vp3dsp.c 9.28 KB
Newer Older
1 2 3
/*
 * Copyright (C) 2004 the ffmpeg project
 *
4
 * This file is part of Libav.
5
 *
6
 * Libav is free software; you can redistribute it and/or
7 8
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
9
 * version 2.1 of the License, or (at your option) any later version.
10
 *
11
 * Libav is distributed in the hope that it will be useful,
12 13 14 15 16
 * but WITHOUT ANY WARRANTY; without even the implied warranty of
 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU
 * Lesser General Public License for more details.
 *
 * You should have received a copy of the GNU Lesser General Public
17
 * License along with Libav; if not, write to the Free Software
18
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
19 20 21
 */

/**
22
 * @file
23
 * Standard C DSP-oriented functions cribbed from the original VP3
24 25 26
 * source code.
 */

27
#include "libavutil/attributes.h"
28
#include "libavutil/common.h"
29
#include "avcodec.h"
30
#include "dsputil.h"
31
#include "vp3dsp.h"
32 33 34 35 36 37 38 39 40 41

#define IdctAdjustBeforeShift 8
#define xC1S7 64277
#define xC2S6 60547
#define xC3S5 54491
#define xC4S4 46341
#define xC5S3 36410
#define xC6S2 25080
#define xC7S1 12785

42 43
#define M(a,b) (((a) * (b))>>16)

44
static av_always_inline void idct(uint8_t *dst, int stride, int16_t *input, int type)
45
{
46
    int16_t *ip = input;
47

48 49
    int A, B, C, D, Ad, Bd, Cd, Dd, E, F, G, H;
    int Ed, Gd, Add, Bdd, Fd, Hd;
50

51
    int i;
52

53 54 55 56
    /* Inverse DCT on the rows now */
    for (i = 0; i < 8; i++) {
        /* Check for non-zero values */
        if ( ip[0] | ip[1] | ip[2] | ip[3] | ip[4] | ip[5] | ip[6] | ip[7] ) {
57 58 59 60
            A = M(xC1S7, ip[1]) + M(xC7S1, ip[7]);
            B = M(xC7S1, ip[1]) - M(xC1S7, ip[7]);
            C = M(xC3S5, ip[3]) + M(xC5S3, ip[5]);
            D = M(xC3S5, ip[5]) - M(xC5S3, ip[3]);
61

62 63
            Ad = M(xC4S4, (A - C));
            Bd = M(xC4S4, (B - D));
64

65 66
            Cd = A + C;
            Dd = B + D;
67

68 69
            E = M(xC4S4, (ip[0] + ip[4]));
            F = M(xC4S4, (ip[0] - ip[4]));
70

71 72
            G = M(xC2S6, ip[2]) + M(xC6S2, ip[6]);
            H = M(xC6S2, ip[2]) - M(xC2S6, ip[6]);
73

74 75
            Ed = E - G;
            Gd = E + G;
76

77 78
            Add = F + Ad;
            Bdd = Bd - H;
79

80 81
            Fd = F - Ad;
            Hd = Bd + H;
82 83

            /*  Final sequence of operations over-write original inputs. */
84 85
            ip[0] = Gd + Cd ;
            ip[7] = Gd - Cd ;
86

87 88
            ip[1] = Add + Hd;
            ip[2] = Add - Hd;
89

90 91
            ip[3] = Ed + Dd ;
            ip[4] = Ed - Dd ;
92

93 94
            ip[5] = Fd + Bdd;
            ip[6] = Fd - Bdd;
95 96 97 98
        }

        ip += 8;            /* next row */
    }
99

100
    ip = input;
101 102 103

    for ( i = 0; i < 8; i++) {
        /* Check for non-zero values (bitwise or faster than ||) */
104
        if ( ip[1 * 8] | ip[2 * 8] | ip[3 * 8] |
105 106
             ip[4 * 8] | ip[5 * 8] | ip[6 * 8] | ip[7 * 8] ) {

107 108 109 110
            A = M(xC1S7, ip[1*8]) + M(xC7S1, ip[7*8]);
            B = M(xC7S1, ip[1*8]) - M(xC1S7, ip[7*8]);
            C = M(xC3S5, ip[3*8]) + M(xC5S3, ip[5*8]);
            D = M(xC3S5, ip[5*8]) - M(xC5S3, ip[3*8]);
111

112 113
            Ad = M(xC4S4, (A - C));
            Bd = M(xC4S4, (B - D));
114

115 116
            Cd = A + C;
            Dd = B + D;
117

Michael Niedermayer's avatar
Michael Niedermayer committed
118 119 120 121 122 123 124
            E = M(xC4S4, (ip[0*8] + ip[4*8])) + 8;
            F = M(xC4S4, (ip[0*8] - ip[4*8])) + 8;

            if(type==1){  //HACK
                E += 16*128;
                F += 16*128;
            }
125

126 127
            G = M(xC2S6, ip[2*8]) + M(xC6S2, ip[6*8]);
            H = M(xC6S2, ip[2*8]) - M(xC2S6, ip[6*8]);
128

129 130
            Ed = E - G;
            Gd = E + G;
131

132 133
            Add = F + Ad;
            Bdd = Bd - H;
134

135 136
            Fd = F - Ad;
            Hd = Bd + H;
137 138

            /* Final sequence of operations over-write original inputs. */
139
            if(type==0){
140 141
                ip[0*8] = (Gd + Cd )  >> 4;
                ip[7*8] = (Gd - Cd )  >> 4;
142

143 144
                ip[1*8] = (Add + Hd ) >> 4;
                ip[2*8] = (Add - Hd ) >> 4;
145

146 147
                ip[3*8] = (Ed + Dd )  >> 4;
                ip[4*8] = (Ed - Dd )  >> 4;
148

149 150
                ip[5*8] = (Fd + Bdd ) >> 4;
                ip[6*8] = (Fd - Bdd ) >> 4;
151
            }else if(type==1){
152 153
                dst[0*stride] = av_clip_uint8((Gd + Cd )  >> 4);
                dst[7*stride] = av_clip_uint8((Gd - Cd )  >> 4);
154

155 156
                dst[1*stride] = av_clip_uint8((Add + Hd ) >> 4);
                dst[2*stride] = av_clip_uint8((Add - Hd ) >> 4);
157

158 159
                dst[3*stride] = av_clip_uint8((Ed + Dd )  >> 4);
                dst[4*stride] = av_clip_uint8((Ed - Dd )  >> 4);
160

161 162
                dst[5*stride] = av_clip_uint8((Fd + Bdd ) >> 4);
                dst[6*stride] = av_clip_uint8((Fd - Bdd ) >> 4);
163
            }else{
164 165
                dst[0*stride] = av_clip_uint8(dst[0*stride] + ((Gd + Cd )  >> 4));
                dst[7*stride] = av_clip_uint8(dst[7*stride] + ((Gd - Cd )  >> 4));
166

167 168
                dst[1*stride] = av_clip_uint8(dst[1*stride] + ((Add + Hd ) >> 4));
                dst[2*stride] = av_clip_uint8(dst[2*stride] + ((Add - Hd ) >> 4));
169

170 171
                dst[3*stride] = av_clip_uint8(dst[3*stride] + ((Ed + Dd )  >> 4));
                dst[4*stride] = av_clip_uint8(dst[4*stride] + ((Ed - Dd )  >> 4));
172

173 174
                dst[5*stride] = av_clip_uint8(dst[5*stride] + ((Fd + Bdd ) >> 4));
                dst[6*stride] = av_clip_uint8(dst[6*stride] + ((Fd - Bdd ) >> 4));
175
            }
176 177

        } else {
178
            if(type==0){
179 180 181 182 183 184
                ip[0*8] =
                ip[1*8] =
                ip[2*8] =
                ip[3*8] =
                ip[4*8] =
                ip[5*8] =
185 186 187 188 189 190 191 192 193 194
                ip[6*8] =
                ip[7*8] = ((xC4S4 * ip[0*8] + (IdctAdjustBeforeShift<<16))>>20);
            }else if(type==1){
                dst[0*stride]=
                dst[1*stride]=
                dst[2*stride]=
                dst[3*stride]=
                dst[4*stride]=
                dst[5*stride]=
                dst[6*stride]=
195
                dst[7*stride]= av_clip_uint8(128 + ((xC4S4 * ip[0*8] + (IdctAdjustBeforeShift<<16))>>20));
196 197 198
            }else{
                if(ip[0*8]){
                    int v= ((xC4S4 * ip[0*8] + (IdctAdjustBeforeShift<<16))>>20);
199 200 201 202 203 204 205 206
                    dst[0*stride] = av_clip_uint8(dst[0*stride] + v);
                    dst[1*stride] = av_clip_uint8(dst[1*stride] + v);
                    dst[2*stride] = av_clip_uint8(dst[2*stride] + v);
                    dst[3*stride] = av_clip_uint8(dst[3*stride] + v);
                    dst[4*stride] = av_clip_uint8(dst[4*stride] + v);
                    dst[5*stride] = av_clip_uint8(dst[5*stride] + v);
                    dst[6*stride] = av_clip_uint8(dst[6*stride] + v);
                    dst[7*stride] = av_clip_uint8(dst[7*stride] + v);
207 208
                }
            }
209 210 211
        }

        ip++;            /* next column */
212
        dst++;
213 214
    }
}
215

216
static void vp3_idct_put_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/){
217 218 219
    idct(dest, line_size, block, 1);
}

220
static void vp3_idct_add_c(uint8_t *dest/*align 8*/, int line_size, DCTELEM *block/*align 16*/){
221 222
    idct(dest, line_size, block, 2);
}
223

224 225
static void vp3_idct_dc_add_c(uint8_t *dest/*align 8*/, int line_size,
                              const DCTELEM *block/*align 16*/){
226
    int i, dc = (block[0] + 15) >> 5;
David Conrad's avatar
David Conrad committed
227 228

    for(i = 0; i < 8; i++){
229 230 231 232 233 234 235 236
        dest[0] = av_clip_uint8(dest[0] + dc);
        dest[1] = av_clip_uint8(dest[1] + dc);
        dest[2] = av_clip_uint8(dest[2] + dc);
        dest[3] = av_clip_uint8(dest[3] + dc);
        dest[4] = av_clip_uint8(dest[4] + dc);
        dest[5] = av_clip_uint8(dest[5] + dc);
        dest[6] = av_clip_uint8(dest[6] + dc);
        dest[7] = av_clip_uint8(dest[7] + dc);
David Conrad's avatar
David Conrad committed
237 238 239 240
        dest += line_size;
    }
}

241 242
static void vp3_v_loop_filter_c(uint8_t *first_pixel, int stride,
                                int *bounding_values)
243 244 245 246 247 248 249 250 251 252 253 254 255 256 257
{
    unsigned char *end;
    int filter_value;
    const int nstride= -stride;

    for (end= first_pixel + 8; first_pixel < end; first_pixel++) {
        filter_value =
            (first_pixel[2 * nstride] - first_pixel[ stride])
         +3*(first_pixel[0          ] - first_pixel[nstride]);
        filter_value = bounding_values[(filter_value + 4) >> 3];
        first_pixel[nstride] = av_clip_uint8(first_pixel[nstride] + filter_value);
        first_pixel[0] = av_clip_uint8(first_pixel[0] - filter_value);
    }
}

258 259
static void vp3_h_loop_filter_c(uint8_t *first_pixel, int stride,
                                int *bounding_values)
260 261 262 263 264 265 266 267 268 269 270 271 272
{
    unsigned char *end;
    int filter_value;

    for (end= first_pixel + 8*stride; first_pixel != end; first_pixel += stride) {
        filter_value =
            (first_pixel[-2] - first_pixel[ 1])
         +3*(first_pixel[ 0] - first_pixel[-1]);
        filter_value = bounding_values[(filter_value + 4) >> 3];
        first_pixel[-1] = av_clip_uint8(first_pixel[-1] + filter_value);
        first_pixel[ 0] = av_clip_uint8(first_pixel[ 0] - filter_value);
    }
}
273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290

av_cold void ff_vp3dsp_init(VP3DSPContext *c, int flags)
{
    c->idct_put      = vp3_idct_put_c;
    c->idct_add      = vp3_idct_add_c;
    c->idct_dc_add   = vp3_idct_dc_add_c;
    c->v_loop_filter = vp3_v_loop_filter_c;
    c->h_loop_filter = vp3_h_loop_filter_c;

    c->idct_perm = FF_NO_IDCT_PERM;

    if (ARCH_ARM)
        ff_vp3dsp_init_arm(c, flags);
    if (ARCH_PPC)
        ff_vp3dsp_init_ppc(c, flags);
    if (ARCH_X86)
        ff_vp3dsp_init_x86(c, flags);
}