2 * Copyright (C) 2010 David Conrad
3 * Copyright (C) 2010 Ronald S. Bultje
4 * Copyright (C) 2014 Peter Ross
6 * This file is part of FFmpeg.
8 * FFmpeg is free software; you can redistribute it and/or
9 * modify it under the terms of the GNU Lesser General Public
10 * License as published by the Free Software Foundation; either
11 * version 2.1 of the License, or (at your option) any later version.
13 * FFmpeg 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 GNU
16 * Lesser General Public License for more details.
18 * You should have received a copy of the GNU Lesser General Public
19 * License along with FFmpeg; if not, write to the Free Software
20 * Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
25 * VP8 compatible video decoder
30 #include "libavutil/common.h"
32 static void vp7_luma_dc_wht_c(int16_t block[4][4][16], int16_t dc[16])
34 int i, a1, b1, c1, d1;
37 for (i = 0; i < 4; i++) {
38 a1 = (dc[i*4+0] + dc[i*4+2]) * 23170;
39 b1 = (dc[i*4+0] - dc[i*4+2]) * 23170;
40 c1 = dc[i*4+1] * 12540 - dc[i*4+3] * 30274;
41 d1 = dc[i*4+1] * 30274 + dc[i*4+3] * 12540;
42 tmp[i*4+0] = (a1 + d1) >> 14;
43 tmp[i*4+3] = (a1 - d1) >> 14;
44 tmp[i*4+1] = (b1 + c1) >> 14;
45 tmp[i*4+2] = (b1 - c1) >> 14;
48 for (i = 0; i < 4; i++) {
49 a1 = (tmp[i + 0] + tmp[i + 8]) * 23170;
50 b1 = (tmp[i + 0] - tmp[i + 8]) * 23170;
51 c1 = tmp[i + 4] * 12540 - tmp[i + 12] * 30274;
52 d1 = tmp[i + 4] * 30274 + tmp[i + 12] * 12540;
53 block[0][i][0] = (a1 + d1 + 0x20000) >> 18;
54 block[3][i][0] = (a1 - d1 + 0x20000) >> 18;
55 block[1][i][0] = (b1 + c1 + 0x20000) >> 18;
56 block[2][i][0] = (b1 - c1 + 0x20000) >> 18;
60 static void vp7_luma_dc_wht_dc_c(int16_t block[4][4][16], int16_t dc[16])
62 int i, val = (23170 * (23170 * dc[0] >> 14) + 0x20000) >> 18;
65 for (i = 0; i < 4; i++) {
73 static void vp7_idct_add_c(uint8_t *dst, int16_t block[16], ptrdiff_t stride)
75 int i, a1, b1, c1, d1;
78 for (i = 0; i < 4; i++) {
79 a1 = (block[i*4+0] + block[i*4+2]) * 23170;
80 b1 = (block[i*4+0] - block[i*4+2]) * 23170;
81 c1 = block[i*4+1] * 12540 - block[i*4+3] * 30274;
82 d1 = block[i*4+1] * 30274 + block[i*4+3] * 12540;
83 tmp[i*4+0] = (a1 + d1) >> 14;
84 tmp[i*4+3] = (a1 - d1) >> 14;
85 tmp[i*4+1] = (b1 + c1) >> 14;
86 tmp[i*4+2] = (b1 - c1) >> 14;
89 for (i = 0; i < 4; i++) {
90 a1 = (tmp[i + 0] + tmp[i + 8]) * 23170;
91 b1 = (tmp[i + 0] - tmp[i + 8]) * 23170;
92 c1 = tmp[i + 4] * 12540 - tmp[i + 12] * 30274;
93 d1 = tmp[i + 4] * 30274 + tmp[i + 12] * 12540;
94 dst[0*stride+i] = av_clip_uint8(dst[0*stride+i] + ((a1 + d1 + 0x20000) >> 18));
95 dst[3*stride+i] = av_clip_uint8(dst[3*stride+i] + ((a1 - d1 + 0x20000) >> 18));
96 dst[1*stride+i] = av_clip_uint8(dst[1*stride+i] + ((b1 + c1 + 0x20000) >> 18));
97 dst[2*stride+i] = av_clip_uint8(dst[2*stride+i] + ((b1 - c1 + 0x20000) >> 18));
101 static void vp7_idct_dc_add_c(uint8_t *dst, int16_t block[16], ptrdiff_t stride)
103 int i, dc = (23170 * (23170 * block[0] >> 14) + 0x20000) >> 18;
106 for (i = 0; i < 4; i++) {
107 dst[0] = av_clip_uint8(dst[0] + dc);
108 dst[1] = av_clip_uint8(dst[1] + dc);
109 dst[2] = av_clip_uint8(dst[2] + dc);
110 dst[3] = av_clip_uint8(dst[3] + dc);
115 // TODO: Maybe add dequant
116 static void vp8_luma_dc_wht_c(int16_t block[4][4][16], int16_t dc[16])
118 int i, t0, t1, t2, t3;
120 for (i = 0; i < 4; i++) {
121 t0 = dc[0*4+i] + dc[3*4+i];
122 t1 = dc[1*4+i] + dc[2*4+i];
123 t2 = dc[1*4+i] - dc[2*4+i];
124 t3 = dc[0*4+i] - dc[3*4+i];
132 for (i = 0; i < 4; i++) {
133 t0 = dc[i*4+0] + dc[i*4+3] + 3; // rounding
134 t1 = dc[i*4+1] + dc[i*4+2];
135 t2 = dc[i*4+1] - dc[i*4+2];
136 t3 = dc[i*4+0] - dc[i*4+3] + 3; // rounding
142 block[i][0][0] = (t0 + t1) >> 3;
143 block[i][1][0] = (t3 + t2) >> 3;
144 block[i][2][0] = (t0 - t1) >> 3;
145 block[i][3][0] = (t3 - t2) >> 3;
149 static void vp8_luma_dc_wht_dc_c(int16_t block[4][4][16], int16_t dc[16])
151 int i, val = (dc[0] + 3) >> 3;
154 for (i = 0; i < 4; i++) {
155 block[i][0][0] = val;
156 block[i][1][0] = val;
157 block[i][2][0] = val;
158 block[i][3][0] = val;
162 #define MUL_20091(a) ((((a)*20091) >> 16) + (a))
163 #define MUL_35468(a) (((a)*35468) >> 16)
165 static void vp8_idct_add_c(uint8_t *dst, int16_t block[16], ptrdiff_t stride)
167 int i, t0, t1, t2, t3;
170 for (i = 0; i < 4; i++) {
171 t0 = block[0*4+i] + block[2*4+i];
172 t1 = block[0*4+i] - block[2*4+i];
173 t2 = MUL_35468(block[1*4+i]) - MUL_20091(block[3*4+i]);
174 t3 = MUL_20091(block[1*4+i]) + MUL_35468(block[3*4+i]);
180 tmp[i*4+0] = t0 + t3;
181 tmp[i*4+1] = t1 + t2;
182 tmp[i*4+2] = t1 - t2;
183 tmp[i*4+3] = t0 - t3;
186 for (i = 0; i < 4; i++) {
187 t0 = tmp[0*4+i] + tmp[2*4+i];
188 t1 = tmp[0*4+i] - tmp[2*4+i];
189 t2 = MUL_35468(tmp[1*4+i]) - MUL_20091(tmp[3*4+i]);
190 t3 = MUL_20091(tmp[1*4+i]) + MUL_35468(tmp[3*4+i]);
192 dst[0] = av_clip_uint8(dst[0] + ((t0 + t3 + 4) >> 3));
193 dst[1] = av_clip_uint8(dst[1] + ((t1 + t2 + 4) >> 3));
194 dst[2] = av_clip_uint8(dst[2] + ((t1 - t2 + 4) >> 3));
195 dst[3] = av_clip_uint8(dst[3] + ((t0 - t3 + 4) >> 3));
200 static void vp8_idct_dc_add_c(uint8_t *dst, int16_t block[16], ptrdiff_t stride)
202 int i, dc = (block[0] + 4) >> 3;
205 for (i = 0; i < 4; i++) {
206 dst[0] = av_clip_uint8(dst[0] + dc);
207 dst[1] = av_clip_uint8(dst[1] + dc);
208 dst[2] = av_clip_uint8(dst[2] + dc);
209 dst[3] = av_clip_uint8(dst[3] + dc);
214 #define MK_IDCT_DC_ADD4_C(name) \
215 static void name ## _idct_dc_add4uv_c(uint8_t *dst, int16_t block[4][16], ptrdiff_t stride)\
217 name ## _idct_dc_add_c(dst+stride*0+0, block[0], stride);\
218 name ## _idct_dc_add_c(dst+stride*0+4, block[1], stride);\
219 name ## _idct_dc_add_c(dst+stride*4+0, block[2], stride);\
220 name ## _idct_dc_add_c(dst+stride*4+4, block[3], stride);\
223 static void name ## _idct_dc_add4y_c(uint8_t *dst, int16_t block[4][16], ptrdiff_t stride)\
225 name ## _idct_dc_add_c(dst+ 0, block[0], stride);\
226 name ## _idct_dc_add_c(dst+ 4, block[1], stride);\
227 name ## _idct_dc_add_c(dst+ 8, block[2], stride);\
228 name ## _idct_dc_add_c(dst+12, block[3], stride);\
231 MK_IDCT_DC_ADD4_C(vp7)
232 MK_IDCT_DC_ADD4_C(vp8)
234 // because I like only having two parameters to pass functions...
236 int av_unused p3 = p[-4*stride];\
237 int av_unused p2 = p[-3*stride];\
238 int av_unused p1 = p[-2*stride];\
239 int av_unused p0 = p[-1*stride];\
240 int av_unused q0 = p[ 0*stride];\
241 int av_unused q1 = p[ 1*stride];\
242 int av_unused q2 = p[ 2*stride];\
243 int av_unused q3 = p[ 3*stride];
245 #define clip_int8(n) (cm[n+0x80]-0x80)
247 static av_always_inline void filter_common(uint8_t *p, ptrdiff_t stride, int is4tap, int vpn)
251 const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
256 a += clip_int8(p1 - q1);
260 // We deviate from the spec here with c(a+3) >> 3
261 // since that's what libvpx does.
262 f1 = FFMIN(a+4, 127) >> 3;
265 f2 = f1 - ((a & 7) == 4);
267 f2 = FFMIN(a+3, 127) >> 3;
269 // Despite what the spec says, we do need to clamp here to
270 // be bitexact with libvpx.
271 p[-1*stride] = cm[p0 + f2];
272 p[ 0*stride] = cm[q0 - f1];
274 // only used for _inner on blocks without high edge variance
277 p[-2*stride] = cm[p1 + a];
278 p[ 1*stride] = cm[q1 - a];
282 static av_always_inline int vp7_simple_limit(uint8_t *p, ptrdiff_t stride, int flim)
285 return FFABS(p0-q0) <= flim;
288 static av_always_inline int vp8_simple_limit(uint8_t *p, ptrdiff_t stride, int flim)
291 return 2*FFABS(p0-q0) + (FFABS(p1-q1) >> 1) <= flim;
295 * E - limit at the macroblock edge
296 * I - limit for interior difference
298 #define NORMAL_LIMIT(vpn) \
299 static av_always_inline int vp ## vpn ## _normal_limit(uint8_t *p, ptrdiff_t stride, int E, int I)\
302 return vp ## vpn ## _simple_limit(p, stride, E)\
303 && FFABS(p3-p2) <= I && FFABS(p2-p1) <= I && FFABS(p1-p0) <= I\
304 && FFABS(q3-q2) <= I && FFABS(q2-q1) <= I && FFABS(q1-q0) <= I;\
310 // high edge variance
311 static av_always_inline int hev(uint8_t *p, ptrdiff_t stride, int thresh)
314 return FFABS(p1-p0) > thresh || FFABS(q1-q0) > thresh;
317 static av_always_inline void filter_mbedge(uint8_t *p, ptrdiff_t stride)
320 const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP;
324 w = clip_int8(p1-q1);
325 w = clip_int8(w + 3*(q0-p0));
327 a0 = (27*w + 63) >> 7;
328 a1 = (18*w + 63) >> 7;
329 a2 = ( 9*w + 63) >> 7;
331 p[-3*stride] = cm[p2 + a2];
332 p[-2*stride] = cm[p1 + a1];
333 p[-1*stride] = cm[p0 + a0];
334 p[ 0*stride] = cm[q0 - a0];
335 p[ 1*stride] = cm[q1 - a1];
336 p[ 2*stride] = cm[q2 - a2];
339 #define LOOP_FILTER(vpn, dir, size, stridea, strideb, maybe_inline) \
340 static maybe_inline void vp ## vpn ## _ ## dir ## _loop_filter ## size ## _c(uint8_t *dst, ptrdiff_t stride,\
341 int flim_E, int flim_I, int hev_thresh)\
345 for (i = 0; i < size; i++)\
346 if (vp ## vpn ## _normal_limit(dst+i*stridea, strideb, flim_E, flim_I)) {\
347 if (hev(dst+i*stridea, strideb, hev_thresh))\
348 filter_common(dst+i*stridea, strideb, 1, vpn);\
350 filter_mbedge(dst+i*stridea, strideb);\
354 static maybe_inline void vp ## vpn ## _ ## dir ## _loop_filter ## size ## _inner_c(uint8_t *dst, ptrdiff_t stride,\
355 int flim_E, int flim_I, int hev_thresh)\
359 for (i = 0; i < size; i++)\
360 if (vp ## vpn ## _normal_limit(dst+i*stridea, strideb, flim_E, flim_I)) {\
361 int hv = hev(dst+i*stridea, strideb, hev_thresh);\
363 filter_common(dst+i*stridea, strideb, 1, vpn);\
365 filter_common(dst+i*stridea, strideb, 0, vpn);\
369 LOOP_FILTER(7, v, 16, 1, stride,)
370 LOOP_FILTER(7, h, 16, stride, 1,)
372 LOOP_FILTER(8, v, 16, 1, stride,)
373 LOOP_FILTER(8, h, 16, stride, 1,)
375 #define UV_LOOP_FILTER(vpn, dir, stridea, strideb) \
376 LOOP_FILTER(vpn, dir, 8, stridea, strideb, av_always_inline) \
377 static void vp ## vpn ## _ ## dir ## _loop_filter8uv_c(uint8_t *dstU, uint8_t *dstV, ptrdiff_t stride,\
378 int fE, int fI, int hev_thresh)\
380 vp ## vpn ## _ ## dir ## _loop_filter8_c(dstU, stride, fE, fI, hev_thresh);\
381 vp ## vpn ## _ ## dir ## _loop_filter8_c(dstV, stride, fE, fI, hev_thresh);\
383 static void vp ## vpn ## _ ## dir ## _loop_filter8uv_inner_c(uint8_t *dstU, uint8_t *dstV, ptrdiff_t stride,\
384 int fE, int fI, int hev_thresh)\
386 vp ## vpn ## _ ## dir ## _loop_filter8_inner_c(dstU, stride, fE, fI, hev_thresh);\
387 vp ## vpn ## _ ## dir ## _loop_filter8_inner_c(dstV, stride, fE, fI, hev_thresh);\
390 UV_LOOP_FILTER(7, v, 1, stride)
391 UV_LOOP_FILTER(7, h, stride, 1)
393 UV_LOOP_FILTER(8, v, 1, stride)
394 UV_LOOP_FILTER(8, h, stride, 1)
396 #define LOOP_FILTER_SIMPLE(vpn) \
397 static void vp ## vpn ## _v_loop_filter_simple_c(uint8_t *dst, ptrdiff_t stride, int flim)\
401 for (i = 0; i < 16; i++)\
402 if (vp ## vpn ## _simple_limit(dst+i, stride, flim))\
403 filter_common(dst+i, stride, 1, vpn);\
406 static void vp ## vpn ## _h_loop_filter_simple_c(uint8_t *dst, ptrdiff_t stride, int flim)\
410 for (i = 0; i < 16; i++)\
411 if (vp ## vpn ## _simple_limit(dst+i*stride, 1, flim))\
412 filter_common(dst+i*stride, 1, 1, vpn);\
415 LOOP_FILTER_SIMPLE(7)
416 LOOP_FILTER_SIMPLE(8)
418 static const uint8_t subpel_filters[7][6] = {
419 { 0, 6, 123, 12, 1, 0 },
420 { 2, 11, 108, 36, 8, 1 },
421 { 0, 9, 93, 50, 6, 0 },
422 { 3, 16, 77, 77, 16, 3 },
423 { 0, 6, 50, 93, 9, 0 },
424 { 1, 8, 36, 108, 11, 2 },
425 { 0, 1, 12, 123, 6, 0 },
428 #define PUT_PIXELS(WIDTH) \
429 static void put_vp8_pixels ## WIDTH ##_c(uint8_t *dst, ptrdiff_t dststride, uint8_t *src, ptrdiff_t srcstride, int h, int x, int y) { \
431 for (i = 0; i < h; i++, dst+= dststride, src+= srcstride) { \
432 memcpy(dst, src, WIDTH); \
440 #define FILTER_6TAP(src, F, stride) \
441 cm[(F[2]*src[x+0*stride] - F[1]*src[x-1*stride] + F[0]*src[x-2*stride] + \
442 F[3]*src[x+1*stride] - F[4]*src[x+2*stride] + F[5]*src[x+3*stride] + 64) >> 7]
444 #define FILTER_4TAP(src, F, stride) \
445 cm[(F[2]*src[x+0*stride] - F[1]*src[x-1*stride] + \
446 F[3]*src[x+1*stride] - F[4]*src[x+2*stride] + 64) >> 7]
448 #define VP8_EPEL_H(SIZE, TAPS) \
449 static void put_vp8_epel ## SIZE ## _h ## TAPS ## _c(uint8_t *dst, ptrdiff_t dststride, uint8_t *src, ptrdiff_t srcstride, int h, int mx, int my) \
451 const uint8_t *filter = subpel_filters[mx-1]; \
452 const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; \
455 for (y = 0; y < h; y++) { \
456 for (x = 0; x < SIZE; x++) \
457 dst[x] = FILTER_ ## TAPS ## TAP(src, filter, 1); \
462 #define VP8_EPEL_V(SIZE, TAPS) \
463 static void put_vp8_epel ## SIZE ## _v ## TAPS ## _c(uint8_t *dst, ptrdiff_t dststride, uint8_t *src, ptrdiff_t srcstride, int h, int mx, int my) \
465 const uint8_t *filter = subpel_filters[my-1]; \
466 const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; \
469 for (y = 0; y < h; y++) { \
470 for (x = 0; x < SIZE; x++) \
471 dst[x] = FILTER_ ## TAPS ## TAP(src, filter, srcstride); \
476 #define VP8_EPEL_HV(SIZE, HTAPS, VTAPS) \
477 static void put_vp8_epel ## SIZE ## _h ## HTAPS ## v ## VTAPS ## _c(uint8_t *dst, ptrdiff_t dststride, uint8_t *src, ptrdiff_t srcstride, int h, int mx, int my) \
479 const uint8_t *filter = subpel_filters[mx-1]; \
480 const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; \
482 uint8_t tmp_array[(2*SIZE+VTAPS-1)*SIZE]; \
483 uint8_t *tmp = tmp_array; \
484 src -= (2-(VTAPS==4))*srcstride; \
486 for (y = 0; y < h+VTAPS-1; y++) { \
487 for (x = 0; x < SIZE; x++) \
488 tmp[x] = FILTER_ ## HTAPS ## TAP(src, filter, 1); \
493 tmp = tmp_array + (2-(VTAPS==4))*SIZE; \
494 filter = subpel_filters[my-1]; \
496 for (y = 0; y < h; y++) { \
497 for (x = 0; x < SIZE; x++) \
498 dst[x] = FILTER_ ## VTAPS ## TAP(tmp, filter, SIZE); \
516 VP8_EPEL_HV(16, 4, 4)
519 VP8_EPEL_HV(16, 4, 6)
522 VP8_EPEL_HV(16, 6, 4)
525 VP8_EPEL_HV(16, 6, 6)
529 #define VP8_BILINEAR(SIZE) \
530 static void put_vp8_bilinear ## SIZE ## _h_c(uint8_t *dst, ptrdiff_t dstride, uint8_t *src, ptrdiff_t sstride, int h, int mx, int my) \
532 int a = 8-mx, b = mx; \
535 for (y = 0; y < h; y++) { \
536 for (x = 0; x < SIZE; x++) \
537 dst[x] = (a*src[x] + b*src[x+1] + 4) >> 3; \
542 static void put_vp8_bilinear ## SIZE ## _v_c(uint8_t *dst, ptrdiff_t dstride, uint8_t *src, ptrdiff_t sstride, int h, int mx, int my) \
544 int c = 8-my, d = my; \
547 for (y = 0; y < h; y++) { \
548 for (x = 0; x < SIZE; x++) \
549 dst[x] = (c*src[x] + d*src[x+sstride] + 4) >> 3; \
555 static void put_vp8_bilinear ## SIZE ## _hv_c(uint8_t *dst, ptrdiff_t dstride, uint8_t *src, ptrdiff_t sstride, int h, int mx, int my) \
557 int a = 8-mx, b = mx; \
558 int c = 8-my, d = my; \
560 uint8_t tmp_array[(2*SIZE+1)*SIZE]; \
561 uint8_t *tmp = tmp_array; \
563 for (y = 0; y < h+1; y++) { \
564 for (x = 0; x < SIZE; x++) \
565 tmp[x] = (a*src[x] + b*src[x+1] + 4) >> 3; \
572 for (y = 0; y < h; y++) { \
573 for (x = 0; x < SIZE; x++) \
574 dst[x] = (c*tmp[x] + d*tmp[x+SIZE] + 4) >> 3; \
584 #define VP8_MC_FUNC(IDX, SIZE) \
585 dsp->put_vp8_epel_pixels_tab[IDX][0][0] = put_vp8_pixels ## SIZE ## _c; \
586 dsp->put_vp8_epel_pixels_tab[IDX][0][1] = put_vp8_epel ## SIZE ## _h4_c; \
587 dsp->put_vp8_epel_pixels_tab[IDX][0][2] = put_vp8_epel ## SIZE ## _h6_c; \
588 dsp->put_vp8_epel_pixels_tab[IDX][1][0] = put_vp8_epel ## SIZE ## _v4_c; \
589 dsp->put_vp8_epel_pixels_tab[IDX][1][1] = put_vp8_epel ## SIZE ## _h4v4_c; \
590 dsp->put_vp8_epel_pixels_tab[IDX][1][2] = put_vp8_epel ## SIZE ## _h6v4_c; \
591 dsp->put_vp8_epel_pixels_tab[IDX][2][0] = put_vp8_epel ## SIZE ## _v6_c; \
592 dsp->put_vp8_epel_pixels_tab[IDX][2][1] = put_vp8_epel ## SIZE ## _h4v6_c; \
593 dsp->put_vp8_epel_pixels_tab[IDX][2][2] = put_vp8_epel ## SIZE ## _h6v6_c
595 #define VP8_BILINEAR_MC_FUNC(IDX, SIZE) \
596 dsp->put_vp8_bilinear_pixels_tab[IDX][0][0] = put_vp8_pixels ## SIZE ## _c; \
597 dsp->put_vp8_bilinear_pixels_tab[IDX][0][1] = put_vp8_bilinear ## SIZE ## _h_c; \
598 dsp->put_vp8_bilinear_pixels_tab[IDX][0][2] = put_vp8_bilinear ## SIZE ## _h_c; \
599 dsp->put_vp8_bilinear_pixels_tab[IDX][1][0] = put_vp8_bilinear ## SIZE ## _v_c; \
600 dsp->put_vp8_bilinear_pixels_tab[IDX][1][1] = put_vp8_bilinear ## SIZE ## _hv_c; \
601 dsp->put_vp8_bilinear_pixels_tab[IDX][1][2] = put_vp8_bilinear ## SIZE ## _hv_c; \
602 dsp->put_vp8_bilinear_pixels_tab[IDX][2][0] = put_vp8_bilinear ## SIZE ## _v_c; \
603 dsp->put_vp8_bilinear_pixels_tab[IDX][2][1] = put_vp8_bilinear ## SIZE ## _hv_c; \
604 dsp->put_vp8_bilinear_pixels_tab[IDX][2][2] = put_vp8_bilinear ## SIZE ## _hv_c
606 av_cold void ff_vp8dsp_init(VP8DSPContext *dsp, int vp7)
608 #define VPX(f) vp7 ? vp7 ## _ ## f : vp8 ## _ ## f
610 dsp->vp8_luma_dc_wht = VPX(luma_dc_wht_c);
611 dsp->vp8_luma_dc_wht_dc = VPX(luma_dc_wht_dc_c);
612 dsp->vp8_idct_add = VPX(idct_add_c);
613 dsp->vp8_idct_dc_add = VPX(idct_dc_add_c);
614 dsp->vp8_idct_dc_add4y = VPX(idct_dc_add4y_c);
615 dsp->vp8_idct_dc_add4uv = VPX(idct_dc_add4uv_c);
617 dsp->vp8_v_loop_filter16y = VPX(v_loop_filter16_c);
618 dsp->vp8_h_loop_filter16y = VPX(h_loop_filter16_c);
619 dsp->vp8_v_loop_filter8uv = VPX(v_loop_filter8uv_c);
620 dsp->vp8_h_loop_filter8uv = VPX(h_loop_filter8uv_c);
622 dsp->vp8_v_loop_filter16y_inner = VPX(v_loop_filter16_inner_c);
623 dsp->vp8_h_loop_filter16y_inner = VPX(h_loop_filter16_inner_c);
624 dsp->vp8_v_loop_filter8uv_inner = VPX(v_loop_filter8uv_inner_c);
625 dsp->vp8_h_loop_filter8uv_inner = VPX(h_loop_filter8uv_inner_c);
627 dsp->vp8_v_loop_filter_simple = VPX(v_loop_filter_simple_c);
628 dsp->vp8_h_loop_filter_simple = VPX(h_loop_filter_simple_c);
634 VP8_BILINEAR_MC_FUNC(0, 16);
635 VP8_BILINEAR_MC_FUNC(1, 8);
636 VP8_BILINEAR_MC_FUNC(2, 4);
639 ff_vp8dsp_init_arm(dsp, vp7);
641 ff_vp8dsp_init_ppc(dsp);
643 ff_vp8dsp_init_x86(dsp, vp7);