Parent Directory
| 
Revision Log
Revision 1.13 - (view) (download)
| 1 : | edgomez | 1.11 | /***************************************************************************** |
| 2 : | * | ||
| 3 : | * XVID MPEG-4 VIDEO CODEC | ||
| 4 : | * - Interpolation related header - | ||
| 5 : | * | ||
| 6 : | * Copyright(C) 2001-2003 Peter Ross <pross@xvid.org> | ||
| 7 : | * | ||
| 8 : | * This program is free software ; you can redistribute it and/or modify | ||
| 9 : | * it under the terms of the GNU General Public License as published by | ||
| 10 : | * the Free Software Foundation ; either version 2 of the License, or | ||
| 11 : | * (at your option) any later version. | ||
| 12 : | * | ||
| 13 : | * This program is distributed in the hope that it will be useful, | ||
| 14 : | * but WITHOUT ANY WARRANTY ; without even the implied warranty of | ||
| 15 : | * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the | ||
| 16 : | * GNU General Public License for more details. | ||
| 17 : | * | ||
| 18 : | * You should have received a copy of the GNU General Public License | ||
| 19 : | * along with this program ; if not, write to the Free Software | ||
| 20 : | * Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA | ||
| 21 : | * | ||
| 22 : | * $Id$ | ||
| 23 : | * | ||
| 24 : | ****************************************************************************/ | ||
| 25 : | |||
| 26 : | edgomez | 1.10 | #ifndef _INTERPOLATE8X8_H_ |
| 27 : | #define _INTERPOLATE8X8_H_ | ||
| 28 : | chl | 1.6 | |
| 29 : | Isibaar | 1.1 | #include "../utils/mem_transfer.h" |
| 30 : | |||
| 31 : | edgomez | 1.3 | typedef void (INTERPOLATE8X8) (uint8_t * const dst, |
| 32 : | const uint8_t * const src, | ||
| 33 : | const uint32_t stride, | ||
| 34 : | const uint32_t rounding); | ||
| 35 : | typedef INTERPOLATE8X8 *INTERPOLATE8X8_PTR; | ||
| 36 : | Isibaar | 1.1 | |
| 37 : | edgomez | 1.10 | typedef void (INTERPOLATE8X8_AVG2) (uint8_t *dst, |
| 38 : | const uint8_t *src1, | ||
| 39 : | const uint8_t *src2, | ||
| 40 : | const uint32_t stride, | ||
| 41 : | const uint32_t rounding, | ||
| 42 : | const uint32_t height); | ||
| 43 : | typedef INTERPOLATE8X8_AVG2 *INTERPOLATE8X8_AVG2_PTR; | ||
| 44 : | |||
| 45 : | typedef void (INTERPOLATE8X8_AVG4) (uint8_t *dst, | ||
| 46 : | const uint8_t *src1, | ||
| 47 : | const uint8_t *src2, | ||
| 48 : | const uint8_t *src3, | ||
| 49 : | const uint8_t *src4, | ||
| 50 : | const uint32_t stride, | ||
| 51 : | const uint32_t rounding); | ||
| 52 : | typedef INTERPOLATE8X8_AVG4 *INTERPOLATE8X8_AVG4_PTR; | ||
| 53 : | |||
| 54 : | typedef void (INTERPOLATE_LOWPASS) (uint8_t *dst, | ||
| 55 : | uint8_t *src, | ||
| 56 : | int32_t stride, | ||
| 57 : | int32_t rounding); | ||
| 58 : | |||
| 59 : | typedef INTERPOLATE_LOWPASS *INTERPOLATE_LOWPASS_PTR; | ||
| 60 : | |||
| 61 : | typedef void (INTERPOLATE_LOWPASS_HV) (uint8_t *dst1, | ||
| 62 : | uint8_t *dst2, | ||
| 63 : | uint8_t *src, | ||
| 64 : | int32_t stride, | ||
| 65 : | int32_t rounding); | ||
| 66 : | |||
| 67 : | typedef INTERPOLATE_LOWPASS_HV *INTERPOLATE_LOWPASS_HV_PTR; | ||
| 68 : | |||
| 69 : | typedef void (INTERPOLATE8X8_6TAP_LOWPASS) (uint8_t *dst, | ||
| 70 : | uint8_t *src, | ||
| 71 : | int32_t stride, | ||
| 72 : | int32_t rounding); | ||
| 73 : | |||
| 74 : | typedef INTERPOLATE8X8_6TAP_LOWPASS *INTERPOLATE8X8_6TAP_LOWPASS_PTR; | ||
| 75 : | |||
| 76 : | edgomez | 1.13 | /* These function do: dst = interpolate(src) */ |
| 77 : | Isibaar | 1.1 | extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_h; |
| 78 : | extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_v; | ||
| 79 : | extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_hv; | ||
| 80 : | |||
| 81 : | edgomez | 1.13 | /* These functions do: dst = (dst+interpolate(src) + 1)/2 |
| 82 : | * Suitable for direct/interpolated bvop prediction block | ||
| 83 : | * building w/o the need for intermediate interpolated result | ||
| 84 : | * storing/reading | ||
| 85 : | * NB: the rounding applies to the interpolation, but not | ||
| 86 : | * the averaging step which will always use rounding=0 */ | ||
| 87 : | extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_add; | ||
| 88 : | extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_h_add; | ||
| 89 : | extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_v_add; | ||
| 90 : | extern INTERPOLATE8X8_PTR interpolate8x8_halfpel_hv_add; | ||
| 91 : | |||
| 92 : | edgomez | 1.10 | extern INTERPOLATE8X8_AVG2_PTR interpolate8x8_avg2; |
| 93 : | extern INTERPOLATE8X8_AVG4_PTR interpolate8x8_avg4; | ||
| 94 : | |||
| 95 : | extern INTERPOLATE_LOWPASS_PTR interpolate8x8_lowpass_h; | ||
| 96 : | extern INTERPOLATE_LOWPASS_PTR interpolate8x8_lowpass_v; | ||
| 97 : | |||
| 98 : | extern INTERPOLATE_LOWPASS_PTR interpolate16x16_lowpass_h; | ||
| 99 : | extern INTERPOLATE_LOWPASS_PTR interpolate16x16_lowpass_v; | ||
| 100 : | |||
| 101 : | extern INTERPOLATE_LOWPASS_HV_PTR interpolate8x8_lowpass_hv; | ||
| 102 : | extern INTERPOLATE_LOWPASS_HV_PTR interpolate16x16_lowpass_hv; | ||
| 103 : | |||
| 104 : | extern INTERPOLATE8X8_6TAP_LOWPASS_PTR interpolate8x8_6tap_lowpass_h; | ||
| 105 : | extern INTERPOLATE8X8_6TAP_LOWPASS_PTR interpolate8x8_6tap_lowpass_v; | ||
| 106 : | |||
| 107 : | Isibaar | 1.1 | INTERPOLATE8X8 interpolate8x8_halfpel_h_c; |
| 108 : | INTERPOLATE8X8 interpolate8x8_halfpel_v_c; | ||
| 109 : | INTERPOLATE8X8 interpolate8x8_halfpel_hv_c; | ||
| 110 : | edgomez | 1.13 | INTERPOLATE8X8 interpolate8x8_halfpel_add_c; |
| 111 : | INTERPOLATE8X8 interpolate8x8_halfpel_h_add_c; | ||
| 112 : | INTERPOLATE8X8 interpolate8x8_halfpel_v_add_c; | ||
| 113 : | INTERPOLATE8X8 interpolate8x8_halfpel_hv_add_c; | ||
| 114 : | Isibaar | 1.1 | |
| 115 : | edgomez | 1.11 | #ifdef ARCH_IS_IA32 |
| 116 : | Isibaar | 1.1 | INTERPOLATE8X8 interpolate8x8_halfpel_h_mmx; |
| 117 : | INTERPOLATE8X8 interpolate8x8_halfpel_v_mmx; | ||
| 118 : | INTERPOLATE8X8 interpolate8x8_halfpel_hv_mmx; | ||
| 119 : | |||
| 120 : | edgomez | 1.13 | INTERPOLATE8X8 interpolate8x8_halfpel_add_mmx; |
| 121 : | INTERPOLATE8X8 interpolate8x8_halfpel_h_add_mmx; | ||
| 122 : | INTERPOLATE8X8 interpolate8x8_halfpel_v_add_mmx; | ||
| 123 : | INTERPOLATE8X8 interpolate8x8_halfpel_hv_add_mmx; | ||
| 124 : | |||
| 125 : | Isibaar | 1.1 | INTERPOLATE8X8 interpolate8x8_halfpel_h_xmm; |
| 126 : | INTERPOLATE8X8 interpolate8x8_halfpel_v_xmm; | ||
| 127 : | INTERPOLATE8X8 interpolate8x8_halfpel_hv_xmm; | ||
| 128 : | |||
| 129 : | edgomez | 1.13 | INTERPOLATE8X8 interpolate8x8_halfpel_add_xmm; |
| 130 : | INTERPOLATE8X8 interpolate8x8_halfpel_h_add_xmm; | ||
| 131 : | INTERPOLATE8X8 interpolate8x8_halfpel_v_add_xmm; | ||
| 132 : | INTERPOLATE8X8 interpolate8x8_halfpel_hv_add_xmm; | ||
| 133 : | |||
| 134 : | Isibaar | 1.1 | INTERPOLATE8X8 interpolate8x8_halfpel_h_3dn; |
| 135 : | INTERPOLATE8X8 interpolate8x8_halfpel_v_3dn; | ||
| 136 : | INTERPOLATE8X8 interpolate8x8_halfpel_hv_3dn; | ||
| 137 : | |||
| 138 : | edgomez | 1.10 | INTERPOLATE8X8 interpolate8x8_halfpel_h_3dne; |
| 139 : | INTERPOLATE8X8 interpolate8x8_halfpel_v_3dne; | ||
| 140 : | INTERPOLATE8X8 interpolate8x8_halfpel_hv_3dne; | ||
| 141 : | edgomez | 1.11 | #endif |
| 142 : | edgomez | 1.10 | |
| 143 : | edgomez | 1.11 | #ifdef ARCH_IS_IA64 |
| 144 : | Isibaar | 1.4 | INTERPOLATE8X8 interpolate8x8_halfpel_h_ia64; |
| 145 : | INTERPOLATE8X8 interpolate8x8_halfpel_v_ia64; | ||
| 146 : | INTERPOLATE8X8 interpolate8x8_halfpel_hv_ia64; | ||
| 147 : | edgomez | 1.11 | #endif |
| 148 : | Isibaar | 1.4 | |
| 149 : | edgomez | 1.12 | #ifdef ARCH_IS_PPC |
| 150 : | INTERPOLATE8X8 interpolate8x8_halfpel_h_altivec_c; | ||
| 151 : | INTERPOLATE8X8 interpolate8x8_halfpel_v_altivec_c; | ||
| 152 : | INTERPOLATE8X8 interpolate8x8_halfpel_hv_altivec_c; | ||
| 153 : | #endif | ||
| 154 : | |||
| 155 : | edgomez | 1.10 | INTERPOLATE8X8_AVG2 interpolate8x8_avg2_c; |
| 156 : | INTERPOLATE8X8_AVG4 interpolate8x8_avg4_c; | ||
| 157 : | |||
| 158 : | edgomez | 1.11 | #ifdef ARCH_IS_IA32 |
| 159 : | edgomez | 1.10 | INTERPOLATE8X8_AVG2 interpolate8x8_avg2_mmx; |
| 160 : | INTERPOLATE8X8_AVG4 interpolate8x8_avg4_mmx; | ||
| 161 : | edgomez | 1.11 | #endif |
| 162 : | edgomez | 1.10 | |
| 163 : | edgomez | 1.12 | #ifdef ARCH_IS_PPC |
| 164 : | INTERPOLATE8X8_AVG2 interpolate8x8_avg2_altivec_c; | ||
| 165 : | INTERPOLATE8X8_AVG4 interpolate8x8_avg4_altivec_c; | ||
| 166 : | #endif | ||
| 167 : | |||
| 168 : | edgomez | 1.10 | INTERPOLATE_LOWPASS interpolate8x8_lowpass_h_c; |
| 169 : | INTERPOLATE_LOWPASS interpolate8x8_lowpass_v_c; | ||
| 170 : | |||
| 171 : | INTERPOLATE_LOWPASS interpolate16x16_lowpass_h_c; | ||
| 172 : | INTERPOLATE_LOWPASS interpolate16x16_lowpass_v_c; | ||
| 173 : | |||
| 174 : | INTERPOLATE_LOWPASS_HV interpolate8x8_lowpass_hv_c; | ||
| 175 : | INTERPOLATE_LOWPASS_HV interpolate16x16_lowpass_hv_c; | ||
| 176 : | |||
| 177 : | INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_h_c; | ||
| 178 : | INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_v_c; | ||
| 179 : | |||
| 180 : | edgomez | 1.11 | #ifdef ARCH_IS_IA32 |
| 181 : | edgomez | 1.10 | INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_h_mmx; |
| 182 : | INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_v_mmx; | ||
| 183 : | edgomez | 1.12 | #endif |
| 184 : | |||
| 185 : | #ifdef ARCH_IS_PPC | ||
| 186 : | INTERPOLATE8X8_6TAP_LOWPASS interpolate8x8_6tap_lowpass_h_altivec_c; | ||
| 187 : | edgomez | 1.11 | #endif |
| 188 : | Isibaar | 1.5 | |
| 189 : | edgomez | 1.3 | static __inline void |
| 190 : | interpolate8x8_switch(uint8_t * const cur, | ||
| 191 : | const uint8_t * const refn, | ||
| 192 : | const uint32_t x, | ||
| 193 : | const uint32_t y, | ||
| 194 : | const int32_t dx, | ||
| 195 : | const int dy, | ||
| 196 : | const uint32_t stride, | ||
| 197 : | const uint32_t rounding) | ||
| 198 : | Isibaar | 1.1 | { |
| 199 : | |||
| 200 : | edgomez | 1.11 | const uint8_t * const src = refn + (int)((y + (dy>>1)) * stride + x + (dx>>1)); |
| 201 : | uint8_t * const dst = cur + (int)(y * stride + x); | ||
| 202 : | |||
| 203 : | switch (((dx & 1) << 1) + (dy & 1)) { /* ((dx%2)?2:0)+((dy%2)?1:0) */ | ||
| 204 : | edgomez | 1.3 | case 0: |
| 205 : | edgomez | 1.11 | transfer8x8_copy(dst, src, stride); |
| 206 : | Isibaar | 1.1 | break; |
| 207 : | edgomez | 1.3 | case 1: |
| 208 : | edgomez | 1.11 | interpolate8x8_halfpel_v(dst, src, stride, rounding); |
| 209 : | Isibaar | 1.1 | break; |
| 210 : | edgomez | 1.3 | case 2: |
| 211 : | edgomez | 1.11 | interpolate8x8_halfpel_h(dst, src, stride, rounding); |
| 212 : | Isibaar | 1.1 | break; |
| 213 : | edgomez | 1.3 | default: |
| 214 : | edgomez | 1.11 | interpolate8x8_halfpel_hv(dst, src, stride, rounding); |
| 215 : | Isibaar | 1.1 | break; |
| 216 : | edgomez | 1.3 | } |
| 217 : | Isibaar | 1.1 | } |
| 218 : | chenm001 | 1.2 | |
| 219 : | edgomez | 1.13 | static __inline void |
| 220 : | interpolate8x8_add_switch(uint8_t * const cur, | ||
| 221 : | const uint8_t * const refn, | ||
| 222 : | const uint32_t x, | ||
| 223 : | const uint32_t y, | ||
| 224 : | const int32_t dx, | ||
| 225 : | const int dy, | ||
| 226 : | const uint32_t stride, | ||
| 227 : | const uint32_t rounding) | ||
| 228 : | { | ||
| 229 : | |||
| 230 : | const uint8_t * const src = refn + (int)((y + (dy>>1)) * stride + x + (dx>>1)); | ||
| 231 : | uint8_t * const dst = cur + (int)(y * stride + x); | ||
| 232 : | |||
| 233 : | switch (((dx & 1) << 1) + (dy & 1)) { /* ((dx%2)?2:0)+((dy%2)?1:0) */ | ||
| 234 : | case 0: | ||
| 235 : | interpolate8x8_halfpel_add(dst, src, stride, rounding); | ||
| 236 : | break; | ||
| 237 : | case 1: | ||
| 238 : | interpolate8x8_halfpel_v_add(dst, src, stride, rounding); | ||
| 239 : | break; | ||
| 240 : | case 2: | ||
| 241 : | interpolate8x8_halfpel_h_add(dst, src, stride, rounding); | ||
| 242 : | break; | ||
| 243 : | default: | ||
| 244 : | interpolate8x8_halfpel_hv_add(dst, src, stride, rounding); | ||
| 245 : | break; | ||
| 246 : | } | ||
| 247 : | } | ||
| 248 : | edgomez | 1.10 | |
| 249 : | static __inline void | ||
| 250 : | interpolate16x16_switch(uint8_t * const cur, | ||
| 251 : | const uint8_t * const refn, | ||
| 252 : | const uint32_t x, | ||
| 253 : | const uint32_t y, | ||
| 254 : | const int32_t dx, | ||
| 255 : | const int dy, | ||
| 256 : | const uint32_t stride, | ||
| 257 : | const uint32_t rounding) | ||
| 258 : | { | ||
| 259 : | interpolate8x8_switch(cur, refn, x, y, dx, dy, stride, rounding); | ||
| 260 : | interpolate8x8_switch(cur, refn, x+8, y, dx, dy, stride, rounding); | ||
| 261 : | interpolate8x8_switch(cur, refn, x, y+8, dx, dy, stride, rounding); | ||
| 262 : | interpolate8x8_switch(cur, refn, x+8, y+8, dx, dy, stride, rounding); | ||
| 263 : | } | ||
| 264 : | |||
| 265 : | edgomez | 1.13 | static __inline void |
| 266 : | interpolate16x16_add_switch(uint8_t * const cur, | ||
| 267 : | const uint8_t * const refn, | ||
| 268 : | const uint32_t x, | ||
| 269 : | const uint32_t y, | ||
| 270 : | const int32_t dx, | ||
| 271 : | const int dy, | ||
| 272 : | const uint32_t stride, | ||
| 273 : | const uint32_t rounding) | ||
| 274 : | { | ||
| 275 : | interpolate8x8_add_switch(cur, refn, x, y, dx, dy, stride, rounding); | ||
| 276 : | interpolate8x8_add_switch(cur, refn, x+8, y, dx, dy, stride, rounding); | ||
| 277 : | interpolate8x8_add_switch(cur, refn, x, y+8, dx, dy, stride, rounding); | ||
| 278 : | interpolate8x8_add_switch(cur, refn, x+8, y+8, dx, dy, stride, rounding); | ||
| 279 : | } | ||
| 280 : | edgomez | 1.10 | |
| 281 : | static __inline void | ||
| 282 : | interpolate32x32_switch(uint8_t * const cur, | ||
| 283 : | const uint8_t * const refn, | ||
| 284 : | const uint32_t x, | ||
| 285 : | const uint32_t y, | ||
| 286 : | const int32_t dx, | ||
| 287 : | const int dy, | ||
| 288 : | const uint32_t stride, | ||
| 289 : | const uint32_t rounding) | ||
| 290 : | { | ||
| 291 : | interpolate16x16_switch(cur, refn, x, y, dx, dy, stride, rounding); | ||
| 292 : | interpolate16x16_switch(cur, refn, x+16, y, dx, dy, stride, rounding); | ||
| 293 : | interpolate16x16_switch(cur, refn, x, y+16, dx, dy, stride, rounding); | ||
| 294 : | interpolate16x16_switch(cur, refn, x+16, y+16, dx, dy, stride, rounding); | ||
| 295 : | } | ||
| 296 : | |||
| 297 : | edgomez | 1.13 | static __inline void |
| 298 : | interpolate32x32_add_switch(uint8_t * const cur, | ||
| 299 : | const uint8_t * const refn, | ||
| 300 : | const uint32_t x, | ||
| 301 : | const uint32_t y, | ||
| 302 : | const int32_t dx, | ||
| 303 : | const int dy, | ||
| 304 : | const uint32_t stride, | ||
| 305 : | const uint32_t rounding) | ||
| 306 : | { | ||
| 307 : | interpolate16x16_add_switch(cur, refn, x, y, dx, dy, stride, rounding); | ||
| 308 : | interpolate16x16_add_switch(cur, refn, x+16, y, dx, dy, stride, rounding); | ||
| 309 : | interpolate16x16_add_switch(cur, refn, x, y+16, dx, dy, stride, rounding); | ||
| 310 : | interpolate16x16_add_switch(cur, refn, x+16, y+16, dx, dy, stride, rounding); | ||
| 311 : | } | ||
| 312 : | edgomez | 1.10 | |
| 313 : | static __inline uint8_t * | ||
| 314 : | interpolate8x8_switch2(uint8_t * const buffer, | ||
| 315 : | const uint8_t * const refn, | ||
| 316 : | edgomez | 1.11 | const int x, |
| 317 : | const int y, | ||
| 318 : | const int dx, | ||
| 319 : | edgomez | 1.10 | const int dy, |
| 320 : | const uint32_t stride, | ||
| 321 : | const uint32_t rounding) | ||
| 322 : | { | ||
| 323 : | |||
| 324 : | edgomez | 1.11 | const uint8_t * const src = refn + (int)((y + (dy>>1)) * stride + x + (dx>>1)); |
| 325 : | |||
| 326 : | switch (((dx & 1) << 1) + (dy & 1)) { /* ((dx%2)?2:0)+((dy%2)?1:0) */ | ||
| 327 : | edgomez | 1.10 | case 0: |
| 328 : | edgomez | 1.11 | return (uint8_t *)src; |
| 329 : | edgomez | 1.10 | case 1: |
| 330 : | edgomez | 1.11 | interpolate8x8_halfpel_v(buffer, src, stride, rounding); |
| 331 : | edgomez | 1.10 | break; |
| 332 : | case 2: | ||
| 333 : | edgomez | 1.11 | interpolate8x8_halfpel_h(buffer, src, stride, rounding); |
| 334 : | edgomez | 1.10 | break; |
| 335 : | default: | ||
| 336 : | edgomez | 1.11 | interpolate8x8_halfpel_hv(buffer, src, stride, rounding); |
| 337 : | edgomez | 1.10 | break; |
| 338 : | } | ||
| 339 : | return buffer; | ||
| 340 : | } | ||
| 341 : | |||
| 342 : | #endif |
| No admin address has been configured | ViewVC Help |
| Powered by ViewVC 1.0.4 |