1 |
|
/***************************************************************************** |
2 |
|
* |
3 |
|
* XVID MPEG-4 VIDEO CODEC |
4 |
|
* - Interpolation related header - |
5 |
|
* |
6 |
|
* Copyright(C) 2001-2003 Peter Ross <pross@xvid.org> |
7 |
|
* |
8 |
|
* This program is free software ; you can redistribute it and/or modify |
9 |
|
* it under the terms of the GNU General Public License as published by |
10 |
|
* the Free Software Foundation ; either version 2 of the License, or |
11 |
|
* (at your option) any later version. |
12 |
|
* |
13 |
|
* This program 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 |
16 |
|
* GNU General Public License for more details. |
17 |
|
* |
18 |
|
* You should have received a copy of the GNU General Public License |
19 |
|
* along with this program ; if not, write to the Free Software |
20 |
|
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
21 |
|
* |
22 |
|
* $Id$ |
23 |
|
* |
24 |
|
****************************************************************************/ |
25 |
|
|
26 |
|
#ifndef _INTERPOLATE8X8_H_ |
27 |
|
#define _INTERPOLATE8X8_H_ |
28 |
|
|
29 |
#include "../utils/mem_transfer.h" |
#include "../utils/mem_transfer.h" |
30 |
|
|
31 |
typedef void (INTERPOLATE8X8)(uint8_t * const dst, |
typedef void (INTERPOLATE8X8)(uint8_t * const dst, |
34 |
const uint32_t rounding); |
const uint32_t rounding); |
35 |
typedef INTERPOLATE8X8 * INTERPOLATE8X8_PTR; |
typedef INTERPOLATE8X8 * INTERPOLATE8X8_PTR; |
36 |
|
|
37 |
|
typedef void (INTERPOLATE8X8_AVG2) (uint8_t *dst, |
38 |
|
const uint8_t *src1, |
39 |
|
const uint8_t *src2, |
40 |
|
const uint32_t stride, |
41 |
|
const uint32_t rounding, |
42 |
|
const uint32_t height); |
43 |
|
typedef INTERPOLATE8X8_AVG2 *INTERPOLATE8X8_AVG2_PTR; |
44 |
|
|
45 |
|
typedef void (INTERPOLATE8X8_AVG4) (uint8_t *dst, |
46 |
|
const uint8_t *src1, |
47 |
|
const uint8_t *src2, |
48 |
|
const uint8_t *src3, |
49 |
|
const uint8_t *src4, |
50 |
|
const uint32_t stride, |
51 |
|
const uint32_t rounding); |
52 |
|
typedef INTERPOLATE8X8_AVG4 *INTERPOLATE8X8_AVG4_PTR; |
53 |
|
|
54 |
|
typedef void (INTERPOLATE_LOWPASS) (uint8_t *dst, |
55 |
|
uint8_t *src, |
56 |
|
int32_t stride, |
57 |
|
int32_t rounding); |
58 |
|
|
59 |
|
typedef INTERPOLATE_LOWPASS *INTERPOLATE_LOWPASS_PTR; |
60 |
|
|
61 |
|
typedef void (INTERPOLATE_LOWPASS_HV) (uint8_t *dst1, |
62 |
|
uint8_t *dst2, |
63 |
|
uint8_t *src, |
64 |
|
int32_t stride, |
65 |
|
int32_t rounding); |
66 |
|
|
67 |
|
typedef INTERPOLATE_LOWPASS_HV *INTERPOLATE_LOWPASS_HV_PTR; |
68 |
|
|
69 |
|
typedef void (INTERPOLATE8X8_6TAP_LOWPASS) (uint8_t *dst, |
70 |
|
uint8_t *src, |
71 |
|
int32_t stride, |
72 |
|
int32_t rounding); |
73 |
|
|
74 |
|
typedef INTERPOLATE8X8_6TAP_LOWPASS *INTERPOLATE8X8_6TAP_LOWPASS_PTR; |
75 |
|
|
76 |
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_h; |
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_h; |
77 |
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_v; |
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_v; |
78 |
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_hv; |
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_hv; |
79 |
|
|
80 |
|
extern INTERPOLATE8X8_AVG2_PTR interpolate8x8_avg2; |
81 |
|
extern INTERPOLATE8X8_AVG4_PTR interpolate8x8_avg4; |
82 |
|
|
83 |
|
extern INTERPOLATE_LOWPASS_PTR interpolate8x8_lowpass_h; |
84 |
|
extern INTERPOLATE_LOWPASS_PTR interpolate8x8_lowpass_v; |
85 |
|
|
86 |
|
extern INTERPOLATE_LOWPASS_PTR interpolate16x16_lowpass_h; |
87 |
|
extern INTERPOLATE_LOWPASS_PTR interpolate16x16_lowpass_v; |
88 |
|
|
89 |
|
extern INTERPOLATE_LOWPASS_HV_PTR interpolate8x8_lowpass_hv; |
90 |
|
extern INTERPOLATE_LOWPASS_HV_PTR interpolate16x16_lowpass_hv; |
91 |
|
|
92 |
|
extern INTERPOLATE8X8_6TAP_LOWPASS_PTR interpolate8x8_6tap_lowpass_h; |
93 |
|
extern INTERPOLATE8X8_6TAP_LOWPASS_PTR interpolate8x8_6tap_lowpass_v; |
94 |
|
|
95 |
INTERPOLATE8X8 interpolate8x8_halfpel_h_c; |
INTERPOLATE8X8 interpolate8x8_halfpel_h_c; |
96 |
INTERPOLATE8X8 interpolate8x8_halfpel_v_c; |
INTERPOLATE8X8 interpolate8x8_halfpel_v_c; |
97 |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_c; |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_c; |
108 |
INTERPOLATE8X8 interpolate8x8_halfpel_v_3dn; |
INTERPOLATE8X8 interpolate8x8_halfpel_v_3dn; |
109 |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_3dn; |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_3dn; |
110 |
|
|
111 |
static __inline void interpolate8x8_switch(uint8_t * const cur, |
INTERPOLATE8X8 interpolate8x8_halfpel_h_3dne; |
112 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_v_3dne; |
113 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_hv_3dne; |
114 |
|
|
115 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_h_ia64; |
116 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_v_ia64; |
117 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_hv_ia64; |
118 |
|
|
119 |
|
INTERPOLATE8X8_AVG2 interpolate8x8_avg2_c; |
120 |
|
INTERPOLATE8X8_AVG4 interpolate8x8_avg4_c; |
121 |
|
|
122 |
|
INTERPOLATE8X8_AVG2 interpolate8x8_avg2_mmx; |
123 |
|
INTERPOLATE8X8_AVG4 interpolate8x8_avg4_mmx; |
124 |
|
|
125 |
|
INTERPOLATE_LOWPASS interpolate8x8_lowpass_h_c; |
126 |
|
INTERPOLATE_LOWPASS interpolate8x8_lowpass_v_c; |
127 |
|
|
128 |
|
INTERPOLATE_LOWPASS interpolate16x16_lowpass_h_c; |
129 |
|
INTERPOLATE_LOWPASS interpolate16x16_lowpass_v_c; |
130 |
|
|
131 |
|
INTERPOLATE_LOWPASS_HV interpolate8x8_lowpass_hv_c; |
132 |
|
INTERPOLATE_LOWPASS_HV interpolate16x16_lowpass_hv_c; |
133 |
|
|
134 |
|
INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_h_c; |
135 |
|
INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_v_c; |
136 |
|
|
137 |
|
INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_h_mmx; |
138 |
|
INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_v_mmx; |
139 |
|
|
140 |
|
static __inline void |
141 |
|
interpolate8x8_switch(uint8_t * const cur, |
142 |
const uint8_t * const refn, |
const uint8_t * const refn, |
143 |
const uint32_t x, const uint32_t y, |
const uint32_t x, |
144 |
const int32_t dx, const int dy, |
const uint32_t y, |
145 |
|
const int32_t dx, |
146 |
|
const int dy, |
147 |
const uint32_t stride, |
const uint32_t stride, |
148 |
const uint32_t rounding) |
const uint32_t rounding) |
149 |
{ |
{ |
150 |
int32_t ddx, ddy; |
int32_t ddx, ddy; |
151 |
|
|
152 |
switch ( ((dx&1)<<1) + (dy&1) ) // ((dx%2)?2:0)+((dy%2)?1:0) |
switch (((dx & 1) << 1) + (dy & 1)) /* ((dx%2)?2:0)+((dy%2)?1:0) */ |
153 |
{ |
{ |
154 |
case 0 : |
case 0 : |
155 |
ddx = dx/2; |
ddx = dx/2; |
156 |
ddy = dy/2; |
ddy = dy/2; |
157 |
transfer8x8_copy(cur + y*stride + x, refn + (y+ddy)*stride + x + ddx, stride); |
transfer8x8_copy(cur + y * stride + x, |
158 |
|
refn + (int)((y + ddy) * stride + x + ddx), stride); |
159 |
break; |
break; |
160 |
|
|
161 |
case 1 : |
case 1 : |
162 |
ddx = dx/2; |
ddx = dx/2; |
163 |
ddy = (dy-1)/2; |
ddy = (dy-1)/2; |
164 |
interpolate8x8_halfpel_v(cur + y*stride + x, |
interpolate8x8_halfpel_v(cur + y*stride + x, |
165 |
refn + (y+ddy)*stride + x + ddx, stride, rounding); |
refn + (int)((y + ddy) * stride + x + ddx), stride, |
166 |
|
rounding); |
167 |
break; |
break; |
168 |
|
|
169 |
case 2 : |
case 2 : |
170 |
ddx = (dx-1)/2; |
ddx = (dx-1)/2; |
171 |
ddy = dy/2; |
ddy = dy/2; |
172 |
interpolate8x8_halfpel_h(cur + y*stride + x, |
interpolate8x8_halfpel_h(cur + y*stride + x, |
173 |
refn + (y+ddy)*stride + x + ddx, stride, rounding); |
refn + (int)((y + ddy) * stride + x + ddx), stride, |
174 |
|
rounding); |
175 |
break; |
break; |
176 |
|
|
177 |
default : |
default : |
178 |
ddx = (dx-1)/2; |
ddx = (dx-1)/2; |
179 |
ddy = (dy-1)/2; |
ddy = (dy-1)/2; |
180 |
interpolate8x8_halfpel_hv(cur + y*stride + x, |
interpolate8x8_halfpel_hv(cur + y*stride + x, |
181 |
refn + (y+ddy)*stride + x + ddx, stride, rounding); |
refn + (int)((y + ddy) * stride + x + ddx), stride, |
182 |
|
rounding); |
183 |
|
break; |
184 |
|
} |
185 |
|
} |
186 |
|
|
187 |
|
|
188 |
|
static __inline void |
189 |
|
interpolate16x16_switch(uint8_t * const cur, |
190 |
|
const uint8_t * const refn, |
191 |
|
const uint32_t x, |
192 |
|
const uint32_t y, |
193 |
|
const int32_t dx, |
194 |
|
const int dy, |
195 |
|
const uint32_t stride, |
196 |
|
const uint32_t rounding) |
197 |
|
{ |
198 |
|
interpolate8x8_switch(cur, refn, x, y, dx, dy, stride, rounding); |
199 |
|
interpolate8x8_switch(cur, refn, x+8, y, dx, dy, stride, rounding); |
200 |
|
interpolate8x8_switch(cur, refn, x, y+8, dx, dy, stride, rounding); |
201 |
|
interpolate8x8_switch(cur, refn, x+8, y+8, dx, dy, stride, rounding); |
202 |
|
} |
203 |
|
|
204 |
|
|
205 |
|
static __inline void |
206 |
|
interpolate32x32_switch(uint8_t * const cur, |
207 |
|
const uint8_t * const refn, |
208 |
|
const uint32_t x, |
209 |
|
const uint32_t y, |
210 |
|
const int32_t dx, |
211 |
|
const int dy, |
212 |
|
const uint32_t stride, |
213 |
|
const uint32_t rounding) |
214 |
|
{ |
215 |
|
interpolate16x16_switch(cur, refn, x, y, dx, dy, stride, rounding); |
216 |
|
interpolate16x16_switch(cur, refn, x+16, y, dx, dy, stride, rounding); |
217 |
|
interpolate16x16_switch(cur, refn, x, y+16, dx, dy, stride, rounding); |
218 |
|
interpolate16x16_switch(cur, refn, x+16, y+16, dx, dy, stride, rounding); |
219 |
|
} |
220 |
|
|
221 |
|
|
222 |
|
static __inline uint8_t * |
223 |
|
interpolate8x8_switch2(uint8_t * const buffer, |
224 |
|
const uint8_t * const refn, |
225 |
|
const uint32_t x, |
226 |
|
const uint32_t y, |
227 |
|
const int32_t dx, |
228 |
|
const int dy, |
229 |
|
const uint32_t stride, |
230 |
|
const uint32_t rounding) |
231 |
|
{ |
232 |
|
int32_t ddx, ddy; |
233 |
|
|
234 |
|
switch (((dx & 1) << 1) + (dy & 1)) /* ((dx%2)?2:0)+((dy%2)?1:0) */ |
235 |
|
{ |
236 |
|
case 0: |
237 |
|
return (uint8_t *)refn + (int)((y + dy/2) * stride + x + dx/2); |
238 |
|
|
239 |
|
case 1: |
240 |
|
ddx = dx / 2; |
241 |
|
ddy = (dy - 1) / 2; |
242 |
|
interpolate8x8_halfpel_v(buffer, |
243 |
|
refn + (int)((y + ddy) * stride + x + ddx), stride, |
244 |
|
rounding); |
245 |
|
break; |
246 |
|
|
247 |
|
case 2: |
248 |
|
ddx = (dx - 1) / 2; |
249 |
|
ddy = dy / 2; |
250 |
|
interpolate8x8_halfpel_h(buffer, |
251 |
|
refn + (int)((y + ddy) * stride + x + ddx), stride, |
252 |
|
rounding); |
253 |
|
break; |
254 |
|
|
255 |
|
default: |
256 |
|
ddx = (dx - 1) / 2; |
257 |
|
ddy = (dy - 1) / 2; |
258 |
|
interpolate8x8_halfpel_hv(buffer, |
259 |
|
refn + (int)((y + ddy) * stride + x + ddx), stride, |
260 |
|
rounding); |
261 |
|
break; |
262 |
|
} |
263 |
|
return buffer; |
264 |
|
} |
265 |
|
|
266 |
|
static __inline void interpolate8x8_quarterpel(uint8_t * const cur, |
267 |
|
uint8_t * const refn, |
268 |
|
uint8_t * const refh, |
269 |
|
uint8_t * const refv, |
270 |
|
uint8_t * const refhv, |
271 |
|
const uint32_t x, const uint32_t y, |
272 |
|
const int32_t dx, const int dy, |
273 |
|
const uint32_t stride, |
274 |
|
const uint32_t rounding) |
275 |
|
{ |
276 |
|
const int32_t xRef = x*4 + dx; |
277 |
|
const int32_t yRef = y*4 + dy; |
278 |
|
|
279 |
|
uint8_t *src, *dst; |
280 |
|
uint8_t *halfpel_h, *halfpel_v, *halfpel_hv; |
281 |
|
int32_t x_int, y_int, x_frac, y_frac; |
282 |
|
|
283 |
|
x_int = xRef/4; |
284 |
|
if (xRef < 0 && xRef % 4) |
285 |
|
x_int--; |
286 |
|
|
287 |
|
x_frac = xRef - (4*x_int); |
288 |
|
|
289 |
|
y_int = yRef/4; |
290 |
|
if (yRef < 0 && yRef % 4) |
291 |
|
y_int--; |
292 |
|
|
293 |
|
y_frac = yRef - (4*y_int); |
294 |
|
|
295 |
|
src = refn + y_int * stride + x_int; |
296 |
|
halfpel_h = refh; |
297 |
|
halfpel_v = refv; |
298 |
|
halfpel_hv = refhv; |
299 |
|
|
300 |
|
dst = cur + y * stride + x; |
301 |
|
|
302 |
|
switch((y_frac << 2) | (x_frac)) { |
303 |
|
|
304 |
|
case 0: |
305 |
|
transfer8x8_copy(dst, src, stride); |
306 |
|
break; |
307 |
|
|
308 |
|
case 1: |
309 |
|
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
310 |
|
interpolate8x8_avg2(dst, src, halfpel_h, stride, rounding, 8); |
311 |
|
break; |
312 |
|
|
313 |
|
case 2: |
314 |
|
interpolate8x8_lowpass_h(dst, src, stride, rounding); |
315 |
|
break; |
316 |
|
|
317 |
|
case 3: |
318 |
|
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
319 |
|
interpolate8x8_avg2(dst, src + 1, halfpel_h, stride, rounding, 8); |
320 |
|
break; |
321 |
|
|
322 |
|
case 4: |
323 |
|
interpolate8x8_lowpass_v(halfpel_v, src, stride, rounding); |
324 |
|
interpolate8x8_avg2(dst, src, halfpel_v, stride, rounding, 8); |
325 |
|
break; |
326 |
|
|
327 |
|
case 5: |
328 |
|
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
329 |
|
interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); |
330 |
|
interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
331 |
|
interpolate8x8_avg2(dst, halfpel_v, halfpel_hv, stride, rounding, 8); |
332 |
|
break; |
333 |
|
|
334 |
|
case 6: |
335 |
|
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
336 |
|
interpolate8x8_avg2(dst, halfpel_h, halfpel_hv, stride, rounding, 8); |
337 |
|
break; |
338 |
|
|
339 |
|
case 7: |
340 |
|
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
341 |
|
interpolate8x8_avg2(halfpel_v, src + 1, halfpel_h, stride, rounding, 9); |
342 |
|
interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
343 |
|
interpolate8x8_avg2(dst, halfpel_v, halfpel_hv, stride, rounding, 8); |
344 |
|
break; |
345 |
|
|
346 |
|
case 8: |
347 |
|
interpolate8x8_lowpass_v(dst, src, stride, rounding); |
348 |
|
break; |
349 |
|
|
350 |
|
case 9: |
351 |
|
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
352 |
|
interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); |
353 |
|
interpolate8x8_lowpass_v(dst, halfpel_v, stride, rounding); |
354 |
|
break; |
355 |
|
|
356 |
|
case 10: |
357 |
|
interpolate8x8_lowpass_hv(dst, halfpel_h, src, stride, rounding); |
358 |
|
break; |
359 |
|
|
360 |
|
case 11: |
361 |
|
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
362 |
|
interpolate8x8_avg2(halfpel_v, src + 1, halfpel_h, stride, rounding, 9); |
363 |
|
interpolate8x8_lowpass_v(dst, halfpel_v, stride, rounding); |
364 |
|
break; |
365 |
|
|
366 |
|
case 12: |
367 |
|
interpolate8x8_lowpass_v(halfpel_v, src, stride, rounding); |
368 |
|
interpolate8x8_avg2(dst, src+stride, halfpel_v, stride, rounding, 8); |
369 |
|
break; |
370 |
|
|
371 |
|
case 13: |
372 |
|
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
373 |
|
interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); |
374 |
|
interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
375 |
|
interpolate8x8_avg2(dst, halfpel_v+stride, halfpel_hv, stride, rounding, 8); |
376 |
|
break; |
377 |
|
|
378 |
|
case 14: |
379 |
|
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
380 |
|
interpolate8x8_avg2(dst, halfpel_h+stride, halfpel_hv, stride, rounding, 8); |
381 |
|
break; |
382 |
|
|
383 |
|
case 15: |
384 |
|
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
385 |
|
interpolate8x8_avg2(halfpel_v, src + 1, halfpel_h, stride, rounding, 9); |
386 |
|
interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
387 |
|
interpolate8x8_avg2(dst, halfpel_hv, halfpel_v + stride, stride, rounding, 8); |
388 |
|
break; |
389 |
|
} |
390 |
|
} |
391 |
|
|
392 |
|
static __inline void interpolate16x16_quarterpel(uint8_t * const cur, |
393 |
|
uint8_t * const refn, |
394 |
|
uint8_t * const refh, |
395 |
|
uint8_t * const refv, |
396 |
|
uint8_t * const refhv, |
397 |
|
const uint32_t x, const uint32_t y, |
398 |
|
const int32_t dx, const int dy, |
399 |
|
const uint32_t stride, |
400 |
|
const uint32_t rounding) |
401 |
|
{ |
402 |
|
const int32_t xRef = x*4 + dx; |
403 |
|
const int32_t yRef = y*4 + dy; |
404 |
|
|
405 |
|
uint8_t *src, *dst; |
406 |
|
uint8_t *halfpel_h, *halfpel_v, *halfpel_hv; |
407 |
|
int32_t x_int, y_int, x_frac, y_frac; |
408 |
|
|
409 |
|
x_int = xRef/4; |
410 |
|
if (xRef < 0 && xRef % 4) |
411 |
|
x_int--; |
412 |
|
|
413 |
|
x_frac = xRef - (4*x_int); |
414 |
|
|
415 |
|
y_int = yRef/4; |
416 |
|
if (yRef < 0 && yRef % 4) |
417 |
|
y_int--; |
418 |
|
|
419 |
|
y_frac = yRef - (4*y_int); |
420 |
|
|
421 |
|
src = refn + y_int * stride + x_int; |
422 |
|
halfpel_h = refh; |
423 |
|
halfpel_v = refv; |
424 |
|
halfpel_hv = refhv; |
425 |
|
|
426 |
|
dst = cur + y * stride + x; |
427 |
|
|
428 |
|
switch((y_frac << 2) | (x_frac)) { |
429 |
|
|
430 |
|
case 0: |
431 |
|
transfer16x16_copy(dst, src, stride); |
432 |
|
break; |
433 |
|
|
434 |
|
case 1: |
435 |
|
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
436 |
|
interpolate8x8_avg2(dst, src, halfpel_h, stride, rounding, 8); |
437 |
|
interpolate8x8_avg2(dst+8, src+8, halfpel_h+8, stride, rounding, 8); |
438 |
|
interpolate8x8_avg2(dst+8*stride, src+8*stride, halfpel_h+8*stride, stride, rounding, 8); |
439 |
|
interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 8); |
440 |
|
break; |
441 |
|
|
442 |
|
case 2: |
443 |
|
interpolate16x16_lowpass_h(dst, src, stride, rounding); |
444 |
|
break; |
445 |
|
|
446 |
|
case 3: |
447 |
|
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
448 |
|
interpolate8x8_avg2(dst, src + 1, halfpel_h, stride, rounding, 8); |
449 |
|
interpolate8x8_avg2(dst+8, src + 8 + 1, halfpel_h+8, stride, rounding, 8); |
450 |
|
interpolate8x8_avg2(dst+8*stride, src + 8*stride + 1, halfpel_h+8*stride, stride, rounding, 8); |
451 |
|
interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8 + 1, halfpel_h+8*stride+8, stride, rounding, 8); |
452 |
|
break; |
453 |
|
|
454 |
|
case 4: |
455 |
|
interpolate16x16_lowpass_v(halfpel_v, src, stride, rounding); |
456 |
|
interpolate8x8_avg2(dst, src, halfpel_v, stride, rounding, 8); |
457 |
|
interpolate8x8_avg2(dst+8, src+8, halfpel_v+8, stride, rounding, 8); |
458 |
|
interpolate8x8_avg2(dst+8*stride, src+8*stride, halfpel_v+8*stride, stride, rounding, 8); |
459 |
|
interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 8); |
460 |
|
break; |
461 |
|
|
462 |
|
case 5: |
463 |
|
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
464 |
|
interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); |
465 |
|
interpolate8x8_avg2(halfpel_v+8, src + 8, halfpel_h+8, stride, rounding, 9); |
466 |
|
interpolate8x8_avg2(halfpel_v+8*stride, src + 8*stride, halfpel_h+8*stride, stride, rounding, 9); |
467 |
|
interpolate8x8_avg2(halfpel_v+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); |
468 |
|
|
469 |
|
interpolate16x16_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
470 |
|
interpolate8x8_avg2(dst, halfpel_hv, halfpel_v, stride, rounding, 8); |
471 |
|
interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+8, stride, rounding, 8); |
472 |
|
interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+8*stride, stride, rounding, 8); |
473 |
|
interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 8); |
474 |
|
break; |
475 |
|
|
476 |
|
case 6: |
477 |
|
interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
478 |
|
interpolate8x8_avg2(dst, halfpel_h, halfpel_hv, stride, rounding, 8); |
479 |
|
interpolate8x8_avg2(dst+8, halfpel_h+8, halfpel_hv+8, stride, rounding, 8); |
480 |
|
interpolate8x8_avg2(dst+8*stride, halfpel_h+8*stride, halfpel_hv+8*stride, stride, rounding, 8); |
481 |
|
interpolate8x8_avg2(dst+8*stride+8, halfpel_h+8*stride+8, halfpel_hv+8*stride+8, stride, rounding, 8); |
482 |
|
break; |
483 |
|
|
484 |
|
case 7: |
485 |
|
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
486 |
|
interpolate8x8_avg2(halfpel_v, src+1, halfpel_h, stride, rounding, 9); |
487 |
|
interpolate8x8_avg2(halfpel_v+8, src+1 + 8, halfpel_h+8, stride, rounding, 9); |
488 |
|
interpolate8x8_avg2(halfpel_v+8*stride, src+1 + 8*stride, halfpel_h+8*stride, stride, rounding, 9); |
489 |
|
interpolate8x8_avg2(halfpel_v+8*stride+8, src+1+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); |
490 |
|
|
491 |
|
interpolate16x16_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
492 |
|
interpolate8x8_avg2(dst, halfpel_hv, halfpel_v, stride, rounding, 8); |
493 |
|
interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+8, stride, rounding, 8); |
494 |
|
interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+8*stride, stride, rounding, 8); |
495 |
|
interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 8); |
496 |
|
break; |
497 |
|
|
498 |
|
case 8: |
499 |
|
interpolate16x16_lowpass_v(dst, src, stride, rounding); |
500 |
|
break; |
501 |
|
|
502 |
|
case 9: |
503 |
|
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
504 |
|
interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); |
505 |
|
interpolate8x8_avg2(halfpel_v+8, src + 8, halfpel_h+8, stride, rounding, 9); |
506 |
|
interpolate8x8_avg2(halfpel_v+8*stride, src + 8*stride, halfpel_h+8*stride, stride, rounding, 9); |
507 |
|
interpolate8x8_avg2(halfpel_v+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); |
508 |
|
interpolate16x16_lowpass_v(dst, halfpel_v, stride, rounding); |
509 |
|
break; |
510 |
|
|
511 |
|
case 10: |
512 |
|
interpolate16x16_lowpass_hv(dst, halfpel_h, src, stride, rounding); |
513 |
|
break; |
514 |
|
|
515 |
|
case 11: |
516 |
|
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
517 |
|
interpolate8x8_avg2(halfpel_v, src+1, halfpel_h, stride, rounding, 9); |
518 |
|
interpolate8x8_avg2(halfpel_v+8, src+1 + 8, halfpel_h+8, stride, rounding, 9); |
519 |
|
interpolate8x8_avg2(halfpel_v+8*stride, src+1 + 8*stride, halfpel_h+8*stride, stride, rounding, 9); |
520 |
|
interpolate8x8_avg2(halfpel_v+8*stride+8, src+1+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); |
521 |
|
interpolate16x16_lowpass_v(dst, halfpel_v, stride, rounding); |
522 |
|
break; |
523 |
|
|
524 |
|
case 12: |
525 |
|
interpolate16x16_lowpass_v(halfpel_v, src, stride, rounding); |
526 |
|
interpolate8x8_avg2(dst, src+stride, halfpel_v, stride, rounding, 8); |
527 |
|
interpolate8x8_avg2(dst+8, src+stride+8, halfpel_v+8, stride, rounding, 8); |
528 |
|
interpolate8x8_avg2(dst+8*stride, src+stride+8*stride, halfpel_v+8*stride, stride, rounding, 8); |
529 |
|
interpolate8x8_avg2(dst+8*stride+8, src+stride+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 8); |
530 |
|
break; |
531 |
|
|
532 |
|
case 13: |
533 |
|
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
534 |
|
interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); |
535 |
|
interpolate8x8_avg2(halfpel_v+8, src + 8, halfpel_h+8, stride, rounding, 9); |
536 |
|
interpolate8x8_avg2(halfpel_v+8*stride, src + 8*stride, halfpel_h+8*stride, stride, rounding, 9); |
537 |
|
interpolate8x8_avg2(halfpel_v+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); |
538 |
|
|
539 |
|
interpolate16x16_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
540 |
|
interpolate8x8_avg2(dst, halfpel_hv, halfpel_v+stride, stride, rounding, 8); |
541 |
|
interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+stride+8, stride, rounding, 8); |
542 |
|
interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+stride+8*stride, stride, rounding, 8); |
543 |
|
interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+stride+8*stride+8, stride, rounding, 8); |
544 |
|
break; |
545 |
|
|
546 |
|
case 14: |
547 |
|
interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
548 |
|
interpolate8x8_avg2(dst, halfpel_h+stride, halfpel_hv, stride, rounding, 8); |
549 |
|
interpolate8x8_avg2(dst+8, halfpel_h+stride+8, halfpel_hv+8, stride, rounding, 8); |
550 |
|
interpolate8x8_avg2(dst+8*stride, halfpel_h+stride+8*stride, halfpel_hv+8*stride, stride, rounding, 8); |
551 |
|
interpolate8x8_avg2(dst+8*stride+8, halfpel_h+stride+8*stride+8, halfpel_hv+8*stride+8, stride, rounding, 8); |
552 |
|
break; |
553 |
|
|
554 |
|
case 15: |
555 |
|
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
556 |
|
interpolate8x8_avg2(halfpel_v, src+1, halfpel_h, stride, rounding, 9); |
557 |
|
interpolate8x8_avg2(halfpel_v+8, src+1 + 8, halfpel_h+8, stride, rounding, 9); |
558 |
|
interpolate8x8_avg2(halfpel_v+8*stride, src+1 + 8*stride, halfpel_h+8*stride, stride, rounding, 9); |
559 |
|
interpolate8x8_avg2(halfpel_v+8*stride+8, src+1+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); |
560 |
|
|
561 |
|
interpolate16x16_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
562 |
|
interpolate8x8_avg2(dst, halfpel_hv, halfpel_v+stride, stride, rounding, 8); |
563 |
|
interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+stride+8, stride, rounding, 8); |
564 |
|
interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+stride+8*stride, stride, rounding, 8); |
565 |
|
interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+stride+8*stride+8, stride, rounding, 8); |
566 |
break; |
break; |
567 |
} |
} |
568 |
} |
} |
569 |
|
|
570 |
|
#endif |