summaryrefslogtreecommitdiff
path: root/libavcodec/faanidct.c
blob: 5cacfdd2a31b61acb23b8c654d0e4c735bdedd32 (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
/*
 * Floating point AAN IDCT
 * Copyright (c) 2008 Michael Niedermayer <michaelni@gmx.at>
 *
 * This file is part of Libav.
 *
 * Libav is free software; you can redistribute it and/or
 * modify it under the terms of the GNU Lesser General Public
 * License as published by the Free Software Foundation; either
 * version 2.1 of the License, or (at your option) any later version.
 *
 * Libav is distributed in the hope that it will be useful,
 * 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
 * License along with Libav; if not, write to the Free Software
 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
 */
#include "faanidct.h"
#include "libavutil/common.h"

/* To allow switching to double. */
#define FLOAT float

#define B0 1.0000000000000000000000
#define B1 1.3870398453221474618216 // cos(pi*1/16)sqrt(2)
#define B2 1.3065629648763765278566 // cos(pi*2/16)sqrt(2)
#define B3 1.1758756024193587169745 // cos(pi*3/16)sqrt(2)
#define B4 1.0000000000000000000000 // cos(pi*4/16)sqrt(2)
#define B5 0.7856949583871021812779 // cos(pi*5/16)sqrt(2)
#define B6 0.5411961001461969843997 // cos(pi*6/16)sqrt(2)
#define B7 0.2758993792829430123360 // cos(pi*7/16)sqrt(2)

#define A4 0.70710678118654752438 // cos(pi*4/16)
#define A2 0.92387953251128675613 // cos(pi*2/16)

static const FLOAT prescale[64]={
B0*B0/8, B0*B1/8, B0*B2/8, B0*B3/8, B0*B4/8, B0*B5/8, B0*B6/8, B0*B7/8,
B1*B0/8, B1*B1/8, B1*B2/8, B1*B3/8, B1*B4/8, B1*B5/8, B1*B6/8, B1*B7/8,
B2*B0/8, B2*B1/8, B2*B2/8, B2*B3/8, B2*B4/8, B2*B5/8, B2*B6/8, B2*B7/8,
B3*B0/8, B3*B1/8, B3*B2/8, B3*B3/8, B3*B4/8, B3*B5/8, B3*B6/8, B3*B7/8,
B4*B0/8, B4*B1/8, B4*B2/8, B4*B3/8, B4*B4/8, B4*B5/8, B4*B6/8, B4*B7/8,
B5*B0/8, B5*B1/8, B5*B2/8, B5*B3/8, B5*B4/8, B5*B5/8, B5*B6/8, B5*B7/8,
B6*B0/8, B6*B1/8, B6*B2/8, B6*B3/8, B6*B4/8, B6*B5/8, B6*B6/8, B6*B7/8,
B7*B0/8, B7*B1/8, B7*B2/8, B7*B3/8, B7*B4/8, B7*B5/8, B7*B6/8, B7*B7/8,
};

static inline void p8idct(int16_t data[64], FLOAT temp[64], uint8_t *dest, int stride, int x, int y, int type){
    int i;
    FLOAT av_unused tmp0;
    FLOAT s04, d04, s17, d17, s26, d26, s53, d53;
    FLOAT os07, os16, os25, os34;
    FLOAT od07, od16, od25, od34;

    for(i=0; i<y*8; i+=y){
        s17= temp[1*x + i] + temp[7*x + i];
        d17= temp[1*x + i] - temp[7*x + i];
        s53= temp[5*x + i] + temp[3*x + i];
        d53= temp[5*x + i] - temp[3*x + i];

        od07=  s17 + s53;
        od25= (s17 - s53)*(2*A4);

#if 0 //these 2 are equivalent
        tmp0= (d17 + d53)*(2*A2);
        od34=  d17*( 2*B6) - tmp0;
        od16=  d53*(-2*B2) + tmp0;
#else
        od34=  d17*(2*(B6-A2)) - d53*(2*A2);
        od16=  d53*(2*(A2-B2)) + d17*(2*A2);
#endif

        od16 -= od07;
        od25 -= od16;
        od34 += od25;

        s26 = temp[2*x + i] + temp[6*x + i];
        d26 = temp[2*x + i] - temp[6*x + i];
        d26*= 2*A4;
        d26-= s26;

        s04= temp[0*x + i] + temp[4*x + i];
        d04= temp[0*x + i] - temp[4*x + i];

        os07= s04 + s26;
        os34= s04 - s26;
        os16= d04 + d26;
        os25= d04 - d26;

        if(type==0){
            temp[0*x + i]= os07 + od07;
            temp[7*x + i]= os07 - od07;
            temp[1*x + i]= os16 + od16;
            temp[6*x + i]= os16 - od16;
            temp[2*x + i]= os25 + od25;
            temp[5*x + i]= os25 - od25;
            temp[3*x + i]= os34 - od34;
            temp[4*x + i]= os34 + od34;
        }else if(type==1){
            data[0*x + i]= lrintf(os07 + od07);
            data[7*x + i]= lrintf(os07 - od07);
            data[1*x + i]= lrintf(os16 + od16);
            data[6*x + i]= lrintf(os16 - od16);
            data[2*x + i]= lrintf(os25 + od25);
            data[5*x + i]= lrintf(os25 - od25);
            data[3*x + i]= lrintf(os34 - od34);
            data[4*x + i]= lrintf(os34 + od34);
        }else if(type==2){
            dest[0*stride + i]= av_clip_uint8(((int)dest[0*stride + i]) + lrintf(os07 + od07));
            dest[7*stride + i]= av_clip_uint8(((int)dest[7*stride + i]) + lrintf(os07 - od07));
            dest[1*stride + i]= av_clip_uint8(((int)dest[1*stride + i]) + lrintf(os16 + od16));
            dest[6*stride + i]= av_clip_uint8(((int)dest[6*stride + i]) + lrintf(os16 - od16));
            dest[2*stride + i]= av_clip_uint8(((int)dest[2*stride + i]) + lrintf(os25 + od25));
            dest[5*stride + i]= av_clip_uint8(((int)dest[5*stride + i]) + lrintf(os25 - od25));
            dest[3*stride + i]= av_clip_uint8(((int)dest[3*stride + i]) + lrintf(os34 - od34));
            dest[4*stride + i]= av_clip_uint8(((int)dest[4*stride + i]) + lrintf(os34 + od34));
        }else{
            dest[0*stride + i]= av_clip_uint8(lrintf(os07 + od07));
            dest[7*stride + i]= av_clip_uint8(lrintf(os07 - od07));
            dest[1*stride + i]= av_clip_uint8(lrintf(os16 + od16));
            dest[6*stride + i]= av_clip_uint8(lrintf(os16 - od16));
            dest[2*stride + i]= av_clip_uint8(lrintf(os25 + od25));
            dest[5*stride + i]= av_clip_uint8(lrintf(os25 - od25));
            dest[3*stride + i]= av_clip_uint8(lrintf(os34 - od34));
            dest[4*stride + i]= av_clip_uint8(lrintf(os34 + od34));
        }
    }
}

void ff_faanidct(int16_t block[64]){
    FLOAT temp[64];
    int i;

    emms_c();

    for(i=0; i<64; i++)
        temp[i] = block[i] * prescale[i];

    p8idct(block, temp, NULL, 0, 1, 8, 0);
    p8idct(block, temp, NULL, 0, 8, 1, 1);
}

void ff_faanidct_add(uint8_t *dest, int line_size, int16_t block[64]){
    FLOAT temp[64];
    int i;

    emms_c();

    for(i=0; i<64; i++)
        temp[i] = block[i] * prescale[i];

    p8idct(block, temp, NULL,         0, 1, 8, 0);
    p8idct(NULL , temp, dest, line_size, 8, 1, 2);
}

void ff_faanidct_put(uint8_t *dest, int line_size, int16_t block[64]){
    FLOAT temp[64];
    int i;

    emms_c();

    for(i=0; i<64; i++)
        temp[i] = block[i] * prescale[i];

    p8idct(block, temp, NULL,         0, 1, 8, 0);
    p8idct(NULL , temp, dest, line_size, 8, 1, 3);
}