4 * Copyright (c) 2001 Michael Niedermayer <michaelni@gmx.at>
6 * This file is part of FFmpeg.
8 * FFmpeg 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.
13 * FFmpeg 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.
18 * You should have received a copy of the GNU Lesser General Public
19 * License along with FFmpeg; if not, write to the Free Software
20 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
28 #include "libavutil/intreadwrite.h"
31 #include "simple_idct.h"
33 #define IN_IDCT_DEPTH 16
36 #include "simple_idct_template.c"
40 #include "simple_idct_template.c"
43 #include "simple_idct_template.c"
49 #include "simple_idct_template.c"
53 #define IN_IDCT_DEPTH 32
55 #include "simple_idct_template.c"
62 #define C_FIX(x) ((int)((x) * (1 << CN_SHIFT) + 0.5))
63 #define C1 C_FIX(0.6532814824)
64 #define C2 C_FIX(0.2705980501)
66 /* row idct is multiple by 16 * sqrt(2.0), col idct4 is normalized,
67 and the butterfly must be multiplied by 0.5 * sqrt(2.0) */
68 #define C_SHIFT (4+1+12)
70 static inline void idct4col_put(uint8_t *dest, ptrdiff_t line_size, const int16_t *col)
72 int c0, c1, c2, c3, a0, a1, a2, a3;
78 c0 = ((a0 + a2) * (1 << CN_SHIFT - 1)) + (1 << (C_SHIFT - 1));
79 c2 = ((a0 - a2) * (1 << CN_SHIFT - 1)) + (1 << (C_SHIFT - 1));
80 c1 = a1 * C1 + a3 * C2;
81 c3 = a1 * C2 - a3 * C1;
82 dest[0] = av_clip_uint8((c0 + c1) >> C_SHIFT);
84 dest[0] = av_clip_uint8((c2 + c3) >> C_SHIFT);
86 dest[0] = av_clip_uint8((c2 - c3) >> C_SHIFT);
88 dest[0] = av_clip_uint8((c0 - c1) >> C_SHIFT);
97 ptr[8 + k] = a0 - a1;\
100 /* only used by DV codec. The input must be interlaced. 128 is added
101 to the pixels before clamping to avoid systematic error
102 (1024*sqrt(2)) offset would be needed otherwise. */
103 /* XXX: I think a 1.0/sqrt(2) normalization should be needed to
104 compensate the extra butterfly stage - I don't have the full DV
106 void ff_simple_idct248_put(uint8_t *dest, ptrdiff_t line_size, int16_t *block)
125 /* IDCT8 on each line */
127 idctRowCondDC_int16_8bit(block + i*8, 0);
130 /* IDCT4 and store */
132 idct4col_put(dest + i, 2 * line_size, block + i);
133 idct4col_put(dest + line_size + i, 2 * line_size, block + 8 + i);
137 /* 8x4 & 4x8 WMV2 IDCT */
144 #define C_FIX(x) ((int)((x) * M_SQRT2 * (1 << CN_SHIFT) + 0.5))
145 #define C1 C_FIX(0.6532814824)
146 #define C2 C_FIX(0.2705980501)
147 #define C3 C_FIX(0.5)
148 #define C_SHIFT (4+1+12)
149 static inline void idct4col_add(uint8_t *dest, ptrdiff_t line_size, const int16_t *col)
151 int c0, c1, c2, c3, a0, a1, a2, a3;
157 c0 = (a0 + a2)*C3 + (1 << (C_SHIFT - 1));
158 c2 = (a0 - a2)*C3 + (1 << (C_SHIFT - 1));
159 c1 = a1 * C1 + a3 * C2;
160 c3 = a1 * C2 - a3 * C1;
161 dest[0] = av_clip_uint8(dest[0] + ((c0 + c1) >> C_SHIFT));
163 dest[0] = av_clip_uint8(dest[0] + ((c2 + c3) >> C_SHIFT));
165 dest[0] = av_clip_uint8(dest[0] + ((c2 - c3) >> C_SHIFT));
167 dest[0] = av_clip_uint8(dest[0] + ((c0 - c1) >> C_SHIFT));
171 #define R_FIX(x) ((int)((x) * M_SQRT2 * (1 << RN_SHIFT) + 0.5))
172 #define R1 R_FIX(0.6532814824)
173 #define R2 R_FIX(0.2705980501)
174 #define R3 R_FIX(0.5)
176 static inline void idct4row(int16_t *row)
178 int c0, c1, c2, c3, a0, a1, a2, a3;
184 c0 = (a0 + a2)*R3 + (1 << (R_SHIFT - 1));
185 c2 = (a0 - a2)*R3 + (1 << (R_SHIFT - 1));
186 c1 = a1 * R1 + a3 * R2;
187 c3 = a1 * R2 - a3 * R1;
188 row[0]= (c0 + c1) >> R_SHIFT;
189 row[1]= (c2 + c3) >> R_SHIFT;
190 row[2]= (c2 - c3) >> R_SHIFT;
191 row[3]= (c0 - c1) >> R_SHIFT;
194 void ff_simple_idct84_add(uint8_t *dest, ptrdiff_t line_size, int16_t *block)
198 /* IDCT8 on each line */
200 idctRowCondDC_int16_8bit(block + i*8, 0);
203 /* IDCT4 and store */
205 idct4col_add(dest + i, line_size, block + i);
209 void ff_simple_idct48_add(uint8_t *dest, ptrdiff_t line_size, int16_t *block)
213 /* IDCT4 on each line */
215 idct4row(block + i*8);
218 /* IDCT8 and store */
220 idctSparseColAdd_int16_8bit(dest + i, line_size, block + i);
224 void ff_simple_idct44_add(uint8_t *dest, ptrdiff_t line_size, int16_t *block)
228 /* IDCT4 on each line */
230 idct4row(block + i*8);
233 /* IDCT4 and store */
235 idct4col_add(dest + i, line_size, block + i);
239 void ff_prores_idct(int16_t *block, const int16_t *qmat)
243 for (i = 0; i < 64; i++)
246 for (i = 0; i < 8; i++)
247 idctRowCondDC_extrashift_10(block + i*8, 2);
249 for (i = 0; i < 8; i++) {
251 idctSparseCol_extrashift_10(block + i);