);
 105             data[6*x + i]= lrintf(os16 - od16);
 106             data[2*x + i]= lrintf(os25 + od25);
 107             data[5*x + i]= lrintf(os25 - od25);
 108             data[3*x + i]= lrintf(os34 - od34);
 109             data[4*x + i]= lrintf(os34 + od34);
 110         }else if(type==2){
 111             dest[0*stride + i]= av_clip_uint8(((int)dest[0*stride + i]) + lrintf(os07 + od07));
 112             dest[7*stride + i]= av_clip_uint8(((int)dest[7*stride + i]) + lrintf(os07 - od07));
 113             dest[1*stride + i]= av_clip_uint8(((int)dest[1*stride + i]) + lrintf(os16 + od16));
 114             dest[6*stride + i]= av_clip_uint8(((int)dest[6*stride + i]) + lrintf(os16 - od16));
 115             dest[2*stride + i]= av_clip_uint8(((int)dest[2*stride + i]) + lrintf(os25 + od25));
 116             dest[5*stride + i]= av_clip_uint8(((int)dest[5*stride + i]) + lrintf(os25 - od25));
 117             dest[3*stride + i]= av_clip_uint8(((int)dest[3*stride + i]) + lrintf(os34 - od34));
 118             dest[4*stride + i]= av_clip_uint8(((int)dest[4*stride + i]) + lrintf(os34 + od34));
 119         }else{
 120             dest[0*stride + i]= av_clip_uint8(lrintf(os07 + od07));
 121             dest[7*stride + i]= av_clip_uint8(lrintf(os07 - od07));
 122             dest[1*stride + i]= av_clip_uint8(lrintf(os16 + od16));
 123             dest[6*stride + i]= av_clip_uint8(lrintf(os16 - od16));
 124             dest[2*stride + i]= av_clip_uint8(lrintf(os25 + od25));
 125             dest[5*stride + i]= av_clip_uint8(lrintf(os25 - od25));
 126             dest[3*stride + i]= av_clip_uint8(lrintf(os34 - od34));
 127             dest[4*stride + i]= av_clip_uint8(lrintf(os34 + od34));
 128         }
 129     }
 130 }
 131
 132 void ff_faanidct(int16_t block[64]){
 133     FLOAT temp[64];
 134     int i;
 135
 136     emms_c();
 137
 138     for(i=0; i<64; i++)
 139         temp[i] = block[i] * prescale[i];
 140
 141     p8idct(block, temp, NULL, 0, 1, 8, 0);
 142     p8idct(block, temp, NULL, 0, 8, 1, 1);
 143 }
 144
 145 void ff_faanidct_add(uint8_t *dest, int line_size, int16_t block[64]){
 146     FLOAT temp[64];
 147     int i;
 148
 149     emms_c();
 150
 151     for(i=0; i<64; i++)
 152         temp[i] = block[i] * prescale[i];
 153
 154     p8idct(block, temp, NULL,         0, 1, 8, 0);
 155     p8idct(NULL , temp, dest, line_size, 8, 1, 2);
 156 }
 157
 158 void ff_faanidct_put(uint8_t *dest, int line_size, int16_t block[64]){
 159     FLOAT temp[64];
 160     int i;
 161
 162     emms_c();
 163
 164     for(i=0; i<64; i++)
 165         temp[i] = block[i] * prescale[i];
 166
 167     p8idct(block, temp, NULL,         0, 1, 8, 0);
 168     p8idct(NULL , temp, dest, line_size, 8, 1, 3);
 169 }