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_ |
#ifndef _INTERPOLATE8X8_H_ |
27 |
#define _INTERPOLATE8X8_H_ |
#define _INTERPOLATE8X8_H_ |
28 |
|
|
38 |
const uint8_t *src1, |
const uint8_t *src1, |
39 |
const uint8_t *src2, |
const uint8_t *src2, |
40 |
const uint32_t stride, |
const uint32_t stride, |
41 |
const uint32_t rounding); |
const uint32_t rounding, |
42 |
|
const uint32_t height); |
43 |
typedef INTERPOLATE8X8_AVG2 *INTERPOLATE8X8_AVG2_PTR; |
typedef INTERPOLATE8X8_AVG2 *INTERPOLATE8X8_AVG2_PTR; |
44 |
|
|
45 |
typedef void (INTERPOLATE8X8_AVG4) (uint8_t *dst, |
typedef void (INTERPOLATE8X8_AVG4) (uint8_t *dst, |
96 |
INTERPOLATE8X8 interpolate8x8_halfpel_v_c; |
INTERPOLATE8X8 interpolate8x8_halfpel_v_c; |
97 |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_c; |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_c; |
98 |
|
|
99 |
|
#ifdef ARCH_IS_IA32 |
100 |
INTERPOLATE8X8 interpolate8x8_halfpel_h_mmx; |
INTERPOLATE8X8 interpolate8x8_halfpel_h_mmx; |
101 |
INTERPOLATE8X8 interpolate8x8_halfpel_v_mmx; |
INTERPOLATE8X8 interpolate8x8_halfpel_v_mmx; |
102 |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_mmx; |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_mmx; |
109 |
INTERPOLATE8X8 interpolate8x8_halfpel_v_3dn; |
INTERPOLATE8X8 interpolate8x8_halfpel_v_3dn; |
110 |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_3dn; |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_3dn; |
111 |
|
|
112 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_h_3dne; |
113 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_v_3dne; |
114 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_hv_3dne; |
115 |
|
#endif |
116 |
|
|
117 |
|
#ifdef ARCH_IS_IA64 |
118 |
INTERPOLATE8X8 interpolate8x8_halfpel_h_ia64; |
INTERPOLATE8X8 interpolate8x8_halfpel_h_ia64; |
119 |
INTERPOLATE8X8 interpolate8x8_halfpel_v_ia64; |
INTERPOLATE8X8 interpolate8x8_halfpel_v_ia64; |
120 |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_ia64; |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_ia64; |
121 |
|
#endif |
122 |
|
|
123 |
|
#ifdef ARCH_IS_PPC |
124 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_h_altivec_c; |
125 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_v_altivec_c; |
126 |
|
INTERPOLATE8X8 interpolate8x8_halfpel_hv_altivec_c; |
127 |
|
#endif |
128 |
|
|
129 |
INTERPOLATE8X8_AVG2 interpolate8x8_avg2_c; |
INTERPOLATE8X8_AVG2 interpolate8x8_avg2_c; |
130 |
INTERPOLATE8X8_AVG4 interpolate8x8_avg4_c; |
INTERPOLATE8X8_AVG4 interpolate8x8_avg4_c; |
131 |
|
|
132 |
|
#ifdef ARCH_IS_IA32 |
133 |
INTERPOLATE8X8_AVG2 interpolate8x8_avg2_mmx; |
INTERPOLATE8X8_AVG2 interpolate8x8_avg2_mmx; |
134 |
INTERPOLATE8X8_AVG4 interpolate8x8_avg4_mmx; |
INTERPOLATE8X8_AVG4 interpolate8x8_avg4_mmx; |
135 |
|
#endif |
136 |
|
|
137 |
|
#ifdef ARCH_IS_PPC |
138 |
|
INTERPOLATE8X8_AVG2 interpolate8x8_avg2_altivec_c; |
139 |
|
INTERPOLATE8X8_AVG4 interpolate8x8_avg4_altivec_c; |
140 |
|
#endif |
141 |
|
|
142 |
INTERPOLATE_LOWPASS interpolate8x8_lowpass_h_c; |
INTERPOLATE_LOWPASS interpolate8x8_lowpass_h_c; |
143 |
INTERPOLATE_LOWPASS interpolate8x8_lowpass_v_c; |
INTERPOLATE_LOWPASS interpolate8x8_lowpass_v_c; |
151 |
INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_h_c; |
INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_h_c; |
152 |
INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_v_c; |
INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_v_c; |
153 |
|
|
154 |
|
#ifdef ARCH_IS_IA32 |
155 |
INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_h_mmx; |
INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_h_mmx; |
156 |
INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_v_mmx; |
INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_v_mmx; |
157 |
|
#endif |
158 |
|
|
159 |
|
#ifdef ARCH_IS_PPC |
160 |
|
INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_h_altivec_c; |
161 |
|
#endif |
162 |
|
|
163 |
static __inline void |
static __inline void |
164 |
interpolate8x8_switch(uint8_t * const cur, |
interpolate8x8_switch(uint8_t * const cur, |
170 |
const uint32_t stride, |
const uint32_t stride, |
171 |
const uint32_t rounding) |
const uint32_t rounding) |
172 |
{ |
{ |
|
int32_t ddx, ddy; |
|
173 |
|
|
174 |
switch (((dx & 1) << 1) + (dy & 1)) // ((dx%2)?2:0)+((dy%2)?1:0) |
const uint8_t * const src = refn + (int)((y + (dy>>1)) * stride + x + (dx>>1)); |
175 |
{ |
uint8_t * const dst = cur + (int)(y * stride + x); |
176 |
|
|
177 |
|
switch (((dx & 1) << 1) + (dy & 1)) { /* ((dx%2)?2:0)+((dy%2)?1:0) */ |
178 |
case 0: |
case 0: |
179 |
ddx = dx / 2; |
transfer8x8_copy(dst, src, stride); |
|
ddy = dy / 2; |
|
|
transfer8x8_copy(cur + y * stride + x, |
|
|
refn + (int)((y + ddy) * stride + x + ddx), stride); |
|
180 |
break; |
break; |
|
|
|
181 |
case 1: |
case 1: |
182 |
ddx = dx / 2; |
interpolate8x8_halfpel_v(dst, src, stride, rounding); |
|
ddy = (dy - 1) / 2; |
|
|
interpolate8x8_halfpel_v(cur + y * stride + x, |
|
|
refn + (int)((y + ddy) * stride + x + ddx), stride, |
|
|
rounding); |
|
183 |
break; |
break; |
|
|
|
184 |
case 2: |
case 2: |
185 |
ddx = (dx - 1) / 2; |
interpolate8x8_halfpel_h(dst, src, stride, rounding); |
|
ddy = dy / 2; |
|
|
interpolate8x8_halfpel_h(cur + y * stride + x, |
|
|
refn + (int)((y + ddy) * stride + x + ddx), stride, |
|
|
rounding); |
|
186 |
break; |
break; |
|
|
|
187 |
default: |
default: |
188 |
ddx = (dx - 1) / 2; |
interpolate8x8_halfpel_hv(dst, src, stride, rounding); |
|
ddy = (dy - 1) / 2; |
|
|
interpolate8x8_halfpel_hv(cur + y * stride + x, |
|
|
refn + (int)((y + ddy) * stride + x + ddx), stride, |
|
|
rounding); |
|
189 |
break; |
break; |
190 |
} |
} |
191 |
} |
} |
192 |
|
|
193 |
static __inline uint8_t * |
|
194 |
interpolate8x8_switch2(uint8_t * const buffer, |
static __inline void |
195 |
|
interpolate16x16_switch(uint8_t * const cur, |
196 |
const uint8_t * const refn, |
const uint8_t * const refn, |
197 |
const uint32_t x, |
const uint32_t x, |
198 |
const uint32_t y, |
const uint32_t y, |
201 |
const uint32_t stride, |
const uint32_t stride, |
202 |
const uint32_t rounding) |
const uint32_t rounding) |
203 |
{ |
{ |
204 |
int32_t ddx, ddy; |
interpolate8x8_switch(cur, refn, x, y, dx, dy, stride, rounding); |
205 |
|
interpolate8x8_switch(cur, refn, x+8, y, dx, dy, stride, rounding); |
206 |
|
interpolate8x8_switch(cur, refn, x, y+8, dx, dy, stride, rounding); |
207 |
|
interpolate8x8_switch(cur, refn, x+8, y+8, dx, dy, stride, rounding); |
208 |
|
} |
209 |
|
|
210 |
switch (((dx & 1) << 1) + (dy & 1)) // ((dx%2)?2:0)+((dy%2)?1:0) |
|
211 |
|
static __inline void |
212 |
|
interpolate32x32_switch(uint8_t * const cur, |
213 |
|
const uint8_t * const refn, |
214 |
|
const uint32_t x, |
215 |
|
const uint32_t y, |
216 |
|
const int32_t dx, |
217 |
|
const int dy, |
218 |
|
const uint32_t stride, |
219 |
|
const uint32_t rounding) |
220 |
{ |
{ |
221 |
case 0: |
interpolate16x16_switch(cur, refn, x, y, dx, dy, stride, rounding); |
222 |
return (uint8_t *)refn + (int)((y + dy/2) * stride + x + dx/2); |
interpolate16x16_switch(cur, refn, x+16, y, dx, dy, stride, rounding); |
223 |
|
interpolate16x16_switch(cur, refn, x, y+16, dx, dy, stride, rounding); |
224 |
|
interpolate16x16_switch(cur, refn, x+16, y+16, dx, dy, stride, rounding); |
225 |
|
} |
226 |
|
|
227 |
|
|
228 |
|
static __inline uint8_t * |
229 |
|
interpolate8x8_switch2(uint8_t * const buffer, |
230 |
|
const uint8_t * const refn, |
231 |
|
const int x, |
232 |
|
const int y, |
233 |
|
const int dx, |
234 |
|
const int dy, |
235 |
|
const uint32_t stride, |
236 |
|
const uint32_t rounding) |
237 |
|
{ |
238 |
|
|
239 |
|
const uint8_t * const src = refn + (int)((y + (dy>>1)) * stride + x + (dx>>1)); |
240 |
|
|
241 |
|
switch (((dx & 1) << 1) + (dy & 1)) { /* ((dx%2)?2:0)+((dy%2)?1:0) */ |
242 |
|
case 0: |
243 |
|
return (uint8_t *)src; |
244 |
case 1: |
case 1: |
245 |
ddx = dx / 2; |
interpolate8x8_halfpel_v(buffer, src, stride, rounding); |
|
ddy = (dy - 1) / 2; |
|
|
interpolate8x8_halfpel_v(buffer, |
|
|
refn + (int)((y + ddy) * stride + x + ddx), stride, |
|
|
rounding); |
|
246 |
break; |
break; |
|
|
|
247 |
case 2: |
case 2: |
248 |
ddx = (dx - 1) / 2; |
interpolate8x8_halfpel_h(buffer, src, stride, rounding); |
|
ddy = dy / 2; |
|
|
interpolate8x8_halfpel_h(buffer, |
|
|
refn + (int)((y + ddy) * stride + x + ddx), stride, |
|
|
rounding); |
|
249 |
break; |
break; |
|
|
|
250 |
default: |
default: |
251 |
ddx = (dx - 1) / 2; |
interpolate8x8_halfpel_hv(buffer, src, stride, rounding); |
|
ddy = (dy - 1) / 2; |
|
|
interpolate8x8_halfpel_hv(buffer, |
|
|
refn + (int)((y + ddy) * stride + x + ddx), stride, |
|
|
rounding); |
|
252 |
break; |
break; |
253 |
} |
} |
254 |
return buffer; |
return buffer; |
255 |
} |
} |
256 |
|
|
|
|
|
257 |
static __inline void interpolate8x8_quarterpel(uint8_t * const cur, |
static __inline void interpolate8x8_quarterpel(uint8_t * const cur, |
258 |
uint8_t * const refn, |
uint8_t * const refn, |
259 |
uint8_t * const refh, |
uint8_t * const refh, |
298 |
|
|
299 |
case 1: |
case 1: |
300 |
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
301 |
interpolate8x8_avg2(dst, src, halfpel_h, stride, rounding); |
interpolate8x8_avg2(dst, src, halfpel_h, stride, rounding, 8); |
302 |
break; |
break; |
303 |
|
|
304 |
case 2: |
case 2: |
307 |
|
|
308 |
case 3: |
case 3: |
309 |
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
310 |
interpolate8x8_avg2(dst, src + 1, halfpel_h, stride, rounding); |
interpolate8x8_avg2(dst, src + 1, halfpel_h, stride, rounding, 8); |
311 |
break; |
break; |
312 |
|
|
313 |
case 4: |
case 4: |
314 |
interpolate8x8_lowpass_v(halfpel_v, src, stride, rounding); |
interpolate8x8_lowpass_v(halfpel_v, src, stride, rounding); |
315 |
interpolate8x8_avg2(dst, src, halfpel_v, stride, rounding); |
interpolate8x8_avg2(dst, src, halfpel_v, stride, rounding, 8); |
316 |
break; |
break; |
317 |
|
|
318 |
case 5: |
case 5: |
319 |
interpolate8x8_lowpass_v(halfpel_v, src, stride, rounding); |
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
320 |
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); |
321 |
interpolate8x8_avg4(dst, src, halfpel_h, halfpel_v, halfpel_hv, stride, rounding); |
interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
322 |
|
interpolate8x8_avg2(dst, halfpel_v, halfpel_hv, stride, rounding, 8); |
323 |
break; |
break; |
324 |
|
|
325 |
case 6: |
case 6: |
326 |
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
327 |
interpolate8x8_avg2(dst, halfpel_h, halfpel_hv, stride, rounding); |
interpolate8x8_avg2(dst, halfpel_h, halfpel_hv, stride, rounding, 8); |
328 |
break; |
break; |
329 |
|
|
330 |
case 7: |
case 7: |
331 |
interpolate8x8_lowpass_v(halfpel_v, src+1, stride, rounding); |
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
332 |
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
interpolate8x8_avg2(halfpel_v, src + 1, halfpel_h, stride, rounding, 9); |
333 |
interpolate8x8_avg4(dst, src+1, halfpel_h, halfpel_v, halfpel_hv, stride, rounding); |
interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
334 |
|
interpolate8x8_avg2(dst, halfpel_v, halfpel_hv, stride, rounding, 8); |
335 |
break; |
break; |
336 |
|
|
337 |
case 8: |
case 8: |
339 |
break; |
break; |
340 |
|
|
341 |
case 9: |
case 9: |
342 |
interpolate8x8_lowpass_v(halfpel_v, src, stride, rounding); |
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
343 |
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); |
344 |
interpolate8x8_avg2(dst, halfpel_v, halfpel_hv, stride, rounding); |
interpolate8x8_lowpass_v(dst, halfpel_v, stride, rounding); |
345 |
break; |
break; |
346 |
|
|
347 |
case 10: |
case 10: |
349 |
break; |
break; |
350 |
|
|
351 |
case 11: |
case 11: |
352 |
interpolate8x8_lowpass_v(halfpel_v, src+1, stride, rounding); |
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
353 |
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
interpolate8x8_avg2(halfpel_v, src + 1, halfpel_h, stride, rounding, 9); |
354 |
interpolate8x8_avg2(dst, halfpel_v, halfpel_hv, stride, rounding); |
interpolate8x8_lowpass_v(dst, halfpel_v, stride, rounding); |
355 |
break; |
break; |
356 |
|
|
357 |
case 12: |
case 12: |
358 |
interpolate8x8_lowpass_v(halfpel_v, src, stride, rounding); |
interpolate8x8_lowpass_v(halfpel_v, src, stride, rounding); |
359 |
interpolate8x8_avg2(dst, src+stride, halfpel_v, stride, rounding); |
interpolate8x8_avg2(dst, src+stride, halfpel_v, stride, rounding, 8); |
360 |
break; |
break; |
361 |
|
|
362 |
case 13: |
case 13: |
363 |
interpolate8x8_lowpass_v(halfpel_v, src, stride, rounding); |
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
364 |
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); |
365 |
interpolate8x8_avg4(dst, src+stride, halfpel_h+stride, halfpel_v, halfpel_hv, stride, rounding); |
interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
366 |
|
interpolate8x8_avg2(dst, halfpel_v+stride, halfpel_hv, stride, rounding, 8); |
367 |
break; |
break; |
368 |
|
|
369 |
case 14: |
case 14: |
370 |
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
371 |
interpolate8x8_avg2(dst, halfpel_h+stride, halfpel_hv, stride, rounding); |
interpolate8x8_avg2(dst, halfpel_h+stride, halfpel_hv, stride, rounding, 8); |
372 |
break; |
break; |
373 |
|
|
374 |
case 15: |
case 15: |
375 |
interpolate8x8_lowpass_v(halfpel_v, src+1, stride, rounding); |
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
376 |
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
interpolate8x8_avg2(halfpel_v, src + 1, halfpel_h, stride, rounding, 9); |
377 |
interpolate8x8_avg4(dst, src+stride+1, halfpel_h+stride, halfpel_v, halfpel_hv, stride, rounding); |
interpolate8x8_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
378 |
|
interpolate8x8_avg2(dst, halfpel_hv, halfpel_v + stride, stride, rounding, 8); |
379 |
break; |
break; |
380 |
} |
} |
381 |
} |
} |
382 |
|
|
|
|
|
383 |
static __inline void interpolate16x16_quarterpel(uint8_t * const cur, |
static __inline void interpolate16x16_quarterpel(uint8_t * const cur, |
384 |
uint8_t * const refn, |
uint8_t * const refn, |
385 |
uint8_t * const refh, |
uint8_t * const refh, |
419 |
switch((y_frac << 2) | (x_frac)) { |
switch((y_frac << 2) | (x_frac)) { |
420 |
|
|
421 |
case 0: |
case 0: |
422 |
transfer8x8_copy(dst, src, stride); |
transfer16x16_copy(dst, src, stride); |
|
transfer8x8_copy(dst+8, src+8, stride); |
|
|
transfer8x8_copy(dst+8*stride, src+8*stride, stride); |
|
|
transfer8x8_copy(dst+8*stride+8, src+8*stride+8, stride); |
|
423 |
break; |
break; |
424 |
|
|
425 |
case 1: |
case 1: |
426 |
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
427 |
interpolate8x8_avg2(dst, src, halfpel_h, stride, rounding); |
interpolate8x8_avg2(dst, src, halfpel_h, stride, rounding, 8); |
428 |
interpolate8x8_avg2(dst+8, src+8, halfpel_h+8, stride, rounding); |
interpolate8x8_avg2(dst+8, src+8, halfpel_h+8, stride, rounding, 8); |
429 |
interpolate8x8_avg2(dst+8*stride, src+8*stride, halfpel_h+8*stride, stride, rounding); |
interpolate8x8_avg2(dst+8*stride, src+8*stride, halfpel_h+8*stride, stride, rounding, 8); |
430 |
interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding); |
interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 8); |
431 |
break; |
break; |
432 |
|
|
433 |
case 2: |
case 2: |
436 |
|
|
437 |
case 3: |
case 3: |
438 |
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
439 |
interpolate8x8_avg2(dst, src + 1, halfpel_h, stride, rounding); |
interpolate8x8_avg2(dst, src + 1, halfpel_h, stride, rounding, 8); |
440 |
interpolate8x8_avg2(dst+8, src + 8 + 1, halfpel_h+8, stride, rounding); |
interpolate8x8_avg2(dst+8, src + 8 + 1, halfpel_h+8, stride, rounding, 8); |
441 |
interpolate8x8_avg2(dst+8*stride, src + 8*stride + 1, halfpel_h+8*stride, stride, rounding); |
interpolate8x8_avg2(dst+8*stride, src + 8*stride + 1, halfpel_h+8*stride, stride, rounding, 8); |
442 |
interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8 + 1, halfpel_h+8*stride+8, stride, rounding); |
interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8 + 1, halfpel_h+8*stride+8, stride, rounding, 8); |
443 |
break; |
break; |
444 |
|
|
445 |
case 4: |
case 4: |
446 |
interpolate16x16_lowpass_v(halfpel_v, src, stride, rounding); |
interpolate16x16_lowpass_v(halfpel_v, src, stride, rounding); |
447 |
interpolate8x8_avg2(dst, src, halfpel_v, stride, rounding); |
interpolate8x8_avg2(dst, src, halfpel_v, stride, rounding, 8); |
448 |
interpolate8x8_avg2(dst+8, src+8, halfpel_v+8, stride, rounding); |
interpolate8x8_avg2(dst+8, src+8, halfpel_v+8, stride, rounding, 8); |
449 |
interpolate8x8_avg2(dst+8*stride, src+8*stride, halfpel_v+8*stride, stride, rounding); |
interpolate8x8_avg2(dst+8*stride, src+8*stride, halfpel_v+8*stride, stride, rounding, 8); |
450 |
interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8, halfpel_v+8*stride+8, stride, rounding); |
interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 8); |
451 |
break; |
break; |
452 |
|
|
453 |
case 5: |
case 5: |
454 |
interpolate16x16_lowpass_v(halfpel_v, src, stride, rounding); |
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
455 |
interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); |
456 |
interpolate8x8_avg4(dst, src, halfpel_h, halfpel_v, halfpel_hv, stride, rounding); |
interpolate8x8_avg2(halfpel_v+8, src + 8, halfpel_h+8, stride, rounding, 9); |
457 |
interpolate8x8_avg4(dst+8, src+8, halfpel_h+8, halfpel_v+8, halfpel_hv+8, stride, rounding); |
interpolate8x8_avg2(halfpel_v+8*stride, src + 8*stride, halfpel_h+8*stride, stride, rounding, 9); |
458 |
interpolate8x8_avg4(dst+8*stride, src+8*stride, halfpel_h+8*stride, halfpel_v+8*stride, halfpel_hv+8*stride, stride, rounding); |
interpolate8x8_avg2(halfpel_v+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); |
459 |
interpolate8x8_avg4(dst+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, halfpel_v+8*stride+8, halfpel_hv+8*stride+8, stride, rounding); |
|
460 |
|
interpolate16x16_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
461 |
|
interpolate8x8_avg2(dst, halfpel_hv, halfpel_v, stride, rounding, 8); |
462 |
|
interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+8, stride, rounding, 8); |
463 |
|
interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+8*stride, stride, rounding, 8); |
464 |
|
interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 8); |
465 |
break; |
break; |
466 |
|
|
467 |
case 6: |
case 6: |
468 |
interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
469 |
interpolate8x8_avg2(dst, halfpel_h, halfpel_hv, stride, rounding); |
interpolate8x8_avg2(dst, halfpel_h, halfpel_hv, stride, rounding, 8); |
470 |
interpolate8x8_avg2(dst+8, halfpel_h+8, halfpel_hv+8, stride, rounding); |
interpolate8x8_avg2(dst+8, halfpel_h+8, halfpel_hv+8, stride, rounding, 8); |
471 |
interpolate8x8_avg2(dst+8*stride, halfpel_h+8*stride, halfpel_hv+8*stride, stride, rounding); |
interpolate8x8_avg2(dst+8*stride, halfpel_h+8*stride, halfpel_hv+8*stride, stride, rounding, 8); |
472 |
interpolate8x8_avg2(dst+8*stride+8, halfpel_h+8*stride+8, halfpel_hv+8*stride+8, stride, rounding); |
interpolate8x8_avg2(dst+8*stride+8, halfpel_h+8*stride+8, halfpel_hv+8*stride+8, stride, rounding, 8); |
473 |
break; |
break; |
474 |
|
|
475 |
case 7: |
case 7: |
476 |
interpolate16x16_lowpass_v(halfpel_v, src+1, stride, rounding); |
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
477 |
interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
interpolate8x8_avg2(halfpel_v, src+1, halfpel_h, stride, rounding, 9); |
478 |
interpolate8x8_avg4(dst, src+1, halfpel_h, halfpel_v, halfpel_hv, stride, rounding); |
interpolate8x8_avg2(halfpel_v+8, src+1 + 8, halfpel_h+8, stride, rounding, 9); |
479 |
interpolate8x8_avg4(dst+8, src+8+1, halfpel_h+8, halfpel_v+8, halfpel_hv+8, stride, rounding); |
interpolate8x8_avg2(halfpel_v+8*stride, src+1 + 8*stride, halfpel_h+8*stride, stride, rounding, 9); |
480 |
interpolate8x8_avg4(dst+8*stride, src+1+8*stride, halfpel_h+8*stride, halfpel_v+8*stride, halfpel_hv+8*stride, stride, rounding); |
interpolate8x8_avg2(halfpel_v+8*stride+8, src+1+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); |
481 |
interpolate8x8_avg4(dst+8*stride+8, src+1+8*stride+8, halfpel_h+8*stride+8, halfpel_v+8*stride+8, halfpel_hv+8*stride+8, stride, rounding); |
|
482 |
|
interpolate16x16_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
483 |
|
interpolate8x8_avg2(dst, halfpel_hv, halfpel_v, stride, rounding, 8); |
484 |
|
interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+8, stride, rounding, 8); |
485 |
|
interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+8*stride, stride, rounding, 8); |
486 |
|
interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 8); |
487 |
break; |
break; |
488 |
|
|
489 |
case 8: |
case 8: |
491 |
break; |
break; |
492 |
|
|
493 |
case 9: |
case 9: |
494 |
interpolate16x16_lowpass_v(halfpel_v, src, stride, rounding); |
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
495 |
interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); |
496 |
interpolate8x8_avg2(dst, halfpel_v, halfpel_hv, stride, rounding); |
interpolate8x8_avg2(halfpel_v+8, src + 8, halfpel_h+8, stride, rounding, 9); |
497 |
interpolate8x8_avg2(dst+8, halfpel_v+8, halfpel_hv+8, stride, rounding); |
interpolate8x8_avg2(halfpel_v+8*stride, src + 8*stride, halfpel_h+8*stride, stride, rounding, 9); |
498 |
interpolate8x8_avg2(dst+8*stride, halfpel_v+8*stride, halfpel_hv+8*stride, stride, rounding); |
interpolate8x8_avg2(halfpel_v+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); |
499 |
interpolate8x8_avg2(dst+8*stride+8, halfpel_v+8*stride+8, halfpel_hv+8*stride+8, stride, rounding); |
interpolate16x16_lowpass_v(dst, halfpel_v, stride, rounding); |
500 |
break; |
break; |
501 |
|
|
502 |
case 10: |
case 10: |
504 |
break; |
break; |
505 |
|
|
506 |
case 11: |
case 11: |
507 |
interpolate16x16_lowpass_v(halfpel_v, src+1, stride, rounding); |
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
508 |
interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
interpolate8x8_avg2(halfpel_v, src+1, halfpel_h, stride, rounding, 9); |
509 |
interpolate8x8_avg2(dst, halfpel_v, halfpel_hv, stride, rounding); |
interpolate8x8_avg2(halfpel_v+8, src+1 + 8, halfpel_h+8, stride, rounding, 9); |
510 |
interpolate8x8_avg2(dst+8, halfpel_v+8, halfpel_hv+8, stride, rounding); |
interpolate8x8_avg2(halfpel_v+8*stride, src+1 + 8*stride, halfpel_h+8*stride, stride, rounding, 9); |
511 |
interpolate8x8_avg2(dst+8*stride, halfpel_v+8*stride, halfpel_hv+8*stride, stride, rounding); |
interpolate8x8_avg2(halfpel_v+8*stride+8, src+1+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); |
512 |
interpolate8x8_avg2(dst+8*stride+8, halfpel_v+8*stride+8, halfpel_hv+8*stride+8, stride, rounding); |
interpolate16x16_lowpass_v(dst, halfpel_v, stride, rounding); |
513 |
break; |
break; |
514 |
|
|
515 |
case 12: |
case 12: |
516 |
interpolate16x16_lowpass_v(halfpel_v, src, stride, rounding); |
interpolate16x16_lowpass_v(halfpel_v, src, stride, rounding); |
517 |
interpolate8x8_avg2(dst, src+stride, halfpel_v, stride, rounding); |
interpolate8x8_avg2(dst, src+stride, halfpel_v, stride, rounding, 8); |
518 |
interpolate8x8_avg2(dst+8, src+stride+8, halfpel_v+8, stride, rounding); |
interpolate8x8_avg2(dst+8, src+stride+8, halfpel_v+8, stride, rounding, 8); |
519 |
interpolate8x8_avg2(dst+8*stride, src+stride+8*stride, halfpel_v+8*stride, stride, rounding); |
interpolate8x8_avg2(dst+8*stride, src+stride+8*stride, halfpel_v+8*stride, stride, rounding, 8); |
520 |
interpolate8x8_avg2(dst+8*stride+8, src+stride+8*stride+8, halfpel_v+8*stride+8, stride, rounding); |
interpolate8x8_avg2(dst+8*stride+8, src+stride+8*stride+8, halfpel_v+8*stride+8, stride, rounding, 8); |
521 |
break; |
break; |
522 |
|
|
523 |
case 13: |
case 13: |
524 |
interpolate16x16_lowpass_v(halfpel_v, src, stride, rounding); |
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
525 |
interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
interpolate8x8_avg2(halfpel_v, src, halfpel_h, stride, rounding, 9); |
526 |
interpolate8x8_avg4(dst, src+stride, halfpel_h+stride, halfpel_v, halfpel_hv, stride, rounding); |
interpolate8x8_avg2(halfpel_v+8, src + 8, halfpel_h+8, stride, rounding, 9); |
527 |
interpolate8x8_avg4(dst+8, src+stride+8, halfpel_h+stride+8, halfpel_v+8, halfpel_hv+8, stride, rounding); |
interpolate8x8_avg2(halfpel_v+8*stride, src + 8*stride, halfpel_h+8*stride, stride, rounding, 9); |
528 |
interpolate8x8_avg4(dst+8*stride, src+stride+8*stride, halfpel_h+stride+8*stride, halfpel_v+8*stride, halfpel_hv+8*stride, stride, rounding); |
interpolate8x8_avg2(halfpel_v+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); |
529 |
interpolate8x8_avg4(dst+8*stride+8, src+stride+8*stride+8, halfpel_h+stride+8*stride+8, halfpel_v+8*stride+8, halfpel_hv+8*stride+8, stride, rounding); |
|
530 |
|
interpolate16x16_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
531 |
|
interpolate8x8_avg2(dst, halfpel_hv, halfpel_v+stride, stride, rounding, 8); |
532 |
|
interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+stride+8, stride, rounding, 8); |
533 |
|
interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+stride+8*stride, stride, rounding, 8); |
534 |
|
interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+stride+8*stride+8, stride, rounding, 8); |
535 |
break; |
break; |
536 |
|
|
537 |
case 14: |
case 14: |
538 |
interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
539 |
interpolate8x8_avg2(dst, halfpel_h+stride, halfpel_hv, stride, rounding); |
interpolate8x8_avg2(dst, halfpel_h+stride, halfpel_hv, stride, rounding, 8); |
540 |
interpolate8x8_avg2(dst+8, halfpel_h+stride+8, halfpel_hv+8, stride, rounding); |
interpolate8x8_avg2(dst+8, halfpel_h+stride+8, halfpel_hv+8, stride, rounding, 8); |
541 |
interpolate8x8_avg2(dst+8*stride, halfpel_h+stride+8*stride, halfpel_hv+8*stride, stride, rounding); |
interpolate8x8_avg2(dst+8*stride, halfpel_h+stride+8*stride, halfpel_hv+8*stride, stride, rounding, 8); |
542 |
interpolate8x8_avg2(dst+8*stride+8, halfpel_h+stride+8*stride+8, halfpel_hv+8*stride+8, stride, rounding); |
interpolate8x8_avg2(dst+8*stride+8, halfpel_h+stride+8*stride+8, halfpel_hv+8*stride+8, stride, rounding, 8); |
543 |
break; |
break; |
544 |
|
|
545 |
case 15: |
case 15: |
546 |
interpolate16x16_lowpass_v(halfpel_v, src+1, stride, rounding); |
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
547 |
interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
interpolate8x8_avg2(halfpel_v, src+1, halfpel_h, stride, rounding, 9); |
548 |
interpolate8x8_avg4(dst, src+stride+1, halfpel_h+stride, halfpel_v, halfpel_hv, stride, rounding); |
interpolate8x8_avg2(halfpel_v+8, src+1 + 8, halfpel_h+8, stride, rounding, 9); |
549 |
interpolate8x8_avg4(dst+8, src+stride+1+8, halfpel_h+stride+8, halfpel_v+8, halfpel_hv+8, stride, rounding); |
interpolate8x8_avg2(halfpel_v+8*stride, src+1 + 8*stride, halfpel_h+8*stride, stride, rounding, 9); |
550 |
interpolate8x8_avg4(dst+8*stride, src+stride+1+8*stride, halfpel_h+stride+8*stride, halfpel_v+8*stride, halfpel_hv+8*stride, stride, rounding); |
interpolate8x8_avg2(halfpel_v+8*stride+8, src+1+8*stride+8, halfpel_h+8*stride+8, stride, rounding, 9); |
551 |
interpolate8x8_avg4(dst+8*stride+8, src+stride+1+8*stride+8, halfpel_h+stride+8*stride+8, halfpel_v+8*stride+8, halfpel_hv+8*stride+8, stride, rounding); |
|
552 |
break; |
interpolate16x16_lowpass_v(halfpel_hv, halfpel_v, stride, rounding); |
553 |
default: |
interpolate8x8_avg2(dst, halfpel_hv, halfpel_v+stride, stride, rounding, 8); |
554 |
interpolate8x8_quarterpel(cur, refn, refh, refv, refhv, x, y, dx, dy, stride, rounding); |
interpolate8x8_avg2(dst+8, halfpel_hv+8, halfpel_v+stride+8, stride, rounding, 8); |
555 |
interpolate8x8_quarterpel(cur, refn, refh, refv, refhv, x+8, y, dx, dy, stride, rounding); |
interpolate8x8_avg2(dst+8*stride, halfpel_hv+8*stride, halfpel_v+stride+8*stride, stride, rounding, 8); |
556 |
interpolate8x8_quarterpel(cur, refn, refh, refv, refhv, x, y+8, dx, dy, stride, rounding); |
interpolate8x8_avg2(dst+8*stride+8, halfpel_hv+8*stride+8, halfpel_v+stride+8*stride+8, stride, rounding, 8); |
|
interpolate8x8_quarterpel(cur, refn, refh, refv, refhv, x+8, y+8, dx, dy, stride, rounding); |
|
|
return; |
|
557 |
break; |
break; |
558 |
} |
} |
559 |
} |
} |