1 |
#ifndef _INTERPOLATE8X8_H_ |
/***************************************************************************** |
2 |
#define _INTERPOLATE8X8_H_ |
* |
3 |
|
* XVID MPEG-4 VIDEO CODEC |
4 |
|
* - 8x8 block-based halfpel interpolation - headers |
5 |
|
* |
6 |
|
* Copyright(C) 2002 Peter Ross <pross@xvid.org> |
7 |
|
* |
8 |
|
* This file is part of XviD, a free MPEG-4 video encoder/decoder |
9 |
|
* |
10 |
|
* XviD is free software; you can redistribute it and/or modify it |
11 |
|
* under the terms of the GNU General Public License as published by |
12 |
|
* the Free Software Foundation; either version 2 of the License, or |
13 |
|
* (at your option) any later version. |
14 |
|
* |
15 |
|
* This program is distributed in the hope that it will be useful, |
16 |
|
* but WITHOUT ANY WARRANTY; without even the implied warranty of |
17 |
|
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
18 |
|
* GNU General Public License for more details. |
19 |
|
* |
20 |
|
* You should have received a copy of the GNU General Public License |
21 |
|
* along with this program; if not, write to the Free Software |
22 |
|
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
23 |
|
* |
24 |
|
* Under section 8 of the GNU General Public License, the copyright |
25 |
|
* holders of XVID explicitly forbid distribution in the following |
26 |
|
* countries: |
27 |
|
* |
28 |
|
* - Japan |
29 |
|
* - United States of America |
30 |
|
* |
31 |
|
* Linking XviD statically or dynamically with other modules is making a |
32 |
|
* combined work based on XviD. Thus, the terms and conditions of the |
33 |
|
* GNU General Public License cover the whole combination. |
34 |
|
* |
35 |
|
* As a special exception, the copyright holders of XviD give you |
36 |
|
* permission to link XviD with independent modules that communicate with |
37 |
|
* XviD solely through the VFW1.1 and DShow interfaces, regardless of the |
38 |
|
* license terms of these independent modules, and to copy and distribute |
39 |
|
* the resulting combined work under terms of your choice, provided that |
40 |
|
* every copy of the combined work is accompanied by a complete copy of |
41 |
|
* the source code of XviD (the version of XviD used to produce the |
42 |
|
* combined work), being distributed under the terms of the GNU General |
43 |
|
* Public License plus this exception. An independent module is a module |
44 |
|
* which is not derived from or based on XviD. |
45 |
|
* |
46 |
|
* Note that people who make modified versions of XviD are not obligated |
47 |
|
* to grant this special exception for their modified versions; it is |
48 |
|
* their choice whether to do so. The GNU General Public License gives |
49 |
|
* permission to release a modified version without this exception; this |
50 |
|
* exception also makes it possible to release a modified version which |
51 |
|
* carries forward this exception. |
52 |
|
* |
53 |
|
* $Id$ |
54 |
|
* |
55 |
|
****************************************************************************/ |
56 |
|
|
57 |
#include "../utils/mem_transfer.h" |
#include "../utils/mem_transfer.h" |
58 |
|
|
62 |
const uint32_t rounding); |
const uint32_t rounding); |
63 |
typedef INTERPOLATE8X8 *INTERPOLATE8X8_PTR; |
typedef INTERPOLATE8X8 *INTERPOLATE8X8_PTR; |
64 |
|
|
|
typedef void (INTERPOLATE8X8_AVG2) (uint8_t *dst, |
|
|
const uint8_t *src1, |
|
|
const uint8_t *src2, |
|
|
const uint32_t stride, |
|
|
const uint32_t rounding); |
|
|
typedef INTERPOLATE8X8_AVG2 *INTERPOLATE8X8_AVG2_PTR; |
|
|
|
|
|
typedef void (INTERPOLATE8X8_AVG4) (uint8_t *dst, |
|
|
const uint8_t *src1, |
|
|
const uint8_t *src2, |
|
|
const uint8_t *src3, |
|
|
const uint8_t *src4, |
|
|
const uint32_t stride, |
|
|
const uint32_t rounding); |
|
|
typedef INTERPOLATE8X8_AVG4 *INTERPOLATE8X8_AVG4_PTR; |
|
|
|
|
|
typedef void (INTERPOLATE_LOWPASS) (uint8_t *dst, |
|
|
uint8_t *src, |
|
|
int32_t stride, |
|
|
int32_t rounding); |
|
|
|
|
|
typedef INTERPOLATE_LOWPASS *INTERPOLATE_LOWPASS_PTR; |
|
|
|
|
|
typedef void (INTERPOLATE_LOWPASS_HV) (uint8_t *dst1, |
|
|
uint8_t *dst2, |
|
|
uint8_t *src, |
|
|
int32_t stride, |
|
|
int32_t rounding); |
|
|
|
|
|
typedef INTERPOLATE_LOWPASS_HV *INTERPOLATE_LOWPASS_HV_PTR; |
|
|
|
|
|
typedef void (INTERPOLATE8X8_6TAP_LOWPASS) (uint8_t *dst, |
|
|
uint8_t *src, |
|
|
int32_t stride, |
|
|
int32_t rounding); |
|
|
|
|
|
typedef INTERPOLATE8X8_6TAP_LOWPASS *INTERPOLATE8X8_6TAP_LOWPASS_PTR; |
|
|
|
|
65 |
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_h; |
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_h; |
66 |
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_v; |
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_v; |
67 |
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_hv; |
extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_hv; |
68 |
|
|
|
extern INTERPOLATE8X8_AVG2_PTR interpolate8x8_avg2; |
|
|
extern INTERPOLATE8X8_AVG4_PTR interpolate8x8_avg4; |
|
|
|
|
|
extern INTERPOLATE_LOWPASS_PTR interpolate8x8_lowpass_h; |
|
|
extern INTERPOLATE_LOWPASS_PTR interpolate8x8_lowpass_v; |
|
|
|
|
|
extern INTERPOLATE_LOWPASS_PTR interpolate16x16_lowpass_h; |
|
|
extern INTERPOLATE_LOWPASS_PTR interpolate16x16_lowpass_v; |
|
|
|
|
|
extern INTERPOLATE_LOWPASS_HV_PTR interpolate8x8_lowpass_hv; |
|
|
extern INTERPOLATE_LOWPASS_HV_PTR interpolate16x16_lowpass_hv; |
|
|
|
|
|
extern INTERPOLATE8X8_6TAP_LOWPASS_PTR interpolate8x8_6tap_lowpass_h; |
|
|
extern INTERPOLATE8X8_6TAP_LOWPASS_PTR interpolate8x8_6tap_lowpass_v; |
|
|
|
|
69 |
INTERPOLATE8X8 interpolate8x8_halfpel_h_c; |
INTERPOLATE8X8 interpolate8x8_halfpel_h_c; |
70 |
INTERPOLATE8X8 interpolate8x8_halfpel_v_c; |
INTERPOLATE8X8 interpolate8x8_halfpel_v_c; |
71 |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_c; |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_c; |
86 |
INTERPOLATE8X8 interpolate8x8_halfpel_v_ia64; |
INTERPOLATE8X8 interpolate8x8_halfpel_v_ia64; |
87 |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_ia64; |
INTERPOLATE8X8 interpolate8x8_halfpel_hv_ia64; |
88 |
|
|
89 |
INTERPOLATE8X8_AVG2 interpolate8x8_avg2_c; |
void interpolate8x8_lowpass_h(uint8_t *dst, uint8_t *src, int32_t dst_stride, int32_t src_stride, int32_t rounding); |
90 |
INTERPOLATE8X8_AVG4 interpolate8x8_avg4_c; |
void interpolate8x8_lowpass_v(uint8_t *dst, uint8_t *src, int32_t dst_stride, int32_t src_stride, int32_t rounding); |
91 |
|
void interpolate8x8_lowpass_hv(uint8_t *dst1, uint8_t *dst2, uint8_t *src, int32_t dst1_stride, int32_t dst2_stride, int32_t src_stride, int32_t rounding); |
92 |
INTERPOLATE8X8_AVG2 interpolate8x8_avg2_mmx; |
void interpolate8x8_bilinear2(uint8_t *dst, uint8_t *src1, uint8_t *src2, int32_t dst_stride, int32_t src_stride, int32_t rounding); |
93 |
INTERPOLATE8X8_AVG4 interpolate8x8_avg4_mmx; |
void interpolate8x8_bilinear4(uint8_t *dst, uint8_t *src1, uint8_t *src2, uint8_t *src3, uint8_t *src4, int32_t stride, int32_t rounding); |
|
|
|
|
INTERPOLATE_LOWPASS interpolate8x8_lowpass_h_c; |
|
|
INTERPOLATE_LOWPASS interpolate8x8_lowpass_v_c; |
|
94 |
|
|
95 |
INTERPOLATE_LOWPASS interpolate16x16_lowpass_h_c; |
void interpolate8x8_c(uint8_t * const dst, |
96 |
INTERPOLATE_LOWPASS interpolate16x16_lowpass_v_c; |
const uint8_t * const src, |
97 |
|
const uint32_t x, |
98 |
INTERPOLATE_LOWPASS_HV interpolate8x8_lowpass_hv_c; |
const uint32_t y, |
99 |
INTERPOLATE_LOWPASS_HV interpolate16x16_lowpass_hv_c; |
const uint32_t stride); |
|
|
|
|
INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_h_c; |
|
|
INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_v_c; |
|
|
|
|
|
INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_h_mmx; |
|
|
INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_v_mmx; |
|
100 |
|
|
101 |
static __inline void |
static __inline void |
102 |
interpolate8x8_switch(uint8_t * const cur, |
interpolate8x8_switch(uint8_t * const cur, |
145 |
} |
} |
146 |
} |
} |
147 |
|
|
|
static __inline uint8_t * |
|
|
interpolate8x8_switch2(uint8_t * const buffer, |
|
|
const uint8_t * const refn, |
|
|
const uint32_t x, |
|
|
const uint32_t y, |
|
|
const int32_t dx, |
|
|
const int dy, |
|
|
const uint32_t stride, |
|
|
const uint32_t rounding) |
|
|
{ |
|
|
int32_t ddx, ddy; |
|
|
|
|
|
switch (((dx & 1) << 1) + (dy & 1)) // ((dx%2)?2:0)+((dy%2)?1:0) |
|
|
{ |
|
|
case 0: |
|
|
return (uint8_t *)refn + (int)((y + dy/2) * stride + x + dx/2); |
|
|
|
|
|
case 1: |
|
|
ddx = dx / 2; |
|
|
ddy = (dy - 1) / 2; |
|
|
interpolate8x8_halfpel_v(buffer, |
|
|
refn + (int)((y + ddy) * stride + x + ddx), stride, |
|
|
rounding); |
|
|
break; |
|
|
|
|
|
case 2: |
|
|
ddx = (dx - 1) / 2; |
|
|
ddy = dy / 2; |
|
|
interpolate8x8_halfpel_h(buffer, |
|
|
refn + (int)((y + ddy) * stride + x + ddx), stride, |
|
|
rounding); |
|
|
break; |
|
|
|
|
|
default: |
|
|
ddx = (dx - 1) / 2; |
|
|
ddy = (dy - 1) / 2; |
|
|
interpolate8x8_halfpel_hv(buffer, |
|
|
refn + (int)((y + ddy) * stride + x + ddx), stride, |
|
|
rounding); |
|
|
break; |
|
|
} |
|
|
return buffer; |
|
|
} |
|
|
|
|
148 |
|
|
149 |
static __inline void interpolate8x8_quarterpel(uint8_t * const cur, |
static __inline void interpolate8x8_quarterpel(uint8_t * const cur, |
150 |
uint8_t * const refn, |
uint8_t * const refn, |
|
uint8_t * const refh, |
|
|
uint8_t * const refv, |
|
|
uint8_t * const refhv, |
|
151 |
const uint32_t x, const uint32_t y, |
const uint32_t x, const uint32_t y, |
152 |
const int32_t dx, const int dy, |
const int32_t dx, const int dy, |
153 |
const uint32_t stride, |
const uint32_t stride, |
157 |
const int32_t yRef = y*4 + dy; |
const int32_t yRef = y*4 + dy; |
158 |
|
|
159 |
uint8_t *src, *dst; |
uint8_t *src, *dst; |
|
uint8_t *halfpel_h, *halfpel_v, *halfpel_hv; |
|
160 |
int32_t x_int, y_int, x_frac, y_frac; |
int32_t x_int, y_int, x_frac, y_frac; |
161 |
|
|
162 |
x_int = xRef/4; |
uint8_t halfpel_h[72]; |
163 |
if (xRef < 0 && xRef % 4) |
uint8_t halfpel_v[64]; |
164 |
x_int--; |
uint8_t halfpel_hv[64]; |
|
|
|
|
x_frac = xRef - (4*x_int); |
|
|
|
|
|
y_int = yRef/4; |
|
|
if (yRef < 0 && yRef % 4) |
|
|
y_int--; |
|
|
|
|
|
y_frac = yRef - (4*y_int); |
|
|
|
|
|
src = refn + y_int * stride + x_int; |
|
|
halfpel_h = refh; |
|
|
halfpel_v = refv; |
|
|
halfpel_hv = refhv; |
|
|
|
|
|
dst = cur + y * stride + x; |
|
|
|
|
|
switch((y_frac << 2) | (x_frac)) { |
|
|
|
|
|
case 0: |
|
|
transfer8x8_copy(dst, src, stride); |
|
|
break; |
|
|
|
|
|
case 1: |
|
|
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg2(dst, src, halfpel_h, stride, rounding); |
|
|
break; |
|
|
|
|
|
case 2: |
|
|
interpolate8x8_lowpass_h(dst, src, stride, rounding); |
|
|
break; |
|
|
|
|
|
case 3: |
|
|
interpolate8x8_lowpass_h(halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg2(dst, src + 1, halfpel_h, stride, rounding); |
|
|
break; |
|
|
|
|
|
case 4: |
|
|
interpolate8x8_lowpass_v(halfpel_v, src, stride, rounding); |
|
|
interpolate8x8_avg2(dst, src, halfpel_v, stride, rounding); |
|
|
break; |
|
|
|
|
|
case 5: |
|
|
interpolate8x8_lowpass_v(halfpel_v, src, stride, rounding); |
|
|
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg4(dst, src, halfpel_h, halfpel_v, halfpel_hv, stride, rounding); |
|
|
break; |
|
|
|
|
|
case 6: |
|
|
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg2(dst, halfpel_h, halfpel_hv, stride, rounding); |
|
|
break; |
|
|
|
|
|
case 7: |
|
|
interpolate8x8_lowpass_v(halfpel_v, src+1, stride, rounding); |
|
|
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg4(dst, src+1, halfpel_h, halfpel_v, halfpel_hv, stride, rounding); |
|
|
break; |
|
|
|
|
|
case 8: |
|
|
interpolate8x8_lowpass_v(dst, src, stride, rounding); |
|
|
break; |
|
|
|
|
|
case 9: |
|
|
interpolate8x8_lowpass_v(halfpel_v, src, stride, rounding); |
|
|
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg2(dst, halfpel_v, halfpel_hv, stride, rounding); |
|
|
break; |
|
|
|
|
|
case 10: |
|
|
interpolate8x8_lowpass_hv(dst, halfpel_h, src, stride, rounding); |
|
|
break; |
|
|
|
|
|
case 11: |
|
|
interpolate8x8_lowpass_v(halfpel_v, src+1, stride, rounding); |
|
|
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg2(dst, halfpel_v, halfpel_hv, stride, rounding); |
|
|
break; |
|
|
|
|
|
case 12: |
|
|
interpolate8x8_lowpass_v(halfpel_v, src, stride, rounding); |
|
|
interpolate8x8_avg2(dst, src+stride, halfpel_v, stride, rounding); |
|
|
break; |
|
|
|
|
|
case 13: |
|
|
interpolate8x8_lowpass_v(halfpel_v, src, stride, rounding); |
|
|
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg4(dst, src+stride, halfpel_h+stride, halfpel_v, halfpel_hv, stride, rounding); |
|
|
break; |
|
|
|
|
|
case 14: |
|
|
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg2(dst, halfpel_h+stride, halfpel_hv, stride, rounding); |
|
|
break; |
|
|
|
|
|
case 15: |
|
|
interpolate8x8_lowpass_v(halfpel_v, src+1, stride, rounding); |
|
|
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
|
|
interpolate8x8_avg4(dst, src+stride+1, halfpel_h+stride, halfpel_v, halfpel_hv, stride, rounding); |
|
|
break; |
|
|
} |
|
|
} |
|
|
|
|
|
|
|
|
static __inline void interpolate16x16_quarterpel(uint8_t * const cur, |
|
|
uint8_t * const refn, |
|
|
uint8_t * const refh, |
|
|
uint8_t * const refv, |
|
|
uint8_t * const refhv, |
|
|
const uint32_t x, const uint32_t y, |
|
|
const int32_t dx, const int dy, |
|
|
const uint32_t stride, |
|
|
const uint32_t rounding) |
|
|
{ |
|
|
const int32_t xRef = x*4 + dx; |
|
|
const int32_t yRef = y*4 + dy; |
|
|
|
|
|
uint8_t *src, *dst; |
|
|
uint8_t *halfpel_h, *halfpel_v, *halfpel_hv; |
|
|
int32_t x_int, y_int, x_frac, y_frac; |
|
165 |
|
|
166 |
x_int = xRef/4; |
x_int = xRef/4; |
167 |
if (xRef < 0 && xRef % 4) |
if (xRef < 0 && xRef % 4) |
176 |
y_frac = yRef - (4*y_int); |
y_frac = yRef - (4*y_int); |
177 |
|
|
178 |
src = refn + y_int * stride + x_int; |
src = refn + y_int * stride + x_int; |
|
halfpel_h = refh; |
|
|
halfpel_v = refv; |
|
|
halfpel_hv = refhv; |
|
|
|
|
179 |
dst = cur + y * stride + x; |
dst = cur + y * stride + x; |
180 |
|
|
181 |
switch((y_frac << 2) | (x_frac)) { |
switch((y_frac << 2) | (x_frac)) { |
182 |
|
|
183 |
case 0: |
case 0: |
184 |
transfer8x8_copy(dst, src, stride); |
transfer8x8_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); |
|
185 |
break; |
break; |
186 |
|
|
187 |
case 1: |
case 1: |
188 |
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
interpolate8x8_lowpass_h(halfpel_h, src, 8, stride, rounding); |
189 |
interpolate8x8_avg2(dst, src, halfpel_h, stride, rounding); |
interpolate8x8_bilinear2(dst, src, halfpel_h, stride, stride, rounding); |
|
interpolate8x8_avg2(dst+8, src+8, halfpel_h+8, stride, rounding); |
|
|
interpolate8x8_avg2(dst+8*stride, src+8*stride, halfpel_h+8*stride, stride, rounding); |
|
|
interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8, halfpel_h+8*stride+8, stride, rounding); |
|
190 |
break; |
break; |
191 |
|
|
192 |
case 2: |
case 2: |
193 |
interpolate16x16_lowpass_h(dst, src, stride, rounding); |
interpolate8x8_lowpass_h(dst, src, stride, stride, rounding); |
194 |
break; |
break; |
195 |
|
|
196 |
case 3: |
case 3: |
197 |
interpolate16x16_lowpass_h(halfpel_h, src, stride, rounding); |
interpolate8x8_lowpass_h(halfpel_h, src, 8, stride, rounding); |
198 |
interpolate8x8_avg2(dst, src + 1, halfpel_h, stride, rounding); |
interpolate8x8_bilinear2(dst, src+1, halfpel_h, stride, stride, rounding); |
|
interpolate8x8_avg2(dst+8, src + 8 + 1, halfpel_h+8, stride, rounding); |
|
|
interpolate8x8_avg2(dst+8*stride, src + 8*stride + 1, halfpel_h+8*stride, stride, rounding); |
|
|
interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8 + 1, halfpel_h+8*stride+8, stride, rounding); |
|
199 |
break; |
break; |
200 |
|
|
201 |
case 4: |
case 4: |
202 |
interpolate16x16_lowpass_v(halfpel_v, src, stride, rounding); |
interpolate8x8_lowpass_v(halfpel_v, src, 8, stride, rounding); |
203 |
interpolate8x8_avg2(dst, src, halfpel_v, stride, rounding); |
interpolate8x8_bilinear2(dst, src, halfpel_v, stride, stride, rounding); |
|
interpolate8x8_avg2(dst+8, src+8, halfpel_v+8, stride, rounding); |
|
|
interpolate8x8_avg2(dst+8*stride, src+8*stride, halfpel_v+8*stride, stride, rounding); |
|
|
interpolate8x8_avg2(dst+8*stride+8, src+8*stride+8, halfpel_v+8*stride+8, stride, rounding); |
|
204 |
break; |
break; |
205 |
|
|
206 |
case 5: |
case 5: |
207 |
interpolate16x16_lowpass_v(halfpel_v, src, stride, rounding); |
interpolate8x8_lowpass_v(halfpel_v, src, 8, stride, rounding); |
208 |
interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, 8, 8, stride, rounding); |
209 |
interpolate8x8_avg4(dst, src, halfpel_h, halfpel_v, halfpel_hv, stride, rounding); |
interpolate8x8_bilinear4(dst, src, halfpel_h, halfpel_v, halfpel_hv, stride, rounding); |
|
interpolate8x8_avg4(dst+8, src+8, halfpel_h+8, halfpel_v+8, halfpel_hv+8, stride, rounding); |
|
|
interpolate8x8_avg4(dst+8*stride, src+8*stride, halfpel_h+8*stride, halfpel_v+8*stride, halfpel_hv+8*stride, stride, rounding); |
|
|
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); |
|
210 |
break; |
break; |
211 |
|
|
212 |
case 6: |
case 6: |
213 |
interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, 8, 8, stride, rounding); |
214 |
interpolate8x8_avg2(dst, halfpel_h, halfpel_hv, stride, rounding); |
interpolate8x8_bilinear2(dst, halfpel_h, halfpel_hv, stride, 8, 1-rounding); |
|
interpolate8x8_avg2(dst+8, halfpel_h+8, halfpel_hv+8, stride, rounding); |
|
|
interpolate8x8_avg2(dst+8*stride, halfpel_h+8*stride, halfpel_hv+8*stride, stride, rounding); |
|
|
interpolate8x8_avg2(dst+8*stride+8, halfpel_h+8*stride+8, halfpel_hv+8*stride+8, stride, rounding); |
|
215 |
break; |
break; |
216 |
|
|
217 |
case 7: |
case 7: |
218 |
interpolate16x16_lowpass_v(halfpel_v, src+1, stride, rounding); |
interpolate8x8_lowpass_v(halfpel_v, src+1, 8, stride, 16-rounding); |
219 |
interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, 8, 8, stride, rounding); |
220 |
interpolate8x8_avg4(dst, src+1, halfpel_h, halfpel_v, halfpel_hv, stride, rounding); |
interpolate8x8_bilinear4(dst, src+1, halfpel_h, halfpel_v, halfpel_hv, stride, rounding); |
|
interpolate8x8_avg4(dst+8, src+8+1, halfpel_h+8, halfpel_v+8, halfpel_hv+8, stride, rounding); |
|
|
interpolate8x8_avg4(dst+8*stride, src+1+8*stride, halfpel_h+8*stride, halfpel_v+8*stride, halfpel_hv+8*stride, stride, rounding); |
|
|
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); |
|
221 |
break; |
break; |
222 |
|
|
223 |
case 8: |
case 8: |
224 |
interpolate16x16_lowpass_v(dst, src, stride, rounding); |
interpolate8x8_lowpass_v(dst, src, stride, stride, rounding); |
225 |
break; |
break; |
226 |
|
|
227 |
case 9: |
case 9: |
228 |
interpolate16x16_lowpass_v(halfpel_v, src, stride, rounding); |
interpolate8x8_lowpass_v(halfpel_v, src, 8, stride, 16-rounding); |
229 |
interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, 8, 8, stride, rounding); |
230 |
interpolate8x8_avg2(dst, halfpel_v, halfpel_hv, stride, rounding); |
interpolate8x8_bilinear2(dst, halfpel_v, halfpel_hv, stride, 8, rounding); |
|
interpolate8x8_avg2(dst+8, halfpel_v+8, halfpel_hv+8, stride, rounding); |
|
|
interpolate8x8_avg2(dst+8*stride, halfpel_v+8*stride, halfpel_hv+8*stride, stride, rounding); |
|
|
interpolate8x8_avg2(dst+8*stride+8, halfpel_v+8*stride+8, halfpel_hv+8*stride+8, stride, rounding); |
|
231 |
break; |
break; |
232 |
|
|
233 |
case 10: |
case 10: |
234 |
interpolate16x16_lowpass_hv(dst, halfpel_h, src, stride, rounding); |
interpolate8x8_lowpass_hv(dst, halfpel_h, src, stride, 8, stride, rounding); |
235 |
break; |
break; |
236 |
|
|
237 |
case 11: |
case 11: |
238 |
interpolate16x16_lowpass_v(halfpel_v, src+1, stride, rounding); |
interpolate8x8_lowpass_v(halfpel_v, src+1, 8, stride, 16-rounding); |
239 |
interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, 8, 8, stride, rounding); |
240 |
interpolate8x8_avg2(dst, halfpel_v, halfpel_hv, stride, rounding); |
interpolate8x8_bilinear2(dst, halfpel_v, halfpel_hv, stride, 8, rounding); |
|
interpolate8x8_avg2(dst+8, halfpel_v+8, halfpel_hv+8, stride, rounding); |
|
|
interpolate8x8_avg2(dst+8*stride, halfpel_v+8*stride, halfpel_hv+8*stride, stride, rounding); |
|
|
interpolate8x8_avg2(dst+8*stride+8, halfpel_v+8*stride+8, halfpel_hv+8*stride+8, stride, rounding); |
|
241 |
break; |
break; |
242 |
|
|
243 |
case 12: |
case 12: |
244 |
interpolate16x16_lowpass_v(halfpel_v, src, stride, rounding); |
interpolate8x8_lowpass_v(halfpel_v, src, 8, stride, rounding); |
245 |
interpolate8x8_avg2(dst, src+stride, halfpel_v, stride, rounding); |
interpolate8x8_bilinear2(dst, src+stride, halfpel_v, stride, stride, rounding); |
|
interpolate8x8_avg2(dst+8, src+stride+8, halfpel_v+8, stride, rounding); |
|
|
interpolate8x8_avg2(dst+8*stride, src+stride+8*stride, halfpel_v+8*stride, stride, rounding); |
|
|
interpolate8x8_avg2(dst+8*stride+8, src+stride+8*stride+8, halfpel_v+8*stride+8, stride, rounding); |
|
246 |
break; |
break; |
247 |
|
|
248 |
case 13: |
case 13: |
249 |
interpolate16x16_lowpass_v(halfpel_v, src, stride, rounding); |
interpolate8x8_lowpass_v(halfpel_v, src, 8, stride, rounding); |
250 |
interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, 8, 8, stride, rounding); |
251 |
interpolate8x8_avg4(dst, src+stride, halfpel_h+stride, halfpel_v, halfpel_hv, stride, rounding); |
interpolate8x8_bilinear4(dst, src+stride, halfpel_h+8, halfpel_v, halfpel_hv, stride, rounding); |
|
interpolate8x8_avg4(dst+8, src+stride+8, halfpel_h+stride+8, halfpel_v+8, halfpel_hv+8, stride, rounding); |
|
|
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_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); |
|
252 |
break; |
break; |
253 |
|
|
254 |
case 14: |
case 14: |
255 |
interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, 8, 8, stride, rounding); |
256 |
interpolate8x8_avg2(dst, halfpel_h+stride, halfpel_hv, stride, rounding); |
interpolate8x8_bilinear2(dst, halfpel_h+8, halfpel_hv, stride, 8, rounding); |
|
interpolate8x8_avg2(dst+8, halfpel_h+stride+8, halfpel_hv+8, stride, rounding); |
|
|
interpolate8x8_avg2(dst+8*stride, halfpel_h+stride+8*stride, halfpel_hv+8*stride, stride, rounding); |
|
|
interpolate8x8_avg2(dst+8*stride+8, halfpel_h+stride+8*stride+8, halfpel_hv+8*stride+8, stride, rounding); |
|
257 |
break; |
break; |
258 |
|
|
259 |
case 15: |
case 15: |
260 |
interpolate16x16_lowpass_v(halfpel_v, src+1, stride, rounding); |
interpolate8x8_lowpass_v(halfpel_v, src+1, 8, stride, rounding); |
261 |
interpolate16x16_lowpass_hv(halfpel_hv, halfpel_h, src, stride, rounding); |
interpolate8x8_lowpass_hv(halfpel_hv, halfpel_h, src, 8, 8, stride, rounding); |
262 |
interpolate8x8_avg4(dst, src+stride+1, halfpel_h+stride, halfpel_v, halfpel_hv, stride, rounding); |
interpolate8x8_bilinear4(dst, src+stride+1, halfpel_h+8, halfpel_v, halfpel_hv, stride, rounding); |
|
interpolate8x8_avg4(dst+8, src+stride+1+8, halfpel_h+stride+8, halfpel_v+8, halfpel_hv+8, stride, rounding); |
|
|
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_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); |
|
|
break; |
|
|
default: |
|
|
interpolate8x8_quarterpel(cur, refn, refh, refv, refhv, x, y, dx, dy, stride, rounding); |
|
|
interpolate8x8_quarterpel(cur, refn, refh, refv, refhv, x+8, y, dx, dy, stride, rounding); |
|
|
interpolate8x8_quarterpel(cur, refn, refh, refv, refhv, x, y+8, dx, dy, stride, rounding); |
|
|
interpolate8x8_quarterpel(cur, refn, refh, refv, refhv, x+8, y+8, dx, dy, stride, rounding); |
|
|
return; |
|
263 |
break; |
break; |
264 |
} |
} |
265 |
} |
} |
|
|
|
|
#endif |
|