Mercurial > libavcodec.hg
annotate vc1dsp.c @ 9440:daee921fb6bb libavcodec
VC1: add and use avg_no_rnd chroma MC functions
author | conrad |
---|---|
date | Tue, 14 Apr 2009 23:56:10 +0000 |
parents | 8aa2e86549cd |
children | a91f60938763 |
rev | line source |
---|---|
3526 | 1 /* |
2 * VC-1 and WMV3 decoder - DSP functions | |
3 * Copyright (c) 2006 Konstantin Shishkov | |
4 * | |
3947
c8c591fe26f8
Change license headers to say 'FFmpeg' instead of 'this program/this library'
diego
parents:
3665
diff
changeset
|
5 * This file is part of FFmpeg. |
c8c591fe26f8
Change license headers to say 'FFmpeg' instead of 'this program/this library'
diego
parents:
3665
diff
changeset
|
6 * |
c8c591fe26f8
Change license headers to say 'FFmpeg' instead of 'this program/this library'
diego
parents:
3665
diff
changeset
|
7 * FFmpeg is free software; you can redistribute it and/or |
3526 | 8 * modify it under the terms of the GNU Lesser General Public |
9 * License as published by the Free Software Foundation; either | |
3947
c8c591fe26f8
Change license headers to say 'FFmpeg' instead of 'this program/this library'
diego
parents:
3665
diff
changeset
|
10 * version 2.1 of the License, or (at your option) any later version. |
3526 | 11 * |
3947
c8c591fe26f8
Change license headers to say 'FFmpeg' instead of 'this program/this library'
diego
parents:
3665
diff
changeset
|
12 * FFmpeg is distributed in the hope that it will be useful, |
3526 | 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 | |
3947
c8c591fe26f8
Change license headers to say 'FFmpeg' instead of 'this program/this library'
diego
parents:
3665
diff
changeset
|
18 * License along with FFmpeg; if not, write to the Free Software |
3526 | 19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA |
20 */ | |
21 | |
22 /** | |
8718
e9d9d946f213
Use full internal pathname in doxygen @file directives.
diego
parents:
8686
diff
changeset
|
23 * @file libavcodec/vc1dsp.c |
3526 | 24 * VC-1 and WMV3 decoder |
25 * | |
26 */ | |
27 | |
28 #include "dsputil.h" | |
29 | |
30 | |
4210 | 31 /** Apply overlap transform to horizontal edge |
3526 | 32 */ |
4239 | 33 static void vc1_v_overlap_c(uint8_t* src, int stride) |
3526 | 34 { |
35 int i; | |
36 int a, b, c, d; | |
4211 | 37 int d1, d2; |
4239 | 38 int rnd = 1; |
3526 | 39 for(i = 0; i < 8; i++) { |
40 a = src[-2*stride]; | |
41 b = src[-stride]; | |
42 c = src[0]; | |
43 d = src[stride]; | |
4211 | 44 d1 = (a - d + 3 + rnd) >> 3; |
45 d2 = (a - d + b - c + 4 - rnd) >> 3; | |
3526 | 46 |
4211 | 47 src[-2*stride] = a - d1; |
8686 | 48 src[-stride] = av_clip_uint8(b - d2); |
49 src[0] = av_clip_uint8(c + d2); | |
4211 | 50 src[stride] = d + d1; |
3526 | 51 src++; |
4239 | 52 rnd = !rnd; |
3526 | 53 } |
54 } | |
55 | |
4210 | 56 /** Apply overlap transform to vertical edge |
3526 | 57 */ |
4239 | 58 static void vc1_h_overlap_c(uint8_t* src, int stride) |
3526 | 59 { |
60 int i; | |
61 int a, b, c, d; | |
4211 | 62 int d1, d2; |
4239 | 63 int rnd = 1; |
3526 | 64 for(i = 0; i < 8; i++) { |
65 a = src[-2]; | |
66 b = src[-1]; | |
67 c = src[0]; | |
68 d = src[1]; | |
4211 | 69 d1 = (a - d + 3 + rnd) >> 3; |
70 d2 = (a - d + b - c + 4 - rnd) >> 3; | |
3526 | 71 |
4211 | 72 src[-2] = a - d1; |
8686 | 73 src[-1] = av_clip_uint8(b - d2); |
74 src[0] = av_clip_uint8(c + d2); | |
4211 | 75 src[1] = d + d1; |
3526 | 76 src += stride; |
4239 | 77 rnd = !rnd; |
3526 | 78 } |
79 } | |
80 | |
81 | |
82 /** Do inverse transform on 8x8 block | |
83 */ | |
84 static void vc1_inv_trans_8x8_c(DCTELEM block[64]) | |
85 { | |
86 int i; | |
87 register int t1,t2,t3,t4,t5,t6,t7,t8; | |
88 DCTELEM *src, *dst; | |
89 | |
90 src = block; | |
91 dst = block; | |
92 for(i = 0; i < 8; i++){ | |
6157 | 93 t1 = 12 * (src[0] + src[4]) + 4; |
94 t2 = 12 * (src[0] - src[4]) + 4; | |
3526 | 95 t3 = 16 * src[2] + 6 * src[6]; |
96 t4 = 6 * src[2] - 16 * src[6]; | |
97 | |
98 t5 = t1 + t3; | |
99 t6 = t2 + t4; | |
100 t7 = t2 - t4; | |
101 t8 = t1 - t3; | |
102 | |
103 t1 = 16 * src[1] + 15 * src[3] + 9 * src[5] + 4 * src[7]; | |
104 t2 = 15 * src[1] - 4 * src[3] - 16 * src[5] - 9 * src[7]; | |
105 t3 = 9 * src[1] - 16 * src[3] + 4 * src[5] + 15 * src[7]; | |
106 t4 = 4 * src[1] - 9 * src[3] + 15 * src[5] - 16 * src[7]; | |
107 | |
6157 | 108 dst[0] = (t5 + t1) >> 3; |
109 dst[1] = (t6 + t2) >> 3; | |
110 dst[2] = (t7 + t3) >> 3; | |
111 dst[3] = (t8 + t4) >> 3; | |
112 dst[4] = (t8 - t4) >> 3; | |
113 dst[5] = (t7 - t3) >> 3; | |
114 dst[6] = (t6 - t2) >> 3; | |
115 dst[7] = (t5 - t1) >> 3; | |
3526 | 116 |
117 src += 8; | |
118 dst += 8; | |
119 } | |
120 | |
121 src = block; | |
122 dst = block; | |
123 for(i = 0; i < 8; i++){ | |
6157 | 124 t1 = 12 * (src[ 0] + src[32]) + 64; |
125 t2 = 12 * (src[ 0] - src[32]) + 64; | |
3526 | 126 t3 = 16 * src[16] + 6 * src[48]; |
127 t4 = 6 * src[16] - 16 * src[48]; | |
128 | |
129 t5 = t1 + t3; | |
130 t6 = t2 + t4; | |
131 t7 = t2 - t4; | |
132 t8 = t1 - t3; | |
133 | |
134 t1 = 16 * src[ 8] + 15 * src[24] + 9 * src[40] + 4 * src[56]; | |
135 t2 = 15 * src[ 8] - 4 * src[24] - 16 * src[40] - 9 * src[56]; | |
136 t3 = 9 * src[ 8] - 16 * src[24] + 4 * src[40] + 15 * src[56]; | |
137 t4 = 4 * src[ 8] - 9 * src[24] + 15 * src[40] - 16 * src[56]; | |
138 | |
6157 | 139 dst[ 0] = (t5 + t1) >> 7; |
140 dst[ 8] = (t6 + t2) >> 7; | |
141 dst[16] = (t7 + t3) >> 7; | |
142 dst[24] = (t8 + t4) >> 7; | |
143 dst[32] = (t8 - t4 + 1) >> 7; | |
144 dst[40] = (t7 - t3 + 1) >> 7; | |
145 dst[48] = (t6 - t2 + 1) >> 7; | |
146 dst[56] = (t5 - t1 + 1) >> 7; | |
3526 | 147 |
148 src++; | |
149 dst++; | |
150 } | |
151 } | |
152 | |
153 /** Do inverse transform on 8x4 part of block | |
154 */ | |
5997
90de28dfd8d6
Switch VC-1 decoder to output decoded residual immediately.
kostya
parents:
5416
diff
changeset
|
155 static void vc1_inv_trans_8x4_c(uint8_t *dest, int linesize, DCTELEM *block) |
3526 | 156 { |
157 int i; | |
158 register int t1,t2,t3,t4,t5,t6,t7,t8; | |
159 DCTELEM *src, *dst; | |
5997
90de28dfd8d6
Switch VC-1 decoder to output decoded residual immediately.
kostya
parents:
5416
diff
changeset
|
160 const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; |
3526 | 161 |
5997
90de28dfd8d6
Switch VC-1 decoder to output decoded residual immediately.
kostya
parents:
5416
diff
changeset
|
162 src = block; |
90de28dfd8d6
Switch VC-1 decoder to output decoded residual immediately.
kostya
parents:
5416
diff
changeset
|
163 dst = block; |
3526 | 164 for(i = 0; i < 4; i++){ |
6157 | 165 t1 = 12 * (src[0] + src[4]) + 4; |
166 t2 = 12 * (src[0] - src[4]) + 4; | |
3526 | 167 t3 = 16 * src[2] + 6 * src[6]; |
168 t4 = 6 * src[2] - 16 * src[6]; | |
169 | |
170 t5 = t1 + t3; | |
171 t6 = t2 + t4; | |
172 t7 = t2 - t4; | |
173 t8 = t1 - t3; | |
174 | |
175 t1 = 16 * src[1] + 15 * src[3] + 9 * src[5] + 4 * src[7]; | |
176 t2 = 15 * src[1] - 4 * src[3] - 16 * src[5] - 9 * src[7]; | |
177 t3 = 9 * src[1] - 16 * src[3] + 4 * src[5] + 15 * src[7]; | |
178 t4 = 4 * src[1] - 9 * src[3] + 15 * src[5] - 16 * src[7]; | |
179 | |
6157 | 180 dst[0] = (t5 + t1) >> 3; |
181 dst[1] = (t6 + t2) >> 3; | |
182 dst[2] = (t7 + t3) >> 3; | |
183 dst[3] = (t8 + t4) >> 3; | |
184 dst[4] = (t8 - t4) >> 3; | |
185 dst[5] = (t7 - t3) >> 3; | |
186 dst[6] = (t6 - t2) >> 3; | |
187 dst[7] = (t5 - t1) >> 3; | |
3526 | 188 |
189 src += 8; | |
190 dst += 8; | |
191 } | |
192 | |
5997
90de28dfd8d6
Switch VC-1 decoder to output decoded residual immediately.
kostya
parents:
5416
diff
changeset
|
193 src = block; |
3526 | 194 for(i = 0; i < 8; i++){ |
6157 | 195 t1 = 17 * (src[ 0] + src[16]) + 64; |
196 t2 = 17 * (src[ 0] - src[16]) + 64; | |
6158 | 197 t3 = 22 * src[ 8] + 10 * src[24]; |
198 t4 = 22 * src[24] - 10 * src[ 8]; | |
3526 | 199 |
6158 | 200 dest[0*linesize] = cm[dest[0*linesize] + ((t1 + t3) >> 7)]; |
201 dest[1*linesize] = cm[dest[1*linesize] + ((t2 - t4) >> 7)]; | |
202 dest[2*linesize] = cm[dest[2*linesize] + ((t2 + t4) >> 7)]; | |
203 dest[3*linesize] = cm[dest[3*linesize] + ((t1 - t3) >> 7)]; | |
3526 | 204 |
205 src ++; | |
5997
90de28dfd8d6
Switch VC-1 decoder to output decoded residual immediately.
kostya
parents:
5416
diff
changeset
|
206 dest++; |
3526 | 207 } |
208 } | |
209 | |
210 /** Do inverse transform on 4x8 parts of block | |
211 */ | |
5997
90de28dfd8d6
Switch VC-1 decoder to output decoded residual immediately.
kostya
parents:
5416
diff
changeset
|
212 static void vc1_inv_trans_4x8_c(uint8_t *dest, int linesize, DCTELEM *block) |
3526 | 213 { |
214 int i; | |
215 register int t1,t2,t3,t4,t5,t6,t7,t8; | |
216 DCTELEM *src, *dst; | |
5997
90de28dfd8d6
Switch VC-1 decoder to output decoded residual immediately.
kostya
parents:
5416
diff
changeset
|
217 const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; |
3526 | 218 |
5997
90de28dfd8d6
Switch VC-1 decoder to output decoded residual immediately.
kostya
parents:
5416
diff
changeset
|
219 src = block; |
90de28dfd8d6
Switch VC-1 decoder to output decoded residual immediately.
kostya
parents:
5416
diff
changeset
|
220 dst = block; |
3526 | 221 for(i = 0; i < 8; i++){ |
6157 | 222 t1 = 17 * (src[0] + src[2]) + 4; |
223 t2 = 17 * (src[0] - src[2]) + 4; | |
6158 | 224 t3 = 22 * src[1] + 10 * src[3]; |
225 t4 = 22 * src[3] - 10 * src[1]; | |
3526 | 226 |
6158 | 227 dst[0] = (t1 + t3) >> 3; |
228 dst[1] = (t2 - t4) >> 3; | |
229 dst[2] = (t2 + t4) >> 3; | |
230 dst[3] = (t1 - t3) >> 3; | |
3526 | 231 |
232 src += 8; | |
233 dst += 8; | |
234 } | |
235 | |
5997
90de28dfd8d6
Switch VC-1 decoder to output decoded residual immediately.
kostya
parents:
5416
diff
changeset
|
236 src = block; |
3526 | 237 for(i = 0; i < 4; i++){ |
6157 | 238 t1 = 12 * (src[ 0] + src[32]) + 64; |
239 t2 = 12 * (src[ 0] - src[32]) + 64; | |
3526 | 240 t3 = 16 * src[16] + 6 * src[48]; |
241 t4 = 6 * src[16] - 16 * src[48]; | |
242 | |
243 t5 = t1 + t3; | |
244 t6 = t2 + t4; | |
245 t7 = t2 - t4; | |
246 t8 = t1 - t3; | |
247 | |
248 t1 = 16 * src[ 8] + 15 * src[24] + 9 * src[40] + 4 * src[56]; | |
249 t2 = 15 * src[ 8] - 4 * src[24] - 16 * src[40] - 9 * src[56]; | |
250 t3 = 9 * src[ 8] - 16 * src[24] + 4 * src[40] + 15 * src[56]; | |
251 t4 = 4 * src[ 8] - 9 * src[24] + 15 * src[40] - 16 * src[56]; | |
252 | |
6157 | 253 dest[0*linesize] = cm[dest[0*linesize] + ((t5 + t1) >> 7)]; |
254 dest[1*linesize] = cm[dest[1*linesize] + ((t6 + t2) >> 7)]; | |
255 dest[2*linesize] = cm[dest[2*linesize] + ((t7 + t3) >> 7)]; | |
256 dest[3*linesize] = cm[dest[3*linesize] + ((t8 + t4) >> 7)]; | |
257 dest[4*linesize] = cm[dest[4*linesize] + ((t8 - t4 + 1) >> 7)]; | |
258 dest[5*linesize] = cm[dest[5*linesize] + ((t7 - t3 + 1) >> 7)]; | |
259 dest[6*linesize] = cm[dest[6*linesize] + ((t6 - t2 + 1) >> 7)]; | |
260 dest[7*linesize] = cm[dest[7*linesize] + ((t5 - t1 + 1) >> 7)]; | |
3526 | 261 |
5997
90de28dfd8d6
Switch VC-1 decoder to output decoded residual immediately.
kostya
parents:
5416
diff
changeset
|
262 src ++; |
90de28dfd8d6
Switch VC-1 decoder to output decoded residual immediately.
kostya
parents:
5416
diff
changeset
|
263 dest++; |
3526 | 264 } |
265 } | |
266 | |
267 /** Do inverse transform on 4x4 part of block | |
268 */ | |
5997
90de28dfd8d6
Switch VC-1 decoder to output decoded residual immediately.
kostya
parents:
5416
diff
changeset
|
269 static void vc1_inv_trans_4x4_c(uint8_t *dest, int linesize, DCTELEM *block) |
3526 | 270 { |
271 int i; | |
6158 | 272 register int t1,t2,t3,t4; |
3526 | 273 DCTELEM *src, *dst; |
5997
90de28dfd8d6
Switch VC-1 decoder to output decoded residual immediately.
kostya
parents:
5416
diff
changeset
|
274 const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; |
3526 | 275 |
5997
90de28dfd8d6
Switch VC-1 decoder to output decoded residual immediately.
kostya
parents:
5416
diff
changeset
|
276 src = block; |
90de28dfd8d6
Switch VC-1 decoder to output decoded residual immediately.
kostya
parents:
5416
diff
changeset
|
277 dst = block; |
3526 | 278 for(i = 0; i < 4; i++){ |
6157 | 279 t1 = 17 * (src[0] + src[2]) + 4; |
280 t2 = 17 * (src[0] - src[2]) + 4; | |
6158 | 281 t3 = 22 * src[1] + 10 * src[3]; |
282 t4 = 22 * src[3] - 10 * src[1]; | |
3526 | 283 |
6158 | 284 dst[0] = (t1 + t3) >> 3; |
285 dst[1] = (t2 - t4) >> 3; | |
286 dst[2] = (t2 + t4) >> 3; | |
287 dst[3] = (t1 - t3) >> 3; | |
3526 | 288 |
289 src += 8; | |
290 dst += 8; | |
291 } | |
292 | |
5997
90de28dfd8d6
Switch VC-1 decoder to output decoded residual immediately.
kostya
parents:
5416
diff
changeset
|
293 src = block; |
3526 | 294 for(i = 0; i < 4; i++){ |
6157 | 295 t1 = 17 * (src[ 0] + src[16]) + 64; |
296 t2 = 17 * (src[ 0] - src[16]) + 64; | |
6158 | 297 t3 = 22 * src[ 8] + 10 * src[24]; |
298 t4 = 22 * src[24] - 10 * src[ 8]; | |
3526 | 299 |
6158 | 300 dest[0*linesize] = cm[dest[0*linesize] + ((t1 + t3) >> 7)]; |
301 dest[1*linesize] = cm[dest[1*linesize] + ((t2 - t4) >> 7)]; | |
302 dest[2*linesize] = cm[dest[2*linesize] + ((t2 + t4) >> 7)]; | |
303 dest[3*linesize] = cm[dest[3*linesize] + ((t1 - t3) >> 7)]; | |
3526 | 304 |
305 src ++; | |
5997
90de28dfd8d6
Switch VC-1 decoder to output decoded residual immediately.
kostya
parents:
5416
diff
changeset
|
306 dest++; |
3526 | 307 } |
308 } | |
309 | |
310 /* motion compensation functions */ | |
5416 | 311 /** Filter in case of 2 filters */ |
312 #define VC1_MSPEL_FILTER_16B(DIR, TYPE) \ | |
313 static av_always_inline int vc1_mspel_ ## DIR ## _filter_16bits(const TYPE *src, int stride, int mode) \ | |
314 { \ | |
315 switch(mode){ \ | |
316 case 0: /* no shift - should not occur */ \ | |
317 return 0; \ | |
318 case 1: /* 1/4 shift */ \ | |
319 return -4*src[-stride] + 53*src[0] + 18*src[stride] - 3*src[stride*2]; \ | |
320 case 2: /* 1/2 shift */ \ | |
321 return -src[-stride] + 9*src[0] + 9*src[stride] - src[stride*2]; \ | |
322 case 3: /* 3/4 shift */ \ | |
323 return -3*src[-stride] + 18*src[0] + 53*src[stride] - 4*src[stride*2]; \ | |
324 } \ | |
325 return 0; /* should not occur */ \ | |
326 } | |
327 | |
328 VC1_MSPEL_FILTER_16B(ver, uint8_t); | |
329 VC1_MSPEL_FILTER_16B(hor, int16_t); | |
330 | |
3526 | 331 |
332 /** Filter used to interpolate fractional pel values | |
333 */ | |
4283
d6f83e2f8804
rename always_inline to av_always_inline and move to common.h
mru
parents:
4239
diff
changeset
|
334 static av_always_inline int vc1_mspel_filter(const uint8_t *src, int stride, int mode, int r) |
3526 | 335 { |
336 switch(mode){ | |
337 case 0: //no shift | |
338 return src[0]; | |
339 case 1: // 1/4 shift | |
340 return (-4*src[-stride] + 53*src[0] + 18*src[stride] - 3*src[stride*2] + 32 - r) >> 6; | |
341 case 2: // 1/2 shift | |
342 return (-src[-stride] + 9*src[0] + 9*src[stride] - src[stride*2] + 8 - r) >> 4; | |
343 case 3: // 3/4 shift | |
344 return (-3*src[-stride] + 18*src[0] + 53*src[stride] - 4*src[stride*2] + 32 - r) >> 6; | |
345 } | |
346 return 0; //should not occur | |
347 } | |
348 | |
349 /** Function used to do motion compensation with bicubic interpolation | |
350 */ | |
9437 | 351 #define VC1_MSPEL_MC(OP, OPNAME)\ |
352 static void OPNAME ## vc1_mspel_mc(uint8_t *dst, const uint8_t *src, int stride, int hmode, int vmode, int rnd)\ | |
353 {\ | |
354 int i, j;\ | |
355 \ | |
356 if (vmode) { /* Horizontal filter to apply */\ | |
357 int r;\ | |
358 \ | |
359 if (hmode) { /* Vertical filter to apply, output to tmp */\ | |
360 static const int shift_value[] = { 0, 5, 1, 5 };\ | |
361 int shift = (shift_value[hmode]+shift_value[vmode])>>1;\ | |
362 int16_t tmp[11*8], *tptr = tmp;\ | |
363 \ | |
364 r = (1<<(shift-1)) + rnd-1;\ | |
365 \ | |
366 src -= 1;\ | |
367 for(j = 0; j < 8; j++) {\ | |
368 for(i = 0; i < 11; i++)\ | |
369 tptr[i] = (vc1_mspel_ver_filter_16bits(src + i, stride, vmode)+r)>>shift;\ | |
370 src += stride;\ | |
371 tptr += 11;\ | |
372 }\ | |
373 \ | |
374 r = 64-rnd;\ | |
375 tptr = tmp+1;\ | |
376 for(j = 0; j < 8; j++) {\ | |
377 for(i = 0; i < 8; i++)\ | |
378 OP(dst[i], (vc1_mspel_hor_filter_16bits(tptr + i, 1, hmode)+r)>>7);\ | |
379 dst += stride;\ | |
380 tptr += 11;\ | |
381 }\ | |
382 \ | |
383 return;\ | |
384 }\ | |
385 else { /* No horizontal filter, output 8 lines to dst */\ | |
386 r = 1-rnd;\ | |
387 \ | |
388 for(j = 0; j < 8; j++) {\ | |
389 for(i = 0; i < 8; i++)\ | |
390 OP(dst[i], vc1_mspel_filter(src + i, stride, vmode, r));\ | |
391 src += stride;\ | |
392 dst += stride;\ | |
393 }\ | |
394 return;\ | |
395 }\ | |
396 }\ | |
397 \ | |
398 /* Horizontal mode with no vertical mode */\ | |
399 for(j = 0; j < 8; j++) {\ | |
400 for(i = 0; i < 8; i++)\ | |
401 OP(dst[i], vc1_mspel_filter(src + i, 1, hmode, rnd));\ | |
402 dst += stride;\ | |
403 src += stride;\ | |
404 }\ | |
405 } | |
3526 | 406 |
9437 | 407 #define op_put(a, b) a = av_clip_uint8(b) |
408 #define op_avg(a, b) a = (a + av_clip_uint8(b) + 1) >> 1 | |
5416 | 409 |
9437 | 410 VC1_MSPEL_MC(op_put, put_) |
411 VC1_MSPEL_MC(op_avg, avg_) | |
3526 | 412 |
413 /* pixel functions - really are entry points to vc1_mspel_mc */ | |
414 | |
415 /* this one is defined in dsputil.c */ | |
416 void ff_put_vc1_mspel_mc00_c(uint8_t *dst, const uint8_t *src, int stride, int rnd); | |
9437 | 417 void ff_avg_vc1_mspel_mc00_c(uint8_t *dst, const uint8_t *src, int stride, int rnd); |
3526 | 418 |
5252 | 419 #define PUT_VC1_MSPEL(a, b)\ |
420 static void put_vc1_mspel_mc ## a ## b ##_c(uint8_t *dst, const uint8_t *src, int stride, int rnd) { \ | |
9437 | 421 put_vc1_mspel_mc(dst, src, stride, a, b, rnd); \ |
422 }\ | |
423 static void avg_vc1_mspel_mc ## a ## b ##_c(uint8_t *dst, const uint8_t *src, int stride, int rnd) { \ | |
424 avg_vc1_mspel_mc(dst, src, stride, a, b, rnd); \ | |
3526 | 425 } |
426 | |
5252 | 427 PUT_VC1_MSPEL(1, 0) |
428 PUT_VC1_MSPEL(2, 0) | |
429 PUT_VC1_MSPEL(3, 0) | |
3526 | 430 |
5252 | 431 PUT_VC1_MSPEL(0, 1) |
432 PUT_VC1_MSPEL(1, 1) | |
433 PUT_VC1_MSPEL(2, 1) | |
434 PUT_VC1_MSPEL(3, 1) | |
3526 | 435 |
5252 | 436 PUT_VC1_MSPEL(0, 2) |
437 PUT_VC1_MSPEL(1, 2) | |
438 PUT_VC1_MSPEL(2, 2) | |
439 PUT_VC1_MSPEL(3, 2) | |
3526 | 440 |
5252 | 441 PUT_VC1_MSPEL(0, 3) |
442 PUT_VC1_MSPEL(1, 3) | |
443 PUT_VC1_MSPEL(2, 3) | |
444 PUT_VC1_MSPEL(3, 3) | |
3526 | 445 |
446 void ff_vc1dsp_init(DSPContext* dsp, AVCodecContext *avctx) { | |
447 dsp->vc1_inv_trans_8x8 = vc1_inv_trans_8x8_c; | |
448 dsp->vc1_inv_trans_4x8 = vc1_inv_trans_4x8_c; | |
449 dsp->vc1_inv_trans_8x4 = vc1_inv_trans_8x4_c; | |
450 dsp->vc1_inv_trans_4x4 = vc1_inv_trans_4x4_c; | |
451 dsp->vc1_h_overlap = vc1_h_overlap_c; | |
452 dsp->vc1_v_overlap = vc1_v_overlap_c; | |
453 | |
454 dsp->put_vc1_mspel_pixels_tab[ 0] = ff_put_vc1_mspel_mc00_c; | |
5251 | 455 dsp->put_vc1_mspel_pixels_tab[ 1] = put_vc1_mspel_mc10_c; |
456 dsp->put_vc1_mspel_pixels_tab[ 2] = put_vc1_mspel_mc20_c; | |
457 dsp->put_vc1_mspel_pixels_tab[ 3] = put_vc1_mspel_mc30_c; | |
458 dsp->put_vc1_mspel_pixels_tab[ 4] = put_vc1_mspel_mc01_c; | |
459 dsp->put_vc1_mspel_pixels_tab[ 5] = put_vc1_mspel_mc11_c; | |
460 dsp->put_vc1_mspel_pixels_tab[ 6] = put_vc1_mspel_mc21_c; | |
461 dsp->put_vc1_mspel_pixels_tab[ 7] = put_vc1_mspel_mc31_c; | |
462 dsp->put_vc1_mspel_pixels_tab[ 8] = put_vc1_mspel_mc02_c; | |
463 dsp->put_vc1_mspel_pixels_tab[ 9] = put_vc1_mspel_mc12_c; | |
464 dsp->put_vc1_mspel_pixels_tab[10] = put_vc1_mspel_mc22_c; | |
465 dsp->put_vc1_mspel_pixels_tab[11] = put_vc1_mspel_mc32_c; | |
466 dsp->put_vc1_mspel_pixels_tab[12] = put_vc1_mspel_mc03_c; | |
467 dsp->put_vc1_mspel_pixels_tab[13] = put_vc1_mspel_mc13_c; | |
468 dsp->put_vc1_mspel_pixels_tab[14] = put_vc1_mspel_mc23_c; | |
469 dsp->put_vc1_mspel_pixels_tab[15] = put_vc1_mspel_mc33_c; | |
9437 | 470 |
471 dsp->avg_vc1_mspel_pixels_tab[ 0] = ff_avg_vc1_mspel_mc00_c; | |
472 dsp->avg_vc1_mspel_pixels_tab[ 1] = avg_vc1_mspel_mc10_c; | |
473 dsp->avg_vc1_mspel_pixels_tab[ 2] = avg_vc1_mspel_mc20_c; | |
474 dsp->avg_vc1_mspel_pixels_tab[ 3] = avg_vc1_mspel_mc30_c; | |
475 dsp->avg_vc1_mspel_pixels_tab[ 4] = avg_vc1_mspel_mc01_c; | |
476 dsp->avg_vc1_mspel_pixels_tab[ 5] = avg_vc1_mspel_mc11_c; | |
477 dsp->avg_vc1_mspel_pixels_tab[ 6] = avg_vc1_mspel_mc21_c; | |
478 dsp->avg_vc1_mspel_pixels_tab[ 7] = avg_vc1_mspel_mc31_c; | |
479 dsp->avg_vc1_mspel_pixels_tab[ 8] = avg_vc1_mspel_mc02_c; | |
480 dsp->avg_vc1_mspel_pixels_tab[ 9] = avg_vc1_mspel_mc12_c; | |
481 dsp->avg_vc1_mspel_pixels_tab[10] = avg_vc1_mspel_mc22_c; | |
482 dsp->avg_vc1_mspel_pixels_tab[11] = avg_vc1_mspel_mc32_c; | |
483 dsp->avg_vc1_mspel_pixels_tab[12] = avg_vc1_mspel_mc03_c; | |
484 dsp->avg_vc1_mspel_pixels_tab[13] = avg_vc1_mspel_mc13_c; | |
485 dsp->avg_vc1_mspel_pixels_tab[14] = avg_vc1_mspel_mc23_c; | |
486 dsp->avg_vc1_mspel_pixels_tab[15] = avg_vc1_mspel_mc33_c; | |
3526 | 487 } |