Parent Directory | Revision Log
Revision 1382 - (view) (download)
1 : | edgomez | 1382 | /***************************************************************************** |
2 : | Isibaar | 333 | * |
3 : | edgomez | 851 | * XVID MPEG-4 VIDEO CODEC |
4 : | edgomez | 1382 | * - 8x8 block-based halfpel interpolation - |
5 : | Isibaar | 333 | * |
6 : | edgomez | 1382 | * Copyright(C) 2001-2003 Peter Ross <pross@xvid.org> |
7 : | Isibaar | 333 | * |
8 : | edgomez | 1382 | * 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 : | Isibaar | 333 | * |
13 : | edgomez | 1382 | * 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 : | Isibaar | 333 | * |
18 : | edgomez | 1382 | * 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 : | Isibaar | 333 | * |
22 : | edgomez | 1382 | * $Id: interpolate8x8.c,v 1.12 2004-03-22 22:36:23 edgomez Exp $ |
23 : | Isibaar | 333 | * |
24 : | edgomez | 1382 | ****************************************************************************/ |
25 : | Isibaar | 333 | |
26 : | #include "../portab.h" | ||
27 : | edgomez | 851 | #include "../global.h" |
28 : | Isibaar | 333 | #include "interpolate8x8.h" |
29 : | |||
30 : | edgomez | 1382 | /* function pointers */ |
31 : | Isibaar | 333 | INTERPOLATE8X8_PTR interpolate8x8_halfpel_h; |
32 : | INTERPOLATE8X8_PTR interpolate8x8_halfpel_v; | ||
33 : | INTERPOLATE8X8_PTR interpolate8x8_halfpel_hv; | ||
34 : | |||
35 : | edgomez | 851 | INTERPOLATE8X8_AVG2_PTR interpolate8x8_avg2; |
36 : | INTERPOLATE8X8_AVG4_PTR interpolate8x8_avg4; | ||
37 : | Isibaar | 333 | |
38 : | edgomez | 851 | INTERPOLATE_LOWPASS_PTR interpolate8x8_lowpass_h; |
39 : | INTERPOLATE_LOWPASS_PTR interpolate8x8_lowpass_v; | ||
40 : | Isibaar | 333 | |
41 : | edgomez | 851 | INTERPOLATE_LOWPASS_PTR interpolate16x16_lowpass_h; |
42 : | INTERPOLATE_LOWPASS_PTR interpolate16x16_lowpass_v; | ||
43 : | |||
44 : | INTERPOLATE_LOWPASS_HV_PTR interpolate8x8_lowpass_hv; | ||
45 : | INTERPOLATE_LOWPASS_HV_PTR interpolate16x16_lowpass_hv; | ||
46 : | |||
47 : | INTERPOLATE8X8_6TAP_LOWPASS_PTR interpolate8x8_6tap_lowpass_h; | ||
48 : | INTERPOLATE8X8_6TAP_LOWPASS_PTR interpolate8x8_6tap_lowpass_v; | ||
49 : | |||
50 : | 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) | ||
51 : | { | ||
52 : | uint32_t i; | ||
53 : | const int32_t round = 1 - rounding; | ||
54 : | |||
55 : | for(i = 0; i < height; i++) | ||
56 : | { | ||
57 : | dst[0] = (src1[0] + src2[0] + round) >> 1; | ||
58 : | dst[1] = (src1[1] + src2[1] + round) >> 1; | ||
59 : | dst[2] = (src1[2] + src2[2] + round) >> 1; | ||
60 : | dst[3] = (src1[3] + src2[3] + round) >> 1; | ||
61 : | dst[4] = (src1[4] + src2[4] + round) >> 1; | ||
62 : | dst[5] = (src1[5] + src2[5] + round) >> 1; | ||
63 : | dst[6] = (src1[6] + src2[6] + round) >> 1; | ||
64 : | dst[7] = (src1[7] + src2[7] + round) >> 1; | ||
65 : | |||
66 : | dst += stride; | ||
67 : | src1 += stride; | ||
68 : | src2 += stride; | ||
69 : | } | ||
70 : | } | ||
71 : | |||
72 : | 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) | ||
73 : | { | ||
74 : | int32_t i; | ||
75 : | const int32_t round = 2 - rounding; | ||
76 : | |||
77 : | for(i = 0; i < 8; i++) | ||
78 : | { | ||
79 : | dst[0] = (src1[0] + src2[0] + src3[0] + src4[0] + round) >> 2; | ||
80 : | dst[1] = (src1[1] + src2[1] + src3[1] + src4[1] + round) >> 2; | ||
81 : | dst[2] = (src1[2] + src2[2] + src3[2] + src4[2] + round) >> 2; | ||
82 : | dst[3] = (src1[3] + src2[3] + src3[3] + src4[3] + round) >> 2; | ||
83 : | dst[4] = (src1[4] + src2[4] + src3[4] + src4[4] + round) >> 2; | ||
84 : | dst[5] = (src1[5] + src2[5] + src3[5] + src4[5] + round) >> 2; | ||
85 : | dst[6] = (src1[6] + src2[6] + src3[6] + src4[6] + round) >> 2; | ||
86 : | dst[7] = (src1[7] + src2[7] + src3[7] + src4[7] + round) >> 2; | ||
87 : | edgomez | 1382 | |
88 : | edgomez | 851 | dst += stride; |
89 : | src1 += stride; | ||
90 : | src2 += stride; | ||
91 : | src3 += stride; | ||
92 : | src4 += stride; | ||
93 : | } | ||
94 : | } | ||
95 : | |||
96 : | edgomez | 1382 | /* dst = interpolate(src) */ |
97 : | edgomez | 851 | |
98 : | Isibaar | 333 | void |
99 : | interpolate8x8_halfpel_h_c(uint8_t * const dst, | ||
100 : | const uint8_t * const src, | ||
101 : | const uint32_t stride, | ||
102 : | const uint32_t rounding) | ||
103 : | { | ||
104 : | edgomez | 1382 | uintptr_t j; |
105 : | |||
106 : | chl | 906 | if (rounding) |
107 : | edgomez | 1382 | for (j = 0; j < 8*stride; j+=stride) |
108 : | chl | 906 | { |
109 : | dst[j + 0] = (uint8_t)((src[j + 0] + src[j + 1] )>>1); | ||
110 : | dst[j + 1] = (uint8_t)((src[j + 1] + src[j + 2] )>>1); | ||
111 : | dst[j + 2] = (uint8_t)((src[j + 2] + src[j + 3] )>>1); | ||
112 : | dst[j + 3] = (uint8_t)((src[j + 3] + src[j + 4] )>>1); | ||
113 : | dst[j + 4] = (uint8_t)((src[j + 4] + src[j + 5] )>>1); | ||
114 : | dst[j + 5] = (uint8_t)((src[j + 5] + src[j + 6] )>>1); | ||
115 : | dst[j + 6] = (uint8_t)((src[j + 6] + src[j + 7] )>>1); | ||
116 : | dst[j + 7] = (uint8_t)((src[j + 7] + src[j + 8] )>>1); | ||
117 : | Isibaar | 333 | } |
118 : | chl | 906 | else |
119 : | for (j = 0; j < 8*stride; j+=stride) /* forward or backwards? Who knows ... */ | ||
120 : | { | ||
121 : | dst[j + 0] = (uint8_t)((src[j + 0] + src[j + 1] + 1)>>1); | ||
122 : | dst[j + 1] = (uint8_t)((src[j + 1] + src[j + 2] + 1)>>1); | ||
123 : | dst[j + 2] = (uint8_t)((src[j + 2] + src[j + 3] + 1)>>1); | ||
124 : | dst[j + 3] = (uint8_t)((src[j + 3] + src[j + 4] + 1)>>1); | ||
125 : | dst[j + 4] = (uint8_t)((src[j + 4] + src[j + 5] + 1)>>1); | ||
126 : | dst[j + 5] = (uint8_t)((src[j + 5] + src[j + 6] + 1)>>1); | ||
127 : | dst[j + 6] = (uint8_t)((src[j + 6] + src[j + 7] + 1)>>1); | ||
128 : | dst[j + 7] = (uint8_t)((src[j + 7] + src[j + 8] + 1)>>1); | ||
129 : | } | ||
130 : | Isibaar | 333 | } |
131 : | |||
132 : | |||
133 : | |||
134 : | void | ||
135 : | interpolate8x8_halfpel_v_c(uint8_t * const dst, | ||
136 : | const uint8_t * const src, | ||
137 : | const uint32_t stride, | ||
138 : | const uint32_t rounding) | ||
139 : | { | ||
140 : | edgomez | 1382 | uintptr_t j; |
141 : | Isibaar | 333 | |
142 : | edgomez | 1382 | |
143 : | chl | 906 | if (rounding) |
144 : | for (j = 0; j < 8*stride; j+=stride) /* forward is better. Some automatic prefetch perhaps. */ | ||
145 : | { | ||
146 : | dst[j + 0] = (uint8_t)((src[j + 0] + src[j + stride + 0] )>>1); | ||
147 : | dst[j + 1] = (uint8_t)((src[j + 1] + src[j + stride + 1] )>>1); | ||
148 : | dst[j + 2] = (uint8_t)((src[j + 2] + src[j + stride + 2] )>>1); | ||
149 : | dst[j + 3] = (uint8_t)((src[j + 3] + src[j + stride + 3] )>>1); | ||
150 : | dst[j + 4] = (uint8_t)((src[j + 4] + src[j + stride + 4] )>>1); | ||
151 : | dst[j + 5] = (uint8_t)((src[j + 5] + src[j + stride + 5] )>>1); | ||
152 : | dst[j + 6] = (uint8_t)((src[j + 6] + src[j + stride + 6] )>>1); | ||
153 : | dst[j + 7] = (uint8_t)((src[j + 7] + src[j + stride + 7] )>>1); | ||
154 : | Isibaar | 333 | } |
155 : | chl | 906 | else |
156 : | for (j = 0; j < 8*stride; j+=stride) | ||
157 : | { | ||
158 : | dst[j + 0] = (uint8_t)((src[j + 0] + src[j + stride + 0] + 1)>>1); | ||
159 : | dst[j + 1] = (uint8_t)((src[j + 1] + src[j + stride + 1] + 1)>>1); | ||
160 : | dst[j + 2] = (uint8_t)((src[j + 2] + src[j + stride + 2] + 1)>>1); | ||
161 : | dst[j + 3] = (uint8_t)((src[j + 3] + src[j + stride + 3] + 1)>>1); | ||
162 : | dst[j + 4] = (uint8_t)((src[j + 4] + src[j + stride + 4] + 1)>>1); | ||
163 : | dst[j + 5] = (uint8_t)((src[j + 5] + src[j + stride + 5] + 1)>>1); | ||
164 : | dst[j + 6] = (uint8_t)((src[j + 6] + src[j + stride + 6] + 1)>>1); | ||
165 : | dst[j + 7] = (uint8_t)((src[j + 7] + src[j + stride + 7] + 1)>>1); | ||
166 : | } | ||
167 : | Isibaar | 333 | } |
168 : | |||
169 : | |||
170 : | void | ||
171 : | interpolate8x8_halfpel_hv_c(uint8_t * const dst, | ||
172 : | const uint8_t * const src, | ||
173 : | const uint32_t stride, | ||
174 : | const uint32_t rounding) | ||
175 : | { | ||
176 : | edgomez | 1382 | uintptr_t j; |
177 : | Isibaar | 333 | |
178 : | chl | 906 | if (rounding) |
179 : | edgomez | 1382 | for (j = 0; j < 8*stride; j+=stride) |
180 : | chl | 906 | { |
181 : | dst[j + 0] = (uint8_t)((src[j+0] + src[j+1] + src[j+stride+0] + src[j+stride+1] +1)>>2); | ||
182 : | dst[j + 1] = (uint8_t)((src[j+1] + src[j+2] + src[j+stride+1] + src[j+stride+2] +1)>>2); | ||
183 : | dst[j + 2] = (uint8_t)((src[j+2] + src[j+3] + src[j+stride+2] + src[j+stride+3] +1)>>2); | ||
184 : | dst[j + 3] = (uint8_t)((src[j+3] + src[j+4] + src[j+stride+3] + src[j+stride+4] +1)>>2); | ||
185 : | dst[j + 4] = (uint8_t)((src[j+4] + src[j+5] + src[j+stride+4] + src[j+stride+5] +1)>>2); | ||
186 : | dst[j + 5] = (uint8_t)((src[j+5] + src[j+6] + src[j+stride+5] + src[j+stride+6] +1)>>2); | ||
187 : | dst[j + 6] = (uint8_t)((src[j+6] + src[j+7] + src[j+stride+6] + src[j+stride+7] +1)>>2); | ||
188 : | dst[j + 7] = (uint8_t)((src[j+7] + src[j+8] + src[j+stride+7] + src[j+stride+8] +1)>>2); | ||
189 : | Isibaar | 333 | } |
190 : | chl | 906 | else |
191 : | edgomez | 1382 | for (j = 0; j < 8*stride; j+=stride) |
192 : | chl | 906 | { |
193 : | dst[j + 0] = (uint8_t)((src[j+0] + src[j+1] + src[j+stride+0] + src[j+stride+1] +2)>>2); | ||
194 : | dst[j + 1] = (uint8_t)((src[j+1] + src[j+2] + src[j+stride+1] + src[j+stride+2] +2)>>2); | ||
195 : | dst[j + 2] = (uint8_t)((src[j+2] + src[j+3] + src[j+stride+2] + src[j+stride+3] +2)>>2); | ||
196 : | dst[j + 3] = (uint8_t)((src[j+3] + src[j+4] + src[j+stride+3] + src[j+stride+4] +2)>>2); | ||
197 : | dst[j + 4] = (uint8_t)((src[j+4] + src[j+5] + src[j+stride+4] + src[j+stride+5] +2)>>2); | ||
198 : | dst[j + 5] = (uint8_t)((src[j+5] + src[j+6] + src[j+stride+5] + src[j+stride+6] +2)>>2); | ||
199 : | dst[j + 6] = (uint8_t)((src[j+6] + src[j+7] + src[j+stride+6] + src[j+stride+7] +2)>>2); | ||
200 : | dst[j + 7] = (uint8_t)((src[j+7] + src[j+8] + src[j+stride+7] + src[j+stride+8] +2)>>2); | ||
201 : | } | ||
202 : | Isibaar | 333 | } |
203 : | |||
204 : | edgomez | 851 | /************************************************************* |
205 : | * QPEL STUFF STARTS HERE * | ||
206 : | *************************************************************/ | ||
207 : | |||
208 : | void interpolate8x8_6tap_lowpass_h_c(uint8_t *dst, uint8_t *src, int32_t stride, int32_t rounding) | ||
209 : | Isibaar | 333 | { |
210 : | edgomez | 851 | int32_t i; |
211 : | uint8_t round_add = 16 - rounding; | ||
212 : | Isibaar | 333 | |
213 : | edgomez | 851 | for(i = 0; i < 8; i++) |
214 : | { | ||
215 : | |||
216 : | dst[0] = CLIP((((src[-2] + src[3]) + 5 * (((src[0] + src[1])<<2) - (src[-1] + src[2])) + round_add) >> 5), 0, 255); | ||
217 : | dst[1] = CLIP((((src[-1] + src[4]) + 5 * (((src[1] + src[2])<<2) - (src[0] + src[3])) + round_add) >> 5), 0, 255); | ||
218 : | dst[2] = CLIP((((src[0] + src[5]) + 5 * (((src[2] + src[3])<<2) - (src[1] + src[4])) + round_add) >> 5), 0, 255); | ||
219 : | dst[3] = CLIP((((src[1] + src[6]) + 5 * (((src[3] + src[4])<<2) - (src[2] + src[5])) + round_add) >> 5), 0, 255); | ||
220 : | dst[4] = CLIP((((src[2] + src[7]) + 5 * (((src[4] + src[5])<<2) - (src[3] + src[6])) + round_add) >> 5), 0, 255); | ||
221 : | dst[5] = CLIP((((src[3] + src[8]) + 5 * (((src[5] + src[6])<<2) - (src[4] + src[7])) + round_add) >> 5), 0, 255); | ||
222 : | dst[6] = CLIP((((src[4] + src[9]) + 5 * (((src[6] + src[7])<<2) - (src[5] + src[8])) + round_add) >> 5), 0, 255); | ||
223 : | dst[7] = CLIP((((src[5] + src[10]) + 5 * (((src[7] + src[8])<<2) - (src[6] + src[9])) + round_add) >> 5), 0, 255); | ||
224 : | |||
225 : | dst += stride; | ||
226 : | src += stride; | ||
227 : | } | ||
228 : | Isibaar | 333 | } |
229 : | edgomez | 851 | |
230 : | void interpolate16x16_lowpass_h_c(uint8_t *dst, uint8_t *src, int32_t stride, int32_t rounding) | ||
231 : | { | ||
232 : | int32_t i; | ||
233 : | uint8_t round_add = 16 - rounding; | ||
234 : | |||
235 : | for(i = 0; i < 17; i++) | ||
236 : | { | ||
237 : | |||
238 : | dst[0] = CLIP(((7 * ((src[0]<<1) - src[2]) + 23 * src[1] + 3 * src[3] - src[4] + round_add) >> 5), 0, 255); | ||
239 : | dst[1] = CLIP(((19 * src[1] + 20 * src[2] - src[5] + 3 * (src[4] - src[0] - (src[3]<<1)) + round_add) >> 5), 0, 255); | ||
240 : | 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); | ||
241 : | |||
242 : | 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); | ||
243 : | 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); | ||
244 : | 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); | ||
245 : | 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); | ||
246 : | 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); | ||
247 : | 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); | ||
248 : | 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); | ||
249 : | 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); | ||
250 : | 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); | ||
251 : | 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); | ||
252 : | |||
253 : | 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); | ||
254 : | dst[14] = CLIP(((19 * src[15] + 20 * src[14] + 3 * (src[12] - src[16] - (src[13] << 1)) - src[11] + round_add) >> 5), 0, 255); | ||
255 : | dst[15] = CLIP(((23 * src[15] + 7 * ((src[16]<<1) - src[14]) + 3 * src[13] - src[12] + round_add) >> 5), 0, 255); | ||
256 : | |||
257 : | dst += stride; | ||
258 : | src += stride; | ||
259 : | } | ||
260 : | } | ||
261 : | |||
262 : | void interpolate8x8_lowpass_h_c(uint8_t *dst, uint8_t *src, int32_t stride, int32_t rounding) | ||
263 : | { | ||
264 : | int32_t i; | ||
265 : | uint8_t round_add = 16 - rounding; | ||
266 : | |||
267 : | for(i = 0; i < 9; i++) | ||
268 : | { | ||
269 : | |||
270 : | dst[0] = CLIP(((7 * ((src[0]<<1) - src[2]) + 23 * src[1] + 3 * src[3] - src[4] + round_add) >> 5), 0, 255); | ||
271 : | dst[1] = CLIP(((19 * src[1] + 20 * src[2] - src[5] + 3 * (src[4] - src[0] - (src[3]<<1)) + round_add) >> 5), 0, 255); | ||
272 : | 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); | ||
273 : | 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); | ||
274 : | 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); | ||
275 : | 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); | ||
276 : | dst[6] = CLIP(((19 * src[7] + 20 * src[6] + 3 * (src[4] - src[8] - (src[5] << 1)) - src[3] + round_add) >> 5), 0, 255); | ||
277 : | dst[7] = CLIP(((23 * src[7] + 7 * ((src[8]<<1) - src[6]) + 3 * src[5] - src[4] + round_add) >> 5), 0, 255); | ||
278 : | |||
279 : | dst += stride; | ||
280 : | src += stride; | ||
281 : | } | ||
282 : | } | ||
283 : | |||
284 : | void interpolate8x8_6tap_lowpass_v_c(uint8_t *dst, uint8_t *src, int32_t stride, int32_t rounding) | ||
285 : | { | ||
286 : | int32_t i; | ||
287 : | uint8_t round_add = 16 - rounding; | ||
288 : | |||
289 : | for(i = 0; i < 8; i++) | ||
290 : | { | ||
291 : | int32_t src_2 = src[-2*stride]; | ||
292 : | int32_t src_1 = src[-stride]; | ||
293 : | int32_t src0 = src[0]; | ||
294 : | int32_t src1 = src[stride]; | ||
295 : | int32_t src2 = src[2 * stride]; | ||
296 : | int32_t src3 = src[3 * stride]; | ||
297 : | int32_t src4 = src[4 * stride]; | ||
298 : | int32_t src5 = src[5 * stride]; | ||
299 : | int32_t src6 = src[6 * stride]; | ||
300 : | int32_t src7 = src[7 * stride]; | ||
301 : | int32_t src8 = src[8 * stride]; | ||
302 : | int32_t src9 = src[9 * stride]; | ||
303 : | int32_t src10 = src[10 * stride]; | ||
304 : | |||
305 : | dst[0] = CLIP((((src_2 + src3) + 5 * (((src0 + src1)<<2) - (src_1 + src2)) + round_add) >> 5), 0, 255); | ||
306 : | dst[stride] = CLIP((((src_1 + src4) + 5 * (((src1 + src2)<<2) - (src0 + src3)) + round_add) >> 5), 0, 255); | ||
307 : | dst[2 * stride] = CLIP((((src0 + src5) + 5 * (((src2 + src3)<<2) - (src1 + src4)) + round_add) >> 5), 0, 255); | ||
308 : | dst[3 * stride] = CLIP((((src1 + src6) + 5 * (((src3 + src4)<<2) - (src2 + src5)) + round_add) >> 5), 0, 255); | ||
309 : | dst[4 * stride] = CLIP((((src2 + src7) + 5 * (((src4 + src5)<<2) - (src3 + src6)) + round_add) >> 5), 0, 255); | ||
310 : | dst[5 * stride] = CLIP((((src3 + src8) + 5 * (((src5 + src6)<<2) - (src4 + src7)) + round_add) >> 5), 0, 255); | ||
311 : | dst[6 * stride] = CLIP((((src4 + src9) + 5 * (((src6 + src7)<<2) - (src5 + src8)) + round_add) >> 5), 0, 255); | ||
312 : | dst[7 * stride] = CLIP((((src5 + src10) + 5 * (((src7 + src8)<<2) - (src6 + src9)) + round_add) >> 5), 0, 255); | ||
313 : | |||
314 : | dst++; | ||
315 : | src++; | ||
316 : | } | ||
317 : | } | ||
318 : | |||
319 : | void interpolate16x16_lowpass_v_c(uint8_t *dst, uint8_t *src, int32_t stride, int32_t rounding) | ||
320 : | { | ||
321 : | int32_t i; | ||
322 : | uint8_t round_add = 16 - rounding; | ||
323 : | |||
324 : | for(i = 0; i < 17; i++) | ||
325 : | { | ||
326 : | int32_t src0 = src[0]; | ||
327 : | int32_t src1 = src[stride]; | ||
328 : | int32_t src2 = src[2 * stride]; | ||
329 : | int32_t src3 = src[3 * stride]; | ||
330 : | int32_t src4 = src[4 * stride]; | ||
331 : | int32_t src5 = src[5 * stride]; | ||
332 : | int32_t src6 = src[6 * stride]; | ||
333 : | int32_t src7 = src[7 * stride]; | ||
334 : | int32_t src8 = src[8 * stride]; | ||
335 : | int32_t src9 = src[9 * stride]; | ||
336 : | int32_t src10 = src[10 * stride]; | ||
337 : | int32_t src11 = src[11 * stride]; | ||
338 : | int32_t src12 = src[12 * stride]; | ||
339 : | int32_t src13 = src[13 * stride]; | ||
340 : | int32_t src14 = src[14 * stride]; | ||
341 : | int32_t src15 = src[15 * stride]; | ||
342 : | int32_t src16 = src[16 * stride]; | ||
343 : | |||
344 : | edgomez | 1382 | |
345 : | edgomez | 851 | dst[0] = CLIP(((7 * ((src0<<1) - src2) + 23 * src1 + 3 * src3 - src4 + round_add) >> 5), 0, 255); |
346 : | dst[stride] = CLIP(((19 * src1 + 20 * src2 - src5 + 3 * (src4 - src0 - (src3<<1)) + round_add) >> 5), 0, 255); | ||
347 : | dst[2*stride] = CLIP(((20 * (src2 + src3) + (src0<<1) + 3 * (src5 - ((src1 + src4)<<1)) - src6 + round_add) >> 5), 0, 255); | ||
348 : | |||
349 : | dst[3*stride] = CLIP(((20 * (src3 + src4) + 3 * ((src6 + src1) - ((src2 + src5)<<1)) - (src0 + src7) + round_add) >> 5), 0, 255); | ||
350 : | dst[4*stride] = CLIP(((20 * (src4 + src5) - 3 * (((src3 + src6)<<1) - (src2 + src7)) - (src1 + src8) + round_add) >> 5), 0, 255); | ||
351 : | dst[5*stride] = CLIP(((20 * (src5 + src6) - 3 * (((src4 + src7)<<1) - (src3 + src8)) - (src2 + src9) + round_add) >> 5), 0, 255); | ||
352 : | dst[6*stride] = CLIP(((20 * (src6 + src7) - 3 * (((src5 + src8)<<1) - (src4 + src9)) - (src3 + src10) + round_add) >> 5), 0, 255); | ||
353 : | dst[7*stride] = CLIP(((20 * (src7 + src8) - 3 * (((src6 + src9)<<1) - (src5 + src10)) - (src4 + src11) + round_add) >> 5), 0, 255); | ||
354 : | dst[8*stride] = CLIP(((20 * (src8 + src9) - 3 * (((src7 + src10)<<1) - (src6 + src11)) - (src5 + src12) + round_add) >> 5), 0, 255); | ||
355 : | dst[9*stride] = CLIP(((20 * (src9 + src10) - 3 * (((src8 + src11)<<1) - (src7 + src12)) - (src6 + src13) + round_add) >> 5), 0, 255); | ||
356 : | dst[10*stride] = CLIP(((20 * (src10 + src11) - 3 * (((src9 + src12)<<1) - (src8 + src13)) - (src7 + src14) + round_add) >> 5), 0, 255); | ||
357 : | dst[11*stride] = CLIP(((20 * (src11 + src12) - 3 * (((src10 + src13)<<1) - (src9 + src14)) - (src8 + src15) + round_add) >> 5), 0, 255); | ||
358 : | dst[12*stride] = CLIP(((20 * (src12 + src13) - 3 * (((src11 + src14)<<1) - (src10 + src15)) - (src9 + src16) + round_add) >> 5), 0, 255); | ||
359 : | |||
360 : | dst[13*stride] = CLIP(((20 * (src13 + src14) + (src16<<1) + 3 * (src11 - ((src12 + src15) << 1)) - src10 + round_add) >> 5), 0, 255); | ||
361 : | dst[14*stride] = CLIP(((19 * src15 + 20 * src14 + 3 * (src12 - src16 - (src13 << 1)) - src11 + round_add) >> 5), 0, 255); | ||
362 : | dst[15*stride] = CLIP(((23 * src15 + 7 * ((src16<<1) - src14) + 3 * src13 - src12 + round_add) >> 5), 0, 255); | ||
363 : | |||
364 : | dst++; | ||
365 : | src++; | ||
366 : | } | ||
367 : | } | ||
368 : | |||
369 : | void interpolate8x8_lowpass_v_c(uint8_t *dst, uint8_t *src, int32_t stride, int32_t rounding) | ||
370 : | { | ||
371 : | int32_t i; | ||
372 : | uint8_t round_add = 16 - rounding; | ||
373 : | |||
374 : | for(i = 0; i < 9; i++) | ||
375 : | { | ||
376 : | int32_t src0 = src[0]; | ||
377 : | int32_t src1 = src[stride]; | ||
378 : | int32_t src2 = src[2 * stride]; | ||
379 : | int32_t src3 = src[3 * stride]; | ||
380 : | int32_t src4 = src[4 * stride]; | ||
381 : | int32_t src5 = src[5 * stride]; | ||
382 : | int32_t src6 = src[6 * stride]; | ||
383 : | int32_t src7 = src[7 * stride]; | ||
384 : | int32_t src8 = src[8 * stride]; | ||
385 : | edgomez | 1382 | |
386 : | edgomez | 851 | dst[0] = CLIP(((7 * ((src0<<1) - src2) + 23 * src1 + 3 * src3 - src4 + round_add) >> 5), 0, 255); |
387 : | dst[stride] = CLIP(((19 * src1 + 20 * src2 - src5 + 3 * (src4 - src0 - (src3 << 1)) + round_add) >> 5), 0, 255); | ||
388 : | dst[2 * stride] = CLIP(((20 * (src2 + src3) + (src0<<1) + 3 * (src5 - ((src1 + src4) <<1 )) - src6 + round_add) >> 5), 0, 255); | ||
389 : | dst[3 * stride] = CLIP(((20 * (src3 + src4) + 3 * ((src6 + src1) - ((src2 + src5)<<1)) - (src0 + src7) + round_add) >> 5), 0, 255); | ||
390 : | dst[4 * stride] = CLIP(((20 * (src4 + src5) + 3 * ((src2 + src7) - ((src3 + src6)<<1)) - (src1 + src8) + round_add) >> 5), 0, 255); | ||
391 : | dst[5 * stride] = CLIP(((20 * (src5 + src6) + (src8<<1) + 3 * (src3 - ((src4 + src7) << 1)) - src2 + round_add) >> 5), 0, 255); | ||
392 : | dst[6 * stride] = CLIP(((19 * src7 + 20 * src6 - src3 + 3 * (src4 - src8 - (src5 << 1)) + round_add) >> 5), 0, 255); | ||
393 : | dst[7 * stride] = CLIP(((7 * ((src8<<1) - src6) + 23 * src7 + 3 * src5 - src4 + round_add) >> 5), 0, 255); | ||
394 : | |||
395 : | dst++; | ||
396 : | src++; | ||
397 : | } | ||
398 : | } | ||
399 : | |||
400 : | void interpolate16x16_lowpass_hv_c(uint8_t *dst1, uint8_t *dst2, uint8_t *src, int32_t stride, int32_t rounding) | ||
401 : | { | ||
402 : | int32_t i; | ||
403 : | uint8_t round_add = 16 - rounding; | ||
404 : | uint8_t *h_ptr = dst2; | ||
405 : | |||
406 : | for(i = 0; i < 17; i++) | ||
407 : | { | ||
408 : | |||
409 : | h_ptr[0] = CLIP(((7 * ((src[0]<<1) - src[2]) + 23 * src[1] + 3 * src[3] - src[4] + round_add) >> 5), 0, 255); | ||
410 : | 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); | ||
411 : | 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); | ||
412 : | |||
413 : | 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); | ||
414 : | 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); | ||
415 : | 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); | ||
416 : | 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); | ||
417 : | 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); | ||
418 : | 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); | ||
419 : | 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); | ||
420 : | 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); | ||
421 : | 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); | ||
422 : | 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); | ||
423 : | |||
424 : | 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); | ||
425 : | 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); | ||
426 : | h_ptr[15] = CLIP(((23 * src[15] + 7 * ((src[16]<<1) - src[14]) + 3 * src[13] - src[12] + round_add) >> 5), 0, 255); | ||
427 : | |||
428 : | h_ptr += stride; | ||
429 : | src += stride; | ||
430 : | } | ||
431 : | |||
432 : | interpolate16x16_lowpass_v_c(dst1, dst2, stride, rounding); | ||
433 : | |||
434 : | } | ||
435 : | |||
436 : | void interpolate8x8_lowpass_hv_c(uint8_t *dst1, uint8_t *dst2, uint8_t *src, int32_t stride, int32_t rounding) | ||
437 : | { | ||
438 : | int32_t i; | ||
439 : | uint8_t round_add = 16 - rounding; | ||
440 : | uint8_t *h_ptr = dst2; | ||
441 : | |||
442 : | for(i = 0; i < 9; i++) | ||
443 : | { | ||
444 : | |||
445 : | h_ptr[0] = CLIP(((7 * ((src[0]<<1) - src[2]) + 23 * src[1] + 3 * src[3] - src[4] + round_add) >> 5), 0, 255); | ||
446 : | 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); | ||
447 : | 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); | ||
448 : | 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); | ||
449 : | 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); | ||
450 : | 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); | ||
451 : | 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); | ||
452 : | h_ptr[7] = CLIP(((23 * src[7] + 7 * ((src[8]<<1) - src[6]) + 3 * src[5] - src[4] + round_add) >> 5), 0, 255); | ||
453 : | |||
454 : | h_ptr += stride; | ||
455 : | src += stride; | ||
456 : | } | ||
457 : | |||
458 : | interpolate8x8_lowpass_v_c(dst1, dst2, stride, rounding); | ||
459 : | |||
460 : | } |
No admin address has been configured | ViewVC Help |
Powered by ViewVC 1.0.4 |