libavcodec/rv34dsp.c

   1 /*
   2  * RV30/40 decoder common dsp functions
   3  * Copyright (c) 2007 Mike Melanson, Konstantin Shishkov
   4  * Copyright (c) 2011 Janne Grunau
   5  *
   6  * This file is part of Libav.
   7  *
   8  * Libav is free software; you can redistribute it and/or
   9  * modify it under the terms of the GNU Lesser General Public
  10  * License as published by the Free Software Foundation; either
  11  * version 2.1 of the License, or (at your option) any later version.
  12  *
  13  * Libav is distributed in the hope that it will be useful,
  14  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  15  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
  16  * Lesser General Public License for more details.
  17  *
  18  * You should have received a copy of the GNU Lesser General Public
  19  * License along with Libav; if not, write to the Free Software
  20  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
  21  */
  22
  23 /**
  24  * @file
  25  * RV30/40 decoder common dsp functions
  26  */
  27
  28 #include "rv34dsp.h"
  29 #include "libavutil/common.h"
  30
  31 /**
  32  * @name RV30/40 inverse transform functions
  33  * @{
  34  */
  35
  36 static av_always_inline void rv34_row_transform(int temp[16], int16_t *block)
  37 {
  38     int i;
  39
  40     for(i = 0; i < 4; i++){
  41         const int z0 = 13*(block[i+4*0] +    block[i+4*2]);
  42         const int z1 = 13*(block[i+4*0] -    block[i+4*2]);
  43         const int z2 =  7* block[i+4*1] - 17*block[i+4*3];
  44         const int z3 = 17* block[i+4*1] +  7*block[i+4*3];
  45
  46         temp[4*i+0] = z0 + z3;
  47         temp[4*i+1] = z1 + z2;
  48         temp[4*i+2] = z1 - z2;
  49         temp[4*i+3] = z0 - z3;
  50     }
  51 }
  52
  53 /**
  54  * Real Video 3.0/4.0 inverse transform + sample reconstruction
  55  * Code is almost the same as in SVQ3, only scaling is different.
  56  */
  57 static void rv34_idct_add_c(uint8_t *dst, ptrdiff_t stride, int16_t *block){
  58     int      temp[16];
  59     int      i;
  60
  61     rv34_row_transform(temp, block);
  62     memset(block, 0, 16*sizeof(int16_t));
  63
  64     for(i = 0; i < 4; i++){
  65         const int z0 = 13*(temp[4*0+i] +    temp[4*2+i]) + 0x200;
  66         const int z1 = 13*(temp[4*0+i] -    temp[4*2+i]) + 0x200;
  67         const int z2 =  7* temp[4*1+i] - 17*temp[4*3+i];
  68         const int z3 = 17* temp[4*1+i] +  7*temp[4*3+i];
  69
  70         dst[0] = av_clip_uint8( dst[0] + ( (z0 + z3) >> 10 ) );
  71         dst[1] = av_clip_uint8( dst[1] + ( (z1 + z2) >> 10 ) );
  72         dst[2] = av_clip_uint8( dst[2] + ( (z1 - z2) >> 10 ) );
  73         dst[3] = av_clip_uint8( dst[3] + ( (z0 - z3) >> 10 ) );
  74
  75         dst  += stride;
  76     }
  77 }
  78
  79 /**
  80  * RealVideo 3.0/4.0 inverse transform for DC block
  81  *
  82  * Code is almost the same as rv34_inv_transform()
  83  * but final coefficients are multiplied by 1.5 and have no rounding.
  84  */
  85 static void rv34_inv_transform_noround_c(int16_t *block){
  86     int temp[16];
  87     int i;
  88
  89     rv34_row_transform(temp, block);
  90
  91     for(i = 0; i < 4; i++){
  92         const int z0 = 39*(temp[4*0+i] +    temp[4*2+i]);
  93         const int z1 = 39*(temp[4*0+i] -    temp[4*2+i]);
  94         const int z2 = 21* temp[4*1+i] - 51*temp[4*3+i];
  95         const int z3 = 51* temp[4*1+i] + 21*temp[4*3+i];
  96
  97         block[i*4+0] = (z0 + z3) >> 11;
  98         block[i*4+1] = (z1 + z2) >> 11;
  99         block[i*4+2] = (z1 - z2) >> 11;
 100         block[i*4+3] = (z0 - z3) >> 11;
 101     }
 102 }
 103
 104 static void rv34_idct_dc_add_c(uint8_t *dst, ptrdiff_t stride, int dc)
 105 {
 106     int i, j;
 107
 108     dc = (13*13*dc + 0x200) >> 10;
 109     for (i = 0; i < 4; i++)
 110     {
 111         for (j = 0; j < 4; j++)
 112             dst[j] = av_clip_uint8( dst[j] + dc );
 113
 114         dst += stride;
 115     }
 116 }
 117
 118 static void rv34_inv_transform_dc_noround_c(int16_t *block)
 119 {
 120     int16_t dc = (13 * 13 * 3 * block[0]) >> 11;
 121     int i, j;
 122
 123     for (i = 0; i < 4; i++, block += 4)
 124         for (j = 0; j < 4; j++)
 125             block[j] = dc;
 126 }
 127
 128 /** @} */ // transform
 129
 130
 131 av_cold void ff_rv34dsp_init(RV34DSPContext *c)
 132 {
 133     c->rv34_inv_transform    = rv34_inv_transform_noround_c;
 134     c->rv34_inv_transform_dc = rv34_inv_transform_dc_noround_c;
 135
 136     c->rv34_idct_add    = rv34_idct_add_c;
 137     c->rv34_idct_dc_add = rv34_idct_dc_add_c;
 138
 139     if (ARCH_ARM)
 140         ff_rv34dsp_init_arm(c);
 141     if (ARCH_X86)
 142         ff_rv34dsp_init_x86(c);
 143 }