faanidct.c (6134B)
1 /* 2 * Floating point AAN IDCT 3 * Copyright (c) 2008 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 #include "faanidct.h" 22 #include "libavutil/common.h" 23 #include "libavutil/emms.h" 24 25 /* To allow switching to double. */ 26 typedef float FLOAT; 27 28 #define B0 1.0000000000000000000000 29 #define B1 1.3870398453221474618216 // cos(pi*1/16)sqrt(2) 30 #define B2 1.3065629648763765278566 // cos(pi*2/16)sqrt(2) 31 #define B3 1.1758756024193587169745 // cos(pi*3/16)sqrt(2) 32 #define B4 1.0000000000000000000000 // cos(pi*4/16)sqrt(2) 33 #define B5 0.7856949583871021812779 // cos(pi*5/16)sqrt(2) 34 #define B6 0.5411961001461969843997 // cos(pi*6/16)sqrt(2) 35 #define B7 0.2758993792829430123360 // cos(pi*7/16)sqrt(2) 36 37 #define A4 0.70710678118654752438 // cos(pi*4/16) 38 #define A2 0.92387953251128675613 // cos(pi*2/16) 39 40 static const FLOAT prescale[64]={ 41 B0*B0/8, B0*B1/8, B0*B2/8, B0*B3/8, B0*B4/8, B0*B5/8, B0*B6/8, B0*B7/8, 42 B1*B0/8, B1*B1/8, B1*B2/8, B1*B3/8, B1*B4/8, B1*B5/8, B1*B6/8, B1*B7/8, 43 B2*B0/8, B2*B1/8, B2*B2/8, B2*B3/8, B2*B4/8, B2*B5/8, B2*B6/8, B2*B7/8, 44 B3*B0/8, B3*B1/8, B3*B2/8, B3*B3/8, B3*B4/8, B3*B5/8, B3*B6/8, B3*B7/8, 45 B4*B0/8, B4*B1/8, B4*B2/8, B4*B3/8, B4*B4/8, B4*B5/8, B4*B6/8, B4*B7/8, 46 B5*B0/8, B5*B1/8, B5*B2/8, B5*B3/8, B5*B4/8, B5*B5/8, B5*B6/8, B5*B7/8, 47 B6*B0/8, B6*B1/8, B6*B2/8, B6*B3/8, B6*B4/8, B6*B5/8, B6*B6/8, B6*B7/8, 48 B7*B0/8, B7*B1/8, B7*B2/8, B7*B3/8, B7*B4/8, B7*B5/8, B7*B6/8, B7*B7/8, 49 }; 50 51 static inline void p8idct(int16_t data[64], FLOAT temp[64], uint8_t *dest, 52 ptrdiff_t stride, int x, int y, int type) 53 { 54 int i; 55 FLOAT s04, d04, s17, d17, s26, d26, s53, d53; 56 FLOAT os07, os16, os25, os34; 57 FLOAT od07, od16, od25, od34; 58 59 for(i=0; i<y*8; i+=y){ 60 s17= temp[1*x + i] + temp[7*x + i]; 61 d17= temp[1*x + i] - temp[7*x + i]; 62 s53= temp[5*x + i] + temp[3*x + i]; 63 d53= temp[5*x + i] - temp[3*x + i]; 64 65 od07= s17 + s53; 66 od25= (s17 - s53)*(2*A4); 67 68 od34= d17*(2*(B6-A2)) - d53*(2*A2); 69 od16= d53*(2*(A2-B2)) + d17*(2*A2); 70 71 od16 -= od07; 72 od25 -= od16; 73 od34 += od25; 74 75 s26 = temp[2*x + i] + temp[6*x + i]; 76 d26 = temp[2*x + i] - temp[6*x + i]; 77 d26*= 2*A4; 78 d26-= s26; 79 80 s04= temp[0*x + i] + temp[4*x + i]; 81 d04= temp[0*x + i] - temp[4*x + i]; 82 83 os07= s04 + s26; 84 os34= s04 - s26; 85 os16= d04 + d26; 86 os25= d04 - d26; 87 88 if(type==0){ 89 temp[0*x + i]= os07 + od07; 90 temp[7*x + i]= os07 - od07; 91 temp[1*x + i]= os16 + od16; 92 temp[6*x + i]= os16 - od16; 93 temp[2*x + i]= os25 + od25; 94 temp[5*x + i]= os25 - od25; 95 temp[3*x + i]= os34 - od34; 96 temp[4*x + i]= os34 + od34; 97 }else if(type==1){ 98 data[0*x + i]= lrintf(os07 + od07); 99 data[7*x + i]= lrintf(os07 - od07); 100 data[1*x + i]= lrintf(os16 + od16); 101 data[6*x + i]= lrintf(os16 - od16); 102 data[2*x + i]= lrintf(os25 + od25); 103 data[5*x + i]= lrintf(os25 - od25); 104 data[3*x + i]= lrintf(os34 - od34); 105 data[4*x + i]= lrintf(os34 + od34); 106 }else if(type==2){ 107 dest[0*stride + i]= av_clip_uint8(((int)dest[0*stride + i]) + lrintf(os07 + od07)); 108 dest[7*stride + i]= av_clip_uint8(((int)dest[7*stride + i]) + lrintf(os07 - od07)); 109 dest[1*stride + i]= av_clip_uint8(((int)dest[1*stride + i]) + lrintf(os16 + od16)); 110 dest[6*stride + i]= av_clip_uint8(((int)dest[6*stride + i]) + lrintf(os16 - od16)); 111 dest[2*stride + i]= av_clip_uint8(((int)dest[2*stride + i]) + lrintf(os25 + od25)); 112 dest[5*stride + i]= av_clip_uint8(((int)dest[5*stride + i]) + lrintf(os25 - od25)); 113 dest[3*stride + i]= av_clip_uint8(((int)dest[3*stride + i]) + lrintf(os34 - od34)); 114 dest[4*stride + i]= av_clip_uint8(((int)dest[4*stride + i]) + lrintf(os34 + od34)); 115 }else{ 116 dest[0*stride + i]= av_clip_uint8(lrintf(os07 + od07)); 117 dest[7*stride + i]= av_clip_uint8(lrintf(os07 - od07)); 118 dest[1*stride + i]= av_clip_uint8(lrintf(os16 + od16)); 119 dest[6*stride + i]= av_clip_uint8(lrintf(os16 - od16)); 120 dest[2*stride + i]= av_clip_uint8(lrintf(os25 + od25)); 121 dest[5*stride + i]= av_clip_uint8(lrintf(os25 - od25)); 122 dest[3*stride + i]= av_clip_uint8(lrintf(os34 - od34)); 123 dest[4*stride + i]= av_clip_uint8(lrintf(os34 + od34)); 124 } 125 } 126 } 127 128 void ff_faanidct(int16_t block[64]){ 129 FLOAT temp[64]; 130 int i; 131 132 emms_c(); 133 134 for(i=0; i<64; i++) 135 temp[i] = block[i] * prescale[i]; 136 137 p8idct(block, temp, NULL, 0, 1, 8, 0); 138 p8idct(block, temp, NULL, 0, 8, 1, 1); 139 } 140 141 void ff_faanidct_add(uint8_t *dest, ptrdiff_t line_size, int16_t block[64]) 142 { 143 FLOAT temp[64]; 144 int i; 145 146 emms_c(); 147 148 for(i=0; i<64; i++) 149 temp[i] = block[i] * prescale[i]; 150 151 p8idct(block, temp, NULL, 0, 1, 8, 0); 152 p8idct(NULL , temp, dest, line_size, 8, 1, 2); 153 } 154 155 void ff_faanidct_put(uint8_t *dest, ptrdiff_t line_size, int16_t block[64]) 156 { 157 FLOAT temp[64]; 158 int i; 159 160 emms_c(); 161 162 for(i=0; i<64; i++) 163 temp[i] = block[i] * prescale[i]; 164 165 p8idct(block, temp, NULL, 0, 1, 8, 0); 166 p8idct(NULL , temp, dest, line_size, 8, 1, 3); 167 }