Parent Directory | Revision Log
Revision 579 - (view) (download)
1 : | Isibaar | 333 | /************************************************************************** |
2 : | * | ||
3 : | * XVID MPEG-4 VIDEO CODEC | ||
4 : | * 8x8 block-based halfpel interpolation | ||
5 : | * | ||
6 : | * 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 : | * | ||
11 : | * 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 : | * | ||
16 : | * 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 : | * | ||
20 : | *************************************************************************/ | ||
21 : | |||
22 : | /************************************************************************** | ||
23 : | * | ||
24 : | * History: | ||
25 : | Isibaar | 579 | * |
26 : | * 05.10.2002 new bilinear and qpel interpolation code - Isibaar | ||
27 : | Isibaar | 333 | * 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" | ||
34 : | #include "interpolate8x8.h" | ||
35 : | |||
36 : | // function pointers | ||
37 : | INTERPOLATE8X8_PTR interpolate8x8_halfpel_h; | ||
38 : | INTERPOLATE8X8_PTR interpolate8x8_halfpel_v; | ||
39 : | INTERPOLATE8X8_PTR interpolate8x8_halfpel_hv; | ||
40 : | |||
41 : | Isibaar | 579 | INTERPOLATE8X8_AVG2_PTR interpolate8x8_avg2; |
42 : | INTERPOLATE8X8_AVG4_PTR interpolate8x8_avg4; | ||
43 : | Isibaar | 333 | |
44 : | Isibaar | 579 | INTERPOLATE8X8_LOWPASS_PTR interpolate8x8_lowpass_h; |
45 : | INTERPOLATE8X8_LOWPASS_PTR interpolate8x8_lowpass_v; | ||
46 : | INTERPOLATE8X8_LOWPASS_HV_PTR interpolate8x8_lowpass_hv; | ||
47 : | |||
48 : | INTERPOLATE8X8_6TAP_LOWPASS_PTR interpolate8x8_6tap_lowpass_h; | ||
49 : | INTERPOLATE8X8_6TAP_LOWPASS_PTR interpolate8x8_6tap_lowpass_v; | ||
50 : | |||
51 : | void interpolate8x8_avg2_c(uint8_t *dst, const uint8_t *src1, const uint8_t *src2, const uint32_t stride, const uint32_t rounding) | ||
52 : | { | ||
53 : | int32_t i; | ||
54 : | const int32_t round = 1 - rounding; | ||
55 : | |||
56 : | for(i = 0; i < 8; i++) | ||
57 : | { | ||
58 : | dst[0] = (src1[0] + src2[0] + round) >> 1; | ||
59 : | dst[1] = (src1[1] + src2[1] + round) >> 1; | ||
60 : | dst[2] = (src1[2] + src2[2] + round) >> 1; | ||
61 : | dst[3] = (src1[3] + src2[3] + round) >> 1; | ||
62 : | dst[4] = (src1[4] + src2[4] + round) >> 1; | ||
63 : | dst[5] = (src1[5] + src2[5] + round) >> 1; | ||
64 : | dst[6] = (src1[6] + src2[6] + round) >> 1; | ||
65 : | dst[7] = (src1[7] + src2[7] + round) >> 1; | ||
66 : | |||
67 : | dst += stride; | ||
68 : | src1 += stride; | ||
69 : | src2 += stride; | ||
70 : | } | ||
71 : | } | ||
72 : | |||
73 : | 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) | ||
74 : | { | ||
75 : | int32_t i; | ||
76 : | const int32_t round = 2 - rounding; | ||
77 : | |||
78 : | for(i = 0; i < 8; i++) | ||
79 : | { | ||
80 : | dst[0] = (src1[0] + src2[0] + src3[0] + src4[0] + round) >> 2; | ||
81 : | dst[1] = (src1[1] + src2[1] + src3[1] + src4[1] + round) >> 2; | ||
82 : | dst[2] = (src1[2] + src2[2] + src3[2] + src4[2] + round) >> 2; | ||
83 : | dst[3] = (src1[3] + src2[3] + src3[3] + src4[3] + round) >> 2; | ||
84 : | dst[4] = (src1[4] + src2[4] + src3[4] + src4[4] + round) >> 2; | ||
85 : | dst[5] = (src1[5] + src2[5] + src3[5] + src4[5] + round) >> 2; | ||
86 : | dst[6] = (src1[6] + src2[6] + src3[6] + src4[6] + round) >> 2; | ||
87 : | dst[7] = (src1[7] + src2[7] + src3[7] + src4[7] + round) >> 2; | ||
88 : | |||
89 : | dst += stride; | ||
90 : | src1 += stride; | ||
91 : | src2 += stride; | ||
92 : | src3 += stride; | ||
93 : | src4 += stride; | ||
94 : | } | ||
95 : | } | ||
96 : | |||
97 : | Isibaar | 333 | // dst = interpolate(src) |
98 : | |||
99 : | void | ||
100 : | interpolate8x8_halfpel_h_c(uint8_t * const dst, | ||
101 : | const uint8_t * const src, | ||
102 : | const uint32_t stride, | ||
103 : | const uint32_t rounding) | ||
104 : | { | ||
105 : | uint32_t i, j; | ||
106 : | |||
107 : | for (j = 0; j < 8; j++) { | ||
108 : | for (i = 0; i < 8; i++) { | ||
109 : | |||
110 : | int16_t tot = | ||
111 : | (int32_t) src[j * stride + i] + (int32_t) src[j * stride + i + | ||
112 : | 1]; | ||
113 : | |||
114 : | tot = (int32_t) ((tot + 1 - rounding) >> 1); | ||
115 : | dst[j * stride + i] = (uint8_t) tot; | ||
116 : | } | ||
117 : | } | ||
118 : | } | ||
119 : | |||
120 : | |||
121 : | |||
122 : | void | ||
123 : | interpolate8x8_halfpel_v_c(uint8_t * const dst, | ||
124 : | const uint8_t * const src, | ||
125 : | const uint32_t stride, | ||
126 : | const uint32_t rounding) | ||
127 : | { | ||
128 : | uint32_t i, j; | ||
129 : | |||
130 : | for (j = 0; j < 8; j++) { | ||
131 : | for (i = 0; i < 8; i++) { | ||
132 : | int16_t tot = src[j * stride + i] + src[j * stride + i + stride]; | ||
133 : | |||
134 : | tot = ((tot + 1 - rounding) >> 1); | ||
135 : | dst[j * stride + i] = (uint8_t) tot; | ||
136 : | } | ||
137 : | } | ||
138 : | } | ||
139 : | |||
140 : | |||
141 : | void | ||
142 : | interpolate8x8_halfpel_hv_c(uint8_t * const dst, | ||
143 : | const uint8_t * const src, | ||
144 : | const uint32_t stride, | ||
145 : | const uint32_t rounding) | ||
146 : | { | ||
147 : | uint32_t i, j; | ||
148 : | |||
149 : | for (j = 0; j < 8; j++) { | ||
150 : | for (i = 0; i < 8; i++) { | ||
151 : | int16_t tot = | ||
152 : | src[j * stride + i] + src[j * stride + i + 1] + | ||
153 : | src[j * stride + i + stride] + src[j * stride + i + stride + | ||
154 : | 1]; | ||
155 : | tot = ((tot + 2 - rounding) >> 2); | ||
156 : | dst[j * stride + i] = (uint8_t) tot; | ||
157 : | } | ||
158 : | } | ||
159 : | } | ||
160 : | |||
161 : | /************************************************************* | ||
162 : | * QPEL STUFF STARTS HERE * | ||
163 : | *************************************************************/ | ||
164 : | |||
165 : | #define CLIP(X,A,B) (X < A) ? (A) : ((X > B) ? (B) : (X)) | ||
166 : | |||
167 : | Isibaar | 579 | void interpolate8x8_6tap_lowpass_h_c(uint8_t *dst, uint8_t *src, int32_t stride, int32_t rounding) |
168 : | Isibaar | 333 | { |
169 : | int32_t i; | ||
170 : | Isibaar | 579 | uint8_t round_add = 16 - rounding; |
171 : | Isibaar | 333 | |
172 : | for(i = 0; i < 8; i++) | ||
173 : | { | ||
174 : | |||
175 : | Isibaar | 579 | dst[0] = CLIP((((src[-2] + src[3]) + 5 * (((src[0] + src[1])<<2) - (src[-1] + src[2])) + round_add) >> 5), 0, 255); |
176 : | dst[1] = CLIP((((src[-1] + src[4]) + 5 * (((src[1] + src[2])<<2) - (src[0] + src[3])) + round_add) >> 5), 0, 255); | ||
177 : | dst[2] = CLIP((((src[0] + src[5]) + 5 * (((src[2] + src[3])<<2) - (src[1] + src[4])) + round_add) >> 5), 0, 255); | ||
178 : | dst[3] = CLIP((((src[1] + src[6]) + 5 * (((src[3] + src[4])<<2) - (src[2] + src[5])) + round_add) >> 5), 0, 255); | ||
179 : | dst[4] = CLIP((((src[2] + src[7]) + 5 * (((src[4] + src[5])<<2) - (src[3] + src[6])) + round_add) >> 5), 0, 255); | ||
180 : | dst[5] = CLIP((((src[3] + src[8]) + 5 * (((src[5] + src[6])<<2) - (src[4] + src[7])) + round_add) >> 5), 0, 255); | ||
181 : | dst[6] = CLIP((((src[4] + src[9]) + 5 * (((src[6] + src[7])<<2) - (src[5] + src[8])) + round_add) >> 5), 0, 255); | ||
182 : | dst[7] = CLIP((((src[5] + src[10]) + 5 * (((src[7] + src[8])<<2) - (src[6] + src[9])) + round_add) >> 5), 0, 255); | ||
183 : | |||
184 : | dst += stride; | ||
185 : | src += stride; | ||
186 : | Isibaar | 333 | } |
187 : | } | ||
188 : | |||
189 : | Isibaar | 579 | void interpolate8x8_lowpass_h_c(uint8_t *dst, uint8_t *src, int32_t stride, int32_t rounding) |
190 : | Isibaar | 333 | { |
191 : | int32_t i; | ||
192 : | Isibaar | 579 | uint8_t round_add = 16 - rounding; |
193 : | Isibaar | 333 | |
194 : | for(i = 0; i < 8; i++) | ||
195 : | { | ||
196 : | Isibaar | 579 | |
197 : | dst[0] = CLIP(((7 * ((src[0]<<1) - src[2]) + 23 * src[1] + 3 * src[3] - src[4] + round_add) >> 5), 0, 255); | ||
198 : | dst[1] = CLIP(((19 * src[1] + 20 * src[2] - src[5] + 3 * (src[4] - src[0] - (src[3]<<1)) + round_add) >> 5), 0, 255); | ||
199 : | 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); | ||
200 : | 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); | ||
201 : | 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); | ||
202 : | 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); | ||
203 : | dst[6] = CLIP(((19 * src[7] + 20 * src[6] + 3 * (src[4] - src[8] - (src[5] << 1)) - src[3] + round_add) >> 5), 0, 255); | ||
204 : | dst[7] = CLIP(((23 * src[7] + 7 * ((src[8]<<1) - src[6]) + 3 * src[5] - src[4] + round_add) >> 5), 0, 255); | ||
205 : | |||
206 : | dst += stride; | ||
207 : | src += stride; | ||
208 : | Isibaar | 333 | } |
209 : | } | ||
210 : | |||
211 : | Isibaar | 579 | void interpolate8x8_6tap_lowpass_v_c(uint8_t *dst, uint8_t *src, int32_t stride, int32_t rounding) |
212 : | Isibaar | 333 | { |
213 : | Isibaar | 579 | int32_t i; |
214 : | uint8_t round_add = 16 - rounding; | ||
215 : | Isibaar | 333 | |
216 : | Isibaar | 579 | for(i = 0; i < 8; i++) |
217 : | Isibaar | 333 | { |
218 : | Isibaar | 579 | int32_t src_2 = src[-2*stride]; |
219 : | int32_t src_1 = src[-stride]; | ||
220 : | int32_t src0 = src[0]; | ||
221 : | int32_t src1 = src[stride]; | ||
222 : | int32_t src2 = src[2 * stride]; | ||
223 : | int32_t src3 = src[3 * stride]; | ||
224 : | int32_t src4 = src[4 * stride]; | ||
225 : | int32_t src5 = src[5 * stride]; | ||
226 : | int32_t src6 = src[6 * stride]; | ||
227 : | int32_t src7 = src[7 * stride]; | ||
228 : | int32_t src8 = src[8 * stride]; | ||
229 : | int32_t src9 = src[9 * stride]; | ||
230 : | int32_t src10 = src[10 * stride]; | ||
231 : | Isibaar | 333 | |
232 : | Isibaar | 579 | dst[0] = CLIP((((src_2 + src3) + 5 * (((src0 + src1)<<2) - (src_1 + src2)) + round_add) >> 5), 0, 255); |
233 : | dst[stride] = CLIP((((src_1 + src4) + 5 * (((src1 + src2)<<2) - (src0 + src3)) + round_add) >> 5), 0, 255); | ||
234 : | dst[2 * stride] = CLIP((((src0 + src5) + 5 * (((src2 + src3)<<2) - (src1 + src4)) + round_add) >> 5), 0, 255); | ||
235 : | dst[3 * stride] = CLIP((((src1 + src6) + 5 * (((src3 + src4)<<2) - (src2 + src5)) + round_add) >> 5), 0, 255); | ||
236 : | dst[4 * stride] = CLIP((((src2 + src7) + 5 * (((src4 + src5)<<2) - (src3 + src6)) + round_add) >> 5), 0, 255); | ||
237 : | dst[5 * stride] = CLIP((((src3 + src8) + 5 * (((src5 + src6)<<2) - (src4 + src7)) + round_add) >> 5), 0, 255); | ||
238 : | dst[6 * stride] = CLIP((((src4 + src9) + 5 * (((src6 + src7)<<2) - (src5 + src8)) + round_add) >> 5), 0, 255); | ||
239 : | dst[7 * stride] = CLIP((((src5 + src10) + 5 * (((src7 + src8)<<2) - (src6 + src9)) + round_add) >> 5), 0, 255); | ||
240 : | Isibaar | 333 | |
241 : | Isibaar | 579 | dst++; |
242 : | src++; | ||
243 : | Isibaar | 333 | } |
244 : | } | ||
245 : | |||
246 : | Isibaar | 579 | void interpolate8x8_lowpass_v_c(uint8_t *dst, uint8_t *src, int32_t stride, int32_t rounding) |
247 : | Isibaar | 333 | { |
248 : | int32_t i; | ||
249 : | Isibaar | 579 | uint8_t round_add = 16 - rounding; |
250 : | Isibaar | 333 | |
251 : | for(i = 0; i < 8; i++) | ||
252 : | { | ||
253 : | Isibaar | 579 | int32_t src0 = src[0]; |
254 : | int32_t src1 = src[stride]; | ||
255 : | int32_t src2 = src[2 * stride]; | ||
256 : | int32_t src3 = src[3 * stride]; | ||
257 : | int32_t src4 = src[4 * stride]; | ||
258 : | int32_t src5 = src[5 * stride]; | ||
259 : | int32_t src6 = src[6 * stride]; | ||
260 : | int32_t src7 = src[7 * stride]; | ||
261 : | int32_t src8 = src[8 * stride]; | ||
262 : | |||
263 : | dst[0] = CLIP(((7 * ((src0<<1) - src2) + 23 * src1 + 3 * src3 - src4 + round_add) >> 5), 0, 255); | ||
264 : | dst[stride] = CLIP(((19 * src1 + 20 * src2 - src5 + 3 * (src4 - src0 - (src3 << 1)) + round_add) >> 5), 0, 255); | ||
265 : | dst[2 * stride] = CLIP(((20 * (src2 + src3) + (src0<<1) + 3 * (src5 - ((src1 + src4) <<1 )) - src6 + round_add) >> 5), 0, 255); | ||
266 : | dst[3 * stride] = CLIP(((20 * (src3 + src4) + 3 * ((src6 + src1) - ((src2 + src5)<<1)) - (src0 + src7) + round_add) >> 5), 0, 255); | ||
267 : | dst[4 * stride] = CLIP(((20 * (src4 + src5) + 3 * ((src2 + src7) - ((src3 + src6)<<1)) - (src1 + src8) + round_add) >> 5), 0, 255); | ||
268 : | dst[5 * stride] = CLIP(((20 * (src5 + src6) + (src8<<1) + 3 * (src3 - ((src4 + src7) << 1)) - src2 + round_add) >> 5), 0, 255); | ||
269 : | dst[6 * stride] = CLIP(((19 * src7 + 20 * src6 - src3 + 3 * (src4 - src8 - (src5 << 1)) + round_add) >> 5), 0, 255); | ||
270 : | dst[7 * stride] = CLIP(((7 * ((src8<<1) - src6) + 23 * src7 + 3 * src5 - src4 + round_add) >> 5), 0, 255); | ||
271 : | Isibaar | 333 | |
272 : | Isibaar | 579 | dst++; |
273 : | src++; | ||
274 : | Isibaar | 333 | } |
275 : | } | ||
276 : | |||
277 : | Isibaar | 579 | void interpolate8x8_lowpass_hv_c(uint8_t *dst1, uint8_t *dst2, uint8_t *src, int32_t stride, int32_t rounding) |
278 : | Isibaar | 333 | { |
279 : | Isibaar | 579 | int32_t i; |
280 : | uint8_t round_add = 16 - rounding; | ||
281 : | uint8_t *h_ptr = dst2; | ||
282 : | Isibaar | 333 | |
283 : | Isibaar | 579 | for(i = 0; i < 9; i++) |
284 : | Isibaar | 333 | { |
285 : | Isibaar | 579 | |
286 : | h_ptr[0] = CLIP(((7 * ((src[0]<<1) - src[2]) + 23 * src[1] + 3 * src[3] - src[4] + round_add) >> 5), 0, 255); | ||
287 : | 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); | ||
288 : | 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); | ||
289 : | 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); | ||
290 : | 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); | ||
291 : | 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); | ||
292 : | 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); | ||
293 : | h_ptr[7] = CLIP(((23 * src[7] + 7 * ((src[8]<<1) - src[6]) + 3 * src[5] - src[4] + round_add) >> 5), 0, 255); | ||
294 : | |||
295 : | h_ptr += stride; | ||
296 : | src += stride; | ||
297 : | Isibaar | 333 | } |
298 : | Isibaar | 579 | |
299 : | interpolate8x8_lowpass_v_c(dst1, dst2, stride, rounding); | ||
300 : | |||
301 : | } |
No admin address has been configured | ViewVC Help |
Powered by ViewVC 1.0.4 |