1/*
2 * RV30 decoder motion compensation functions
3 * Copyright (c) 2007 Konstantin Shishkov
4 *
5 * This file is part of Libav.
6 *
7 * Libav 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 * Libav 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 Libav; 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 "avcodec.h"
28#include "dsputil.h"
29#include "rv34dsp.h"
30
31#define RV30_LOWPASS(OPNAME, OP) \
32static av_unused void OPNAME ## rv30_tpel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, const int C1, const int C2){\
33    const int h = 8;\
34    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
35    int i;\
36    for(i = 0; i < h; i++)\
37    {\
38        OP(dst[0], (-(src[-1]+src[2]) + src[0]*C1 + src[1]*C2 + 8)>>4);\
39        OP(dst[1], (-(src[ 0]+src[3]) + src[1]*C1 + src[2]*C2 + 8)>>4);\
40        OP(dst[2], (-(src[ 1]+src[4]) + src[2]*C1 + src[3]*C2 + 8)>>4);\
41        OP(dst[3], (-(src[ 2]+src[5]) + src[3]*C1 + src[4]*C2 + 8)>>4);\
42        OP(dst[4], (-(src[ 3]+src[6]) + src[4]*C1 + src[5]*C2 + 8)>>4);\
43        OP(dst[5], (-(src[ 4]+src[7]) + src[5]*C1 + src[6]*C2 + 8)>>4);\
44        OP(dst[6], (-(src[ 5]+src[8]) + src[6]*C1 + src[7]*C2 + 8)>>4);\
45        OP(dst[7], (-(src[ 6]+src[9]) + src[7]*C1 + src[8]*C2 + 8)>>4);\
46        dst += dstStride;\
47        src += srcStride;\
48    }\
49}\
50\
51static void OPNAME ## rv30_tpel8_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, const int C1, const int C2){\
52    const int w = 8;\
53    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
54    int i;\
55    for(i = 0; i < w; i++)\
56    {\
57        const int srcA = src[-1*srcStride];\
58        const int src0 = src[0 *srcStride];\
59        const int src1 = src[1 *srcStride];\
60        const int src2 = src[2 *srcStride];\
61        const int src3 = src[3 *srcStride];\
62        const int src4 = src[4 *srcStride];\
63        const int src5 = src[5 *srcStride];\
64        const int src6 = src[6 *srcStride];\
65        const int src7 = src[7 *srcStride];\
66        const int src8 = src[8 *srcStride];\
67        const int src9 = src[9 *srcStride];\
68        OP(dst[0*dstStride], (-(srcA+src2) + src0*C1 + src1*C2 + 8)>>4);\
69        OP(dst[1*dstStride], (-(src0+src3) + src1*C1 + src2*C2 + 8)>>4);\
70        OP(dst[2*dstStride], (-(src1+src4) + src2*C1 + src3*C2 + 8)>>4);\
71        OP(dst[3*dstStride], (-(src2+src5) + src3*C1 + src4*C2 + 8)>>4);\
72        OP(dst[4*dstStride], (-(src3+src6) + src4*C1 + src5*C2 + 8)>>4);\
73        OP(dst[5*dstStride], (-(src4+src7) + src5*C1 + src6*C2 + 8)>>4);\
74        OP(dst[6*dstStride], (-(src5+src8) + src6*C1 + src7*C2 + 8)>>4);\
75        OP(dst[7*dstStride], (-(src6+src9) + src7*C1 + src8*C2 + 8)>>4);\
76        dst++;\
77        src++;\
78    }\
79}\
80\
81static void OPNAME ## rv30_tpel8_hv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
82    const int w = 8;\
83    const int h = 8;\
84    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
85    int i, j;\
86    for(j = 0; j < h; j++){\
87        for(i = 0; i < w; i++){\
88            OP(dst[i], (\
89                  src[srcStride*-1+i-1]  -12*src[srcStride*-1+i]  -6*src[srcStride*-1+i+1]    +src[srcStride*-1+i+2]+\
90              -12*src[srcStride* 0+i-1] +144*src[srcStride* 0+i] +72*src[srcStride* 0+i+1] -12*src[srcStride* 0+i+2] +\
91               -6*src[srcStride* 1+i-1]  +72*src[srcStride* 1+i] +36*src[srcStride* 1+i+1]  -6*src[srcStride* 1+i+2] +\
92                  src[srcStride* 2+i-1]  -12*src[srcStride* 2+i]  -6*src[srcStride* 2+i+1]    +src[srcStride* 2+i+2] +\
93                  128)>>8);\
94        }\
95        src += srcStride;\
96        dst += dstStride;\
97    }\
98}\
99\
100static void OPNAME ## rv30_tpel8_hhv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
101    const int w = 8;\
102    const int h = 8;\
103    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
104    int i, j;\
105    for(j = 0; j < h; j++){\
106        for(i = 0; i < w; i++){\
107            OP(dst[i], (\
108                  src[srcStride*-1+i-1]  -12*src[srcStride*-1+i+1]  -6*src[srcStride*-1+i]    +src[srcStride*-1+i+2]+\
109              -12*src[srcStride* 0+i-1] +144*src[srcStride* 0+i+1] +72*src[srcStride* 0+i] -12*src[srcStride* 0+i+2]+\
110               -6*src[srcStride* 1+i-1]  +72*src[srcStride* 1+i+1] +36*src[srcStride* 1+i]  -6*src[srcStride* 1+i+2]+\
111                  src[srcStride* 2+i-1]  -12*src[srcStride* 2+i+1]  -6*src[srcStride* 2+i]    +src[srcStride* 2+i+2]+\
112                  128)>>8);\
113        }\
114        src += srcStride;\
115        dst += dstStride;\
116    }\
117}\
118\
119static void OPNAME ## rv30_tpel8_hvv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
120    const int w = 8;\
121    const int h = 8;\
122    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
123    int i, j;\
124    for(j = 0; j < h; j++){\
125        for(i = 0; i < w; i++){\
126            OP(dst[i], (\
127                  src[srcStride*-1+i-1]  -12*src[srcStride*-1+i]  -6*src[srcStride*-1+i+1]    +src[srcStride*-1+i+2]+\
128               -6*src[srcStride* 0+i-1]  +72*src[srcStride* 0+i] +36*src[srcStride* 0+i+1]  -6*src[srcStride* 0+i+2]+\
129              -12*src[srcStride* 1+i-1] +144*src[srcStride* 1+i] +72*src[srcStride* 1+i+1] -12*src[srcStride* 1+i+2]+\
130                  src[srcStride* 2+i-1]  -12*src[srcStride* 2+i]  -6*src[srcStride* 2+i+1]    +src[srcStride* 2+i+2]+\
131                  128)>>8);\
132        }\
133        src += srcStride;\
134        dst += dstStride;\
135    }\
136}\
137\
138static void OPNAME ## rv30_tpel8_hhvv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
139    const int w = 8;\
140    const int h = 8;\
141    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
142    int i, j;\
143    for(j = 0; j < h; j++){\
144        for(i = 0; i < w; i++){\
145            OP(dst[i], (\
146               36*src[i+srcStride*0] +54*src[i+1+srcStride*0] +6*src[i+2+srcStride*0]+\
147               54*src[i+srcStride*1] +81*src[i+1+srcStride*1] +9*src[i+2+srcStride*1]+\
148                6*src[i+srcStride*2] + 9*src[i+1+srcStride*2] +  src[i+2+srcStride*2]+\
149               128)>>8);\
150        }\
151        src += srcStride;\
152        dst += dstStride;\
153    }\
154}\
155\
156static void OPNAME ## rv30_tpel16_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, const int C1, const int C2){\
157    OPNAME ## rv30_tpel8_v_lowpass(dst  , src  , dstStride, srcStride, C1, C2);\
158    OPNAME ## rv30_tpel8_v_lowpass(dst+8, src+8, dstStride, srcStride, C1, C2);\
159    src += 8*srcStride;\
160    dst += 8*dstStride;\
161    OPNAME ## rv30_tpel8_v_lowpass(dst  , src  , dstStride, srcStride, C1, C2);\
162    OPNAME ## rv30_tpel8_v_lowpass(dst+8, src+8, dstStride, srcStride, C1, C2);\
163}\
164\
165static void OPNAME ## rv30_tpel16_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, const int C1, const int C2){\
166    OPNAME ## rv30_tpel8_h_lowpass(dst  , src  , dstStride, srcStride, C1, C2);\
167    OPNAME ## rv30_tpel8_h_lowpass(dst+8, src+8, dstStride, srcStride, C1, C2);\
168    src += 8*srcStride;\
169    dst += 8*dstStride;\
170    OPNAME ## rv30_tpel8_h_lowpass(dst  , src  , dstStride, srcStride, C1, C2);\
171    OPNAME ## rv30_tpel8_h_lowpass(dst+8, src+8, dstStride, srcStride, C1, C2);\
172}\
173\
174static void OPNAME ## rv30_tpel16_hv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
175    OPNAME ## rv30_tpel8_hv_lowpass(dst  , src  , dstStride, srcStride);\
176    OPNAME ## rv30_tpel8_hv_lowpass(dst+8, src+8, dstStride, srcStride);\
177    src += 8*srcStride;\
178    dst += 8*dstStride;\
179    OPNAME ## rv30_tpel8_hv_lowpass(dst  , src  , dstStride, srcStride);\
180    OPNAME ## rv30_tpel8_hv_lowpass(dst+8, src+8, dstStride, srcStride);\
181}\
182\
183static void OPNAME ## rv30_tpel16_hhv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
184    OPNAME ## rv30_tpel8_hhv_lowpass(dst  , src  , dstStride, srcStride);\
185    OPNAME ## rv30_tpel8_hhv_lowpass(dst+8, src+8, dstStride, srcStride);\
186    src += 8*srcStride;\
187    dst += 8*dstStride;\
188    OPNAME ## rv30_tpel8_hhv_lowpass(dst  , src  , dstStride, srcStride);\
189    OPNAME ## rv30_tpel8_hhv_lowpass(dst+8, src+8, dstStride, srcStride);\
190}\
191\
192static void OPNAME ## rv30_tpel16_hvv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
193    OPNAME ## rv30_tpel8_hvv_lowpass(dst  , src  , dstStride, srcStride);\
194    OPNAME ## rv30_tpel8_hvv_lowpass(dst+8, src+8, dstStride, srcStride);\
195    src += 8*srcStride;\
196    dst += 8*dstStride;\
197    OPNAME ## rv30_tpel8_hvv_lowpass(dst  , src  , dstStride, srcStride);\
198    OPNAME ## rv30_tpel8_hvv_lowpass(dst+8, src+8, dstStride, srcStride);\
199}\
200\
201static void OPNAME ## rv30_tpel16_hhvv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
202    OPNAME ## rv30_tpel8_hhvv_lowpass(dst  , src  , dstStride, srcStride);\
203    OPNAME ## rv30_tpel8_hhvv_lowpass(dst+8, src+8, dstStride, srcStride);\
204    src += 8*srcStride;\
205    dst += 8*dstStride;\
206    OPNAME ## rv30_tpel8_hhvv_lowpass(dst  , src  , dstStride, srcStride);\
207    OPNAME ## rv30_tpel8_hhvv_lowpass(dst+8, src+8, dstStride, srcStride);\
208}\
209\
210
211#define RV30_MC(OPNAME, SIZE) \
212static void OPNAME ## rv30_tpel ## SIZE ## _mc10_c(uint8_t *dst, uint8_t *src, int stride){\
213    OPNAME ## rv30_tpel ## SIZE ## _h_lowpass(dst, src, stride, stride, 12, 6);\
214}\
215\
216static void OPNAME ## rv30_tpel ## SIZE ## _mc20_c(uint8_t *dst, uint8_t *src, int stride){\
217    OPNAME ## rv30_tpel ## SIZE ## _h_lowpass(dst, src, stride, stride, 6, 12);\
218}\
219\
220static void OPNAME ## rv30_tpel ## SIZE ## _mc01_c(uint8_t *dst, uint8_t *src, int stride){\
221    OPNAME ## rv30_tpel ## SIZE ## _v_lowpass(dst, src, stride, stride, 12, 6);\
222}\
223\
224static void OPNAME ## rv30_tpel ## SIZE ## _mc02_c(uint8_t *dst, uint8_t *src, int stride){\
225    OPNAME ## rv30_tpel ## SIZE ## _v_lowpass(dst, src, stride, stride, 6, 12);\
226}\
227\
228static void OPNAME ## rv30_tpel ## SIZE ## _mc11_c(uint8_t *dst, uint8_t *src, int stride){\
229    OPNAME ## rv30_tpel ## SIZE ## _hv_lowpass(dst, src, stride, stride);\
230}\
231\
232static void OPNAME ## rv30_tpel ## SIZE ## _mc12_c(uint8_t *dst, uint8_t *src, int stride){\
233    OPNAME ## rv30_tpel ## SIZE ## _hvv_lowpass(dst, src, stride, stride);\
234}\
235\
236static void OPNAME ## rv30_tpel ## SIZE ## _mc21_c(uint8_t *dst, uint8_t *src, int stride){\
237    OPNAME ## rv30_tpel ## SIZE ## _hhv_lowpass(dst, src, stride, stride);\
238}\
239\
240static void OPNAME ## rv30_tpel ## SIZE ## _mc22_c(uint8_t *dst, uint8_t *src, int stride){\
241    OPNAME ## rv30_tpel ## SIZE ## _hhvv_lowpass(dst, src, stride, stride);\
242}\
243\
244
245#define op_avg(a, b)  a = (((a)+cm[b]+1)>>1)
246#define op_put(a, b)  a = cm[b]
247
248RV30_LOWPASS(put_       , op_put)
249RV30_LOWPASS(avg_       , op_avg)
250RV30_MC(put_, 8)
251RV30_MC(put_, 16)
252RV30_MC(avg_, 8)
253RV30_MC(avg_, 16)
254
255av_cold void ff_rv30dsp_init(RV34DSPContext *c, DSPContext* dsp) {
256
257    ff_rv34dsp_init(c, dsp);
258
259    c->put_pixels_tab[0][ 0] = dsp->put_h264_qpel_pixels_tab[0][0];
260    c->put_pixels_tab[0][ 1] = put_rv30_tpel16_mc10_c;
261    c->put_pixels_tab[0][ 2] = put_rv30_tpel16_mc20_c;
262    c->put_pixels_tab[0][ 4] = put_rv30_tpel16_mc01_c;
263    c->put_pixels_tab[0][ 5] = put_rv30_tpel16_mc11_c;
264    c->put_pixels_tab[0][ 6] = put_rv30_tpel16_mc21_c;
265    c->put_pixels_tab[0][ 8] = put_rv30_tpel16_mc02_c;
266    c->put_pixels_tab[0][ 9] = put_rv30_tpel16_mc12_c;
267    c->put_pixels_tab[0][10] = put_rv30_tpel16_mc22_c;
268    c->avg_pixels_tab[0][ 0] = dsp->avg_h264_qpel_pixels_tab[0][0];
269    c->avg_pixels_tab[0][ 1] = avg_rv30_tpel16_mc10_c;
270    c->avg_pixels_tab[0][ 2] = avg_rv30_tpel16_mc20_c;
271    c->avg_pixels_tab[0][ 4] = avg_rv30_tpel16_mc01_c;
272    c->avg_pixels_tab[0][ 5] = avg_rv30_tpel16_mc11_c;
273    c->avg_pixels_tab[0][ 6] = avg_rv30_tpel16_mc21_c;
274    c->avg_pixels_tab[0][ 8] = avg_rv30_tpel16_mc02_c;
275    c->avg_pixels_tab[0][ 9] = avg_rv30_tpel16_mc12_c;
276    c->avg_pixels_tab[0][10] = avg_rv30_tpel16_mc22_c;
277    c->put_pixels_tab[1][ 0] = dsp->put_h264_qpel_pixels_tab[1][0];
278    c->put_pixels_tab[1][ 1] = put_rv30_tpel8_mc10_c;
279    c->put_pixels_tab[1][ 2] = put_rv30_tpel8_mc20_c;
280    c->put_pixels_tab[1][ 4] = put_rv30_tpel8_mc01_c;
281    c->put_pixels_tab[1][ 5] = put_rv30_tpel8_mc11_c;
282    c->put_pixels_tab[1][ 6] = put_rv30_tpel8_mc21_c;
283    c->put_pixels_tab[1][ 8] = put_rv30_tpel8_mc02_c;
284    c->put_pixels_tab[1][ 9] = put_rv30_tpel8_mc12_c;
285    c->put_pixels_tab[1][10] = put_rv30_tpel8_mc22_c;
286    c->avg_pixels_tab[1][ 0] = dsp->avg_h264_qpel_pixels_tab[1][0];
287    c->avg_pixels_tab[1][ 1] = avg_rv30_tpel8_mc10_c;
288    c->avg_pixels_tab[1][ 2] = avg_rv30_tpel8_mc20_c;
289    c->avg_pixels_tab[1][ 4] = avg_rv30_tpel8_mc01_c;
290    c->avg_pixels_tab[1][ 5] = avg_rv30_tpel8_mc11_c;
291    c->avg_pixels_tab[1][ 6] = avg_rv30_tpel8_mc21_c;
292    c->avg_pixels_tab[1][ 8] = avg_rv30_tpel8_mc02_c;
293    c->avg_pixels_tab[1][ 9] = avg_rv30_tpel8_mc12_c;
294    c->avg_pixels_tab[1][10] = avg_rv30_tpel8_mc22_c;
295
296    c->put_chroma_pixels_tab[0] = dsp->put_h264_chroma_pixels_tab[0];
297    c->put_chroma_pixels_tab[1] = dsp->put_h264_chroma_pixels_tab[1];
298    c->avg_chroma_pixels_tab[0] = dsp->avg_h264_chroma_pixels_tab[0];
299    c->avg_chroma_pixels_tab[1] = dsp->avg_h264_chroma_pixels_tab[1];
300}
301