FFmpeg: libavcodec/rv30dsp.c Source File

00001 /*
00002  * RV30 decoder motion compensation functions
00003  * Copyright (c) 2007 Konstantin Shishkov
00004  *
00005  * This file is part of FFmpeg.
00006  *
00007  * FFmpeg is free software; you can redistribute it and/or
00008  * modify it under the terms of the GNU Lesser General Public
00009  * License as published by the Free Software Foundation; either
00010  * version 2.1 of the License, or (at your option) any later version.
00011  *
00012  * FFmpeg is distributed in the hope that it will be useful,
00013  * but WITHOUT ANY WARRANTY; without even the implied warranty of
00014  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
00015  * Lesser General Public License for more details.
00016  *
00017  * You should have received a copy of the GNU Lesser General Public
00018  * License along with FFmpeg; if not, write to the Free Software
00019  * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
00020  */
00021 
00027 #include "avcodec.h"
00028 #include "dsputil.h"
00029 #include "rv34dsp.h"
00030 
00031 #define RV30_LOWPASS(OPNAME, OP) \
00032 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){\
00033     const int h = 8;\
00034     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
00035     int i;\
00036     for(i = 0; i < h; i++)\
00037     {\
00038         OP(dst[0], (-(src[-1]+src[2]) + src[0]*C1 + src[1]*C2 + 8)>>4);\
00039         OP(dst[1], (-(src[ 0]+src[3]) + src[1]*C1 + src[2]*C2 + 8)>>4);\
00040         OP(dst[2], (-(src[ 1]+src[4]) + src[2]*C1 + src[3]*C2 + 8)>>4);\
00041         OP(dst[3], (-(src[ 2]+src[5]) + src[3]*C1 + src[4]*C2 + 8)>>4);\
00042         OP(dst[4], (-(src[ 3]+src[6]) + src[4]*C1 + src[5]*C2 + 8)>>4);\
00043         OP(dst[5], (-(src[ 4]+src[7]) + src[5]*C1 + src[6]*C2 + 8)>>4);\
00044         OP(dst[6], (-(src[ 5]+src[8]) + src[6]*C1 + src[7]*C2 + 8)>>4);\
00045         OP(dst[7], (-(src[ 6]+src[9]) + src[7]*C1 + src[8]*C2 + 8)>>4);\
00046         dst += dstStride;\
00047         src += srcStride;\
00048     }\
00049 }\
00050 \
00051 static void OPNAME ## rv30_tpel8_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, const int C1, const int C2){\
00052     const int w = 8;\
00053     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
00054     int i;\
00055     for(i = 0; i < w; i++)\
00056     {\
00057         const int srcA = src[-1*srcStride];\
00058         const int src0 = src[0 *srcStride];\
00059         const int src1 = src[1 *srcStride];\
00060         const int src2 = src[2 *srcStride];\
00061         const int src3 = src[3 *srcStride];\
00062         const int src4 = src[4 *srcStride];\
00063         const int src5 = src[5 *srcStride];\
00064         const int src6 = src[6 *srcStride];\
00065         const int src7 = src[7 *srcStride];\
00066         const int src8 = src[8 *srcStride];\
00067         const int src9 = src[9 *srcStride];\
00068         OP(dst[0*dstStride], (-(srcA+src2) + src0*C1 + src1*C2 + 8)>>4);\
00069         OP(dst[1*dstStride], (-(src0+src3) + src1*C1 + src2*C2 + 8)>>4);\
00070         OP(dst[2*dstStride], (-(src1+src4) + src2*C1 + src3*C2 + 8)>>4);\
00071         OP(dst[3*dstStride], (-(src2+src5) + src3*C1 + src4*C2 + 8)>>4);\
00072         OP(dst[4*dstStride], (-(src3+src6) + src4*C1 + src5*C2 + 8)>>4);\
00073         OP(dst[5*dstStride], (-(src4+src7) + src5*C1 + src6*C2 + 8)>>4);\
00074         OP(dst[6*dstStride], (-(src5+src8) + src6*C1 + src7*C2 + 8)>>4);\
00075         OP(dst[7*dstStride], (-(src6+src9) + src7*C1 + src8*C2 + 8)>>4);\
00076         dst++;\
00077         src++;\
00078     }\
00079 }\
00080 \
00081 static void OPNAME ## rv30_tpel8_hv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
00082     const int w = 8;\
00083     const int h = 8;\
00084     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
00085     int i, j;\
00086     for(j = 0; j < h; j++){\
00087         for(i = 0; i < w; i++){\
00088             OP(dst[i], (\
00089                   src[srcStride*-1+i-1]  -12*src[srcStride*-1+i]  -6*src[srcStride*-1+i+1]    +src[srcStride*-1+i+2]+\
00090               -12*src[srcStride* 0+i-1] +144*src[srcStride* 0+i] +72*src[srcStride* 0+i+1] -12*src[srcStride* 0+i+2] +\
00091                -6*src[srcStride* 1+i-1]  +72*src[srcStride* 1+i] +36*src[srcStride* 1+i+1]  -6*src[srcStride* 1+i+2] +\
00092                   src[srcStride* 2+i-1]  -12*src[srcStride* 2+i]  -6*src[srcStride* 2+i+1]    +src[srcStride* 2+i+2] +\
00093                   128)>>8);\
00094         }\
00095         src += srcStride;\
00096         dst += dstStride;\
00097     }\
00098 }\
00099 \
00100 static void OPNAME ## rv30_tpel8_hhv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
00101     const int w = 8;\
00102     const int h = 8;\
00103     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
00104     int i, j;\
00105     for(j = 0; j < h; j++){\
00106         for(i = 0; i < w; i++){\
00107             OP(dst[i], (\
00108                   src[srcStride*-1+i-1]  -12*src[srcStride*-1+i+1]  -6*src[srcStride*-1+i]    +src[srcStride*-1+i+2]+\
00109               -12*src[srcStride* 0+i-1] +144*src[srcStride* 0+i+1] +72*src[srcStride* 0+i] -12*src[srcStride* 0+i+2]+\
00110                -6*src[srcStride* 1+i-1]  +72*src[srcStride* 1+i+1] +36*src[srcStride* 1+i]  -6*src[srcStride* 1+i+2]+\
00111                   src[srcStride* 2+i-1]  -12*src[srcStride* 2+i+1]  -6*src[srcStride* 2+i]    +src[srcStride* 2+i+2]+\
00112                   128)>>8);\
00113         }\
00114         src += srcStride;\
00115         dst += dstStride;\
00116     }\
00117 }\
00118 \
00119 static void OPNAME ## rv30_tpel8_hvv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
00120     const int w = 8;\
00121     const int h = 8;\
00122     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
00123     int i, j;\
00124     for(j = 0; j < h; j++){\
00125         for(i = 0; i < w; i++){\
00126             OP(dst[i], (\
00127                   src[srcStride*-1+i-1]  -12*src[srcStride*-1+i]  -6*src[srcStride*-1+i+1]    +src[srcStride*-1+i+2]+\
00128                -6*src[srcStride* 0+i-1]  +72*src[srcStride* 0+i] +36*src[srcStride* 0+i+1]  -6*src[srcStride* 0+i+2]+\
00129               -12*src[srcStride* 1+i-1] +144*src[srcStride* 1+i] +72*src[srcStride* 1+i+1] -12*src[srcStride* 1+i+2]+\
00130                   src[srcStride* 2+i-1]  -12*src[srcStride* 2+i]  -6*src[srcStride* 2+i+1]    +src[srcStride* 2+i+2]+\
00131                   128)>>8);\
00132         }\
00133         src += srcStride;\
00134         dst += dstStride;\
00135     }\
00136 }\
00137 \
00138 static void OPNAME ## rv30_tpel8_hhvv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
00139     const int w = 8;\
00140     const int h = 8;\
00141     uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;\
00142     int i, j;\
00143     for(j = 0; j < h; j++){\
00144         for(i = 0; i < w; i++){\
00145             OP(dst[i], (\
00146                36*src[i+srcStride*0] +54*src[i+1+srcStride*0] +6*src[i+2+srcStride*0]+\
00147                54*src[i+srcStride*1] +81*src[i+1+srcStride*1] +9*src[i+2+srcStride*1]+\
00148                 6*src[i+srcStride*2] + 9*src[i+1+srcStride*2] +  src[i+2+srcStride*2]+\
00149                128)>>8);\
00150         }\
00151         src += srcStride;\
00152         dst += dstStride;\
00153     }\
00154 }\
00155 \
00156 static void OPNAME ## rv30_tpel16_v_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, const int C1, const int C2){\
00157     OPNAME ## rv30_tpel8_v_lowpass(dst  , src  , dstStride, srcStride, C1, C2);\
00158     OPNAME ## rv30_tpel8_v_lowpass(dst+8, src+8, dstStride, srcStride, C1, C2);\
00159     src += 8*srcStride;\
00160     dst += 8*dstStride;\
00161     OPNAME ## rv30_tpel8_v_lowpass(dst  , src  , dstStride, srcStride, C1, C2);\
00162     OPNAME ## rv30_tpel8_v_lowpass(dst+8, src+8, dstStride, srcStride, C1, C2);\
00163 }\
00164 \
00165 static void OPNAME ## rv30_tpel16_h_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride, const int C1, const int C2){\
00166     OPNAME ## rv30_tpel8_h_lowpass(dst  , src  , dstStride, srcStride, C1, C2);\
00167     OPNAME ## rv30_tpel8_h_lowpass(dst+8, src+8, dstStride, srcStride, C1, C2);\
00168     src += 8*srcStride;\
00169     dst += 8*dstStride;\
00170     OPNAME ## rv30_tpel8_h_lowpass(dst  , src  , dstStride, srcStride, C1, C2);\
00171     OPNAME ## rv30_tpel8_h_lowpass(dst+8, src+8, dstStride, srcStride, C1, C2);\
00172 }\
00173 \
00174 static void OPNAME ## rv30_tpel16_hv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
00175     OPNAME ## rv30_tpel8_hv_lowpass(dst  , src  , dstStride, srcStride);\
00176     OPNAME ## rv30_tpel8_hv_lowpass(dst+8, src+8, dstStride, srcStride);\
00177     src += 8*srcStride;\
00178     dst += 8*dstStride;\
00179     OPNAME ## rv30_tpel8_hv_lowpass(dst  , src  , dstStride, srcStride);\
00180     OPNAME ## rv30_tpel8_hv_lowpass(dst+8, src+8, dstStride, srcStride);\
00181 }\
00182 \
00183 static void OPNAME ## rv30_tpel16_hhv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
00184     OPNAME ## rv30_tpel8_hhv_lowpass(dst  , src  , dstStride, srcStride);\
00185     OPNAME ## rv30_tpel8_hhv_lowpass(dst+8, src+8, dstStride, srcStride);\
00186     src += 8*srcStride;\
00187     dst += 8*dstStride;\
00188     OPNAME ## rv30_tpel8_hhv_lowpass(dst  , src  , dstStride, srcStride);\
00189     OPNAME ## rv30_tpel8_hhv_lowpass(dst+8, src+8, dstStride, srcStride);\
00190 }\
00191 \
00192 static void OPNAME ## rv30_tpel16_hvv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
00193     OPNAME ## rv30_tpel8_hvv_lowpass(dst  , src  , dstStride, srcStride);\
00194     OPNAME ## rv30_tpel8_hvv_lowpass(dst+8, src+8, dstStride, srcStride);\
00195     src += 8*srcStride;\
00196     dst += 8*dstStride;\
00197     OPNAME ## rv30_tpel8_hvv_lowpass(dst  , src  , dstStride, srcStride);\
00198     OPNAME ## rv30_tpel8_hvv_lowpass(dst+8, src+8, dstStride, srcStride);\
00199 }\
00200 \
00201 static void OPNAME ## rv30_tpel16_hhvv_lowpass(uint8_t *dst, uint8_t *src, int dstStride, int srcStride){\
00202     OPNAME ## rv30_tpel8_hhvv_lowpass(dst  , src  , dstStride, srcStride);\
00203     OPNAME ## rv30_tpel8_hhvv_lowpass(dst+8, src+8, dstStride, srcStride);\
00204     src += 8*srcStride;\
00205     dst += 8*dstStride;\
00206     OPNAME ## rv30_tpel8_hhvv_lowpass(dst  , src  , dstStride, srcStride);\
00207     OPNAME ## rv30_tpel8_hhvv_lowpass(dst+8, src+8, dstStride, srcStride);\
00208 }\
00209 \
00210 
00211 #define RV30_MC(OPNAME, SIZE) \
00212 static void OPNAME ## rv30_tpel ## SIZE ## _mc10_c(uint8_t *dst, uint8_t *src, int stride){\
00213     OPNAME ## rv30_tpel ## SIZE ## _h_lowpass(dst, src, stride, stride, 12, 6);\
00214 }\
00215 \
00216 static void OPNAME ## rv30_tpel ## SIZE ## _mc20_c(uint8_t *dst, uint8_t *src, int stride){\
00217     OPNAME ## rv30_tpel ## SIZE ## _h_lowpass(dst, src, stride, stride, 6, 12);\
00218 }\
00219 \
00220 static void OPNAME ## rv30_tpel ## SIZE ## _mc01_c(uint8_t *dst, uint8_t *src, int stride){\
00221     OPNAME ## rv30_tpel ## SIZE ## _v_lowpass(dst, src, stride, stride, 12, 6);\
00222 }\
00223 \
00224 static void OPNAME ## rv30_tpel ## SIZE ## _mc02_c(uint8_t *dst, uint8_t *src, int stride){\
00225     OPNAME ## rv30_tpel ## SIZE ## _v_lowpass(dst, src, stride, stride, 6, 12);\
00226 }\
00227 \
00228 static void OPNAME ## rv30_tpel ## SIZE ## _mc11_c(uint8_t *dst, uint8_t *src, int stride){\
00229     OPNAME ## rv30_tpel ## SIZE ## _hv_lowpass(dst, src, stride, stride);\
00230 }\
00231 \
00232 static void OPNAME ## rv30_tpel ## SIZE ## _mc12_c(uint8_t *dst, uint8_t *src, int stride){\
00233     OPNAME ## rv30_tpel ## SIZE ## _hvv_lowpass(dst, src, stride, stride);\
00234 }\
00235 \
00236 static void OPNAME ## rv30_tpel ## SIZE ## _mc21_c(uint8_t *dst, uint8_t *src, int stride){\
00237     OPNAME ## rv30_tpel ## SIZE ## _hhv_lowpass(dst, src, stride, stride);\
00238 }\
00239 \
00240 static void OPNAME ## rv30_tpel ## SIZE ## _mc22_c(uint8_t *dst, uint8_t *src, int stride){\
00241     OPNAME ## rv30_tpel ## SIZE ## _hhvv_lowpass(dst, src, stride, stride);\
00242 }\
00243 \
00244 
00245 #define op_avg(a, b)  a = (((a)+cm[b]+1)>>1)
00246 #define op_put(a, b)  a = cm[b]
00247 
00248 RV30_LOWPASS(put_       , op_put)
00249 RV30_LOWPASS(avg_       , op_avg)
00250 RV30_MC(put_, 8)
00251 RV30_MC(put_, 16)
00252 RV30_MC(avg_, 8)
00253 RV30_MC(avg_, 16)
00254 
00255 av_cold void ff_rv30dsp_init(RV34DSPContext *c, DSPContext* dsp) {
00256 
00257     ff_rv34dsp_init(c, dsp);
00258 
00259     c->put_pixels_tab[0][ 0] = dsp->put_h264_qpel_pixels_tab[0][0];
00260     c->put_pixels_tab[0][ 1] = put_rv30_tpel16_mc10_c;
00261     c->put_pixels_tab[0][ 2] = put_rv30_tpel16_mc20_c;
00262     c->put_pixels_tab[0][ 4] = put_rv30_tpel16_mc01_c;
00263     c->put_pixels_tab[0][ 5] = put_rv30_tpel16_mc11_c;
00264     c->put_pixels_tab[0][ 6] = put_rv30_tpel16_mc21_c;
00265     c->put_pixels_tab[0][ 8] = put_rv30_tpel16_mc02_c;
00266     c->put_pixels_tab[0][ 9] = put_rv30_tpel16_mc12_c;
00267     c->put_pixels_tab[0][10] = put_rv30_tpel16_mc22_c;
00268     c->avg_pixels_tab[0][ 0] = dsp->avg_h264_qpel_pixels_tab[0][0];
00269     c->avg_pixels_tab[0][ 1] = avg_rv30_tpel16_mc10_c;
00270     c->avg_pixels_tab[0][ 2] = avg_rv30_tpel16_mc20_c;
00271     c->avg_pixels_tab[0][ 4] = avg_rv30_tpel16_mc01_c;
00272     c->avg_pixels_tab[0][ 5] = avg_rv30_tpel16_mc11_c;
00273     c->avg_pixels_tab[0][ 6] = avg_rv30_tpel16_mc21_c;
00274     c->avg_pixels_tab[0][ 8] = avg_rv30_tpel16_mc02_c;
00275     c->avg_pixels_tab[0][ 9] = avg_rv30_tpel16_mc12_c;
00276     c->avg_pixels_tab[0][10] = avg_rv30_tpel16_mc22_c;
00277     c->put_pixels_tab[1][ 0] = dsp->put_h264_qpel_pixels_tab[1][0];
00278     c->put_pixels_tab[1][ 1] = put_rv30_tpel8_mc10_c;
00279     c->put_pixels_tab[1][ 2] = put_rv30_tpel8_mc20_c;
00280     c->put_pixels_tab[1][ 4] = put_rv30_tpel8_mc01_c;
00281     c->put_pixels_tab[1][ 5] = put_rv30_tpel8_mc11_c;
00282     c->put_pixels_tab[1][ 6] = put_rv30_tpel8_mc21_c;
00283     c->put_pixels_tab[1][ 8] = put_rv30_tpel8_mc02_c;
00284     c->put_pixels_tab[1][ 9] = put_rv30_tpel8_mc12_c;
00285     c->put_pixels_tab[1][10] = put_rv30_tpel8_mc22_c;
00286     c->avg_pixels_tab[1][ 0] = dsp->avg_h264_qpel_pixels_tab[1][0];
00287     c->avg_pixels_tab[1][ 1] = avg_rv30_tpel8_mc10_c;
00288     c->avg_pixels_tab[1][ 2] = avg_rv30_tpel8_mc20_c;
00289     c->avg_pixels_tab[1][ 4] = avg_rv30_tpel8_mc01_c;
00290     c->avg_pixels_tab[1][ 5] = avg_rv30_tpel8_mc11_c;
00291     c->avg_pixels_tab[1][ 6] = avg_rv30_tpel8_mc21_c;
00292     c->avg_pixels_tab[1][ 8] = avg_rv30_tpel8_mc02_c;
00293     c->avg_pixels_tab[1][ 9] = avg_rv30_tpel8_mc12_c;
00294     c->avg_pixels_tab[1][10] = avg_rv30_tpel8_mc22_c;
00295 
00296     c->put_chroma_pixels_tab[0] = dsp->put_h264_chroma_pixels_tab[0];
00297     c->put_chroma_pixels_tab[1] = dsp->put_h264_chroma_pixels_tab[1];
00298     c->avg_chroma_pixels_tab[0] = dsp->avg_h264_chroma_pixels_tab[0];
00299     c->avg_chroma_pixels_tab[1] = dsp->avg_h264_chroma_pixels_tab[1];
00300 }