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 "avcodec.h"
28#include "dsputil.h"
29
30#define RV30_LOWPASS(OPNAME, OP) \
31static av_unused void OPNAME ## rv30_tpel8_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, const int C1, const int C2){\
32    const int h=8;\
33    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
34    int i;\
35    for(i=0; i<h; i++)\
36    {\
37        OP(dst[0], (-(src[-1]+src[2]) + src[0]*C1 + src[1]*C2 + 8)>>4);\
38        OP(dst[1], (-(src[ 0]+src[3]) + src[1]*C1 + src[2]*C2 + 8)>>4);\
39        OP(dst[2], (-(src[ 1]+src[4]) + src[2]*C1 + src[3]*C2 + 8)>>4);\
40        OP(dst[3], (-(src[ 2]+src[5]) + src[3]*C1 + src[4]*C2 + 8)>>4);\
41        OP(dst[4], (-(src[ 3]+src[6]) + src[4]*C1 + src[5]*C2 + 8)>>4);\
42        OP(dst[5], (-(src[ 4]+src[7]) + src[5]*C1 + src[6]*C2 + 8)>>4);\
43        OP(dst[6], (-(src[ 5]+src[8]) + src[6]*C1 + src[7]*C2 + 8)>>4);\
44        OP(dst[7], (-(src[ 6]+src[9]) + src[7]*C1 + src[8]*C2 + 8)>>4);\
45        dst+=dstStride;\
46        src+=srcStride;\
47    }\
48}\
49\
50static void OPNAME ## rv30_tpel8_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, const int C1, const int C2){\
51    const int w=8;\
52    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
53    int i;\
54    for(i=0; i<w; i++)\
55    {\
56        const int srcA= src[-1*srcStride];\
57        const int src0= src[0 *srcStride];\
58        const int src1= src[1 *srcStride];\
59        const int src2= src[2 *srcStride];\
60        const int src3= src[3 *srcStride];\
61        const int src4= src[4 *srcStride];\
62        const int src5= src[5 *srcStride];\
63        const int src6= src[6 *srcStride];\
64        const int src7= src[7 *srcStride];\
65        const int src8= src[8 *srcStride];\
66        const int src9= src[9 *srcStride];\
67        OP(dst[0*dstStride], (-(srcA+src2) + src0*C1 + src1*C2 + 8)>>4);\
68        OP(dst[1*dstStride], (-(src0+src3) + src1*C1 + src2*C2 + 8)>>4);\
69        OP(dst[2*dstStride], (-(src1+src4) + src2*C1 + src3*C2 + 8)>>4);\
70        OP(dst[3*dstStride], (-(src2+src5) + src3*C1 + src4*C2 + 8)>>4);\
71        OP(dst[4*dstStride], (-(src3+src6) + src4*C1 + src5*C2 + 8)>>4);\
72        OP(dst[5*dstStride], (-(src4+src7) + src5*C1 + src6*C2 + 8)>>4);\
73        OP(dst[6*dstStride], (-(src5+src8) + src6*C1 + src7*C2 + 8)>>4);\
74        OP(dst[7*dstStride], (-(src6+src9) + src7*C1 + src8*C2 + 8)>>4);\
75        dst++;\
76        src++;\
77    }\
78}\
79\
80static void OPNAME ## rv30_tpel8_hv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
81    const int w = 8;\
82    const int h = 8;\
83    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
84    int i, j;\
85    for(j = 0; j < h; j++){\
86        for(i = 0; i < w; i++){\
87            OP(dst[i], (\
88                  src[srcStride*-1+i-1]  -12*src[srcStride*-1+i]  -6*src[srcStride*-1+i+1]    +src[srcStride*-1+i+2]+\
89              -12*src[srcStride* 0+i-1] +144*src[srcStride* 0+i] +72*src[srcStride* 0+i+1] -12*src[srcStride* 0+i+2] +\
90               -6*src[srcStride* 1+i-1]  +72*src[srcStride* 1+i] +36*src[srcStride* 1+i+1]  -6*src[srcStride* 1+i+2] +\
91                  src[srcStride* 2+i-1]  -12*src[srcStride* 2+i]  -6*src[srcStride* 2+i+1]    +src[srcStride* 2+i+2] +\
92                  128)>>8);\
93        }\
94        src += srcStride;\
95        dst += dstStride;\
96    }\
97}\
98\
99static void OPNAME ## rv30_tpel8_hhv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
100    const int w = 8;\
101    const int h = 8;\
102    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
103    int i, j;\
104    for(j = 0; j < h; j++){\
105        for(i = 0; i < w; i++){\
106            OP(dst[i], (\
107                  src[srcStride*-1+i-1]  -12*src[srcStride*-1+i+1]  -6*src[srcStride*-1+i]    +src[srcStride*-1+i+2]+\
108              -12*src[srcStride* 0+i-1] +144*src[srcStride* 0+i+1] +72*src[srcStride* 0+i] -12*src[srcStride* 0+i+2]+\
109               -6*src[srcStride* 1+i-1]  +72*src[srcStride* 1+i+1] +36*src[srcStride* 1+i]  -6*src[srcStride* 1+i+2]+\
110                  src[srcStride* 2+i-1]  -12*src[srcStride* 2+i+1]  -6*src[srcStride* 2+i]    +src[srcStride* 2+i+2]+\
111                  128)>>8);\
112        }\
113        src += srcStride;\
114        dst += dstStride;\
115    }\
116}\
117\
118static void OPNAME ## rv30_tpel8_hvv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
119    const int w = 8;\
120    const int h = 8;\
121    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
122    int i, j;\
123    for(j = 0; j < h; j++){\
124        for(i = 0; i < w; i++){\
125            OP(dst[i], (\
126                  src[srcStride*-1+i-1]  -12*src[srcStride*-1+i]  -6*src[srcStride*-1+i+1]    +src[srcStride*-1+i+2]+\
127               -6*src[srcStride* 0+i-1]  +72*src[srcStride* 0+i] +36*src[srcStride* 0+i+1]  -6*src[srcStride* 0+i+2]+\
128              -12*src[srcStride* 1+i-1] +144*src[srcStride* 1+i] +72*src[srcStride* 1+i+1] -12*src[srcStride* 1+i+2]+\
129                  src[srcStride* 2+i-1]  -12*src[srcStride* 2+i]  -6*src[srcStride* 2+i+1]    +src[srcStride* 2+i+2]+\
130                  128)>>8);\
131        }\
132        src += srcStride;\
133        dst += dstStride;\
134    }\
135}\
136\
137static void OPNAME ## rv30_tpel8_hhvv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
138    const int w = 8;\
139    const int h = 8;\
140    uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
141    int i, j;\
142    for(j = 0; j < h; j++){\
143        for(i = 0; i < w; i++){\
144            OP(dst[i], (\
145               36*src[i+srcStride*0] +54*src[i+1+srcStride*0] +6*src[i+2+srcStride*0]+\
146               54*src[i+srcStride*1] +81*src[i+1+srcStride*1] +9*src[i+2+srcStride*1]+\
147                6*src[i+srcStride*2] + 9*src[i+1+srcStride*2] +  src[i+2+srcStride*2]+\
148               128)>>8);\
149        }\
150        src += srcStride;\
151        dst += dstStride;\
152    }\
153}\
154\
155static void OPNAME ## rv30_tpel16_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, const int C1, const int C2){\
156    OPNAME ## rv30_tpel8_v_lowpass(dst  , src  , dstStride, srcStride, C1, C2);\
157    OPNAME ## rv30_tpel8_v_lowpass(dst+8, src+8, dstStride, srcStride, C1, C2);\
158    src += 8*srcStride;\
159    dst += 8*dstStride;\
160    OPNAME ## rv30_tpel8_v_lowpass(dst  , src  , dstStride, srcStride, C1, C2);\
161    OPNAME ## rv30_tpel8_v_lowpass(dst+8, src+8, dstStride, srcStride, C1, C2);\
162}\
163\
164static void OPNAME ## rv30_tpel16_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, const int C1, const int C2){\
165    OPNAME ## rv30_tpel8_h_lowpass(dst  , src  , dstStride, srcStride, C1, C2);\
166    OPNAME ## rv30_tpel8_h_lowpass(dst+8, src+8, dstStride, srcStride, C1, C2);\
167    src += 8*srcStride;\
168    dst += 8*dstStride;\
169    OPNAME ## rv30_tpel8_h_lowpass(dst  , src  , dstStride, srcStride, C1, C2);\
170    OPNAME ## rv30_tpel8_h_lowpass(dst+8, src+8, dstStride, srcStride, C1, C2);\
171}\
172\
173static void OPNAME ## rv30_tpel16_hv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
174    OPNAME ## rv30_tpel8_hv_lowpass(dst  , src  , dstStride, srcStride);\
175    OPNAME ## rv30_tpel8_hv_lowpass(dst+8, src+8, dstStride, srcStride);\
176    src += 8*srcStride;\
177    dst += 8*dstStride;\
178    OPNAME ## rv30_tpel8_hv_lowpass(dst  , src  , dstStride, srcStride);\
179    OPNAME ## rv30_tpel8_hv_lowpass(dst+8, src+8, dstStride, srcStride);\
180}\
181\
182static void OPNAME ## rv30_tpel16_hhv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
183    OPNAME ## rv30_tpel8_hhv_lowpass(dst  , src  , dstStride, srcStride);\
184    OPNAME ## rv30_tpel8_hhv_lowpass(dst+8, src+8, dstStride, srcStride);\
185    src += 8*srcStride;\
186    dst += 8*dstStride;\
187    OPNAME ## rv30_tpel8_hhv_lowpass(dst  , src  , dstStride, srcStride);\
188    OPNAME ## rv30_tpel8_hhv_lowpass(dst+8, src+8, dstStride, srcStride);\
189}\
190\
191static void OPNAME ## rv30_tpel16_hvv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
192    OPNAME ## rv30_tpel8_hvv_lowpass(dst  , src  , dstStride, srcStride);\
193    OPNAME ## rv30_tpel8_hvv_lowpass(dst+8, src+8, dstStride, srcStride);\
194    src += 8*srcStride;\
195    dst += 8*dstStride;\
196    OPNAME ## rv30_tpel8_hvv_lowpass(dst  , src  , dstStride, srcStride);\
197    OPNAME ## rv30_tpel8_hvv_lowpass(dst+8, src+8, dstStride, srcStride);\
198}\
199\
200static void OPNAME ## rv30_tpel16_hhvv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
201    OPNAME ## rv30_tpel8_hhvv_lowpass(dst  , src  , dstStride, srcStride);\
202    OPNAME ## rv30_tpel8_hhvv_lowpass(dst+8, src+8, dstStride, srcStride);\
203    src += 8*srcStride;\
204    dst += 8*dstStride;\
205    OPNAME ## rv30_tpel8_hhvv_lowpass(dst  , src  , dstStride, srcStride);\
206    OPNAME ## rv30_tpel8_hhvv_lowpass(dst+8, src+8, dstStride, srcStride);\
207}\
208\
209
210#define RV30_MC(OPNAME, SIZE) \
211static void OPNAME ## rv30_tpel ## SIZE ## _mc10_c(uint8_t *dst, uint8_t *src, int stride){\
212    OPNAME ## rv30_tpel ## SIZE ## _h_lowpass(dst, src, stride, stride, 12, 6);\
213}\
214\
215static void OPNAME ## rv30_tpel ## SIZE ## _mc20_c(uint8_t *dst, uint8_t *src, int stride){\
216    OPNAME ## rv30_tpel ## SIZE ## _h_lowpass(dst, src, stride, stride, 6, 12);\
217}\
218\
219static void OPNAME ## rv30_tpel ## SIZE ## _mc01_c(uint8_t *dst, uint8_t *src, int stride){\
220    OPNAME ## rv30_tpel ## SIZE ## _v_lowpass(dst, src, stride, stride, 12, 6);\
221}\
222\
223static void OPNAME ## rv30_tpel ## SIZE ## _mc02_c(uint8_t *dst, uint8_t *src, int stride){\
224    OPNAME ## rv30_tpel ## SIZE ## _v_lowpass(dst, src, stride, stride, 6, 12);\
225}\
226\
227static void OPNAME ## rv30_tpel ## SIZE ## _mc11_c(uint8_t *dst, uint8_t *src, int stride){\
228    OPNAME ## rv30_tpel ## SIZE ## _hv_lowpass(dst, src, stride, stride);\
229}\
230\
231static void OPNAME ## rv30_tpel ## SIZE ## _mc12_c(uint8_t *dst, uint8_t *src, int stride){\
232    OPNAME ## rv30_tpel ## SIZE ## _hvv_lowpass(dst, src, stride, stride);\
233}\
234\
235static void OPNAME ## rv30_tpel ## SIZE ## _mc21_c(uint8_t *dst, uint8_t *src, int stride){\
236    OPNAME ## rv30_tpel ## SIZE ## _hhv_lowpass(dst, src, stride, stride);\
237}\
238\
239static void OPNAME ## rv30_tpel ## SIZE ## _mc22_c(uint8_t *dst, uint8_t *src, int stride){\
240    OPNAME ## rv30_tpel ## SIZE ## _hhvv_lowpass(dst, src, stride, stride);\
241}\
242\
243
244#define op_avg(a, b)  a = (((a)+cm[b]+1)>>1)
245#define op_put(a, b)  a = cm[b]
246
247RV30_LOWPASS(put_       , op_put)
248RV30_LOWPASS(avg_       , op_avg)
249RV30_MC(put_, 8)
250RV30_MC(put_, 16)
251RV30_MC(avg_, 8)
252RV30_MC(avg_, 16)
253
254av_cold void ff_rv30dsp_init(DSPContext* c, AVCodecContext *avctx) {
255    c->put_rv30_tpel_pixels_tab[0][ 0] = c->put_h264_qpel_pixels_tab[0][0];
256    c->put_rv30_tpel_pixels_tab[0][ 1] = put_rv30_tpel16_mc10_c;
257    c->put_rv30_tpel_pixels_tab[0][ 2] = put_rv30_tpel16_mc20_c;
258    c->put_rv30_tpel_pixels_tab[0][ 4] = put_rv30_tpel16_mc01_c;
259    c->put_rv30_tpel_pixels_tab[0][ 5] = put_rv30_tpel16_mc11_c;
260    c->put_rv30_tpel_pixels_tab[0][ 6] = put_rv30_tpel16_mc21_c;
261    c->put_rv30_tpel_pixels_tab[0][ 8] = put_rv30_tpel16_mc02_c;
262    c->put_rv30_tpel_pixels_tab[0][ 9] = put_rv30_tpel16_mc12_c;
263    c->put_rv30_tpel_pixels_tab[0][10] = put_rv30_tpel16_mc22_c;
264    c->avg_rv30_tpel_pixels_tab[0][ 0] = c->avg_h264_qpel_pixels_tab[0][0];
265    c->avg_rv30_tpel_pixels_tab[0][ 1] = avg_rv30_tpel16_mc10_c;
266    c->avg_rv30_tpel_pixels_tab[0][ 2] = avg_rv30_tpel16_mc20_c;
267    c->avg_rv30_tpel_pixels_tab[0][ 4] = avg_rv30_tpel16_mc01_c;
268    c->avg_rv30_tpel_pixels_tab[0][ 5] = avg_rv30_tpel16_mc11_c;
269    c->avg_rv30_tpel_pixels_tab[0][ 6] = avg_rv30_tpel16_mc21_c;
270    c->avg_rv30_tpel_pixels_tab[0][ 8] = avg_rv30_tpel16_mc02_c;
271    c->avg_rv30_tpel_pixels_tab[0][ 9] = avg_rv30_tpel16_mc12_c;
272    c->avg_rv30_tpel_pixels_tab[0][10] = avg_rv30_tpel16_mc22_c;
273    c->put_rv30_tpel_pixels_tab[1][ 0] = c->put_h264_qpel_pixels_tab[1][0];
274    c->put_rv30_tpel_pixels_tab[1][ 1] = put_rv30_tpel8_mc10_c;
275    c->put_rv30_tpel_pixels_tab[1][ 2] = put_rv30_tpel8_mc20_c;
276    c->put_rv30_tpel_pixels_tab[1][ 4] = put_rv30_tpel8_mc01_c;
277    c->put_rv30_tpel_pixels_tab[1][ 5] = put_rv30_tpel8_mc11_c;
278    c->put_rv30_tpel_pixels_tab[1][ 6] = put_rv30_tpel8_mc21_c;
279    c->put_rv30_tpel_pixels_tab[1][ 8] = put_rv30_tpel8_mc02_c;
280    c->put_rv30_tpel_pixels_tab[1][ 9] = put_rv30_tpel8_mc12_c;
281    c->put_rv30_tpel_pixels_tab[1][10] = put_rv30_tpel8_mc22_c;
282    c->avg_rv30_tpel_pixels_tab[1][ 0] = c->avg_h264_qpel_pixels_tab[1][0];
283    c->avg_rv30_tpel_pixels_tab[1][ 1] = avg_rv30_tpel8_mc10_c;
284    c->avg_rv30_tpel_pixels_tab[1][ 2] = avg_rv30_tpel8_mc20_c;
285    c->avg_rv30_tpel_pixels_tab[1][ 4] = avg_rv30_tpel8_mc01_c;
286    c->avg_rv30_tpel_pixels_tab[1][ 5] = avg_rv30_tpel8_mc11_c;
287    c->avg_rv30_tpel_pixels_tab[1][ 6] = avg_rv30_tpel8_mc21_c;
288    c->avg_rv30_tpel_pixels_tab[1][ 8] = avg_rv30_tpel8_mc02_c;
289    c->avg_rv30_tpel_pixels_tab[1][ 9] = avg_rv30_tpel8_mc12_c;
290    c->avg_rv30_tpel_pixels_tab[1][10] = avg_rv30_tpel8_mc22_c;
291}
292