1 |
/***************************************************************************** |
/************************************************************************** |
2 |
* |
* |
3 |
* XVID MPEG-4 VIDEO CODEC |
* XVID MPEG-4 VIDEO CODEC |
4 |
* - 8x8 block-based halfpel interpolation - |
* 8x8 block-based halfpel interpolation |
5 |
* |
* |
6 |
* Copyright(C) 2002 Peter Ross <pross@xvid.org> |
* This program is free software; you can redistribute it and/or modify |
7 |
* Copyright(C) 2002 MinChen <chenm002@163.com> |
* it under the terms of the GNU General Public License as published by |
|
* |
|
|
* This file is part of XviD, a free MPEG-4 video encoder/decoder |
|
|
* |
|
|
* XviD is free software; you can redistribute it and/or modify it |
|
|
* under the terms of the GNU General Public License as published by |
|
8 |
* the Free Software Foundation; either version 2 of the License, or |
* the Free Software Foundation; either version 2 of the License, or |
9 |
* (at your option) any later version. |
* (at your option) any later version. |
10 |
* |
* |
15 |
* |
* |
16 |
* You should have received a copy of the GNU General Public License |
* You should have received a copy of the GNU General Public License |
17 |
* along with this program; if not, write to the Free Software |
* along with this program; if not, write to the Free Software |
18 |
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. |
|
* |
|
|
* Under section 8 of the GNU General Public License, the copyright |
|
|
* holders of XVID explicitly forbid distribution in the following |
|
|
* countries: |
|
19 |
* |
* |
20 |
* - Japan |
*************************************************************************/ |
21 |
* - United States of America |
|
22 |
* |
/************************************************************************** |
|
* Linking XviD statically or dynamically with other modules is making a |
|
|
* combined work based on XviD. Thus, the terms and conditions of the |
|
|
* GNU General Public License cover the whole combination. |
|
|
* |
|
|
* As a special exception, the copyright holders of XviD give you |
|
|
* permission to link XviD with independent modules that communicate with |
|
|
* XviD solely through the VFW1.1 and DShow interfaces, regardless of the |
|
|
* license terms of these independent modules, and to copy and distribute |
|
|
* the resulting combined work under terms of your choice, provided that |
|
|
* every copy of the combined work is accompanied by a complete copy of |
|
|
* the source code of XviD (the version of XviD used to produce the |
|
|
* combined work), being distributed under the terms of the GNU General |
|
|
* Public License plus this exception. An independent module is a module |
|
|
* which is not derived from or based on XviD. |
|
23 |
* |
* |
24 |
* Note that people who make modified versions of XviD are not obligated |
* History: |
|
* to grant this special exception for their modified versions; it is |
|
|
* their choice whether to do so. The GNU General Public License gives |
|
|
* permission to release a modified version without this exception; this |
|
|
* exception also makes it possible to release a modified version which |
|
|
* carries forward this exception. |
|
25 |
* |
* |
26 |
* $Id$ |
* 05.10.2002 new bilinear and qpel interpolation code - Isibaar |
27 |
|
* 27.12.2001 modified "compensate_halfpel" |
28 |
|
* 05.11.2001 initial version; (c)2001 peter ross <pross@cs.rmit.edu.au> |
29 |
* |
* |
30 |
****************************************************************************/ |
*************************************************************************/ |
31 |
|
|
32 |
|
|
33 |
#include "../portab.h" |
#include "../portab.h" |
34 |
|
#include "../global.h" |
35 |
#include "interpolate8x8.h" |
#include "interpolate8x8.h" |
36 |
|
|
37 |
/* function pointers */ |
// function pointers |
38 |
INTERPOLATE8X8_PTR interpolate8x8_halfpel_h; |
INTERPOLATE8X8_PTR interpolate8x8_halfpel_h; |
39 |
INTERPOLATE8X8_PTR interpolate8x8_halfpel_v; |
INTERPOLATE8X8_PTR interpolate8x8_halfpel_v; |
40 |
INTERPOLATE8X8_PTR interpolate8x8_halfpel_hv; |
INTERPOLATE8X8_PTR interpolate8x8_halfpel_hv; |
41 |
|
|
42 |
|
INTERPOLATE8X8_AVG2_PTR interpolate8x8_avg2; |
43 |
|
INTERPOLATE8X8_AVG4_PTR interpolate8x8_avg4; |
44 |
|
|
45 |
|
INTERPOLATE_LOWPASS_PTR interpolate8x8_lowpass_h; |
46 |
|
INTERPOLATE_LOWPASS_PTR interpolate8x8_lowpass_v; |
47 |
|
|
48 |
|
INTERPOLATE_LOWPASS_PTR interpolate16x16_lowpass_h; |
49 |
|
INTERPOLATE_LOWPASS_PTR interpolate16x16_lowpass_v; |
50 |
|
|
51 |
|
INTERPOLATE_LOWPASS_HV_PTR interpolate8x8_lowpass_hv; |
52 |
|
INTERPOLATE_LOWPASS_HV_PTR interpolate16x16_lowpass_hv; |
53 |
|
|
54 |
|
INTERPOLATE8X8_6TAP_LOWPASS_PTR interpolate8x8_6tap_lowpass_h; |
55 |
|
INTERPOLATE8X8_6TAP_LOWPASS_PTR interpolate8x8_6tap_lowpass_v; |
56 |
|
|
57 |
|
void interpolate8x8_avg2_c(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, const uint32_t stride, const uint32_t rounding, const uint32_t height) |
58 |
|
{ |
59 |
|
uint32_t i; |
60 |
|
const int32_t round = 1 - rounding; |
61 |
|
|
62 |
|
for(i = 0; i < height; i++) |
63 |
|
{ |
64 |
|
dst[0] = (src1[0] + src2[0] + round) >> 1; |
65 |
|
dst[1] = (src1[1] + src2[1] + round) >> 1; |
66 |
|
dst[2] = (src1[2] + src2[2] + round) >> 1; |
67 |
|
dst[3] = (src1[3] + src2[3] + round) >> 1; |
68 |
|
dst[4] = (src1[4] + src2[4] + round) >> 1; |
69 |
|
dst[5] = (src1[5] + src2[5] + round) >> 1; |
70 |
|
dst[6] = (src1[6] + src2[6] + round) >> 1; |
71 |
|
dst[7] = (src1[7] + src2[7] + round) >> 1; |
72 |
|
|
73 |
|
dst += stride; |
74 |
|
src1 += stride; |
75 |
|
src2 += stride; |
76 |
|
} |
77 |
|
} |
78 |
|
|
79 |
|
void interpolate8x8_avg4_c(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) |
80 |
|
{ |
81 |
|
int32_t i; |
82 |
|
const int32_t round = 2 - rounding; |
83 |
|
|
84 |
|
for(i = 0; i < 8; i++) |
85 |
|
{ |
86 |
|
dst[0] = (src1[0] + src2[0] + src3[0] + src4[0] + round) >> 2; |
87 |
|
dst[1] = (src1[1] + src2[1] + src3[1] + src4[1] + round) >> 2; |
88 |
|
dst[2] = (src1[2] + src2[2] + src3[2] + src4[2] + round) >> 2; |
89 |
|
dst[3] = (src1[3] + src2[3] + src3[3] + src4[3] + round) >> 2; |
90 |
|
dst[4] = (src1[4] + src2[4] + src3[4] + src4[4] + round) >> 2; |
91 |
|
dst[5] = (src1[5] + src2[5] + src3[5] + src4[5] + round) >> 2; |
92 |
|
dst[6] = (src1[6] + src2[6] + src3[6] + src4[6] + round) >> 2; |
93 |
|
dst[7] = (src1[7] + src2[7] + src3[7] + src4[7] + round) >> 2; |
94 |
|
|
95 |
|
dst += stride; |
96 |
|
src1 += stride; |
97 |
|
src2 += stride; |
98 |
|
src3 += stride; |
99 |
|
src4 += stride; |
100 |
|
} |
101 |
|
} |
102 |
|
|
103 |
/* dst = interpolate(src) */ |
// dst = interpolate(src) |
104 |
|
|
105 |
void |
void |
106 |
interpolate8x8_halfpel_h_c(uint8_t * const dst, |
interpolate8x8_halfpel_h_c(uint8_t * const dst, |
164 |
} |
} |
165 |
} |
} |
166 |
|
|
167 |
/* add by MinChen <chenm001@163.com> */ |
/************************************************************* |
168 |
/* interpolate8x8 two pred block */ |
* QPEL STUFF STARTS HERE * |
169 |
void |
*************************************************************/ |
170 |
interpolate8x8_c(uint8_t * const dst, |
|
171 |
const uint8_t * const src, |
void interpolate8x8_6tap_lowpass_h_c(uint8_t *dst, uint8_t *src, int32_t stride, int32_t rounding) |
|
const uint32_t x, |
|
|
const uint32_t y, |
|
|
const uint32_t stride) |
|
172 |
{ |
{ |
173 |
uint32_t i, j; |
int32_t i; |
174 |
|
uint8_t round_add = 16 - rounding; |
175 |
|
|
176 |
for (j = 0; j < 8; j++) { |
for(i = 0; i < 8; i++) |
177 |
for (i = 0; i < 8; i++) { |
{ |
178 |
int32_t tot = |
|
179 |
((src[(y + j) * stride + x + i] + |
dst[0] = CLIP((((src[-2] + src[3]) + 5 * (((src[0] + src[1])<<2) - (src[-1] + src[2])) + round_add) >> 5), 0, 255); |
180 |
dst[(y + j) * stride + x + i] + 1) >> 1); |
dst[1] = CLIP((((src[-1] + src[4]) + 5 * (((src[1] + src[2])<<2) - (src[0] + src[3])) + round_add) >> 5), 0, 255); |
181 |
dst[(y + j) * stride + x + i] = (uint8_t) tot; |
dst[2] = CLIP((((src[0] + src[5]) + 5 * (((src[2] + src[3])<<2) - (src[1] + src[4])) + round_add) >> 5), 0, 255); |
182 |
} |
dst[3] = CLIP((((src[1] + src[6]) + 5 * (((src[3] + src[4])<<2) - (src[2] + src[5])) + round_add) >> 5), 0, 255); |
183 |
|
dst[4] = CLIP((((src[2] + src[7]) + 5 * (((src[4] + src[5])<<2) - (src[3] + src[6])) + round_add) >> 5), 0, 255); |
184 |
|
dst[5] = CLIP((((src[3] + src[8]) + 5 * (((src[5] + src[6])<<2) - (src[4] + src[7])) + round_add) >> 5), 0, 255); |
185 |
|
dst[6] = CLIP((((src[4] + src[9]) + 5 * (((src[6] + src[7])<<2) - (src[5] + src[8])) + round_add) >> 5), 0, 255); |
186 |
|
dst[7] = CLIP((((src[5] + src[10]) + 5 * (((src[7] + src[8])<<2) - (src[6] + src[9])) + round_add) >> 5), 0, 255); |
187 |
|
|
188 |
|
dst += stride; |
189 |
|
src += stride; |
190 |
} |
} |
191 |
} |
} |
192 |
|
|
193 |
/************************************************************* |
void interpolate16x16_lowpass_h_c(uint8_t *dst, uint8_t *src, int32_t stride, int32_t rounding) |
194 |
* QPEL STUFF STARTS HERE * |
{ |
195 |
*************************************************************/ |
int32_t i; |
196 |
|
uint8_t round_add = 16 - rounding; |
197 |
|
|
198 |
#define CLIP(X,A,B) (X < A) ? (A) : ((X > B) ? (B) : (X)) |
for(i = 0; i < 17; i++) |
199 |
|
{ |
200 |
|
|
201 |
|
dst[0] = CLIP(((7 * ((src[0]<<1) - src[2]) + 23 * src[1] + 3 * src[3] - src[4] + round_add) >> 5), 0, 255); |
202 |
|
dst[1] = CLIP(((19 * src[1] + 20 * src[2] - src[5] + 3 * (src[4] - src[0] - (src[3]<<1)) + round_add) >> 5), 0, 255); |
203 |
|
dst[2] = CLIP(((20 * (src[2] + src[3]) + (src[0]<<1) + 3 * (src[5] - ((src[1] + src[4])<<1)) - src[6] + round_add) >> 5), 0, 255); |
204 |
|
|
205 |
|
dst[3] = CLIP(((20 * (src[3] + src[4]) + 3 * ((src[6] + src[1]) - ((src[2] + src[5])<<1)) - (src[0] + src[7]) + round_add) >> 5), 0, 255); |
206 |
|
dst[4] = CLIP(((20 * (src[4] + src[5]) - 3 * (((src[3] + src[6])<<1) - (src[2] + src[7])) - (src[1] + src[8]) + round_add) >> 5), 0, 255); |
207 |
|
dst[5] = CLIP(((20 * (src[5] + src[6]) - 3 * (((src[4] + src[7])<<1) - (src[3] + src[8])) - (src[2] + src[9]) + round_add) >> 5), 0, 255); |
208 |
|
dst[6] = CLIP(((20 * (src[6] + src[7]) - 3 * (((src[5] + src[8])<<1) - (src[4] + src[9])) - (src[3] + src[10]) + round_add) >> 5), 0, 255); |
209 |
|
dst[7] = CLIP(((20 * (src[7] + src[8]) - 3 * (((src[6] + src[9])<<1) - (src[5] + src[10])) - (src[4] + src[11]) + round_add) >> 5), 0, 255); |
210 |
|
dst[8] = CLIP(((20 * (src[8] + src[9]) - 3 * (((src[7] + src[10])<<1) - (src[6] + src[11])) - (src[5] + src[12]) + round_add) >> 5), 0, 255); |
211 |
|
dst[9] = CLIP(((20 * (src[9] + src[10]) - 3 * (((src[8] + src[11])<<1) - (src[7] + src[12])) - (src[6] + src[13]) + round_add) >> 5), 0, 255); |
212 |
|
dst[10] = CLIP(((20 * (src[10] + src[11]) - 3 * (((src[9] + src[12])<<1) - (src[8] + src[13])) - (src[7] + src[14]) + round_add) >> 5), 0, 255); |
213 |
|
dst[11] = CLIP(((20 * (src[11] + src[12]) - 3 * (((src[10] + src[13])<<1) - (src[9] + src[14])) - (src[8] + src[15]) + round_add) >> 5), 0, 255); |
214 |
|
dst[12] = CLIP(((20 * (src[12] + src[13]) - 3 * (((src[11] + src[14])<<1) - (src[10] + src[15])) - (src[9] + src[16]) + round_add) >> 5), 0, 255); |
215 |
|
|
216 |
|
dst[13] = CLIP(((20 * (src[13] + src[14]) + (src[16]<<1) + 3 * (src[11] - ((src[12] + src[15]) << 1)) - src[10] + round_add) >> 5), 0, 255); |
217 |
|
dst[14] = CLIP(((19 * src[15] + 20 * src[14] + 3 * (src[12] - src[16] - (src[13] << 1)) - src[11] + round_add) >> 5), 0, 255); |
218 |
|
dst[15] = CLIP(((23 * src[15] + 7 * ((src[16]<<1) - src[14]) + 3 * src[13] - src[12] + round_add) >> 5), 0, 255); |
219 |
|
|
220 |
|
dst += stride; |
221 |
|
src += stride; |
222 |
|
} |
223 |
|
} |
224 |
|
|
225 |
void interpolate8x8_lowpass_h(uint8_t *dst, uint8_t *src, int32_t dst_stride, int32_t src_stride, int32_t rounding) |
void interpolate8x8_lowpass_h_c(uint8_t *dst, uint8_t *src, int32_t stride, int32_t rounding) |
226 |
{ |
{ |
227 |
int32_t i; |
int32_t i; |
228 |
|
uint8_t round_add = 16 - rounding; |
229 |
|
|
230 |
for(i = 0; i < 8; i++) |
for(i = 0; i < 9; i++) |
231 |
{ |
{ |
|
dst[0] = CLIP((((src[0] + src[1]) * 160 - (src[0] + src[2]) * 48 + (src[1] + src[3]) * 24 - (src[2] + src[4]) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst[1] = CLIP((((src[1] + src[2]) * 160 - (src[0] + src[3]) * 48 + (src[0] + src[4]) * 24 - (src[1] + src[5]) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst[2] = CLIP((((src[2] + src[3]) * 160 - (src[1] + src[4]) * 48 + (src[0] + src[5]) * 24 - (src[0] + src[6]) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst[3] = CLIP((((src[3] + src[4]) * 160 - (src[2] + src[5]) * 48 + (src[1] + src[6]) * 24 - (src[0] + src[7]) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst[4] = CLIP((((src[4] + src[5]) * 160 - (src[3] + src[6]) * 48 + (src[2] + src[7]) * 24 - (src[1] + src[8]) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst[5] = CLIP((((src[5] + src[6]) * 160 - (src[4] + src[7]) * 48 + (src[3] + src[8]) * 24 - (src[2] + src[8]) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst[6] = CLIP((((src[6] + src[7]) * 160 - (src[5] + src[8]) * 48 + (src[4] + src[8]) * 24 - (src[3] + src[7]) * 8 + (128 - rounding)) / 256), 0, 255); |
|
|
dst[7] = CLIP((((src[7] + src[8]) * 160 - (src[6] + src[8]) * 48 + (src[5] + src[7]) * 24 - (src[4] + src[6]) * 8 + (128 - rounding)) / 256), 0, 255); |
|
232 |
|
|
233 |
dst += dst_stride; |
dst[0] = CLIP(((7 * ((src[0]<<1) - src[2]) + 23 * src[1] + 3 * src[3] - src[4] + round_add) >> 5), 0, 255); |
234 |
src += src_stride; |
dst[1] = CLIP(((19 * src[1] + 20 * src[2] - src[5] + 3 * (src[4] - src[0] - (src[3]<<1)) + round_add) >> 5), 0, 255); |
235 |
|
dst[2] = CLIP(((20 * (src[2] + src[3]) + (src[0]<<1) + 3 * (src[5] - ((src[1] + src[4])<<1)) - src[6] + round_add) >> 5), 0, 255); |
236 |
|
dst[3] = CLIP(((20 * (src[3] + src[4]) + 3 * ((src[6] + src[1]) - ((src[2] + src[5])<<1)) - (src[0] + src[7]) + round_add) >> 5), 0, 255); |
237 |
|
dst[4] = CLIP(((20 * (src[4] + src[5]) - 3 * (((src[3] + src[6])<<1) - (src[2] + src[7])) - (src[1] + src[8]) + round_add) >> 5), 0, 255); |
238 |
|
dst[5] = CLIP(((20 * (src[5] + src[6]) + (src[8]<<1) + 3 * (src[3] - ((src[4] + src[7]) << 1)) - src[2] + round_add) >> 5), 0, 255); |
239 |
|
dst[6] = CLIP(((19 * src[7] + 20 * src[6] + 3 * (src[4] - src[8] - (src[5] << 1)) - src[3] + round_add) >> 5), 0, 255); |
240 |
|
dst[7] = CLIP(((23 * src[7] + 7 * ((src[8]<<1) - src[6]) + 3 * src[5] - src[4] + round_add) >> 5), 0, 255); |
241 |
|
|
242 |
|
dst += stride; |
243 |
|
src += stride; |
244 |
} |
} |
245 |
} |
} |
246 |
|
|
247 |
void interpolate8x8_lowpass_v(uint8_t *dst, uint8_t *src, int32_t dst_stride, int32_t src_stride, int32_t rounding) |
void interpolate8x8_6tap_lowpass_v_c(uint8_t *dst, uint8_t *src, int32_t stride, int32_t rounding) |
248 |
{ |
{ |
249 |
int32_t i; |
int32_t i; |
250 |
|
uint8_t round_add = 16 - rounding; |
251 |
|
|
252 |
for(i = 0; i < 8; i++) |
for(i = 0; i < 8; i++) |
253 |
{ |
{ |
254 |
|
int32_t src_2 = src[-2*stride]; |
255 |
|
int32_t src_1 = src[-stride]; |
256 |
int32_t src0 = src[0]; |
int32_t src0 = src[0]; |
257 |
int32_t src1 = src[src_stride]; |
int32_t src1 = src[stride]; |
258 |
int32_t src2 = src[2 * src_stride]; |
int32_t src2 = src[2 * stride]; |
259 |
int32_t src3 = src[3 * src_stride]; |
int32_t src3 = src[3 * stride]; |
260 |
int32_t src4 = src[4 * src_stride]; |
int32_t src4 = src[4 * stride]; |
261 |
int32_t src5 = src[5 * src_stride]; |
int32_t src5 = src[5 * stride]; |
262 |
int32_t src6 = src[6 * src_stride]; |
int32_t src6 = src[6 * stride]; |
263 |
int32_t src7 = src[7 * src_stride]; |
int32_t src7 = src[7 * stride]; |
264 |
int32_t src8 = src[8 * src_stride]; |
int32_t src8 = src[8 * stride]; |
265 |
|
int32_t src9 = src[9 * stride]; |
266 |
dst[0] = CLIP((((src0 + src1) * 160 - (src0 + src2) * 48 + (src1 + src3) * 24 - (src2 + src4) * 8 + (128 - rounding)) / 256), 0, 255); |
int32_t src10 = src[10 * stride]; |
267 |
dst[dst_stride] = CLIP((((src1 + src2) * 160 - (src0 + src3) * 48 + (src0 + src4) * 24 - (src1 + src5) * 8 + (128 - rounding)) / 256), 0, 255); |
|
268 |
dst[2 * dst_stride] = CLIP((((src2 + src3) * 160 - (src1 + src4) * 48 + (src0 + src5) * 24 - (src0 + src6) * 8 + (128 - rounding)) / 256), 0, 255); |
dst[0] = CLIP((((src_2 + src3) + 5 * (((src0 + src1)<<2) - (src_1 + src2)) + round_add) >> 5), 0, 255); |
269 |
dst[3 * dst_stride] = CLIP((((src3 + src4) * 160 - (src2 + src5) * 48 + (src1 + src6) * 24 - (src0 + src7) * 8 + (128 - rounding)) / 256), 0, 255); |
dst[stride] = CLIP((((src_1 + src4) + 5 * (((src1 + src2)<<2) - (src0 + src3)) + round_add) >> 5), 0, 255); |
270 |
dst[4 * dst_stride] = CLIP((((src4 + src5) * 160 - (src3 + src6) * 48 + (src2 + src7) * 24 - (src1 + src8) * 8 + (128 - rounding)) / 256), 0, 255); |
dst[2 * stride] = CLIP((((src0 + src5) + 5 * (((src2 + src3)<<2) - (src1 + src4)) + round_add) >> 5), 0, 255); |
271 |
dst[5 * dst_stride] = CLIP((((src5 + src6) * 160 - (src4 + src7) * 48 + (src3 + src8) * 24 - (src2 + src8) * 8 + (128 - rounding)) / 256), 0, 255); |
dst[3 * stride] = CLIP((((src1 + src6) + 5 * (((src3 + src4)<<2) - (src2 + src5)) + round_add) >> 5), 0, 255); |
272 |
dst[6 * dst_stride] = CLIP((((src6 + src7) * 160 - (src5 + src8) * 48 + (src4 + src8) * 24 - (src3 + src7) * 8 + (128 - rounding)) / 256), 0, 255); |
dst[4 * stride] = CLIP((((src2 + src7) + 5 * (((src4 + src5)<<2) - (src3 + src6)) + round_add) >> 5), 0, 255); |
273 |
dst[7 * dst_stride] = CLIP((((src7 + src8) * 160 - (src6 + src8) * 48 + (src5 + src7) * 24 - (src4 + src6) * 8 + (128 - rounding)) / 256), 0, 255); |
dst[5 * stride] = CLIP((((src3 + src8) + 5 * (((src5 + src6)<<2) - (src4 + src7)) + round_add) >> 5), 0, 255); |
274 |
|
dst[6 * stride] = CLIP((((src4 + src9) + 5 * (((src6 + src7)<<2) - (src5 + src8)) + round_add) >> 5), 0, 255); |
275 |
|
dst[7 * stride] = CLIP((((src5 + src10) + 5 * (((src7 + src8)<<2) - (src6 + src9)) + round_add) >> 5), 0, 255); |
276 |
|
|
277 |
dst++; |
dst++; |
278 |
src++; |
src++; |
279 |
} |
} |
280 |
} |
} |
281 |
|
|
282 |
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) |
void interpolate16x16_lowpass_v_c(uint8_t *dst, uint8_t *src, int32_t stride, int32_t rounding) |
283 |
{ |
{ |
|
uint8_t data[72]; |
|
|
|
|
284 |
int32_t i; |
int32_t i; |
285 |
|
uint8_t round_add = 16 - rounding; |
286 |
|
|
287 |
|
for(i = 0; i < 17; i++) |
|
for(i = 0; i < 9; i++) |
|
288 |
{ |
{ |
289 |
dst2[0] = data[8 * i + 0] = CLIP((((src[0] + src[1]) * 160 - (src[0] + src[2]) * 48 + (src[1] + src[3]) * 24 - (src[2] + src[4]) * 8 + (128 - rounding)) / 256), 0, 255); |
int32_t src0 = src[0]; |
290 |
dst2[1] = data[8 * i + 1] = CLIP((((src[1] + src[2]) * 160 - (src[0] + src[3]) * 48 + (src[0] + src[4]) * 24 - (src[1] + src[5]) * 8 + (128 - rounding)) / 256), 0, 255); |
int32_t src1 = src[stride]; |
291 |
dst2[2] = data[8 * i + 2] = CLIP((((src[2] + src[3]) * 160 - (src[1] + src[4]) * 48 + (src[0] + src[5]) * 24 - (src[0] + src[6]) * 8 + (128 - rounding)) / 256), 0, 255); |
int32_t src2 = src[2 * stride]; |
292 |
dst2[3] = data[8 * i + 3] = CLIP((((src[3] + src[4]) * 160 - (src[2] + src[5]) * 48 + (src[1] + src[6]) * 24 - (src[0] + src[7]) * 8 + (128 - rounding)) / 256), 0, 255); |
int32_t src3 = src[3 * stride]; |
293 |
dst2[4] = data[8 * i + 4] = CLIP((((src[4] + src[5]) * 160 - (src[3] + src[6]) * 48 + (src[2] + src[7]) * 24 - (src[1] + src[8]) * 8 + (128 - rounding)) / 256), 0, 255); |
int32_t src4 = src[4 * stride]; |
294 |
dst2[5] = data[8 * i + 5] = CLIP((((src[5] + src[6]) * 160 - (src[4] + src[7]) * 48 + (src[3] + src[8]) * 24 - (src[2] + src[8]) * 8 + (128 - rounding)) / 256), 0, 255); |
int32_t src5 = src[5 * stride]; |
295 |
dst2[6] = data[8 * i + 6] = CLIP((((src[6] + src[7]) * 160 - (src[5] + src[8]) * 48 + (src[4] + src[8]) * 24 - (src[3] + src[7]) * 8 + (128 - rounding)) / 256), 0, 255); |
int32_t src6 = src[6 * stride]; |
296 |
dst2[7] = data[8 * i + 7] = CLIP((((src[7] + src[8]) * 160 - (src[6] + src[8]) * 48 + (src[5] + src[7]) * 24 - (src[4] + src[6]) * 8 + (128 - rounding)) / 256), 0, 255); |
int32_t src7 = src[7 * stride]; |
297 |
|
int32_t src8 = src[8 * stride]; |
298 |
|
int32_t src9 = src[9 * stride]; |
299 |
|
int32_t src10 = src[10 * stride]; |
300 |
|
int32_t src11 = src[11 * stride]; |
301 |
|
int32_t src12 = src[12 * stride]; |
302 |
|
int32_t src13 = src[13 * stride]; |
303 |
|
int32_t src14 = src[14 * stride]; |
304 |
|
int32_t src15 = src[15 * stride]; |
305 |
|
int32_t src16 = src[16 * stride]; |
306 |
|
|
307 |
|
|
308 |
|
dst[0] = CLIP(((7 * ((src0<<1) - src2) + 23 * src1 + 3 * src3 - src4 + round_add) >> 5), 0, 255); |
309 |
|
dst[stride] = CLIP(((19 * src1 + 20 * src2 - src5 + 3 * (src4 - src0 - (src3<<1)) + round_add) >> 5), 0, 255); |
310 |
|
dst[2*stride] = CLIP(((20 * (src2 + src3) + (src0<<1) + 3 * (src5 - ((src1 + src4)<<1)) - src6 + round_add) >> 5), 0, 255); |
311 |
|
|
312 |
|
dst[3*stride] = CLIP(((20 * (src3 + src4) + 3 * ((src6 + src1) - ((src2 + src5)<<1)) - (src0 + src7) + round_add) >> 5), 0, 255); |
313 |
|
dst[4*stride] = CLIP(((20 * (src4 + src5) - 3 * (((src3 + src6)<<1) - (src2 + src7)) - (src1 + src8) + round_add) >> 5), 0, 255); |
314 |
|
dst[5*stride] = CLIP(((20 * (src5 + src6) - 3 * (((src4 + src7)<<1) - (src3 + src8)) - (src2 + src9) + round_add) >> 5), 0, 255); |
315 |
|
dst[6*stride] = CLIP(((20 * (src6 + src7) - 3 * (((src5 + src8)<<1) - (src4 + src9)) - (src3 + src10) + round_add) >> 5), 0, 255); |
316 |
|
dst[7*stride] = CLIP(((20 * (src7 + src8) - 3 * (((src6 + src9)<<1) - (src5 + src10)) - (src4 + src11) + round_add) >> 5), 0, 255); |
317 |
|
dst[8*stride] = CLIP(((20 * (src8 + src9) - 3 * (((src7 + src10)<<1) - (src6 + src11)) - (src5 + src12) + round_add) >> 5), 0, 255); |
318 |
|
dst[9*stride] = CLIP(((20 * (src9 + src10) - 3 * (((src8 + src11)<<1) - (src7 + src12)) - (src6 + src13) + round_add) >> 5), 0, 255); |
319 |
|
dst[10*stride] = CLIP(((20 * (src10 + src11) - 3 * (((src9 + src12)<<1) - (src8 + src13)) - (src7 + src14) + round_add) >> 5), 0, 255); |
320 |
|
dst[11*stride] = CLIP(((20 * (src11 + src12) - 3 * (((src10 + src13)<<1) - (src9 + src14)) - (src8 + src15) + round_add) >> 5), 0, 255); |
321 |
|
dst[12*stride] = CLIP(((20 * (src12 + src13) - 3 * (((src11 + src14)<<1) - (src10 + src15)) - (src9 + src16) + round_add) >> 5), 0, 255); |
322 |
|
|
323 |
|
dst[13*stride] = CLIP(((20 * (src13 + src14) + (src16<<1) + 3 * (src11 - ((src12 + src15) << 1)) - src10 + round_add) >> 5), 0, 255); |
324 |
|
dst[14*stride] = CLIP(((19 * src15 + 20 * src14 + 3 * (src12 - src16 - (src13 << 1)) - src11 + round_add) >> 5), 0, 255); |
325 |
|
dst[15*stride] = CLIP(((23 * src15 + 7 * ((src16<<1) - src14) + 3 * src13 - src12 + round_add) >> 5), 0, 255); |
326 |
|
|
327 |
src += src_stride; |
dst++; |
328 |
dst2 += dst2_stride; |
src++; |
329 |
|
} |
330 |
} |
} |
331 |
|
|
332 |
for(i = 0; i < 8; i++) |
void interpolate8x8_lowpass_v_c(uint8_t *dst, uint8_t *src, int32_t stride, int32_t rounding) |
333 |
|
{ |
334 |
|
int32_t i; |
335 |
|
uint8_t round_add = 16 - rounding; |
336 |
|
|
337 |
|
for(i = 0; i < 9; i++) |
338 |
{ |
{ |
339 |
int32_t src0 = data[i]; |
int32_t src0 = src[0]; |
340 |
int32_t src1 = data[8 + i]; |
int32_t src1 = src[stride]; |
341 |
int32_t src2 = data[2 * 8 + i]; |
int32_t src2 = src[2 * stride]; |
342 |
int32_t src3 = data[3 * 8 + i]; |
int32_t src3 = src[3 * stride]; |
343 |
int32_t src4 = data[4 * 8 + i]; |
int32_t src4 = src[4 * stride]; |
344 |
int32_t src5 = data[5 * 8 + i]; |
int32_t src5 = src[5 * stride]; |
345 |
int32_t src6 = data[6 * 8 + i]; |
int32_t src6 = src[6 * stride]; |
346 |
int32_t src7 = data[7 * 8 + i]; |
int32_t src7 = src[7 * stride]; |
347 |
int32_t src8 = data[8 * 8 + i]; |
int32_t src8 = src[8 * stride]; |
348 |
|
|
349 |
dst1[0] = CLIP((((src0 + src1) * 160 - (src0 + src2) * 48 + (src1 + src3) * 24 - (src2 + src4) * 8 + (128 - rounding)) / 256), 0, 255); |
dst[0] = CLIP(((7 * ((src0<<1) - src2) + 23 * src1 + 3 * src3 - src4 + round_add) >> 5), 0, 255); |
350 |
dst1[dst1_stride] = CLIP((((src1 + src2) * 160 - (src0 + src3) * 48 + (src0 + src4) * 24 - (src1 + src5) * 8 + (128 - rounding)) / 256), 0, 255); |
dst[stride] = CLIP(((19 * src1 + 20 * src2 - src5 + 3 * (src4 - src0 - (src3 << 1)) + round_add) >> 5), 0, 255); |
351 |
dst1[2 * dst1_stride] = CLIP((((src2 + src3) * 160 - (src1 + src4) * 48 + (src0 + src5) * 24 - (src0 + src6) * 8 + (128 - rounding)) / 256), 0, 255); |
dst[2 * stride] = CLIP(((20 * (src2 + src3) + (src0<<1) + 3 * (src5 - ((src1 + src4) <<1 )) - src6 + round_add) >> 5), 0, 255); |
352 |
dst1[3 * dst1_stride] = CLIP((((src3 + src4) * 160 - (src2 + src5) * 48 + (src1 + src6) * 24 - (src0 + src7) * 8 + (128 - rounding)) / 256), 0, 255); |
dst[3 * stride] = CLIP(((20 * (src3 + src4) + 3 * ((src6 + src1) - ((src2 + src5)<<1)) - (src0 + src7) + round_add) >> 5), 0, 255); |
353 |
dst1[4 * dst1_stride] = CLIP((((src4 + src5) * 160 - (src3 + src6) * 48 + (src2 + src7) * 24 - (src1 + src8) * 8 + (128 - rounding)) / 256), 0, 255); |
dst[4 * stride] = CLIP(((20 * (src4 + src5) + 3 * ((src2 + src7) - ((src3 + src6)<<1)) - (src1 + src8) + round_add) >> 5), 0, 255); |
354 |
dst1[5 * dst1_stride] = CLIP((((src5 + src6) * 160 - (src4 + src7) * 48 + (src3 + src8) * 24 - (src2 + src8) * 8 + (128 - rounding)) / 256), 0, 255); |
dst[5 * stride] = CLIP(((20 * (src5 + src6) + (src8<<1) + 3 * (src3 - ((src4 + src7) << 1)) - src2 + round_add) >> 5), 0, 255); |
355 |
dst1[6 * dst1_stride] = CLIP((((src6 + src7) * 160 - (src5 + src8) * 48 + (src4 + src8) * 24 - (src3 + src7) * 8 + (128 - rounding)) / 256), 0, 255); |
dst[6 * stride] = CLIP(((19 * src7 + 20 * src6 - src3 + 3 * (src4 - src8 - (src5 << 1)) + round_add) >> 5), 0, 255); |
356 |
dst1[7 * dst1_stride] = CLIP((((src7 + src8) * 160 - (src6 + src8) * 48 + (src5 + src7) * 24 - (src4 + src6) * 8 + (128 - rounding)) / 256), 0, 255); |
dst[7 * stride] = CLIP(((7 * ((src8<<1) - src6) + 23 * src7 + 3 * src5 - src4 + round_add) >> 5), 0, 255); |
357 |
|
|
358 |
dst1++; |
dst++; |
359 |
|
src++; |
360 |
} |
} |
361 |
} |
} |
362 |
|
|
363 |
void interpolate8x8_bilinear2(uint8_t *dst, uint8_t *src1, uint8_t *src2, int32_t dst_stride, int32_t src_stride, int32_t rounding) |
void interpolate16x16_lowpass_hv_c(uint8_t *dst1, uint8_t *dst2, uint8_t *src, int32_t stride, int32_t rounding) |
364 |
{ |
{ |
365 |
int32_t i; |
int32_t i; |
366 |
|
uint8_t round_add = 16 - rounding; |
367 |
|
uint8_t *h_ptr = dst2; |
368 |
|
|
369 |
for(i = 0; i < 8; i++) |
for(i = 0; i < 17; i++) |
370 |
{ |
{ |
371 |
dst[0] = (src1[0] + src2[0] + (1 - rounding)) >> 1; |
|
372 |
dst[1] = (src1[1] + src2[1] + (1 - rounding)) >> 1; |
h_ptr[0] = CLIP(((7 * ((src[0]<<1) - src[2]) + 23 * src[1] + 3 * src[3] - src[4] + round_add) >> 5), 0, 255); |
373 |
dst[2] = (src1[2] + src2[2] + (1 - rounding)) >> 1; |
h_ptr[1] = CLIP(((19 * src[1] + 20 * src[2] - src[5] + 3 * (src[4] - src[0] - (src[3]<<1)) + round_add) >> 5), 0, 255); |
374 |
dst[3] = (src1[3] + src2[3] + (1 - rounding)) >> 1; |
h_ptr[2] = CLIP(((20 * (src[2] + src[3]) + (src[0]<<1) + 3 * (src[5] - ((src[1] + src[4])<<1)) - src[6] + round_add) >> 5), 0, 255); |
375 |
dst[4] = (src1[4] + src2[4] + (1 - rounding)) >> 1; |
|
376 |
dst[5] = (src1[5] + src2[5] + (1 - rounding)) >> 1; |
h_ptr[3] = CLIP(((20 * (src[3] + src[4]) + 3 * ((src[6] + src[1]) - ((src[2] + src[5])<<1)) - (src[0] + src[7]) + round_add) >> 5), 0, 255); |
377 |
dst[6] = (src1[6] + src2[6] + (1 - rounding)) >> 1; |
h_ptr[4] = CLIP(((20 * (src[4] + src[5]) - 3 * (((src[3] + src[6])<<1) - (src[2] + src[7])) - (src[1] + src[8]) + round_add) >> 5), 0, 255); |
378 |
dst[7] = (src1[7] + src2[7] + (1 - rounding)) >> 1; |
h_ptr[5] = CLIP(((20 * (src[5] + src[6]) - 3 * (((src[4] + src[7])<<1) - (src[3] + src[8])) - (src[2] + src[9]) + round_add) >> 5), 0, 255); |
379 |
|
h_ptr[6] = CLIP(((20 * (src[6] + src[7]) - 3 * (((src[5] + src[8])<<1) - (src[4] + src[9])) - (src[3] + src[10]) + round_add) >> 5), 0, 255); |
380 |
dst += dst_stride; |
h_ptr[7] = CLIP(((20 * (src[7] + src[8]) - 3 * (((src[6] + src[9])<<1) - (src[5] + src[10])) - (src[4] + src[11]) + round_add) >> 5), 0, 255); |
381 |
src1 += src_stride; |
h_ptr[8] = CLIP(((20 * (src[8] + src[9]) - 3 * (((src[7] + src[10])<<1) - (src[6] + src[11])) - (src[5] + src[12]) + round_add) >> 5), 0, 255); |
382 |
src2 += 8; |
h_ptr[9] = CLIP(((20 * (src[9] + src[10]) - 3 * (((src[8] + src[11])<<1) - (src[7] + src[12])) - (src[6] + src[13]) + round_add) >> 5), 0, 255); |
383 |
|
h_ptr[10] = CLIP(((20 * (src[10] + src[11]) - 3 * (((src[9] + src[12])<<1) - (src[8] + src[13])) - (src[7] + src[14]) + round_add) >> 5), 0, 255); |
384 |
|
h_ptr[11] = CLIP(((20 * (src[11] + src[12]) - 3 * (((src[10] + src[13])<<1) - (src[9] + src[14])) - (src[8] + src[15]) + round_add) >> 5), 0, 255); |
385 |
|
h_ptr[12] = CLIP(((20 * (src[12] + src[13]) - 3 * (((src[11] + src[14])<<1) - (src[10] + src[15])) - (src[9] + src[16]) + round_add) >> 5), 0, 255); |
386 |
|
|
387 |
|
h_ptr[13] = CLIP(((20 * (src[13] + src[14]) + (src[16]<<1) + 3 * (src[11] - ((src[12] + src[15]) << 1)) - src[10] + round_add) >> 5), 0, 255); |
388 |
|
h_ptr[14] = CLIP(((19 * src[15] + 20 * src[14] + 3 * (src[12] - src[16] - (src[13] << 1)) - src[11] + round_add) >> 5), 0, 255); |
389 |
|
h_ptr[15] = CLIP(((23 * src[15] + 7 * ((src[16]<<1) - src[14]) + 3 * src[13] - src[12] + round_add) >> 5), 0, 255); |
390 |
|
|
391 |
|
h_ptr += stride; |
392 |
|
src += stride; |
393 |
} |
} |
394 |
|
|
395 |
|
interpolate16x16_lowpass_v_c(dst1, dst2, stride, rounding); |
396 |
|
|
397 |
} |
} |
398 |
|
|
399 |
void interpolate8x8_bilinear4(uint8_t *dst, uint8_t *src1, uint8_t *src2, uint8_t *src3, uint8_t *src4, int32_t stride, int32_t rounding) |
void interpolate8x8_lowpass_hv_c(uint8_t *dst1, uint8_t *dst2, uint8_t *src, int32_t stride, int32_t rounding) |
400 |
{ |
{ |
401 |
int32_t i; |
int32_t i; |
402 |
|
uint8_t round_add = 16 - rounding; |
403 |
|
uint8_t *h_ptr = dst2; |
404 |
|
|
405 |
for(i = 0; i < 8; i++) |
for(i = 0; i < 9; i++) |
406 |
{ |
{ |
|
dst[0] = (src1[0] + src2[0] + src3[0] + src4[0] + (2 - rounding)) >> 2; |
|
|
dst[1] = (src1[1] + src2[1] + src3[1] + src4[1] + (2 - rounding)) >> 2; |
|
|
dst[2] = (src1[2] + src2[2] + src3[2] + src4[2] + (2 - rounding)) >> 2; |
|
|
dst[3] = (src1[3] + src2[3] + src3[3] + src4[3] + (2 - rounding)) >> 2; |
|
|
dst[4] = (src1[4] + src2[4] + src3[4] + src4[4] + (2 - rounding)) >> 2; |
|
|
dst[5] = (src1[5] + src2[5] + src3[5] + src4[5] + (2 - rounding)) >> 2; |
|
|
dst[6] = (src1[6] + src2[6] + src3[6] + src4[6] + (2 - rounding)) >> 2; |
|
|
dst[7] = (src1[7] + src2[7] + src3[7] + src4[7] + (2 - rounding)) >> 2; |
|
407 |
|
|
408 |
dst += stride; |
h_ptr[0] = CLIP(((7 * ((src[0]<<1) - src[2]) + 23 * src[1] + 3 * src[3] - src[4] + round_add) >> 5), 0, 255); |
409 |
src1 += stride; |
h_ptr[1] = CLIP(((19 * src[1] + 20 * src[2] - src[5] + 3 * (src[4] - src[0] - (src[3]<<1)) + round_add) >> 5), 0, 255); |
410 |
src2 += 8; |
h_ptr[2] = CLIP(((20 * (src[2] + src[3]) + (src[0]<<1) + 3 * (src[5] - ((src[1] + src[4])<<1)) - src[6] + round_add) >> 5), 0, 255); |
411 |
src3 += 8; |
h_ptr[3] = CLIP(((20 * (src[3] + src[4]) + 3 * ((src[6] + src[1]) - ((src[2] + src[5])<<1)) - (src[0] + src[7]) + round_add) >> 5), 0, 255); |
412 |
src4 += 8; |
h_ptr[4] = CLIP(((20 * (src[4] + src[5]) - 3 * (((src[3] + src[6])<<1) - (src[2] + src[7])) - (src[1] + src[8]) + round_add) >> 5), 0, 255); |
413 |
|
h_ptr[5] = CLIP(((20 * (src[5] + src[6]) + (src[8]<<1) + 3 * (src[3] - ((src[4] + src[7]) << 1)) - src[2] + round_add) >> 5), 0, 255); |
414 |
|
h_ptr[6] = CLIP(((19 * src[7] + 20 * src[6] + 3 * (src[4] - src[8] - (src[5] << 1)) - src[3] + round_add) >> 5), 0, 255); |
415 |
|
h_ptr[7] = CLIP(((23 * src[7] + 7 * ((src[8]<<1) - src[6]) + 3 * src[5] - src[4] + round_add) >> 5), 0, 255); |
416 |
|
|
417 |
|
h_ptr += stride; |
418 |
|
src += stride; |
419 |
} |
} |
420 |
|
|
421 |
|
interpolate8x8_lowpass_v_c(dst1, dst2, stride, rounding); |
422 |
|
|
423 |
} |
} |