73
73
void ff_h264_idct8_add_c(uint8_t *dst, DCTELEM *block, int stride){
75
DCTELEM (*src)[8] = (DCTELEM(*)[8])block;
76
75
uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
80
79
for( i = 0; i < 8; i++ )
82
const int a0 = src[i][0] + src[i][4];
83
const int a2 = src[i][0] - src[i][4];
84
const int a4 = (src[i][2]>>1) - src[i][6];
85
const int a6 = (src[i][6]>>1) + src[i][2];
81
const int a0 = block[0+i*8] + block[4+i*8];
82
const int a2 = block[0+i*8] - block[4+i*8];
83
const int a4 = (block[2+i*8]>>1) - block[6+i*8];
84
const int a6 = (block[6+i*8]>>1) + block[2+i*8];
87
86
const int b0 = a0 + a6;
88
87
const int b2 = a2 + a4;
89
88
const int b4 = a2 - a4;
90
89
const int b6 = a0 - a6;
92
const int a1 = -src[i][3] + src[i][5] - src[i][7] - (src[i][7]>>1);
93
const int a3 = src[i][1] + src[i][7] - src[i][3] - (src[i][3]>>1);
94
const int a5 = -src[i][1] + src[i][7] + src[i][5] + (src[i][5]>>1);
95
const int a7 = src[i][3] + src[i][5] + src[i][1] + (src[i][1]>>1);
91
const int a1 = -block[3+i*8] + block[5+i*8] - block[7+i*8] - (block[7+i*8]>>1);
92
const int a3 = block[1+i*8] + block[7+i*8] - block[3+i*8] - (block[3+i*8]>>1);
93
const int a5 = -block[1+i*8] + block[7+i*8] + block[5+i*8] + (block[5+i*8]>>1);
94
const int a7 = block[3+i*8] + block[5+i*8] + block[1+i*8] + (block[1+i*8]>>1);
97
96
const int b1 = (a7>>2) + a1;
98
97
const int b3 = a3 + (a5>>2);
99
98
const int b5 = (a3>>2) - a5;
100
99
const int b7 = a7 - (a1>>2);
101
block[0+i*8] = b0 + b7;
102
block[7+i*8] = b0 - b7;
103
block[1+i*8] = b2 + b5;
104
block[6+i*8] = b2 - b5;
105
block[2+i*8] = b4 + b3;
106
block[5+i*8] = b4 - b3;
107
block[3+i*8] = b6 + b1;
108
block[4+i*8] = b6 - b1;
111
110
for( i = 0; i < 8; i++ )
113
const int a0 = src[0][i] + src[4][i];
114
const int a2 = src[0][i] - src[4][i];
115
const int a4 = (src[2][i]>>1) - src[6][i];
116
const int a6 = (src[6][i]>>1) + src[2][i];
112
const int a0 = block[i+0*8] + block[i+4*8];
113
const int a2 = block[i+0*8] - block[i+4*8];
114
const int a4 = (block[i+2*8]>>1) - block[i+6*8];
115
const int a6 = (block[i+6*8]>>1) + block[i+2*8];
118
117
const int b0 = a0 + a6;
119
118
const int b2 = a2 + a4;
120
119
const int b4 = a2 - a4;
121
120
const int b6 = a0 - a6;
123
const int a1 = -src[3][i] + src[5][i] - src[7][i] - (src[7][i]>>1);
124
const int a3 = src[1][i] + src[7][i] - src[3][i] - (src[3][i]>>1);
125
const int a5 = -src[1][i] + src[7][i] + src[5][i] + (src[5][i]>>1);
126
const int a7 = src[3][i] + src[5][i] + src[1][i] + (src[1][i]>>1);
122
const int a1 = -block[i+3*8] + block[i+5*8] - block[i+7*8] - (block[i+7*8]>>1);
123
const int a3 = block[i+1*8] + block[i+7*8] - block[i+3*8] - (block[i+3*8]>>1);
124
const int a5 = -block[i+1*8] + block[i+7*8] + block[i+5*8] + (block[i+5*8]>>1);
125
const int a7 = block[i+3*8] + block[i+5*8] + block[i+1*8] + (block[i+1*8]>>1);
128
127
const int b1 = (a7>>2) + a1;
129
128
const int b3 = a3 + (a5>>2);
168
//FIXME this table is a duplicate from h264data.h, and will be removed once the tables from, h264 have been split
169
static const uint8_t scan8[16 + 2*4]={
170
4+1*8, 5+1*8, 4+2*8, 5+2*8,
171
6+1*8, 7+1*8, 6+2*8, 7+2*8,
172
4+3*8, 5+3*8, 4+4*8, 5+4*8,
173
6+3*8, 7+3*8, 6+4*8, 7+4*8,
180
void ff_h264_idct_add16_c(uint8_t *dst, const int *block_offset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]){
183
int nnz = nnzc[ scan8[i] ];
185
if(nnz==1 && block[i*16]) ff_h264_idct_dc_add_c(dst + block_offset[i], block + i*16, stride);
186
else idct_internal (dst + block_offset[i], block + i*16, stride, 4, 6, 1);
191
void ff_h264_idct_add16intra_c(uint8_t *dst, const int *block_offset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]){
194
if(nnzc[ scan8[i] ]) idct_internal (dst + block_offset[i], block + i*16, stride, 4, 6, 1);
195
else if(block[i*16]) ff_h264_idct_dc_add_c(dst + block_offset[i], block + i*16, stride);
199
void ff_h264_idct8_add4_c(uint8_t *dst, const int *block_offset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]){
201
for(i=0; i<16; i+=4){
202
int nnz = nnzc[ scan8[i] ];
204
if(nnz==1 && block[i*16]) ff_h264_idct8_dc_add_c(dst + block_offset[i], block + i*16, stride);
205
else ff_h264_idct8_add_c (dst + block_offset[i], block + i*16, stride);
210
void ff_h264_idct_add8_c(uint8_t **dest, const int *block_offset, DCTELEM *block, int stride, const uint8_t nnzc[6*8]){
212
for(i=16; i<16+8; i++){
214
ff_h264_idct_add_c (dest[(i&4)>>2] + block_offset[i], block + i*16, stride);
216
ff_h264_idct_dc_add_c(dest[(i&4)>>2] + block_offset[i], block + i*16, stride);