2001-12-09 12:39:54 +00:00
|
|
|
/*
|
2002-05-25 22:45:33 +00:00
|
|
|
* Simple IDCT
|
|
|
|
*
|
|
|
|
* Copyright (c) 2001 Michael Niedermayer <michaelni@gmx.at>
|
|
|
|
*
|
2006-10-07 15:30:46 +00:00
|
|
|
* This file is part of FFmpeg.
|
|
|
|
*
|
|
|
|
* FFmpeg is free software; you can redistribute it and/or
|
2002-05-25 22:45:33 +00:00
|
|
|
* modify it under the terms of the GNU Lesser General Public
|
|
|
|
* License as published by the Free Software Foundation; either
|
2006-10-07 15:30:46 +00:00
|
|
|
* version 2.1 of the License, or (at your option) any later version.
|
2002-05-25 22:45:33 +00:00
|
|
|
*
|
2006-10-07 15:30:46 +00:00
|
|
|
* FFmpeg is distributed in the hope that it will be useful,
|
2002-05-25 22:45:33 +00:00
|
|
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of
|
|
|
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
|
|
|
|
* Lesser General Public License for more details.
|
|
|
|
*
|
|
|
|
* You should have received a copy of the GNU Lesser General Public
|
2006-10-07 15:30:46 +00:00
|
|
|
* License along with FFmpeg; if not, write to the Free Software
|
2006-01-12 22:43:26 +00:00
|
|
|
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
|
2002-05-25 22:45:33 +00:00
|
|
|
*/
|
2005-12-17 18:14:38 +00:00
|
|
|
|
2003-03-06 11:32:04 +00:00
|
|
|
/**
|
2010-04-20 14:45:34 +00:00
|
|
|
* @file
|
2003-03-06 11:32:04 +00:00
|
|
|
* simpleidct in C.
|
|
|
|
*/
|
2005-12-17 18:14:38 +00:00
|
|
|
|
2011-07-18 15:04:25 +00:00
|
|
|
#include "libavutil/intreadwrite.h"
|
2002-05-18 23:03:29 +00:00
|
|
|
#include "avcodec.h"
|
2002-06-05 18:46:25 +00:00
|
|
|
#include "dsputil.h"
|
2008-12-27 16:25:36 +00:00
|
|
|
#include "mathops.h"
|
2001-12-09 12:39:54 +00:00
|
|
|
#include "simple_idct.h"
|
|
|
|
|
2011-07-20 15:05:05 +00:00
|
|
|
#define BIT_DEPTH 8
|
|
|
|
#include "simple_idct_template.c"
|
|
|
|
#undef BIT_DEPTH
|
2001-12-09 12:39:54 +00:00
|
|
|
|
2011-07-20 15:05:05 +00:00
|
|
|
#define BIT_DEPTH 10
|
|
|
|
#include "simple_idct_template.c"
|
|
|
|
#undef BIT_DEPTH
|
2002-09-01 09:33:05 +00:00
|
|
|
|
2002-10-03 13:41:33 +00:00
|
|
|
/* 2x4x8 idct */
|
|
|
|
|
|
|
|
#define CN_SHIFT 12
|
|
|
|
#define C_FIX(x) ((int)((x) * (1 << CN_SHIFT) + 0.5))
|
2002-10-03 21:03:59 +00:00
|
|
|
#define C1 C_FIX(0.6532814824)
|
|
|
|
#define C2 C_FIX(0.2705980501)
|
2002-10-03 13:41:33 +00:00
|
|
|
|
2002-10-03 21:03:59 +00:00
|
|
|
/* row idct is multiple by 16 * sqrt(2.0), col idct4 is normalized,
|
|
|
|
and the butterfly must be multiplied by 0.5 * sqrt(2.0) */
|
|
|
|
#define C_SHIFT (4+1+12)
|
2002-10-03 13:41:33 +00:00
|
|
|
|
2013-01-20 00:02:29 +00:00
|
|
|
static inline void idct4col_put(uint8_t *dest, int line_size, const int16_t *col)
|
2002-10-03 13:41:33 +00:00
|
|
|
{
|
|
|
|
int c0, c1, c2, c3, a0, a1, a2, a3;
|
|
|
|
|
|
|
|
a0 = col[8*0];
|
|
|
|
a1 = col[8*2];
|
|
|
|
a2 = col[8*4];
|
|
|
|
a3 = col[8*6];
|
2002-10-03 21:03:59 +00:00
|
|
|
c0 = ((a0 + a2) << (CN_SHIFT - 1)) + (1 << (C_SHIFT - 1));
|
|
|
|
c2 = ((a0 - a2) << (CN_SHIFT - 1)) + (1 << (C_SHIFT - 1));
|
2002-10-03 13:41:33 +00:00
|
|
|
c1 = a1 * C1 + a3 * C2;
|
|
|
|
c3 = a1 * C2 - a3 * C1;
|
2012-03-02 21:09:44 +00:00
|
|
|
dest[0] = av_clip_uint8((c0 + c1) >> C_SHIFT);
|
2002-10-03 13:41:33 +00:00
|
|
|
dest += line_size;
|
2012-03-02 21:09:44 +00:00
|
|
|
dest[0] = av_clip_uint8((c2 + c3) >> C_SHIFT);
|
2002-10-03 13:41:33 +00:00
|
|
|
dest += line_size;
|
2012-03-02 21:09:44 +00:00
|
|
|
dest[0] = av_clip_uint8((c2 - c3) >> C_SHIFT);
|
2002-10-03 13:41:33 +00:00
|
|
|
dest += line_size;
|
2012-03-02 21:09:44 +00:00
|
|
|
dest[0] = av_clip_uint8((c0 - c1) >> C_SHIFT);
|
2002-10-03 13:41:33 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
#define BF(k) \
|
|
|
|
{\
|
|
|
|
int a0, a1;\
|
|
|
|
a0 = ptr[k];\
|
|
|
|
a1 = ptr[8 + k];\
|
|
|
|
ptr[k] = a0 + a1;\
|
|
|
|
ptr[8 + k] = a0 - a1;\
|
|
|
|
}
|
|
|
|
|
|
|
|
/* only used by DV codec. The input must be interlaced. 128 is added
|
|
|
|
to the pixels before clamping to avoid systematic error
|
|
|
|
(1024*sqrt(2)) offset would be needed otherwise. */
|
|
|
|
/* XXX: I think a 1.0/sqrt(2) normalization should be needed to
|
|
|
|
compensate the extra butterfly stage - I don't have the full DV
|
|
|
|
specification */
|
2013-01-20 00:02:29 +00:00
|
|
|
void ff_simple_idct248_put(uint8_t *dest, int line_size, int16_t *block)
|
2002-10-03 13:41:33 +00:00
|
|
|
{
|
|
|
|
int i;
|
2013-01-20 00:02:29 +00:00
|
|
|
int16_t *ptr;
|
2005-12-17 18:14:38 +00:00
|
|
|
|
2002-10-03 13:41:33 +00:00
|
|
|
/* butterfly */
|
|
|
|
ptr = block;
|
|
|
|
for(i=0;i<4;i++) {
|
|
|
|
BF(0);
|
|
|
|
BF(1);
|
|
|
|
BF(2);
|
|
|
|
BF(3);
|
|
|
|
BF(4);
|
|
|
|
BF(5);
|
|
|
|
BF(6);
|
|
|
|
BF(7);
|
|
|
|
ptr += 2 * 8;
|
|
|
|
}
|
|
|
|
|
|
|
|
/* IDCT8 on each line */
|
|
|
|
for(i=0; i<8; i++) {
|
2011-10-14 16:27:59 +00:00
|
|
|
idctRowCondDC_8(block + i*8, 0);
|
2002-10-03 13:41:33 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/* IDCT4 and store */
|
|
|
|
for(i=0;i<8;i++) {
|
2007-12-03 06:59:48 +00:00
|
|
|
idct4col_put(dest + i, 2 * line_size, block + i);
|
|
|
|
idct4col_put(dest + line_size + i, 2 * line_size, block + 8 + i);
|
2002-10-03 13:41:33 +00:00
|
|
|
}
|
|
|
|
}
|
2002-12-27 23:51:46 +00:00
|
|
|
|
|
|
|
/* 8x4 & 4x8 WMV2 IDCT */
|
|
|
|
#undef CN_SHIFT
|
|
|
|
#undef C_SHIFT
|
|
|
|
#undef C_FIX
|
|
|
|
#undef C1
|
|
|
|
#undef C2
|
|
|
|
#define CN_SHIFT 12
|
|
|
|
#define C_FIX(x) ((int)((x) * 1.414213562 * (1 << CN_SHIFT) + 0.5))
|
|
|
|
#define C1 C_FIX(0.6532814824)
|
|
|
|
#define C2 C_FIX(0.2705980501)
|
|
|
|
#define C3 C_FIX(0.5)
|
|
|
|
#define C_SHIFT (4+1+12)
|
2013-01-20 00:02:29 +00:00
|
|
|
static inline void idct4col_add(uint8_t *dest, int line_size, const int16_t *col)
|
2002-12-27 23:51:46 +00:00
|
|
|
{
|
|
|
|
int c0, c1, c2, c3, a0, a1, a2, a3;
|
|
|
|
|
|
|
|
a0 = col[8*0];
|
|
|
|
a1 = col[8*1];
|
|
|
|
a2 = col[8*2];
|
|
|
|
a3 = col[8*3];
|
|
|
|
c0 = (a0 + a2)*C3 + (1 << (C_SHIFT - 1));
|
|
|
|
c2 = (a0 - a2)*C3 + (1 << (C_SHIFT - 1));
|
|
|
|
c1 = a1 * C1 + a3 * C2;
|
|
|
|
c3 = a1 * C2 - a3 * C1;
|
2012-03-06 00:01:19 +00:00
|
|
|
dest[0] = av_clip_uint8(dest[0] + ((c0 + c1) >> C_SHIFT));
|
2002-12-27 23:51:46 +00:00
|
|
|
dest += line_size;
|
2012-03-06 00:01:19 +00:00
|
|
|
dest[0] = av_clip_uint8(dest[0] + ((c2 + c3) >> C_SHIFT));
|
2002-12-27 23:51:46 +00:00
|
|
|
dest += line_size;
|
2012-03-06 00:01:19 +00:00
|
|
|
dest[0] = av_clip_uint8(dest[0] + ((c2 - c3) >> C_SHIFT));
|
2002-12-27 23:51:46 +00:00
|
|
|
dest += line_size;
|
2012-03-06 00:01:19 +00:00
|
|
|
dest[0] = av_clip_uint8(dest[0] + ((c0 - c1) >> C_SHIFT));
|
2002-12-27 23:51:46 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
#define RN_SHIFT 15
|
|
|
|
#define R_FIX(x) ((int)((x) * 1.414213562 * (1 << RN_SHIFT) + 0.5))
|
|
|
|
#define R1 R_FIX(0.6532814824)
|
|
|
|
#define R2 R_FIX(0.2705980501)
|
|
|
|
#define R3 R_FIX(0.5)
|
|
|
|
#define R_SHIFT 11
|
2013-01-20 00:02:29 +00:00
|
|
|
static inline void idct4row(int16_t *row)
|
2002-12-27 23:51:46 +00:00
|
|
|
{
|
|
|
|
int c0, c1, c2, c3, a0, a1, a2, a3;
|
|
|
|
|
|
|
|
a0 = row[0];
|
|
|
|
a1 = row[1];
|
|
|
|
a2 = row[2];
|
|
|
|
a3 = row[3];
|
|
|
|
c0 = (a0 + a2)*R3 + (1 << (R_SHIFT - 1));
|
|
|
|
c2 = (a0 - a2)*R3 + (1 << (R_SHIFT - 1));
|
|
|
|
c1 = a1 * R1 + a3 * R2;
|
|
|
|
c3 = a1 * R2 - a3 * R1;
|
|
|
|
row[0]= (c0 + c1) >> R_SHIFT;
|
|
|
|
row[1]= (c2 + c3) >> R_SHIFT;
|
|
|
|
row[2]= (c2 - c3) >> R_SHIFT;
|
|
|
|
row[3]= (c0 - c1) >> R_SHIFT;
|
|
|
|
}
|
|
|
|
|
2013-01-20 00:02:29 +00:00
|
|
|
void ff_simple_idct84_add(uint8_t *dest, int line_size, int16_t *block)
|
2002-12-27 23:51:46 +00:00
|
|
|
{
|
|
|
|
int i;
|
|
|
|
|
|
|
|
/* IDCT8 on each line */
|
|
|
|
for(i=0; i<4; i++) {
|
2011-10-14 16:27:59 +00:00
|
|
|
idctRowCondDC_8(block + i*8, 0);
|
2002-12-27 23:51:46 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
/* IDCT4 and store */
|
|
|
|
for(i=0;i<8;i++) {
|
|
|
|
idct4col_add(dest + i, line_size, block + i);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2013-01-20 00:02:29 +00:00
|
|
|
void ff_simple_idct48_add(uint8_t *dest, int line_size, int16_t *block)
|
2002-12-27 23:51:46 +00:00
|
|
|
{
|
|
|
|
int i;
|
|
|
|
|
|
|
|
/* IDCT4 on each line */
|
|
|
|
for(i=0; i<8; i++) {
|
|
|
|
idct4row(block + i*8);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* IDCT8 and store */
|
|
|
|
for(i=0; i<4; i++){
|
2011-07-20 15:05:05 +00:00
|
|
|
idctSparseColAdd_8(dest + i, line_size, block + i);
|
2002-12-27 23:51:46 +00:00
|
|
|
}
|
|
|
|
}
|
|
|
|
|
2013-01-20 00:02:29 +00:00
|
|
|
void ff_simple_idct44_add(uint8_t *dest, int line_size, int16_t *block)
|
2007-12-09 05:54:59 +00:00
|
|
|
{
|
|
|
|
int i;
|
|
|
|
|
|
|
|
/* IDCT4 on each line */
|
|
|
|
for(i=0; i<4; i++) {
|
|
|
|
idct4row(block + i*8);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* IDCT4 and store */
|
|
|
|
for(i=0; i<4; i++){
|
|
|
|
idct4col_add(dest + i, line_size, block + i);
|
|
|
|
}
|
|
|
|
}
|
2011-10-06 15:03:38 +00:00
|
|
|
|
2013-01-20 00:02:29 +00:00
|
|
|
void ff_prores_idct(int16_t *block, const int16_t *qmat)
|
2011-10-06 15:03:38 +00:00
|
|
|
{
|
|
|
|
int i;
|
|
|
|
|
|
|
|
for (i = 0; i < 64; i++)
|
|
|
|
block[i] *= qmat[i];
|
|
|
|
|
|
|
|
for (i = 0; i < 8; i++)
|
2011-10-14 16:27:59 +00:00
|
|
|
idctRowCondDC_10(block + i*8, 2);
|
2011-10-06 15:03:38 +00:00
|
|
|
|
|
|
|
for (i = 0; i < 8; i++)
|
|
|
|
idctSparseCol_10(block + i);
|
|
|
|
}
|