Parent Directory | Revision Log
Revision 906 - (view) (download)
1 : | edgomez | 851 | /************************************************************************** |
2 : | Isibaar | 333 | * |
3 : | edgomez | 851 | * XVID MPEG-4 VIDEO CODEC |
4 : | * 8x8 block-based halfpel interpolation | ||
5 : | Isibaar | 333 | * |
6 : | edgomez | 851 | * This program is free software; you can redistribute it and/or modify |
7 : | * 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 | ||
9 : | * (at your option) any later version. | ||
10 : | Isibaar | 333 | * |
11 : | edgomez | 851 | * This program is distributed in the hope that it will be useful, |
12 : | * but WITHOUT ANY WARRANTY; without even the implied warranty of | ||
13 : | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | ||
14 : | * GNU General Public License for more details. | ||
15 : | Isibaar | 333 | * |
16 : | edgomez | 851 | * You should have received a copy of the GNU General Public License |
17 : | * along with this program; if not, write to the Free Software | ||
18 : | * Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. | ||
19 : | Isibaar | 333 | * |
20 : | edgomez | 851 | *************************************************************************/ |
21 : | |||
22 : | /************************************************************************** | ||
23 : | Isibaar | 333 | * |
24 : | edgomez | 851 | * History: |
25 : | Isibaar | 333 | * |
26 : | edgomez | 851 | * 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 : | edgomez | 651 | * |
30 : | edgomez | 851 | *************************************************************************/ |
31 : | Isibaar | 333 | |
32 : | edgomez | 851 | |
33 : | Isibaar | 333 | #include "../portab.h" |
34 : | edgomez | 851 | #include "../global.h" |
35 : | Isibaar | 333 | #include "interpolate8x8.h" |
36 : | |||
37 : | edgomez | 851 | // function pointers |
38 : | Isibaar | 333 | INTERPOLATE8X8_PTR interpolate8x8_halfpel_h; |
39 : | INTERPOLATE8X8_PTR interpolate8x8_halfpel_v; | ||
40 : | INTERPOLATE8X8_PTR interpolate8x8_halfpel_hv; | ||
41 : | |||
42 : | edgomez | 851 | INTERPOLATE8X8_AVG2_PTR interpolate8x8_avg2; |
43 : | INTERPOLATE8X8_AVG4_PTR interpolate8x8_avg4; | ||
44 : | Isibaar | 333 | |
45 : | edgomez | 851 | INTERPOLATE_LOWPASS_PTR interpolate8x8_lowpass_h; |
46 : | INTERPOLATE_LOWPASS_PTR interpolate8x8_lowpass_v; | ||
47 : | Isibaar | 333 | |
48 : | edgomez | 851 | 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) | ||
104 : | |||
105 : | Isibaar | 333 | void |
106 : | interpolate8x8_halfpel_h_c(uint8_t * const dst, | ||
107 : | const uint8_t * const src, | ||
108 : | const uint32_t stride, | ||
109 : | const uint32_t rounding) | ||
110 : | { | ||
111 : | chl | 906 | intptr_t j; |
112 : | |||
113 : | if (rounding) | ||
114 : | for (j = 7*stride; j >= 0; j-=stride) | ||
115 : | { | ||
116 : | dst[j + 0] = (uint8_t)((src[j + 0] + src[j + 1] )>>1); | ||
117 : | dst[j + 1] = (uint8_t)((src[j + 1] + src[j + 2] )>>1); | ||
118 : | dst[j + 2] = (uint8_t)((src[j + 2] + src[j + 3] )>>1); | ||
119 : | dst[j + 3] = (uint8_t)((src[j + 3] + src[j + 4] )>>1); | ||
120 : | dst[j + 4] = (uint8_t)((src[j + 4] + src[j + 5] )>>1); | ||
121 : | dst[j + 5] = (uint8_t)((src[j + 5] + src[j + 6] )>>1); | ||
122 : | dst[j + 6] = (uint8_t)((src[j + 6] + src[j + 7] )>>1); | ||
123 : | dst[j + 7] = (uint8_t)((src[j + 7] + src[j + 8] )>>1); | ||
124 : | Isibaar | 333 | } |
125 : | chl | 906 | else |
126 : | for (j = 0; j < 8*stride; j+=stride) /* forward or backwards? Who knows ... */ | ||
127 : | { | ||
128 : | dst[j + 0] = (uint8_t)((src[j + 0] + src[j + 1] + 1)>>1); | ||
129 : | dst[j + 1] = (uint8_t)((src[j + 1] + src[j + 2] + 1)>>1); | ||
130 : | dst[j + 2] = (uint8_t)((src[j + 2] + src[j + 3] + 1)>>1); | ||
131 : | dst[j + 3] = (uint8_t)((src[j + 3] + src[j + 4] + 1)>>1); | ||
132 : | dst[j + 4] = (uint8_t)((src[j + 4] + src[j + 5] + 1)>>1); | ||
133 : | dst[j + 5] = (uint8_t)((src[j + 5] + src[j + 6] + 1)>>1); | ||
134 : | dst[j + 6] = (uint8_t)((src[j + 6] + src[j + 7] + 1)>>1); | ||
135 : | dst[j + 7] = (uint8_t)((src[j + 7] + src[j + 8] + 1)>>1); | ||
136 : | } | ||
137 : | Isibaar | 333 | } |
138 : | |||
139 : | |||
140 : | |||
141 : | void | ||
142 : | interpolate8x8_halfpel_v_c(uint8_t * const dst, | ||
143 : | const uint8_t * const src, | ||
144 : | const uint32_t stride, | ||
145 : | const uint32_t rounding) | ||
146 : | { | ||
147 : | chl | 906 | intptr_t j; |
148 : | // const uint8_t * const src2 = src+stride; /* using a second pointer is _not_ faster here */ | ||
149 : | Isibaar | 333 | |
150 : | chl | 906 | if (rounding) |
151 : | for (j = 0; j < 8*stride; j+=stride) /* forward is better. Some automatic prefetch perhaps. */ | ||
152 : | { | ||
153 : | dst[j + 0] = (uint8_t)((src[j + 0] + src[j + stride + 0] )>>1); | ||
154 : | dst[j + 1] = (uint8_t)((src[j + 1] + src[j + stride + 1] )>>1); | ||
155 : | dst[j + 2] = (uint8_t)((src[j + 2] + src[j + stride + 2] )>>1); | ||
156 : | dst[j + 3] = (uint8_t)((src[j + 3] + src[j + stride + 3] )>>1); | ||
157 : | dst[j + 4] = (uint8_t)((src[j + 4] + src[j + stride + 4] )>>1); | ||
158 : | dst[j + 5] = (uint8_t)((src[j + 5] + src[j + stride + 5] )>>1); | ||
159 : | dst[j + 6] = (uint8_t)((src[j + 6] + src[j + stride + 6] )>>1); | ||
160 : | dst[j + 7] = (uint8_t)((src[j + 7] + src[j + stride + 7] )>>1); | ||
161 : | Isibaar | 333 | } |
162 : | chl | 906 | else |
163 : | for (j = 0; j < 8*stride; j+=stride) | ||
164 : | { | ||
165 : | dst[j + 0] = (uint8_t)((src[j + 0] + src[j + stride + 0] + 1)>>1); | ||
166 : | dst[j + 1] = (uint8_t)((src[j + 1] + src[j + stride + 1] + 1)>>1); | ||
167 : | dst[j + 2] = (uint8_t)((src[j + 2] + src[j + stride + 2] + 1)>>1); | ||
168 : | dst[j + 3] = (uint8_t)((src[j + 3] + src[j + stride + 3] + 1)>>1); | ||
169 : | dst[j + 4] = (uint8_t)((src[j + 4] + src[j + stride + 4] + 1)>>1); | ||
170 : | dst[j + 5] = (uint8_t)((src[j + 5] + src[j + stride + 5] + 1)>>1); | ||
171 : | dst[j + 6] = (uint8_t)((src[j + 6] + src[j + stride + 6] + 1)>>1); | ||
172 : | dst[j + 7] = (uint8_t)((src[j + 7] + src[j + stride + 7] + 1)>>1); | ||
173 : | } | ||
174 : | Isibaar | 333 | } |
175 : | |||
176 : | |||
177 : | void | ||
178 : | interpolate8x8_halfpel_hv_c(uint8_t * const dst, | ||
179 : | const uint8_t * const src, | ||
180 : | const uint32_t stride, | ||
181 : | const uint32_t rounding) | ||
182 : | { | ||
183 : | chl | 906 | intptr_t j; |
184 : | Isibaar | 333 | |
185 : | chl | 906 | if (rounding) |
186 : | for (j = 7*stride; j >= 0; j-=stride) | ||
187 : | { | ||
188 : | dst[j + 0] = (uint8_t)((src[j+0] + src[j+1] + src[j+stride+0] + src[j+stride+1] +1)>>2); | ||
189 : | dst[j + 1] = (uint8_t)((src[j+1] + src[j+2] + src[j+stride+1] + src[j+stride+2] +1)>>2); | ||
190 : | dst[j + 2] = (uint8_t)((src[j+2] + src[j+3] + src[j+stride+2] + src[j+stride+3] +1)>>2); | ||
191 : | dst[j + 3] = (uint8_t)((src[j+3] + src[j+4] + src[j+stride+3] + src[j+stride+4] +1)>>2); | ||
192 : | dst[j + 4] = (uint8_t)((src[j+4] + src[j+5] + src[j+stride+4] + src[j+stride+5] +1)>>2); | ||
193 : | dst[j + 5] = (uint8_t)((src[j+5] + src[j+6] + src[j+stride+5] + src[j+stride+6] +1)>>2); | ||
194 : | dst[j + 6] = (uint8_t)((src[j+6] + src[j+7] + src[j+stride+6] + src[j+stride+7] +1)>>2); | ||
195 : | dst[j + 7] = (uint8_t)((src[j+7] + src[j+8] + src[j+stride+7] + src[j+stride+8] +1)>>2); | ||
196 : | Isibaar | 333 | } |
197 : | chl | 906 | else |
198 : | for (j = 7*stride; j >= 0; j-=stride) | ||
199 : | { | ||
200 : | dst[j + 0] = (uint8_t)((src[j+0] + src[j+1] + src[j+stride+0] + src[j+stride+1] +2)>>2); | ||
201 : | dst[j + 1] = (uint8_t)((src[j+1] + src[j+2] + src[j+stride+1] + src[j+stride+2] +2)>>2); | ||
202 : | dst[j + 2] = (uint8_t)((src[j+2] + src[j+3] + src[j+stride+2] + src[j+stride+3] +2)>>2); | ||
203 : | dst[j + 3] = (uint8_t)((src[j+3] + src[j+4] + src[j+stride+3] + src[j+stride+4] +2)>>2); | ||
204 : | dst[j + 4] = (uint8_t)((src[j+4] + src[j+5] + src[j+stride+4] + src[j+stride+5] +2)>>2); | ||
205 : | dst[j + 5] = (uint8_t)((src[j+5] + src[j+6] + src[j+stride+5] + src[j+stride+6] +2)>>2); | ||
206 : | dst[j + 6] = (uint8_t)((src[j+6] + src[j+7] + src[j+stride+6] + src[j+stride+7] +2)>>2); | ||
207 : | dst[j + 7] = (uint8_t)((src[j+7] + src[j+8] + src[j+stride+7] + src[j+stride+8] +2)>>2); | ||
208 : | } | ||
209 : | Isibaar | 333 | } |
210 : | |||
211 : | chl | 906 | |
212 : | |||
213 : | |||
214 : | edgomez | 851 | /************************************************************* |
215 : | * QPEL STUFF STARTS HERE * | ||
216 : | *************************************************************/ | ||
217 : | |||
218 : | void interpolate8x8_6tap_lowpass_h_c(uint8_t *dst, uint8_t *src, int32_t stride, int32_t rounding) | ||
219 : | Isibaar | 333 | { |
220 : | edgomez | 851 | int32_t i; |
221 : | uint8_t round_add = 16 - rounding; | ||
222 : | Isibaar | 333 | |
223 : | edgomez | 851 | for(i = 0; i < 8; i++) |
224 : | { | ||
225 : | |||
226 : | dst[0] = CLIP((((src[-2] + src[3]) + 5 * (((src[0] + src[1])<<2) - (src[-1] + src[2])) + round_add) >> 5), 0, 255); | ||
227 : | dst[1] = CLIP((((src[-1] + src[4]) + 5 * (((src[1] + src[2])<<2) - (src[0] + src[3])) + round_add) >> 5), 0, 255); | ||
228 : | dst[2] = CLIP((((src[0] + src[5]) + 5 * (((src[2] + src[3])<<2) - (src[1] + src[4])) + round_add) >> 5), 0, 255); | ||
229 : | dst[3] = CLIP((((src[1] + src[6]) + 5 * (((src[3] + src[4])<<2) - (src[2] + src[5])) + round_add) >> 5), 0, 255); | ||
230 : | dst[4] = CLIP((((src[2] + src[7]) + 5 * (((src[4] + src[5])<<2) - (src[3] + src[6])) + round_add) >> 5), 0, 255); | ||
231 : | dst[5] = CLIP((((src[3] + src[8]) + 5 * (((src[5] + src[6])<<2) - (src[4] + src[7])) + round_add) >> 5), 0, 255); | ||
232 : | dst[6] = CLIP((((src[4] + src[9]) + 5 * (((src[6] + src[7])<<2) - (src[5] + src[8])) + round_add) >> 5), 0, 255); | ||
233 : | dst[7] = CLIP((((src[5] + src[10]) + 5 * (((src[7] + src[8])<<2) - (src[6] + src[9])) + round_add) >> 5), 0, 255); | ||
234 : | |||
235 : | dst += stride; | ||
236 : | src += stride; | ||
237 : | } | ||
238 : | Isibaar | 333 | } |
239 : | edgomez | 851 | |
240 : | void interpolate16x16_lowpass_h_c(uint8_t *dst, uint8_t *src, int32_t stride, int32_t rounding) | ||
241 : | { | ||
242 : | int32_t i; | ||
243 : | uint8_t round_add = 16 - rounding; | ||
244 : | |||
245 : | for(i = 0; i < 17; i++) | ||
246 : | { | ||
247 : | |||
248 : | dst[0] = CLIP(((7 * ((src[0]<<1) - src[2]) + 23 * src[1] + 3 * src[3] - src[4] + round_add) >> 5), 0, 255); | ||
249 : | dst[1] = CLIP(((19 * src[1] + 20 * src[2] - src[5] + 3 * (src[4] - src[0] - (src[3]<<1)) + round_add) >> 5), 0, 255); | ||
250 : | 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); | ||
251 : | |||
252 : | 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); | ||
253 : | 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); | ||
254 : | 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); | ||
255 : | 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); | ||
256 : | 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); | ||
257 : | 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); | ||
258 : | 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); | ||
259 : | 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); | ||
260 : | 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); | ||
261 : | 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); | ||
262 : | |||
263 : | 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); | ||
264 : | dst[14] = CLIP(((19 * src[15] + 20 * src[14] + 3 * (src[12] - src[16] - (src[13] << 1)) - src[11] + round_add) >> 5), 0, 255); | ||
265 : | dst[15] = CLIP(((23 * src[15] + 7 * ((src[16]<<1) - src[14]) + 3 * src[13] - src[12] + round_add) >> 5), 0, 255); | ||
266 : | |||
267 : | dst += stride; | ||
268 : | src += stride; | ||
269 : | } | ||
270 : | } | ||
271 : | |||
272 : | void interpolate8x8_lowpass_h_c(uint8_t *dst, uint8_t *src, int32_t stride, int32_t rounding) | ||
273 : | { | ||
274 : | int32_t i; | ||
275 : | uint8_t round_add = 16 - rounding; | ||
276 : | |||
277 : | for(i = 0; i < 9; i++) | ||
278 : | { | ||
279 : | |||
280 : | dst[0] = CLIP(((7 * ((src[0]<<1) - src[2]) + 23 * src[1] + 3 * src[3] - src[4] + round_add) >> 5), 0, 255); | ||
281 : | dst[1] = CLIP(((19 * src[1] + 20 * src[2] - src[5] + 3 * (src[4] - src[0] - (src[3]<<1)) + round_add) >> 5), 0, 255); | ||
282 : | 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); | ||
283 : | 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); | ||
284 : | 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); | ||
285 : | 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); | ||
286 : | dst[6] = CLIP(((19 * src[7] + 20 * src[6] + 3 * (src[4] - src[8] - (src[5] << 1)) - src[3] + round_add) >> 5), 0, 255); | ||
287 : | dst[7] = CLIP(((23 * src[7] + 7 * ((src[8]<<1) - src[6]) + 3 * src[5] - src[4] + round_add) >> 5), 0, 255); | ||
288 : | |||
289 : | dst += stride; | ||
290 : | src += stride; | ||
291 : | } | ||
292 : | } | ||
293 : | |||
294 : | void interpolate8x8_6tap_lowpass_v_c(uint8_t *dst, uint8_t *src, int32_t stride, int32_t rounding) | ||
295 : | { | ||
296 : | int32_t i; | ||
297 : | uint8_t round_add = 16 - rounding; | ||
298 : | |||
299 : | for(i = 0; i < 8; i++) | ||
300 : | { | ||
301 : | int32_t src_2 = src[-2*stride]; | ||
302 : | int32_t src_1 = src[-stride]; | ||
303 : | int32_t src0 = src[0]; | ||
304 : | int32_t src1 = src[stride]; | ||
305 : | int32_t src2 = src[2 * stride]; | ||
306 : | int32_t src3 = src[3 * stride]; | ||
307 : | int32_t src4 = src[4 * stride]; | ||
308 : | int32_t src5 = src[5 * stride]; | ||
309 : | int32_t src6 = src[6 * stride]; | ||
310 : | int32_t src7 = src[7 * stride]; | ||
311 : | int32_t src8 = src[8 * stride]; | ||
312 : | int32_t src9 = src[9 * stride]; | ||
313 : | int32_t src10 = src[10 * stride]; | ||
314 : | |||
315 : | dst[0] = CLIP((((src_2 + src3) + 5 * (((src0 + src1)<<2) - (src_1 + src2)) + round_add) >> 5), 0, 255); | ||
316 : | dst[stride] = CLIP((((src_1 + src4) + 5 * (((src1 + src2)<<2) - (src0 + src3)) + round_add) >> 5), 0, 255); | ||
317 : | dst[2 * stride] = CLIP((((src0 + src5) + 5 * (((src2 + src3)<<2) - (src1 + src4)) + round_add) >> 5), 0, 255); | ||
318 : | dst[3 * stride] = CLIP((((src1 + src6) + 5 * (((src3 + src4)<<2) - (src2 + src5)) + round_add) >> 5), 0, 255); | ||
319 : | dst[4 * stride] = CLIP((((src2 + src7) + 5 * (((src4 + src5)<<2) - (src3 + src6)) + round_add) >> 5), 0, 255); | ||
320 : | dst[5 * stride] = CLIP((((src3 + src8) + 5 * (((src5 + src6)<<2) - (src4 + src7)) + round_add) >> 5), 0, 255); | ||
321 : | dst[6 * stride] = CLIP((((src4 + src9) + 5 * (((src6 + src7)<<2) - (src5 + src8)) + round_add) >> 5), 0, 255); | ||
322 : | dst[7 * stride] = CLIP((((src5 + src10) + 5 * (((src7 + src8)<<2) - (src6 + src9)) + round_add) >> 5), 0, 255); | ||
323 : | |||
324 : | dst++; | ||
325 : | src++; | ||
326 : | } | ||
327 : | } | ||
328 : | |||
329 : | void interpolate16x16_lowpass_v_c(uint8_t *dst, uint8_t *src, int32_t stride, int32_t rounding) | ||
330 : | { | ||
331 : | int32_t i; | ||
332 : | uint8_t round_add = 16 - rounding; | ||
333 : | |||
334 : | for(i = 0; i < 17; i++) | ||
335 : | { | ||
336 : | int32_t src0 = src[0]; | ||
337 : | int32_t src1 = src[stride]; | ||
338 : | int32_t src2 = src[2 * stride]; | ||
339 : | int32_t src3 = src[3 * stride]; | ||
340 : | int32_t src4 = src[4 * stride]; | ||
341 : | int32_t src5 = src[5 * stride]; | ||
342 : | int32_t src6 = src[6 * stride]; | ||
343 : | int32_t src7 = src[7 * stride]; | ||
344 : | int32_t src8 = src[8 * stride]; | ||
345 : | int32_t src9 = src[9 * stride]; | ||
346 : | int32_t src10 = src[10 * stride]; | ||
347 : | int32_t src11 = src[11 * stride]; | ||
348 : | int32_t src12 = src[12 * stride]; | ||
349 : | int32_t src13 = src[13 * stride]; | ||
350 : | int32_t src14 = src[14 * stride]; | ||
351 : | int32_t src15 = src[15 * stride]; | ||
352 : | int32_t src16 = src[16 * stride]; | ||
353 : | |||
354 : | |||
355 : | dst[0] = CLIP(((7 * ((src0<<1) - src2) + 23 * src1 + 3 * src3 - src4 + round_add) >> 5), 0, 255); | ||
356 : | dst[stride] = CLIP(((19 * src1 + 20 * src2 - src5 + 3 * (src4 - src0 - (src3<<1)) + round_add) >> 5), 0, 255); | ||
357 : | dst[2*stride] = CLIP(((20 * (src2 + src3) + (src0<<1) + 3 * (src5 - ((src1 + src4)<<1)) - src6 + round_add) >> 5), 0, 255); | ||
358 : | |||
359 : | dst[3*stride] = CLIP(((20 * (src3 + src4) + 3 * ((src6 + src1) - ((src2 + src5)<<1)) - (src0 + src7) + round_add) >> 5), 0, 255); | ||
360 : | dst[4*stride] = CLIP(((20 * (src4 + src5) - 3 * (((src3 + src6)<<1) - (src2 + src7)) - (src1 + src8) + round_add) >> 5), 0, 255); | ||
361 : | dst[5*stride] = CLIP(((20 * (src5 + src6) - 3 * (((src4 + src7)<<1) - (src3 + src8)) - (src2 + src9) + round_add) >> 5), 0, 255); | ||
362 : | dst[6*stride] = CLIP(((20 * (src6 + src7) - 3 * (((src5 + src8)<<1) - (src4 + src9)) - (src3 + src10) + round_add) >> 5), 0, 255); | ||
363 : | dst[7*stride] = CLIP(((20 * (src7 + src8) - 3 * (((src6 + src9)<<1) - (src5 + src10)) - (src4 + src11) + round_add) >> 5), 0, 255); | ||
364 : | dst[8*stride] = CLIP(((20 * (src8 + src9) - 3 * (((src7 + src10)<<1) - (src6 + src11)) - (src5 + src12) + round_add) >> 5), 0, 255); | ||
365 : | dst[9*stride] = CLIP(((20 * (src9 + src10) - 3 * (((src8 + src11)<<1) - (src7 + src12)) - (src6 + src13) + round_add) >> 5), 0, 255); | ||
366 : | dst[10*stride] = CLIP(((20 * (src10 + src11) - 3 * (((src9 + src12)<<1) - (src8 + src13)) - (src7 + src14) + round_add) >> 5), 0, 255); | ||
367 : | dst[11*stride] = CLIP(((20 * (src11 + src12) - 3 * (((src10 + src13)<<1) - (src9 + src14)) - (src8 + src15) + round_add) >> 5), 0, 255); | ||
368 : | dst[12*stride] = CLIP(((20 * (src12 + src13) - 3 * (((src11 + src14)<<1) - (src10 + src15)) - (src9 + src16) + round_add) >> 5), 0, 255); | ||
369 : | |||
370 : | dst[13*stride] = CLIP(((20 * (src13 + src14) + (src16<<1) + 3 * (src11 - ((src12 + src15) << 1)) - src10 + round_add) >> 5), 0, 255); | ||
371 : | dst[14*stride] = CLIP(((19 * src15 + 20 * src14 + 3 * (src12 - src16 - (src13 << 1)) - src11 + round_add) >> 5), 0, 255); | ||
372 : | dst[15*stride] = CLIP(((23 * src15 + 7 * ((src16<<1) - src14) + 3 * src13 - src12 + round_add) >> 5), 0, 255); | ||
373 : | |||
374 : | dst++; | ||
375 : | src++; | ||
376 : | } | ||
377 : | } | ||
378 : | |||
379 : | void interpolate8x8_lowpass_v_c(uint8_t *dst, uint8_t *src, int32_t stride, int32_t rounding) | ||
380 : | { | ||
381 : | int32_t i; | ||
382 : | uint8_t round_add = 16 - rounding; | ||
383 : | |||
384 : | for(i = 0; i < 9; i++) | ||
385 : | { | ||
386 : | int32_t src0 = src[0]; | ||
387 : | int32_t src1 = src[stride]; | ||
388 : | int32_t src2 = src[2 * stride]; | ||
389 : | int32_t src3 = src[3 * stride]; | ||
390 : | int32_t src4 = src[4 * stride]; | ||
391 : | int32_t src5 = src[5 * stride]; | ||
392 : | int32_t src6 = src[6 * stride]; | ||
393 : | int32_t src7 = src[7 * stride]; | ||
394 : | int32_t src8 = src[8 * stride]; | ||
395 : | |||
396 : | dst[0] = CLIP(((7 * ((src0<<1) - src2) + 23 * src1 + 3 * src3 - src4 + round_add) >> 5), 0, 255); | ||
397 : | dst[stride] = CLIP(((19 * src1 + 20 * src2 - src5 + 3 * (src4 - src0 - (src3 << 1)) + round_add) >> 5), 0, 255); | ||
398 : | dst[2 * stride] = CLIP(((20 * (src2 + src3) + (src0<<1) + 3 * (src5 - ((src1 + src4) <<1 )) - src6 + round_add) >> 5), 0, 255); | ||
399 : | dst[3 * stride] = CLIP(((20 * (src3 + src4) + 3 * ((src6 + src1) - ((src2 + src5)<<1)) - (src0 + src7) + round_add) >> 5), 0, 255); | ||
400 : | dst[4 * stride] = CLIP(((20 * (src4 + src5) + 3 * ((src2 + src7) - ((src3 + src6)<<1)) - (src1 + src8) + round_add) >> 5), 0, 255); | ||
401 : | dst[5 * stride] = CLIP(((20 * (src5 + src6) + (src8<<1) + 3 * (src3 - ((src4 + src7) << 1)) - src2 + round_add) >> 5), 0, 255); | ||
402 : | dst[6 * stride] = CLIP(((19 * src7 + 20 * src6 - src3 + 3 * (src4 - src8 - (src5 << 1)) + round_add) >> 5), 0, 255); | ||
403 : | dst[7 * stride] = CLIP(((7 * ((src8<<1) - src6) + 23 * src7 + 3 * src5 - src4 + round_add) >> 5), 0, 255); | ||
404 : | |||
405 : | dst++; | ||
406 : | src++; | ||
407 : | } | ||
408 : | } | ||
409 : | |||
410 : | void interpolate16x16_lowpass_hv_c(uint8_t *dst1, uint8_t *dst2, uint8_t *src, int32_t stride, int32_t rounding) | ||
411 : | { | ||
412 : | int32_t i; | ||
413 : | uint8_t round_add = 16 - rounding; | ||
414 : | uint8_t *h_ptr = dst2; | ||
415 : | |||
416 : | for(i = 0; i < 17; i++) | ||
417 : | { | ||
418 : | |||
419 : | h_ptr[0] = CLIP(((7 * ((src[0]<<1) - src[2]) + 23 * src[1] + 3 * src[3] - src[4] + round_add) >> 5), 0, 255); | ||
420 : | 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); | ||
421 : | 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); | ||
422 : | |||
423 : | 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); | ||
424 : | 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); | ||
425 : | 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); | ||
426 : | 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); | ||
427 : | 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); | ||
428 : | 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); | ||
429 : | 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); | ||
430 : | 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); | ||
431 : | 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); | ||
432 : | 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); | ||
433 : | |||
434 : | 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); | ||
435 : | 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); | ||
436 : | h_ptr[15] = CLIP(((23 * src[15] + 7 * ((src[16]<<1) - src[14]) + 3 * src[13] - src[12] + round_add) >> 5), 0, 255); | ||
437 : | |||
438 : | h_ptr += stride; | ||
439 : | src += stride; | ||
440 : | } | ||
441 : | |||
442 : | interpolate16x16_lowpass_v_c(dst1, dst2, stride, rounding); | ||
443 : | |||
444 : | } | ||
445 : | |||
446 : | void interpolate8x8_lowpass_hv_c(uint8_t *dst1, uint8_t *dst2, uint8_t *src, int32_t stride, int32_t rounding) | ||
447 : | { | ||
448 : | int32_t i; | ||
449 : | uint8_t round_add = 16 - rounding; | ||
450 : | uint8_t *h_ptr = dst2; | ||
451 : | |||
452 : | for(i = 0; i < 9; i++) | ||
453 : | { | ||
454 : | |||
455 : | h_ptr[0] = CLIP(((7 * ((src[0]<<1) - src[2]) + 23 * src[1] + 3 * src[3] - src[4] + round_add) >> 5), 0, 255); | ||
456 : | 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); | ||
457 : | 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); | ||
458 : | 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); | ||
459 : | 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); | ||
460 : | 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); | ||
461 : | 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); | ||
462 : | h_ptr[7] = CLIP(((23 * src[7] + 7 * ((src[8]<<1) - src[6]) + 3 * src[5] - src[4] + round_add) >> 5), 0, 255); | ||
463 : | |||
464 : | h_ptr += stride; | ||
465 : | src += stride; | ||
466 : | } | ||
467 : | |||
468 : | interpolate8x8_lowpass_v_c(dst1, dst2, stride, rounding); | ||
469 : | |||
470 : | } |
No admin address has been configured | ViewVC Help |
Powered by ViewVC 1.0.4 |