libavcodec/binkdsp.c

   1 /*
   2  * Bink DSP routines
   3  * Copyright (c) 2009 Konstantin Shishkov
   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
  24  * Bink DSP routines
  25  */
  26
  27 #include "libavutil/attributes.h"
  28 #include "binkdsp.h"
  29
  30 #define A1  2896 /* (1/sqrt(2))<<12 */
  31 #define A2  2217
  32 #define A3  3784
  33 #define A4 -5352
  34
  35 #define MUL(X,Y) ((int)((unsigned)(X) * (Y)) >> 11)
  36
  37 #define IDCT_TRANSFORM(dest,s0,s1,s2,s3,s4,s5,s6,s7,d0,d1,d2,d3,d4,d5,d6,d7,munge,src) {\
  38     const int a0 = (src)[s0] + (src)[s4]; \
  39     const int a1 = (src)[s0] - (src)[s4]; \
  40     const int a2 = (src)[s2] + (src)[s6]; \
  41     const int a3 = MUL(A1, (src)[s2] - (src)[s6]); \
  42     const int a4 = (src)[s5] + (src)[s3]; \
  43     const int a5 = (src)[s5] - (src)[s3]; \
  44     const int a6 = (src)[s1] + (src)[s7]; \
  45     const int a7 = (src)[s1] - (src)[s7]; \
  46     const int b0 = a4 + a6; \
  47     const int b1 = MUL(A3, a5 + a7); \
  48     const int b2 = MUL(A4, a5) - b0 + b1; \
  49     const int b3 = MUL(A1, a6 - a4) - b2; \
  50     const int b4 = MUL(A2, a7) + b3 - b1; \
  51     (dest)[d0] = munge(a0+a2   +b0); \
  52     (dest)[d1] = munge(a1+a3-a2+b2); \
  53     (dest)[d2] = munge(a1-a3+a2+b3); \
  54     (dest)[d3] = munge(a0-a2   -b4); \
  55     (dest)[d4] = munge(a0-a2   +b4); \
  56     (dest)[d5] = munge(a1-a3+a2-b3); \
  57     (dest)[d6] = munge(a1+a3-a2-b2); \
  58     (dest)[d7] = munge(a0+a2   -b0); \
  59 }
  60 /* end IDCT_TRANSFORM macro */
  61
  62 #define MUNGE_NONE(x) (x)
  63 #define IDCT_COL(dest,src) IDCT_TRANSFORM(dest,0,8,16,24,32,40,48,56,0,8,16,24,32,40,48,56,MUNGE_NONE,src)
  64
  65 #define MUNGE_ROW(x) (((x) + 0x7F)>>8)
  66 #define IDCT_ROW(dest,src) IDCT_TRANSFORM(dest,0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7,MUNGE_ROW,src)
  67
  68 static inline void bink_idct_col(int *dest, const int32_t *src)
  69 {
  70     if ((src[8]|src[16]|src[24]|src[32]|src[40]|src[48]|src[56])==0) {
  71         dest[0]  =
  72         dest[8]  =
  73         dest[16] =
  74         dest[24] =
  75         dest[32] =
  76         dest[40] =
  77         dest[48] =
  78         dest[56] = src[0];
  79     } else {
  80         IDCT_COL(dest, src);
  81     }
  82 }
  83
  84 static void bink_idct_c(int32_t *block)
  85 {
  86     int i;
  87     int temp[64];
  88
  89     for (i = 0; i < 8; i++)
  90         bink_idct_col(&temp[i], &block[i]);
  91     for (i = 0; i < 8; i++) {
  92         IDCT_ROW( (&block[8*i]), (&temp[8*i]) );
  93     }
  94 }
  95
  96 static void bink_idct_add_c(uint8_t *dest, int linesize, int32_t *block)
  97 {
  98     int i, j;
  99
 100     bink_idct_c(block);
 101     for (i = 0; i < 8; i++, dest += linesize, block += 8)
 102         for (j = 0; j < 8; j++)
 103              dest[j] += block[j];
 104 }
 105
 106 static void bink_idct_put_c(uint8_t *dest, int linesize, int32_t *block)
 107 {
 108     int i;
 109     int temp[64];
 110     for (i = 0; i < 8; i++)
 111         bink_idct_col(&temp[i], &block[i]);
 112     for (i = 0; i < 8; i++) {
 113         IDCT_ROW( (&dest[i*linesize]), (&temp[8*i]) );
 114     }
 115 }
 116
 117 static void scale_block_c(const uint8_t src[64]/*align 8*/, uint8_t *dst/*align 8*/, int linesize)
 118 {
 119     int i, j;
 120     uint16_t *dst1 = (uint16_t *) dst;
 121     uint16_t *dst2 = (uint16_t *)(dst + linesize);
 122
 123     for (j = 0; j < 8; j++) {
 124         for (i = 0; i < 8; i++) {
 125             dst1[i] = dst2[i] = src[i] * 0x0101;
 126         }
 127         src  += 8;
 128         dst1 += linesize;
 129         dst2 += linesize;
 130     }
 131 }
 132
 133 static void add_pixels8_c(uint8_t *restrict pixels, int16_t *block,
 134                           int line_size)
 135 {
 136     int i;
 137
 138     for (i = 0; i < 8; i++) {
 139         pixels[0] += block[0];
 140         pixels[1] += block[1];
 141         pixels[2] += block[2];
 142         pixels[3] += block[3];
 143         pixels[4] += block[4];
 144         pixels[5] += block[5];
 145         pixels[6] += block[6];
 146         pixels[7] += block[7];
 147         pixels    += line_size;
 148         block     += 8;
 149     }
 150 }
 151
 152 av_cold void ff_binkdsp_init(BinkDSPContext *c)
 153 {
 154     c->idct_add    = bink_idct_add_c;
 155     c->idct_put    = bink_idct_put_c;
 156     c->scale_block = scale_block_c;
 157     c->add_pixels8 = add_pixels8_c;
 158 }