File size: 11,964 Bytes
8ead80b |
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 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 |
/*
* Copyright (C) 2019 Philip Langdale <[email protected]>
*
* This file is part of FFmpeg.
*
* FFmpeg 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.
*
* FFmpeg 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 FFmpeg; if not, write to the Free Software
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
*/
__device__ static const int coef_lf[2] = { 4309, 213 };
__device__ static const int coef_hf[3] = { 5570, 3801, 1016 };
__device__ static const int coef_sp[2] = { 5077, 981 };
template<typename T>
__inline__ __device__ T max3(T a, T b, T c)
{
T x = max(a, b);
return max(x, c);
}
template<typename T>
__inline__ __device__ T min3(T a, T b, T c)
{
T x = min(a, b);
return min(x, c);
}
template<typename T>
__inline__ __device__ T clip(T a, T min, T max)
{
if (a < min) {
return min;
} else if (a > max) {
return max;
} else {
return a;
}
}
template<typename T>
__inline__ __device__ T filter_intra(T cur_prefs3, T cur_prefs,
T cur_mrefs, T cur_mrefs3,
int clip_max)
{
int final = (coef_sp[0] * (cur_mrefs + cur_prefs) -
coef_sp[1] * (cur_mrefs3 + cur_prefs3)) >> 13;
return clip(final, 0, clip_max);
}
template<typename T>
__inline__ __device__ T filter(T cur_prefs3, T cur_prefs, T cur_mrefs, T cur_mrefs3,
T prev2_prefs4, T prev2_prefs2, T prev2_0, T prev2_mrefs2, T prev2_mrefs4,
T prev_prefs, T prev_mrefs, T next_prefs, T next_mrefs,
T next2_prefs4, T next2_prefs2, T next2_0, T next2_mrefs2, T next2_mrefs4,
int clip_max)
{
T final;
int c = cur_mrefs;
int d = (prev2_0 + next2_0) >> 1;
int e = cur_prefs;
int temporal_diff0 = abs(prev2_0 - next2_0);
int temporal_diff1 = (abs(prev_mrefs - c) + abs(prev_prefs - e)) >> 1;
int temporal_diff2 = (abs(next_mrefs - c) + abs(next_prefs - e)) >> 1;
int diff = max3(temporal_diff0 >> 1, temporal_diff1, temporal_diff2);
if (!diff) {
final = d;
} else {
int b = ((prev2_mrefs2 + next2_mrefs2) >> 1) - c;
int f = ((prev2_prefs2 + next2_prefs2) >> 1) - e;
int dc = d - c;
int de = d - e;
int mmax = max3(de, dc, min(b, f));
int mmin = min3(de, dc, max(b, f));
diff = max3(diff, mmin, -mmax);
int interpol;
if (abs(c - e) > temporal_diff0) {
interpol = (((coef_hf[0] * (prev2_0 + next2_0)
- coef_hf[1] * (prev2_mrefs2 + next2_mrefs2 + prev2_prefs2 + next2_prefs2)
+ coef_hf[2] * (prev2_mrefs4 + next2_mrefs4 + prev2_prefs4 + next2_mrefs4)) >> 2)
+ coef_lf[0] * (c + e) - coef_lf[1] * (cur_mrefs3 + cur_prefs3)) >> 13;
} else {
interpol = (coef_sp[0] * (c + e) - coef_sp[1] * (cur_mrefs3 + cur_prefs3)) >> 13;
}
if (interpol > d + diff) {
interpol = d + diff;
} else if (interpol < d - diff) {
interpol = d - diff;
}
final = clip(interpol, 0, clip_max);
}
return final;
}
template<typename T>
__inline__ __device__ void bwdif_single(T *dst,
cudaTextureObject_t prev,
cudaTextureObject_t cur,
cudaTextureObject_t next,
int dst_width, int dst_height, int dst_pitch,
int src_width, int src_height,
int parity, int tff,
int is_field_end, int clip_max)
{
// Identify location
int xo = blockIdx.x * blockDim.x + threadIdx.x;
int yo = blockIdx.y * blockDim.y + threadIdx.y;
if (xo >= dst_width || yo >= dst_height) {
return;
}
// Don't modify the primary field
if (yo % 2 == parity) {
dst[yo*dst_pitch+xo] = tex2D<T>(cur, xo, yo);
return;
}
T cur_prefs3 = tex2D<T>(cur, xo, yo + 3);
T cur_prefs = tex2D<T>(cur, xo, yo + 1);
T cur_mrefs = tex2D<T>(cur, xo, yo - 1);
T cur_mrefs3 = tex2D<T>(cur, xo, yo - 3);
if (is_field_end) {
dst[yo*dst_pitch+xo] =
filter_intra(cur_prefs3, cur_prefs, cur_mrefs, cur_mrefs3, clip_max);
return;
}
// Calculate temporal prediction
int is_second_field = !(parity ^ tff);
cudaTextureObject_t prev2 = prev;
cudaTextureObject_t prev1 = is_second_field ? cur : prev;
cudaTextureObject_t next1 = is_second_field ? next : cur;
cudaTextureObject_t next2 = next;
T prev2_prefs4 = tex2D<T>(prev2, xo, yo + 4);
T prev2_prefs2 = tex2D<T>(prev2, xo, yo + 2);
T prev2_0 = tex2D<T>(prev2, xo, yo + 0);
T prev2_mrefs2 = tex2D<T>(prev2, xo, yo - 2);
T prev2_mrefs4 = tex2D<T>(prev2, xo, yo - 4);
T prev_prefs = tex2D<T>(prev1, xo, yo + 1);
T prev_mrefs = tex2D<T>(prev1, xo, yo - 1);
T next_prefs = tex2D<T>(next1, xo, yo + 1);
T next_mrefs = tex2D<T>(next1, xo, yo - 1);
T next2_prefs4 = tex2D<T>(next2, xo, yo + 4);
T next2_prefs2 = tex2D<T>(next2, xo, yo + 2);
T next2_0 = tex2D<T>(next2, xo, yo + 0);
T next2_mrefs2 = tex2D<T>(next2, xo, yo - 2);
T next2_mrefs4 = tex2D<T>(next2, xo, yo - 4);
dst[yo*dst_pitch+xo] = filter(cur_prefs3, cur_prefs, cur_mrefs, cur_mrefs3,
prev2_prefs4, prev2_prefs2, prev2_0, prev2_mrefs2, prev2_mrefs4,
prev_prefs, prev_mrefs, next_prefs, next_mrefs,
next2_prefs4, next2_prefs2, next2_0, next2_mrefs2, next2_mrefs4,
clip_max);
}
template <typename T>
__inline__ __device__ void bwdif_double(T *dst,
cudaTextureObject_t prev,
cudaTextureObject_t cur,
cudaTextureObject_t next,
int dst_width, int dst_height, int dst_pitch,
int src_width, int src_height,
int parity, int tff,
int is_field_end, int clip_max)
{
int xo = blockIdx.x * blockDim.x + threadIdx.x;
int yo = blockIdx.y * blockDim.y + threadIdx.y;
if (xo >= dst_width || yo >= dst_height) {
return;
}
if (yo % 2 == parity) {
// Don't modify the primary field
dst[yo*dst_pitch+xo] = tex2D<T>(cur, xo, yo);
return;
}
T cur_prefs3 = tex2D<T>(cur, xo, yo + 3);
T cur_prefs = tex2D<T>(cur, xo, yo + 1);
T cur_mrefs = tex2D<T>(cur, xo, yo - 1);
T cur_mrefs3 = tex2D<T>(cur, xo, yo - 3);
if (is_field_end) {
T final;
final.x = filter_intra(cur_prefs3.x, cur_prefs.x, cur_mrefs.x, cur_mrefs3.x,
clip_max);
final.y = filter_intra(cur_prefs3.y, cur_prefs.y, cur_mrefs.y, cur_mrefs3.y,
clip_max);
dst[yo*dst_pitch+xo] = final;
return;
}
int is_second_field = !(parity ^ tff);
cudaTextureObject_t prev2 = prev;
cudaTextureObject_t prev1 = is_second_field ? cur : prev;
cudaTextureObject_t next1 = is_second_field ? next : cur;
cudaTextureObject_t next2 = next;
T prev2_prefs4 = tex2D<T>(prev2, xo, yo + 4);
T prev2_prefs2 = tex2D<T>(prev2, xo, yo + 2);
T prev2_0 = tex2D<T>(prev2, xo, yo + 0);
T prev2_mrefs2 = tex2D<T>(prev2, xo, yo - 2);
T prev2_mrefs4 = tex2D<T>(prev2, xo, yo - 4);
T prev_prefs = tex2D<T>(prev1, xo, yo + 1);
T prev_mrefs = tex2D<T>(prev1, xo, yo - 1);
T next_prefs = tex2D<T>(next1, xo, yo + 1);
T next_mrefs = tex2D<T>(next1, xo, yo - 1);
T next2_prefs4 = tex2D<T>(next2, xo, yo + 4);
T next2_prefs2 = tex2D<T>(next2, xo, yo + 2);
T next2_0 = tex2D<T>(next2, xo, yo + 0);
T next2_mrefs2 = tex2D<T>(next2, xo, yo - 2);
T next2_mrefs4 = tex2D<T>(next2, xo, yo - 4);
T final;
final.x = filter(cur_prefs3.x, cur_prefs.x, cur_mrefs.x, cur_mrefs3.x,
prev2_prefs4.x, prev2_prefs2.x, prev2_0.x, prev2_mrefs2.x, prev2_mrefs4.x,
prev_prefs.x, prev_mrefs.x, next_prefs.x, next_mrefs.x,
next2_prefs4.x, next2_prefs2.x, next2_0.x, next2_mrefs2.x, next2_mrefs4.x,
clip_max);
final.y = filter(cur_prefs3.y, cur_prefs.y, cur_mrefs.y, cur_mrefs3.y,
prev2_prefs4.y, prev2_prefs2.y, prev2_0.y, prev2_mrefs2.y, prev2_mrefs4.y,
prev_prefs.y, prev_mrefs.y, next_prefs.y, next_mrefs.y,
next2_prefs4.y, next2_prefs2.y, next2_0.y, next2_mrefs2.y, next2_mrefs4.y,
clip_max);
dst[yo*dst_pitch+xo] = final;
}
extern "C" {
__global__ void bwdif_uchar(unsigned char *dst,
cudaTextureObject_t prev,
cudaTextureObject_t cur,
cudaTextureObject_t next,
int dst_width, int dst_height, int dst_pitch,
int src_width, int src_height,
int parity, int tff, int is_field_end, int clip_max)
{
bwdif_single(dst, prev, cur, next,
dst_width, dst_height, dst_pitch,
src_width, src_height,
parity, tff, is_field_end, clip_max);
}
__global__ void bwdif_ushort(unsigned short *dst,
cudaTextureObject_t prev,
cudaTextureObject_t cur,
cudaTextureObject_t next,
int dst_width, int dst_height, int dst_pitch,
int src_width, int src_height,
int parity, int tff, int is_field_end, int clip_max)
{
bwdif_single(dst, prev, cur, next,
dst_width, dst_height, dst_pitch,
src_width, src_height,
parity, tff, is_field_end, clip_max);
}
__global__ void bwdif_uchar2(uchar2 *dst,
cudaTextureObject_t prev,
cudaTextureObject_t cur,
cudaTextureObject_t next,
int dst_width, int dst_height, int dst_pitch,
int src_width, int src_height,
int parity, int tff, int is_field_end, int clip_max)
{
bwdif_double(dst, prev, cur, next,
dst_width, dst_height, dst_pitch,
src_width, src_height,
parity, tff, is_field_end, clip_max);
}
__global__ void bwdif_ushort2(ushort2 *dst,
cudaTextureObject_t prev,
cudaTextureObject_t cur,
cudaTextureObject_t next,
int dst_width, int dst_height, int dst_pitch,
int src_width, int src_height,
int parity, int tff, int is_field_end, int clip_max)
{
bwdif_double(dst, prev, cur, next,
dst_width, dst_height, dst_pitch,
src_width, src_height,
parity, tff, is_field_end, clip_max);
}
} /* extern "C" */
|