2 * RV30 decoder motion compensation functions
3 * Copyright (c) 2007 Konstantin Shishkov
5 * This file is part of FFmpeg.
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.
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.
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
24 * RV30 decoder motion compensation functions
28 #include "h264chroma.h"
32 #define RV30_LOWPASS(OPNAME, OP) \
33 static av_unused void OPNAME ## rv30_tpel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, const int C1, const int C2){\
35 const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
37 for(i = 0; i < h; i++)\
39 OP(dst[0], (-(src[-1]+src[2]) + src[0]*C1 + src[1]*C2 + 8)>>4);\
40 OP(dst[1], (-(src[ 0]+src[3]) + src[1]*C1 + src[2]*C2 + 8)>>4);\
41 OP(dst[2], (-(src[ 1]+src[4]) + src[2]*C1 + src[3]*C2 + 8)>>4);\
42 OP(dst[3], (-(src[ 2]+src[5]) + src[3]*C1 + src[4]*C2 + 8)>>4);\
43 OP(dst[4], (-(src[ 3]+src[6]) + src[4]*C1 + src[5]*C2 + 8)>>4);\
44 OP(dst[5], (-(src[ 4]+src[7]) + src[5]*C1 + src[6]*C2 + 8)>>4);\
45 OP(dst[6], (-(src[ 5]+src[8]) + src[6]*C1 + src[7]*C2 + 8)>>4);\
46 OP(dst[7], (-(src[ 6]+src[9]) + src[7]*C1 + src[8]*C2 + 8)>>4);\
52 static void OPNAME ## rv30_tpel8_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, const int C1, const int C2){\
54 const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
56 for(i = 0; i < w; i++)\
58 const int srcA = src[-1*srcStride];\
59 const int src0 = src[0 *srcStride];\
60 const int src1 = src[1 *srcStride];\
61 const int src2 = src[2 *srcStride];\
62 const int src3 = src[3 *srcStride];\
63 const int src4 = src[4 *srcStride];\
64 const int src5 = src[5 *srcStride];\
65 const int src6 = src[6 *srcStride];\
66 const int src7 = src[7 *srcStride];\
67 const int src8 = src[8 *srcStride];\
68 const int src9 = src[9 *srcStride];\
69 OP(dst[0*dstStride], (-(srcA+src2) + src0*C1 + src1*C2 + 8)>>4);\
70 OP(dst[1*dstStride], (-(src0+src3) + src1*C1 + src2*C2 + 8)>>4);\
71 OP(dst[2*dstStride], (-(src1+src4) + src2*C1 + src3*C2 + 8)>>4);\
72 OP(dst[3*dstStride], (-(src2+src5) + src3*C1 + src4*C2 + 8)>>4);\
73 OP(dst[4*dstStride], (-(src3+src6) + src4*C1 + src5*C2 + 8)>>4);\
74 OP(dst[5*dstStride], (-(src4+src7) + src5*C1 + src6*C2 + 8)>>4);\
75 OP(dst[6*dstStride], (-(src5+src8) + src6*C1 + src7*C2 + 8)>>4);\
76 OP(dst[7*dstStride], (-(src6+src9) + src7*C1 + src8*C2 + 8)>>4);\
82 static void OPNAME ## rv30_tpel8_hv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
85 const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
87 for(j = 0; j < h; j++){\
88 for(i = 0; i < w; i++){\
90 src[srcStride*-1+i-1] -12*src[srcStride*-1+i] -6*src[srcStride*-1+i+1] +src[srcStride*-1+i+2]+\
91 -12*src[srcStride* 0+i-1] +144*src[srcStride* 0+i] +72*src[srcStride* 0+i+1] -12*src[srcStride* 0+i+2] +\
92 -6*src[srcStride* 1+i-1] +72*src[srcStride* 1+i] +36*src[srcStride* 1+i+1] -6*src[srcStride* 1+i+2] +\
93 src[srcStride* 2+i-1] -12*src[srcStride* 2+i] -6*src[srcStride* 2+i+1] +src[srcStride* 2+i+2] +\
101 static void OPNAME ## rv30_tpel8_hhv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
104 const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
106 for(j = 0; j < h; j++){\
107 for(i = 0; i < w; i++){\
109 src[srcStride*-1+i-1] -12*src[srcStride*-1+i+1] -6*src[srcStride*-1+i] +src[srcStride*-1+i+2]+\
110 -12*src[srcStride* 0+i-1] +144*src[srcStride* 0+i+1] +72*src[srcStride* 0+i] -12*src[srcStride* 0+i+2]+\
111 -6*src[srcStride* 1+i-1] +72*src[srcStride* 1+i+1] +36*src[srcStride* 1+i] -6*src[srcStride* 1+i+2]+\
112 src[srcStride* 2+i-1] -12*src[srcStride* 2+i+1] -6*src[srcStride* 2+i] +src[srcStride* 2+i+2]+\
120 static void OPNAME ## rv30_tpel8_hvv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
123 const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
125 for(j = 0; j < h; j++){\
126 for(i = 0; i < w; i++){\
128 src[srcStride*-1+i-1] -12*src[srcStride*-1+i] -6*src[srcStride*-1+i+1] +src[srcStride*-1+i+2]+\
129 -6*src[srcStride* 0+i-1] +72*src[srcStride* 0+i] +36*src[srcStride* 0+i+1] -6*src[srcStride* 0+i+2]+\
130 -12*src[srcStride* 1+i-1] +144*src[srcStride* 1+i] +72*src[srcStride* 1+i+1] -12*src[srcStride* 1+i+2]+\
131 src[srcStride* 2+i-1] -12*src[srcStride* 2+i] -6*src[srcStride* 2+i+1] +src[srcStride* 2+i+2]+\
139 static void OPNAME ## rv30_tpel8_hhvv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
142 const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
144 for(j = 0; j < h; j++){\
145 for(i = 0; i < w; i++){\
147 36*src[i+srcStride*0] +54*src[i+1+srcStride*0] +6*src[i+2+srcStride*0]+\
148 54*src[i+srcStride*1] +81*src[i+1+srcStride*1] +9*src[i+2+srcStride*1]+\
149 6*src[i+srcStride*2] + 9*src[i+1+srcStride*2] + src[i+2+srcStride*2]+\
157 static void OPNAME ## rv30_tpel16_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, const int C1, const int C2){\
158 OPNAME ## rv30_tpel8_v_lowpass(dst , src , dstStride, srcStride, C1, C2);\
159 OPNAME ## rv30_tpel8_v_lowpass(dst+8, src+8, dstStride, srcStride, C1, 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);\
166 static void OPNAME ## rv30_tpel16_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, const int C1, const int C2){\
167 OPNAME ## rv30_tpel8_h_lowpass(dst , src , dstStride, srcStride, C1, C2);\
168 OPNAME ## rv30_tpel8_h_lowpass(dst+8, src+8, dstStride, srcStride, C1, 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);\
175 static void OPNAME ## rv30_tpel16_hv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
176 OPNAME ## rv30_tpel8_hv_lowpass(dst , src , dstStride, srcStride);\
177 OPNAME ## rv30_tpel8_hv_lowpass(dst+8, src+8, dstStride, srcStride);\
180 OPNAME ## rv30_tpel8_hv_lowpass(dst , src , dstStride, srcStride);\
181 OPNAME ## rv30_tpel8_hv_lowpass(dst+8, src+8, dstStride, srcStride);\
184 static void OPNAME ## rv30_tpel16_hhv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
185 OPNAME ## rv30_tpel8_hhv_lowpass(dst , src , dstStride, srcStride);\
186 OPNAME ## rv30_tpel8_hhv_lowpass(dst+8, src+8, dstStride, srcStride);\
189 OPNAME ## rv30_tpel8_hhv_lowpass(dst , src , dstStride, srcStride);\
190 OPNAME ## rv30_tpel8_hhv_lowpass(dst+8, src+8, dstStride, srcStride);\
193 static void OPNAME ## rv30_tpel16_hvv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
194 OPNAME ## rv30_tpel8_hvv_lowpass(dst , src , dstStride, srcStride);\
195 OPNAME ## rv30_tpel8_hvv_lowpass(dst+8, src+8, dstStride, srcStride);\
198 OPNAME ## rv30_tpel8_hvv_lowpass(dst , src , dstStride, srcStride);\
199 OPNAME ## rv30_tpel8_hvv_lowpass(dst+8, src+8, dstStride, srcStride);\
202 static void OPNAME ## rv30_tpel16_hhvv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
203 OPNAME ## rv30_tpel8_hhvv_lowpass(dst , src , dstStride, srcStride);\
204 OPNAME ## rv30_tpel8_hhvv_lowpass(dst+8, src+8, dstStride, srcStride);\
207 OPNAME ## rv30_tpel8_hhvv_lowpass(dst , src , dstStride, srcStride);\
208 OPNAME ## rv30_tpel8_hhvv_lowpass(dst+8, src+8, dstStride, srcStride);\
212 #define RV30_MC(OPNAME, SIZE) \
213 static void OPNAME ## rv30_tpel ## SIZE ## _mc10_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)\
215 OPNAME ## rv30_tpel ## SIZE ## _h_lowpass(dst, src, stride, stride, 12, 6);\
218 static void OPNAME ## rv30_tpel ## SIZE ## _mc20_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)\
220 OPNAME ## rv30_tpel ## SIZE ## _h_lowpass(dst, src, stride, stride, 6, 12);\
223 static void OPNAME ## rv30_tpel ## SIZE ## _mc01_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)\
225 OPNAME ## rv30_tpel ## SIZE ## _v_lowpass(dst, src, stride, stride, 12, 6);\
228 static void OPNAME ## rv30_tpel ## SIZE ## _mc02_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)\
230 OPNAME ## rv30_tpel ## SIZE ## _v_lowpass(dst, src, stride, stride, 6, 12);\
233 static void OPNAME ## rv30_tpel ## SIZE ## _mc11_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)\
235 OPNAME ## rv30_tpel ## SIZE ## _hv_lowpass(dst, src, stride, stride);\
238 static void OPNAME ## rv30_tpel ## SIZE ## _mc12_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)\
240 OPNAME ## rv30_tpel ## SIZE ## _hvv_lowpass(dst, src, stride, stride);\
243 static void OPNAME ## rv30_tpel ## SIZE ## _mc21_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)\
245 OPNAME ## rv30_tpel ## SIZE ## _hhv_lowpass(dst, src, stride, stride);\
248 static void OPNAME ## rv30_tpel ## SIZE ## _mc22_c(uint8_t *dst, uint8_t *src, ptrdiff_t stride)\
250 OPNAME ## rv30_tpel ## SIZE ## _hhvv_lowpass(dst, src, stride, stride);\
254 #define op_avg(a, b) a = (((a)+cm[b]+1)>>1)
255 #define op_put(a, b) a = cm[b]
257 RV30_LOWPASS(put_ , op_put)
258 RV30_LOWPASS(avg_ , op_avg)
264 av_cold void ff_rv30dsp_init(RV34DSPContext *c)
266 H264ChromaContext h264chroma;
267 H264QpelContext qpel;
270 ff_h264chroma_init(&h264chroma, 8);
271 ff_h264qpel_init(&qpel, 8);
273 c->put_pixels_tab[0][ 0] = qpel.put_h264_qpel_pixels_tab[0][0];
274 c->put_pixels_tab[0][ 1] = put_rv30_tpel16_mc10_c;
275 c->put_pixels_tab[0][ 2] = put_rv30_tpel16_mc20_c;
276 c->put_pixels_tab[0][ 4] = put_rv30_tpel16_mc01_c;
277 c->put_pixels_tab[0][ 5] = put_rv30_tpel16_mc11_c;
278 c->put_pixels_tab[0][ 6] = put_rv30_tpel16_mc21_c;
279 c->put_pixels_tab[0][ 8] = put_rv30_tpel16_mc02_c;
280 c->put_pixels_tab[0][ 9] = put_rv30_tpel16_mc12_c;
281 c->put_pixels_tab[0][10] = put_rv30_tpel16_mc22_c;
282 c->avg_pixels_tab[0][ 0] = qpel.avg_h264_qpel_pixels_tab[0][0];
283 c->avg_pixels_tab[0][ 1] = avg_rv30_tpel16_mc10_c;
284 c->avg_pixels_tab[0][ 2] = avg_rv30_tpel16_mc20_c;
285 c->avg_pixels_tab[0][ 4] = avg_rv30_tpel16_mc01_c;
286 c->avg_pixels_tab[0][ 5] = avg_rv30_tpel16_mc11_c;
287 c->avg_pixels_tab[0][ 6] = avg_rv30_tpel16_mc21_c;
288 c->avg_pixels_tab[0][ 8] = avg_rv30_tpel16_mc02_c;
289 c->avg_pixels_tab[0][ 9] = avg_rv30_tpel16_mc12_c;
290 c->avg_pixels_tab[0][10] = avg_rv30_tpel16_mc22_c;
291 c->put_pixels_tab[1][ 0] = qpel.put_h264_qpel_pixels_tab[1][0];
292 c->put_pixels_tab[1][ 1] = put_rv30_tpel8_mc10_c;
293 c->put_pixels_tab[1][ 2] = put_rv30_tpel8_mc20_c;
294 c->put_pixels_tab[1][ 4] = put_rv30_tpel8_mc01_c;
295 c->put_pixels_tab[1][ 5] = put_rv30_tpel8_mc11_c;
296 c->put_pixels_tab[1][ 6] = put_rv30_tpel8_mc21_c;
297 c->put_pixels_tab[1][ 8] = put_rv30_tpel8_mc02_c;
298 c->put_pixels_tab[1][ 9] = put_rv30_tpel8_mc12_c;
299 c->put_pixels_tab[1][10] = put_rv30_tpel8_mc22_c;
300 c->avg_pixels_tab[1][ 0] = qpel.avg_h264_qpel_pixels_tab[1][0];
301 c->avg_pixels_tab[1][ 1] = avg_rv30_tpel8_mc10_c;
302 c->avg_pixels_tab[1][ 2] = avg_rv30_tpel8_mc20_c;
303 c->avg_pixels_tab[1][ 4] = avg_rv30_tpel8_mc01_c;
304 c->avg_pixels_tab[1][ 5] = avg_rv30_tpel8_mc11_c;
305 c->avg_pixels_tab[1][ 6] = avg_rv30_tpel8_mc21_c;
306 c->avg_pixels_tab[1][ 8] = avg_rv30_tpel8_mc02_c;
307 c->avg_pixels_tab[1][ 9] = avg_rv30_tpel8_mc12_c;
308 c->avg_pixels_tab[1][10] = avg_rv30_tpel8_mc22_c;
310 c->put_chroma_pixels_tab[0] = h264chroma.put_h264_chroma_pixels_tab[0];
311 c->put_chroma_pixels_tab[1] = h264chroma.put_h264_chroma_pixels_tab[1];
312 c->avg_chroma_pixels_tab[0] = h264chroma.avg_h264_chroma_pixels_tab[0];
313 c->avg_chroma_pixels_tab[1] = h264chroma.avg_h264_chroma_pixels_tab[1];