Line | Branch | Exec | Source |
---|---|---|---|
1 | /* | ||
2 | * Simple IDCT | ||
3 | * | ||
4 | * Copyright (c) 2001 Michael Niedermayer <michaelni@gmx.at> | ||
5 | * | ||
6 | * This file is part of FFmpeg. | ||
7 | * | ||
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. | ||
12 | * | ||
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. | ||
17 | * | ||
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 | ||
21 | */ | ||
22 | |||
23 | /** | ||
24 | * @file | ||
25 | * simpleidct in C. | ||
26 | */ | ||
27 | |||
28 | #include "libavutil/intreadwrite.h" | ||
29 | #include "mathops.h" | ||
30 | #include "simple_idct.h" | ||
31 | |||
32 | #define IN_IDCT_DEPTH 16 | ||
33 | |||
34 | #define BIT_DEPTH 8 | ||
35 | #include "simple_idct_template.c" | ||
36 | #undef BIT_DEPTH | ||
37 | |||
38 | #define BIT_DEPTH 10 | ||
39 | #include "simple_idct_template.c" | ||
40 | |||
41 | #define EXTRA_SHIFT 2 | ||
42 | #include "simple_idct_template.c" | ||
43 | |||
44 | #undef EXTRA_SHIFT | ||
45 | #undef BIT_DEPTH | ||
46 | |||
47 | #define BIT_DEPTH 12 | ||
48 | #include "simple_idct_template.c" | ||
49 | #undef BIT_DEPTH | ||
50 | #undef IN_IDCT_DEPTH | ||
51 | |||
52 | #define IN_IDCT_DEPTH 32 | ||
53 | #define BIT_DEPTH 10 | ||
54 | #include "simple_idct_template.c" | ||
55 | #undef BIT_DEPTH | ||
56 | #undef IN_IDCT_DEPTH | ||
57 | |||
58 | /* 2x4x8 idct */ | ||
59 | |||
60 | #define CN_SHIFT 12 | ||
61 | #define C_FIX(x) ((int)((x) * (1 << CN_SHIFT) + 0.5)) | ||
62 | #define C1 C_FIX(0.6532814824) | ||
63 | #define C2 C_FIX(0.2705980501) | ||
64 | |||
65 | /* row idct is multiple by 16 * sqrt(2.0), col idct4 is normalized, | ||
66 | and the butterfly must be multiplied by 0.5 * sqrt(2.0) */ | ||
67 | #define C_SHIFT (4+1+12) | ||
68 | |||
69 | 441216 | static inline void idct4col_put(uint8_t *dest, ptrdiff_t line_size, const int16_t *col) | |
70 | { | ||
71 | int c0, c1, c2, c3, a0, a1, a2, a3; | ||
72 | |||
73 | 441216 | a0 = col[8*0]; | |
74 | 441216 | a1 = col[8*2]; | |
75 | 441216 | a2 = col[8*4]; | |
76 | 441216 | a3 = col[8*6]; | |
77 | 441216 | c0 = ((a0 + a2) * (1 << CN_SHIFT - 1)) + (1 << (C_SHIFT - 1)); | |
78 | 441216 | c2 = ((a0 - a2) * (1 << CN_SHIFT - 1)) + (1 << (C_SHIFT - 1)); | |
79 | 441216 | c1 = a1 * C1 + a3 * C2; | |
80 | 441216 | c3 = a1 * C2 - a3 * C1; | |
81 | 441216 | dest[0] = av_clip_uint8((c0 + c1) >> C_SHIFT); | |
82 | 441216 | dest += line_size; | |
83 | 441216 | dest[0] = av_clip_uint8((c2 + c3) >> C_SHIFT); | |
84 | 441216 | dest += line_size; | |
85 | 441216 | dest[0] = av_clip_uint8((c2 - c3) >> C_SHIFT); | |
86 | 441216 | dest += line_size; | |
87 | 441216 | dest[0] = av_clip_uint8((c0 - c1) >> C_SHIFT); | |
88 | 441216 | } | |
89 | |||
90 | #define BF(k) \ | ||
91 | {\ | ||
92 | int a0, a1;\ | ||
93 | a0 = ptr[k];\ | ||
94 | a1 = ptr[8 + k];\ | ||
95 | ptr[k] = a0 + a1;\ | ||
96 | ptr[8 + k] = a0 - a1;\ | ||
97 | } | ||
98 | |||
99 | /* only used by DV codec. The input must be interlaced. 128 is added | ||
100 | to the pixels before clamping to avoid systematic error | ||
101 | (1024*sqrt(2)) offset would be needed otherwise. */ | ||
102 | /* XXX: I think a 1.0/sqrt(2) normalization should be needed to | ||
103 | compensate the extra butterfly stage - I don't have the full DV | ||
104 | specification */ | ||
105 | 27576 | void ff_simple_idct248_put(uint8_t *dest, ptrdiff_t line_size, int16_t *block) | |
106 | { | ||
107 | int i; | ||
108 | int16_t *ptr; | ||
109 | |||
110 | /* butterfly */ | ||
111 | 27576 | ptr = block; | |
112 |
2/2✓ Branch 0 taken 110304 times.
✓ Branch 1 taken 27576 times.
|
137880 | for(i=0;i<4;i++) { |
113 | 110304 | BF(0); | |
114 | 110304 | BF(1); | |
115 | 110304 | BF(2); | |
116 | 110304 | BF(3); | |
117 | 110304 | BF(4); | |
118 | 110304 | BF(5); | |
119 | 110304 | BF(6); | |
120 | 110304 | BF(7); | |
121 | 110304 | ptr += 2 * 8; | |
122 | } | ||
123 | |||
124 | /* IDCT8 on each line */ | ||
125 |
2/2✓ Branch 0 taken 220608 times.
✓ Branch 1 taken 27576 times.
|
248184 | for(i=0; i<8; i++) { |
126 | 220608 | idctRowCondDC_int16_8bit(block + i*8, 0); | |
127 | } | ||
128 | |||
129 | /* IDCT4 and store */ | ||
130 |
2/2✓ Branch 0 taken 220608 times.
✓ Branch 1 taken 27576 times.
|
248184 | for(i=0;i<8;i++) { |
131 | 220608 | idct4col_put(dest + i, 2 * line_size, block + i); | |
132 | 220608 | idct4col_put(dest + line_size + i, 2 * line_size, block + 8 + i); | |
133 | } | ||
134 | 27576 | } | |
135 | |||
136 | /* 8x4 & 4x8 WMV2 IDCT */ | ||
137 | #undef CN_SHIFT | ||
138 | #undef C_SHIFT | ||
139 | #undef C_FIX | ||
140 | #undef C1 | ||
141 | #undef C2 | ||
142 | #define CN_SHIFT 12 | ||
143 | #define C_FIX(x) ((int)((x) * M_SQRT2 * (1 << CN_SHIFT) + 0.5)) | ||
144 | #define C1 C_FIX(0.6532814824) | ||
145 | #define C2 C_FIX(0.2705980501) | ||
146 | #define C3 C_FIX(0.5) | ||
147 | #define C_SHIFT (4+1+12) | ||
148 | 263504 | static inline void idct4col_add(uint8_t *dest, ptrdiff_t line_size, const int16_t *col) | |
149 | { | ||
150 | int c0, c1, c2, c3, a0, a1, a2, a3; | ||
151 | |||
152 | 263504 | a0 = col[8*0]; | |
153 | 263504 | a1 = col[8*1]; | |
154 | 263504 | a2 = col[8*2]; | |
155 | 263504 | a3 = col[8*3]; | |
156 | 263504 | c0 = (a0 + a2)*C3 + (1 << (C_SHIFT - 1)); | |
157 | 263504 | c2 = (a0 - a2)*C3 + (1 << (C_SHIFT - 1)); | |
158 | 263504 | c1 = a1 * C1 + a3 * C2; | |
159 | 263504 | c3 = a1 * C2 - a3 * C1; | |
160 | 263504 | dest[0] = av_clip_uint8(dest[0] + ((c0 + c1) >> C_SHIFT)); | |
161 | 263504 | dest += line_size; | |
162 | 263504 | dest[0] = av_clip_uint8(dest[0] + ((c2 + c3) >> C_SHIFT)); | |
163 | 263504 | dest += line_size; | |
164 | 263504 | dest[0] = av_clip_uint8(dest[0] + ((c2 - c3) >> C_SHIFT)); | |
165 | 263504 | dest += line_size; | |
166 | 263504 | dest[0] = av_clip_uint8(dest[0] + ((c0 - c1) >> C_SHIFT)); | |
167 | 263504 | } | |
168 | |||
169 | #define RN_SHIFT 15 | ||
170 | #define R_FIX(x) ((int)((x) * M_SQRT2 * (1 << RN_SHIFT) + 0.5)) | ||
171 | #define R1 R_FIX(0.6532814824) | ||
172 | #define R2 R_FIX(0.2705980501) | ||
173 | #define R3 R_FIX(0.5) | ||
174 | #define R_SHIFT 11 | ||
175 | 239648 | static inline void idct4row(int16_t *row) | |
176 | { | ||
177 | unsigned c0, c1, c2, c3; | ||
178 | int a0, a1, a2, a3; | ||
179 | |||
180 | 239648 | a0 = row[0]; | |
181 | 239648 | a1 = row[1]; | |
182 | 239648 | a2 = row[2]; | |
183 | 239648 | a3 = row[3]; | |
184 | 239648 | c0 = (a0 + a2)*R3 + (1 << (R_SHIFT - 1)); | |
185 | 239648 | c2 = (a0 - a2)*R3 + (1 << (R_SHIFT - 1)); | |
186 | 239648 | c1 = a1 * R1 + a3 * R2; | |
187 | 239648 | c3 = a1 * R2 - a3 * R1; | |
188 | 239648 | row[0]= (c0 + c1) >> R_SHIFT; | |
189 | 239648 | row[1]= (c2 + c3) >> R_SHIFT; | |
190 | 239648 | row[2]= (c2 - c3) >> R_SHIFT; | |
191 | 239648 | row[3]= (c0 - c1) >> R_SHIFT; | |
192 | 239648 | } | |
193 | |||
194 | 32938 | void ff_simple_idct84_add(uint8_t *dest, ptrdiff_t line_size, int16_t *block) | |
195 | { | ||
196 | int i; | ||
197 | |||
198 | /* IDCT8 on each line */ | ||
199 |
2/2✓ Branch 0 taken 131752 times.
✓ Branch 1 taken 32938 times.
|
164690 | for(i=0; i<4; i++) { |
200 | 131752 | idctRowCondDC_int16_8bit(block + i*8, 0); | |
201 | } | ||
202 | |||
203 | /* IDCT4 and store */ | ||
204 |
2/2✓ Branch 0 taken 263504 times.
✓ Branch 1 taken 32938 times.
|
296442 | for(i=0;i<8;i++) { |
205 | 263504 | idct4col_add(dest + i, line_size, block + i); | |
206 | } | ||
207 | 32938 | } | |
208 | |||
209 | 29956 | void ff_simple_idct48_add(uint8_t *dest, ptrdiff_t line_size, int16_t *block) | |
210 | { | ||
211 | int i; | ||
212 | |||
213 | /* IDCT4 on each line */ | ||
214 |
2/2✓ Branch 0 taken 239648 times.
✓ Branch 1 taken 29956 times.
|
269604 | for(i=0; i<8; i++) { |
215 | 239648 | idct4row(block + i*8); | |
216 | } | ||
217 | |||
218 | /* IDCT8 and store */ | ||
219 |
2/2✓ Branch 0 taken 119824 times.
✓ Branch 1 taken 29956 times.
|
149780 | for(i=0; i<4; i++){ |
220 | 119824 | idctSparseColAdd_int16_8bit(dest + i, line_size, block + i); | |
221 | } | ||
222 | 29956 | } | |
223 | |||
224 | ✗ | void ff_simple_idct44_add(uint8_t *dest, ptrdiff_t line_size, int16_t *block) | |
225 | { | ||
226 | int i; | ||
227 | |||
228 | /* IDCT4 on each line */ | ||
229 | ✗ | for(i=0; i<4; i++) { | |
230 | ✗ | idct4row(block + i*8); | |
231 | } | ||
232 | |||
233 | /* IDCT4 and store */ | ||
234 | ✗ | for(i=0; i<4; i++){ | |
235 | ✗ | idct4col_add(dest + i, line_size, block + i); | |
236 | } | ||
237 | ✗ | } | |
238 | |||
239 | 2740512 | void ff_prores_idct_10(int16_t *block, const int16_t *qmat) | |
240 | { | ||
241 | int i; | ||
242 | |||
243 |
2/2✓ Branch 0 taken 175392768 times.
✓ Branch 1 taken 2740512 times.
|
178133280 | for (i = 0; i < 64; i++) |
244 | 175392768 | block[i] *= qmat[i]; | |
245 | |||
246 |
2/2✓ Branch 0 taken 21924096 times.
✓ Branch 1 taken 2740512 times.
|
24664608 | for (i = 0; i < 8; i++) |
247 | 21924096 | idctRowCondDC_extrashift_10(block + i*8, 2); | |
248 | |||
249 |
2/2✓ Branch 0 taken 21924096 times.
✓ Branch 1 taken 2740512 times.
|
24664608 | for (i = 0; i < 8; i++) { |
250 | 21924096 | block[i] += 8192; | |
251 | 21924096 | idctSparseCol_extrashift_10(block + i); | |
252 | } | ||
253 | 2740512 | } | |
254 | |||
255 | 2446164 | void ff_prores_idct_12(int16_t *block, const int16_t *qmat) | |
256 | { | ||
257 | int i; | ||
258 | |||
259 |
2/2✓ Branch 0 taken 156554496 times.
✓ Branch 1 taken 2446164 times.
|
159000660 | for (i = 0; i < 64; i++) |
260 | 156554496 | block[i] *= qmat[i]; | |
261 | |||
262 |
2/2✓ Branch 0 taken 19569312 times.
✓ Branch 1 taken 2446164 times.
|
22015476 | for (i = 0; i < 8; i++) |
263 | 19569312 | idctRowCondDC_int16_12bit(block + i*8, 0); | |
264 | |||
265 |
2/2✓ Branch 0 taken 19569312 times.
✓ Branch 1 taken 2446164 times.
|
22015476 | for (i = 0; i < 8; i++) { |
266 | 19569312 | block[i] += 8192; | |
267 | 19569312 | idctSparseCol_int16_12bit(block + i); | |
268 | } | ||
269 | 2446164 | } | |
270 |