1 /*
2 * RV30 decoder motion compensation functions
3 * Copyright (c) 2007 Konstantin Shishkov
4 *
5 * This file is part of FFmpeg.
6 *
7 * FFmpeg is free software; you can redistribute it and/or
8 * modify it under the terms of the GNU Lesser General Public
9 * License as published by the Free Software Foundation; either
10 * version 2.1 of the License, or (at your option) any later version.
11 *
12 * FFmpeg is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
15 * Lesser General Public License for more details.
16 *
17 * You should have received a copy of the GNU Lesser General Public
18 * License along with FFmpeg; if not, write to the Free Software
19 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
20 */
21
22 /**
23 * @file
24 * RV30 decoder motion compensation functions
25 */
26
27 #include <stddef.h>
28 #include <stdint.h>
29 #include "libavutil/attributes.h"
30 #include "h264chroma.h"
31 #include "h264qpel.h"
32 #include "mathops.h"
33 #include "qpeldsp.h"
34 #include "rv34dsp.h"
35
36 #define RV30_LOWPASS(OPNAME, OP) \
37 static void OPNAME ## rv30_tpel8_h_lowpass(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, const int C1, const int C2){\
38 const int h = 8;\
39 const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP;\
40 int i;\
41 for(i = 0; i < h; i++)\
42 {\
43 OP(dst[0], (-(src[-1]+src[2]) + src[0]*C1 + src[1]*C2 + 8)>>4);\
44 OP(dst[1], (-(src[ 0]+src[3]) + src[1]*C1 + src[2]*C2 + 8)>>4);\
45 OP(dst[2], (-(src[ 1]+src[4]) + src[2]*C1 + src[3]*C2 + 8)>>4);\
46 OP(dst[3], (-(src[ 2]+src[5]) + src[3]*C1 + src[4]*C2 + 8)>>4);\
47 OP(dst[4], (-(src[ 3]+src[6]) + src[4]*C1 + src[5]*C2 + 8)>>4);\
48 OP(dst[5], (-(src[ 4]+src[7]) + src[5]*C1 + src[6]*C2 + 8)>>4);\
49 OP(dst[6], (-(src[ 5]+src[8]) + src[6]*C1 + src[7]*C2 + 8)>>4);\
50 OP(dst[7], (-(src[ 6]+src[9]) + src[7]*C1 + src[8]*C2 + 8)>>4);\
51 dst += dstStride;\
52 src += srcStride;\
53 }\
54 }\
55 \
56 static void OPNAME ## rv30_tpel8_v_lowpass(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, const int C1, const int C2){\
57 const int w = 8;\
58 const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP;\
59 int i;\
60 for(i = 0; i < w; i++)\
61 {\
62 const int srcA = src[-1*srcStride];\
63 const int src0 = src[0 *srcStride];\
64 const int src1 = src[1 *srcStride];\
65 const int src2 = src[2 *srcStride];\
66 const int src3 = src[3 *srcStride];\
67 const int src4 = src[4 *srcStride];\
68 const int src5 = src[5 *srcStride];\
69 const int src6 = src[6 *srcStride];\
70 const int src7 = src[7 *srcStride];\
71 const int src8 = src[8 *srcStride];\
72 const int src9 = src[9 *srcStride];\
73 OP(dst[0*dstStride], (-(srcA+src2) + src0*C1 + src1*C2 + 8)>>4);\
74 OP(dst[1*dstStride], (-(src0+src3) + src1*C1 + src2*C2 + 8)>>4);\
75 OP(dst[2*dstStride], (-(src1+src4) + src2*C1 + src3*C2 + 8)>>4);\
76 OP(dst[3*dstStride], (-(src2+src5) + src3*C1 + src4*C2 + 8)>>4);\
77 OP(dst[4*dstStride], (-(src3+src6) + src4*C1 + src5*C2 + 8)>>4);\
78 OP(dst[5*dstStride], (-(src4+src7) + src5*C1 + src6*C2 + 8)>>4);\
79 OP(dst[6*dstStride], (-(src5+src8) + src6*C1 + src7*C2 + 8)>>4);\
80 OP(dst[7*dstStride], (-(src6+src9) + src7*C1 + src8*C2 + 8)>>4);\
81 dst++;\
82 src++;\
83 }\
84 }\
85 \
86 static void OPNAME ## rv30_tpel8_hv_lowpass(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride){\
87 const int w = 8;\
88 const int h = 8;\
89 const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP;\
90 int i, j;\
91 for(j = 0; j < h; j++){\
92 for(i = 0; i < w; i++){\
93 OP(dst[i], (\
94 src[srcStride*-1+i-1] -12*src[srcStride*-1+i] -6*src[srcStride*-1+i+1] +src[srcStride*-1+i+2]+\
95 -12*src[srcStride* 0+i-1] +144*src[srcStride* 0+i] +72*src[srcStride* 0+i+1] -12*src[srcStride* 0+i+2] +\
96 -6*src[srcStride* 1+i-1] +72*src[srcStride* 1+i] +36*src[srcStride* 1+i+1] -6*src[srcStride* 1+i+2] +\
97 src[srcStride* 2+i-1] -12*src[srcStride* 2+i] -6*src[srcStride* 2+i+1] +src[srcStride* 2+i+2] +\
98 128)>>8);\
99 }\
100 src += srcStride;\
101 dst += dstStride;\
102 }\
103 }\
104 \
105 static void OPNAME ## rv30_tpel8_hhv_lowpass(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride){\
106 const int w = 8;\
107 const int h = 8;\
108 const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP;\
109 int i, j;\
110 for(j = 0; j < h; j++){\
111 for(i = 0; i < w; i++){\
112 OP(dst[i], (\
113 src[srcStride*-1+i-1] -12*src[srcStride*-1+i+1] -6*src[srcStride*-1+i] +src[srcStride*-1+i+2]+\
114 -12*src[srcStride* 0+i-1] +144*src[srcStride* 0+i+1] +72*src[srcStride* 0+i] -12*src[srcStride* 0+i+2]+\
115 -6*src[srcStride* 1+i-1] +72*src[srcStride* 1+i+1] +36*src[srcStride* 1+i] -6*src[srcStride* 1+i+2]+\
116 src[srcStride* 2+i-1] -12*src[srcStride* 2+i+1] -6*src[srcStride* 2+i] +src[srcStride* 2+i+2]+\
117 128)>>8);\
118 }\
119 src += srcStride;\
120 dst += dstStride;\
121 }\
122 }\
123 \
124 static void OPNAME ## rv30_tpel8_hvv_lowpass(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride){\
125 const int w = 8;\
126 const int h = 8;\
127 const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP;\
128 int i, j;\
129 for(j = 0; j < h; j++){\
130 for(i = 0; i < w; i++){\
131 OP(dst[i], (\
132 src[srcStride*-1+i-1] -12*src[srcStride*-1+i] -6*src[srcStride*-1+i+1] +src[srcStride*-1+i+2]+\
133 -6*src[srcStride* 0+i-1] +72*src[srcStride* 0+i] +36*src[srcStride* 0+i+1] -6*src[srcStride* 0+i+2]+\
134 -12*src[srcStride* 1+i-1] +144*src[srcStride* 1+i] +72*src[srcStride* 1+i+1] -12*src[srcStride* 1+i+2]+\
135 src[srcStride* 2+i-1] -12*src[srcStride* 2+i] -6*src[srcStride* 2+i+1] +src[srcStride* 2+i+2]+\
136 128)>>8);\
137 }\
138 src += srcStride;\
139 dst += dstStride;\
140 }\
141 }\
142 \
143 static void OPNAME ## rv30_tpel8_hhvv_lowpass(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride){\
144 const int w = 8;\
145 const int h = 8;\
146 const uint8_t *cm = ff_crop_tab + MAX_NEG_CROP;\
147 int i, j;\
148 for(j = 0; j < h; j++){\
149 for(i = 0; i < w; i++){\
150 OP(dst[i], (\
151 36*src[i+srcStride*0] +54*src[i+1+srcStride*0] +6*src[i+2+srcStride*0]+\
152 54*src[i+srcStride*1] +81*src[i+1+srcStride*1] +9*src[i+2+srcStride*1]+\
153 6*src[i+srcStride*2] + 9*src[i+1+srcStride*2] + src[i+2+srcStride*2]+\
154 128)>>8);\
155 }\
156 src += srcStride;\
157 dst += dstStride;\
158 }\
159 }\
160 \
161 static void OPNAME ## rv30_tpel16_v_lowpass(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, const int C1, const int C2){\
162 OPNAME ## rv30_tpel8_v_lowpass(dst , src , dstStride, srcStride, C1, C2);\
163 OPNAME ## rv30_tpel8_v_lowpass(dst+8, src+8, dstStride, srcStride, C1, C2);\
164 src += 8*srcStride;\
165 dst += 8*dstStride;\
166 OPNAME ## rv30_tpel8_v_lowpass(dst , src , dstStride, srcStride, C1, C2);\
167 OPNAME ## rv30_tpel8_v_lowpass(dst+8, src+8, dstStride, srcStride, C1, C2);\
168 }\
169 \
170 static void OPNAME ## rv30_tpel16_h_lowpass(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride, const int C1, const int C2){\
171 OPNAME ## rv30_tpel8_h_lowpass(dst , src , dstStride, srcStride, C1, C2);\
172 OPNAME ## rv30_tpel8_h_lowpass(dst+8, src+8, dstStride, srcStride, C1, C2);\
173 src += 8*srcStride;\
174 dst += 8*dstStride;\
175 OPNAME ## rv30_tpel8_h_lowpass(dst , src , dstStride, srcStride, C1, C2);\
176 OPNAME ## rv30_tpel8_h_lowpass(dst+8, src+8, dstStride, srcStride, C1, C2);\
177 }\
178 \
179 static void OPNAME ## rv30_tpel16_hv_lowpass(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride){\
180 OPNAME ## rv30_tpel8_hv_lowpass(dst , src , dstStride, srcStride);\
181 OPNAME ## rv30_tpel8_hv_lowpass(dst+8, src+8, dstStride, srcStride);\
182 src += 8*srcStride;\
183 dst += 8*dstStride;\
184 OPNAME ## rv30_tpel8_hv_lowpass(dst , src , dstStride, srcStride);\
185 OPNAME ## rv30_tpel8_hv_lowpass(dst+8, src+8, dstStride, srcStride);\
186 }\
187 \
188 static void OPNAME ## rv30_tpel16_hhv_lowpass(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride){\
189 OPNAME ## rv30_tpel8_hhv_lowpass(dst , src , dstStride, srcStride);\
190 OPNAME ## rv30_tpel8_hhv_lowpass(dst+8, src+8, dstStride, srcStride);\
191 src += 8*srcStride;\
192 dst += 8*dstStride;\
193 OPNAME ## rv30_tpel8_hhv_lowpass(dst , src , dstStride, srcStride);\
194 OPNAME ## rv30_tpel8_hhv_lowpass(dst+8, src+8, dstStride, srcStride);\
195 }\
196 \
197 static void OPNAME ## rv30_tpel16_hvv_lowpass(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride){\
198 OPNAME ## rv30_tpel8_hvv_lowpass(dst , src , dstStride, srcStride);\
199 OPNAME ## rv30_tpel8_hvv_lowpass(dst+8, src+8, dstStride, srcStride);\
200 src += 8*srcStride;\
201 dst += 8*dstStride;\
202 OPNAME ## rv30_tpel8_hvv_lowpass(dst , src , dstStride, srcStride);\
203 OPNAME ## rv30_tpel8_hvv_lowpass(dst+8, src+8, dstStride, srcStride);\
204 }\
205 \
206 static void OPNAME ## rv30_tpel16_hhvv_lowpass(uint8_t *dst, const uint8_t *src, int dstStride, int srcStride){\
207 OPNAME ## rv30_tpel8_hhvv_lowpass(dst , src , dstStride, srcStride);\
208 OPNAME ## rv30_tpel8_hhvv_lowpass(dst+8, src+8, dstStride, srcStride);\
209 src += 8*srcStride;\
210 dst += 8*dstStride;\
211 OPNAME ## rv30_tpel8_hhvv_lowpass(dst , src , dstStride, srcStride);\
212 OPNAME ## rv30_tpel8_hhvv_lowpass(dst+8, src+8, dstStride, srcStride);\
213 }\
214 \
215
216 #define RV30_MC(OPNAME, SIZE) \
217 static void OPNAME ## rv30_tpel ## SIZE ## _mc10_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
218 {\
219 OPNAME ## rv30_tpel ## SIZE ## _h_lowpass(dst, src, stride, stride, 12, 6);\
220 }\
221 \
222 static void OPNAME ## rv30_tpel ## SIZE ## _mc20_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
223 {\
224 OPNAME ## rv30_tpel ## SIZE ## _h_lowpass(dst, src, stride, stride, 6, 12);\
225 }\
226 \
227 static void OPNAME ## rv30_tpel ## SIZE ## _mc01_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
228 {\
229 OPNAME ## rv30_tpel ## SIZE ## _v_lowpass(dst, src, stride, stride, 12, 6);\
230 }\
231 \
232 static void OPNAME ## rv30_tpel ## SIZE ## _mc02_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
233 {\
234 OPNAME ## rv30_tpel ## SIZE ## _v_lowpass(dst, src, stride, stride, 6, 12);\
235 }\
236 \
237 static void OPNAME ## rv30_tpel ## SIZE ## _mc11_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
238 {\
239 OPNAME ## rv30_tpel ## SIZE ## _hv_lowpass(dst, src, stride, stride);\
240 }\
241 \
242 static void OPNAME ## rv30_tpel ## SIZE ## _mc12_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
243 {\
244 OPNAME ## rv30_tpel ## SIZE ## _hvv_lowpass(dst, src, stride, stride);\
245 }\
246 \
247 static void OPNAME ## rv30_tpel ## SIZE ## _mc21_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
248 {\
249 OPNAME ## rv30_tpel ## SIZE ## _hhv_lowpass(dst, src, stride, stride);\
250 }\
251 \
252 static void OPNAME ## rv30_tpel ## SIZE ## _mc22_c(uint8_t *dst, const uint8_t *src, ptrdiff_t stride)\
253 {\
254 OPNAME ## rv30_tpel ## SIZE ## _hhvv_lowpass(dst, src, stride, stride);\
255 }\
256 \
257
258 #define op_avg(a, b) a = (((a)+cm[b]+1)>>1)
259 #define op_put(a, b) a = cm[b]
260
RV30_LOWPASS(put_,op_put)261 RV30_LOWPASS(put_ , op_put)
262 RV30_LOWPASS(avg_ , op_avg)
263 RV30_MC(put_, 8)
264 RV30_MC(put_, 16)
265 RV30_MC(avg_, 8)
266 RV30_MC(avg_, 16)
267
268 av_cold void ff_rv30dsp_init(RV34DSPContext *c)
269 {
270 H264ChromaContext h264chroma;
271 H264QpelContext qpel;
272
273 ff_rv34dsp_init(c);
274 ff_h264chroma_init(&h264chroma, 8);
275 ff_h264qpel_init(&qpel, 8);
276
277 c->put_pixels_tab[0][ 0] = qpel.put_h264_qpel_pixels_tab[0][0];
278 c->put_pixels_tab[0][ 1] = put_rv30_tpel16_mc10_c;
279 c->put_pixels_tab[0][ 2] = put_rv30_tpel16_mc20_c;
280 c->put_pixels_tab[0][ 4] = put_rv30_tpel16_mc01_c;
281 c->put_pixels_tab[0][ 5] = put_rv30_tpel16_mc11_c;
282 c->put_pixels_tab[0][ 6] = put_rv30_tpel16_mc21_c;
283 c->put_pixels_tab[0][ 8] = put_rv30_tpel16_mc02_c;
284 c->put_pixels_tab[0][ 9] = put_rv30_tpel16_mc12_c;
285 c->put_pixels_tab[0][10] = put_rv30_tpel16_mc22_c;
286 c->avg_pixels_tab[0][ 0] = qpel.avg_h264_qpel_pixels_tab[0][0];
287 c->avg_pixels_tab[0][ 1] = avg_rv30_tpel16_mc10_c;
288 c->avg_pixels_tab[0][ 2] = avg_rv30_tpel16_mc20_c;
289 c->avg_pixels_tab[0][ 4] = avg_rv30_tpel16_mc01_c;
290 c->avg_pixels_tab[0][ 5] = avg_rv30_tpel16_mc11_c;
291 c->avg_pixels_tab[0][ 6] = avg_rv30_tpel16_mc21_c;
292 c->avg_pixels_tab[0][ 8] = avg_rv30_tpel16_mc02_c;
293 c->avg_pixels_tab[0][ 9] = avg_rv30_tpel16_mc12_c;
294 c->avg_pixels_tab[0][10] = avg_rv30_tpel16_mc22_c;
295 c->put_pixels_tab[1][ 0] = qpel.put_h264_qpel_pixels_tab[1][0];
296 c->put_pixels_tab[1][ 1] = put_rv30_tpel8_mc10_c;
297 c->put_pixels_tab[1][ 2] = put_rv30_tpel8_mc20_c;
298 c->put_pixels_tab[1][ 4] = put_rv30_tpel8_mc01_c;
299 c->put_pixels_tab[1][ 5] = put_rv30_tpel8_mc11_c;
300 c->put_pixels_tab[1][ 6] = put_rv30_tpel8_mc21_c;
301 c->put_pixels_tab[1][ 8] = put_rv30_tpel8_mc02_c;
302 c->put_pixels_tab[1][ 9] = put_rv30_tpel8_mc12_c;
303 c->put_pixels_tab[1][10] = put_rv30_tpel8_mc22_c;
304 c->avg_pixels_tab[1][ 0] = qpel.avg_h264_qpel_pixels_tab[1][0];
305 c->avg_pixels_tab[1][ 1] = avg_rv30_tpel8_mc10_c;
306 c->avg_pixels_tab[1][ 2] = avg_rv30_tpel8_mc20_c;
307 c->avg_pixels_tab[1][ 4] = avg_rv30_tpel8_mc01_c;
308 c->avg_pixels_tab[1][ 5] = avg_rv30_tpel8_mc11_c;
309 c->avg_pixels_tab[1][ 6] = avg_rv30_tpel8_mc21_c;
310 c->avg_pixels_tab[1][ 8] = avg_rv30_tpel8_mc02_c;
311 c->avg_pixels_tab[1][ 9] = avg_rv30_tpel8_mc12_c;
312 c->avg_pixels_tab[1][10] = avg_rv30_tpel8_mc22_c;
313
314 c->put_chroma_pixels_tab[0] = h264chroma.put_h264_chroma_pixels_tab[0];
315 c->put_chroma_pixels_tab[1] = h264chroma.put_h264_chroma_pixels_tab[1];
316 c->avg_chroma_pixels_tab[0] = h264chroma.avg_h264_chroma_pixels_tab[0];
317 c->avg_chroma_pixels_tab[1] = h264chroma.avg_h264_chroma_pixels_tab[1];
318 }
319