36 for (i = 0; i < 4; i++) {
37 t0 = dc[0*4+i] + dc[3*4+i];
38 t1 = dc[1*4+i] + dc[2*4+i];
39 t2 = dc[1*4+i] - dc[2*4+i];
40 t3 = dc[0*4+i] - dc[3*4+i];
48 for (i = 0; i < 4; i++) {
49 t0 = dc[i*4+0] + dc[i*4+3] + 3;
50 t1 = dc[i*4+1] + dc[i*4+2];
51 t2 = dc[i*4+1] - dc[i*4+2];
52 t3 = dc[i*4+0] - dc[i*4+3] + 3;
58 block[i][0][0] = (t0 +
t1) >> 3;
59 block[i][1][0] = (t3 +
t2) >> 3;
60 block[i][2][0] = (t0 -
t1) >> 3;
61 block[i][3][0] = (t3 -
t2) >> 3;
67 int i,
val = (dc[0] + 3) >> 3;
70 for (i = 0; i < 4; i++) {
78 #define MUL_20091(a) ((((a)*20091) >> 16) + (a))
79 #define MUL_35468(a) (((a)*35468) >> 16)
86 for (i = 0; i < 4; i++) {
87 t0 = block[0*4+i] + block[2*4+i];
88 t1 = block[0*4+i] - block[2*4+i];
102 for (i = 0; i < 4; i++) {
103 t0 = tmp[0*4+i] + tmp[2*4+i];
104 t1 = tmp[0*4+i] - tmp[2*4+i];
108 dst[0] = av_clip_uint8(dst[0] + ((t0 + t3 + 4) >> 3));
109 dst[1] = av_clip_uint8(dst[1] + ((t1 + t2 + 4) >> 3));
110 dst[2] = av_clip_uint8(dst[2] + ((t1 - t2 + 4) >> 3));
111 dst[3] = av_clip_uint8(dst[3] + ((t0 - t3 + 4) >> 3));
118 int i,
dc = (block[0] + 4) >> 3;
121 for (i = 0; i < 4; i++) {
122 dst[0] = av_clip_uint8(dst[0] + dc);
123 dst[1] = av_clip_uint8(dst[1] + dc);
124 dst[2] = av_clip_uint8(dst[2] + dc);
125 dst[3] = av_clip_uint8(dst[3] + dc);
148 int av_unused p3 = p[-4*stride];\
149 int av_unused p2 = p[-3*stride];\
150 int av_unused p1 = p[-2*stride];\
151 int av_unused p0 = p[-1*stride];\
152 int av_unused q0 = p[ 0*stride];\
153 int av_unused q1 = p[ 1*stride];\
154 int av_unused q2 = p[ 2*stride];\
155 int av_unused q3 = p[ 3*stride];
157 #define clip_int8(n) (cm[n+0x80]-0x80)
174 f1 =
FFMIN(a+4, 127) >> 3;
175 f2 =
FFMIN(a+3, 127) >> 3;
179 p[-1*
stride] = cm[p0 + f2];
180 p[ 0*
stride] = cm[q0 - f1];
193 return 2*
FFABS(p0-q0) + (
FFABS(p1-q1) >> 1) <= flim;
212 return FFABS(p1-p0) > thresh ||
FFABS(q1-q0) > thresh;
225 a0 = (27*w + 63) >> 7;
226 a1 = (18*w + 63) >> 7;
227 a2 = ( 9*w + 63) >> 7;
237 #define LOOP_FILTER(dir, size, stridea, strideb, maybe_inline) \
238 static maybe_inline void vp8_ ## dir ## _loop_filter ## size ## _c(uint8_t *dst, ptrdiff_t stride,\
239 int flim_E, int flim_I, int hev_thresh)\
243 for (i = 0; i < size; i++)\
244 if (normal_limit(dst+i*stridea, strideb, flim_E, flim_I)) {\
245 if (hev(dst+i*stridea, strideb, hev_thresh))\
246 filter_common(dst+i*stridea, strideb, 1);\
248 filter_mbedge(dst+i*stridea, strideb);\
252 static maybe_inline void vp8_ ## dir ## _loop_filter ## size ## _inner_c(uint8_t *dst, ptrdiff_t stride,\
253 int flim_E, int flim_I, int hev_thresh)\
257 for (i = 0; i < size; i++)\
258 if (normal_limit(dst+i*stridea, strideb, flim_E, flim_I)) {\
259 int hv = hev(dst+i*stridea, strideb, hev_thresh);\
261 filter_common(dst+i*stridea, strideb, 1);\
263 filter_common(dst+i*stridea, strideb, 0);\
270 #define UV_LOOP_FILTER(dir, stridea, strideb) \
271 LOOP_FILTER(dir, 8, stridea, strideb, av_always_inline) \
272 static void vp8_ ## dir ## _loop_filter8uv_c(uint8_t *dstU, uint8_t *dstV, ptrdiff_t stride,\
273 int fE, int fI, int hev_thresh)\
275 vp8_ ## dir ## _loop_filter8_c(dstU, stride, fE, fI, hev_thresh);\
276 vp8_ ## dir ## _loop_filter8_c(dstV, stride, fE, fI, hev_thresh);\
278 static void vp8_ ## dir ## _loop_filter8uv_inner_c(uint8_t *dstU, uint8_t *dstV, ptrdiff_t stride,\
279 int fE, int fI, int hev_thresh)\
281 vp8_ ## dir ## _loop_filter8_inner_c(dstU, stride, fE, fI, hev_thresh);\
282 vp8_ ## dir ## _loop_filter8_inner_c(dstV, stride, fE, fI, hev_thresh);\
292 for (i = 0; i < 16; i++)
301 for (i = 0; i < 16; i++)
307 { 0, 6, 123, 12, 1, 0 },
308 { 2, 11, 108, 36, 8, 1 },
309 { 0, 9, 93, 50, 6, 0 },
310 { 3, 16, 77, 77, 16, 3 },
311 { 0, 6, 50, 93, 9, 0 },
312 { 1, 8, 36, 108, 11, 2 },
313 { 0, 1, 12, 123, 6, 0 },
316 #define PUT_PIXELS(WIDTH) \
317 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) { \
319 for (i = 0; i < h; i++, dst+= dststride, src+= srcstride) { \
320 memcpy(dst, src, WIDTH); \
328 #define FILTER_6TAP(src, F, stride) \
329 cm[(F[2]*src[x+0*stride] - F[1]*src[x-1*stride] + F[0]*src[x-2*stride] + \
330 F[3]*src[x+1*stride] - F[4]*src[x+2*stride] + F[5]*src[x+3*stride] + 64) >> 7]
332 #define FILTER_4TAP(src, F, stride) \
333 cm[(F[2]*src[x+0*stride] - F[1]*src[x-1*stride] + \
334 F[3]*src[x+1*stride] - F[4]*src[x+2*stride] + 64) >> 7]
336 #define VP8_EPEL_H(SIZE, TAPS) \
337 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) \
339 const uint8_t *filter = subpel_filters[mx-1]; \
340 const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; \
343 for (y = 0; y < h; y++) { \
344 for (x = 0; x < SIZE; x++) \
345 dst[x] = FILTER_ ## TAPS ## TAP(src, filter, 1); \
350 #define VP8_EPEL_V(SIZE, TAPS) \
351 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) \
353 const uint8_t *filter = subpel_filters[my-1]; \
354 const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; \
357 for (y = 0; y < h; y++) { \
358 for (x = 0; x < SIZE; x++) \
359 dst[x] = FILTER_ ## TAPS ## TAP(src, filter, srcstride); \
364 #define VP8_EPEL_HV(SIZE, HTAPS, VTAPS) \
365 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) \
367 const uint8_t *filter = subpel_filters[mx-1]; \
368 const uint8_t *cm = ff_cropTbl + MAX_NEG_CROP; \
370 uint8_t tmp_array[(2*SIZE+VTAPS-1)*SIZE]; \
371 uint8_t *tmp = tmp_array; \
372 src -= (2-(VTAPS==4))*srcstride; \
374 for (y = 0; y < h+VTAPS-1; y++) { \
375 for (x = 0; x < SIZE; x++) \
376 tmp[x] = FILTER_ ## HTAPS ## TAP(src, filter, 1); \
381 tmp = tmp_array + (2-(VTAPS==4))*SIZE; \
382 filter = subpel_filters[my-1]; \
384 for (y = 0; y < h; y++) { \
385 for (x = 0; x < SIZE; x++) \
386 dst[x] = FILTER_ ## VTAPS ## TAP(tmp, filter, SIZE); \
417 #define VP8_BILINEAR(SIZE) \
418 static void put_vp8_bilinear ## SIZE ## _h_c(uint8_t *dst, ptrdiff_t stride, uint8_t *src, ptrdiff_t s2, int h, int mx, int my) \
420 int a = 8-mx, b = mx; \
423 for (y = 0; y < h; y++) { \
424 for (x = 0; x < SIZE; x++) \
425 dst[x] = (a*src[x] + b*src[x+1] + 4) >> 3; \
430 static void put_vp8_bilinear ## SIZE ## _v_c(uint8_t *dst, ptrdiff_t stride, uint8_t *src, ptrdiff_t s2, int h, int mx, int my) \
432 int c = 8-my, d = my; \
435 for (y = 0; y < h; y++) { \
436 for (x = 0; x < SIZE; x++) \
437 dst[x] = (c*src[x] + d*src[x+stride] + 4) >> 3; \
443 static void put_vp8_bilinear ## SIZE ## _hv_c(uint8_t *dst, ptrdiff_t stride, uint8_t *src, ptrdiff_t s2, int h, int mx, int my) \
445 int a = 8-mx, b = mx; \
446 int c = 8-my, d = my; \
448 uint8_t tmp_array[(2*SIZE+1)*SIZE]; \
449 uint8_t *tmp = tmp_array; \
451 for (y = 0; y < h+1; y++) { \
452 for (x = 0; x < SIZE; x++) \
453 tmp[x] = (a*src[x] + b*src[x+1] + 4) >> 3; \
460 for (y = 0; y < h; y++) { \
461 for (x = 0; x < SIZE; x++) \
462 dst[x] = (c*tmp[x] + d*tmp[x+SIZE] + 4) >> 3; \
472 #define VP8_MC_FUNC(IDX, SIZE) \
473 dsp->put_vp8_epel_pixels_tab[IDX][0][0] = put_vp8_pixels ## SIZE ## _c; \
474 dsp->put_vp8_epel_pixels_tab[IDX][0][1] = put_vp8_epel ## SIZE ## _h4_c; \
475 dsp->put_vp8_epel_pixels_tab[IDX][0][2] = put_vp8_epel ## SIZE ## _h6_c; \
476 dsp->put_vp8_epel_pixels_tab[IDX][1][0] = put_vp8_epel ## SIZE ## _v4_c; \
477 dsp->put_vp8_epel_pixels_tab[IDX][1][1] = put_vp8_epel ## SIZE ## _h4v4_c; \
478 dsp->put_vp8_epel_pixels_tab[IDX][1][2] = put_vp8_epel ## SIZE ## _h6v4_c; \
479 dsp->put_vp8_epel_pixels_tab[IDX][2][0] = put_vp8_epel ## SIZE ## _v6_c; \
480 dsp->put_vp8_epel_pixels_tab[IDX][2][1] = put_vp8_epel ## SIZE ## _h4v6_c; \
481 dsp->put_vp8_epel_pixels_tab[IDX][2][2] = put_vp8_epel ## SIZE ## _h6v6_c
483 #define VP8_BILINEAR_MC_FUNC(IDX, SIZE) \
484 dsp->put_vp8_bilinear_pixels_tab[IDX][0][0] = put_vp8_pixels ## SIZE ## _c; \
485 dsp->put_vp8_bilinear_pixels_tab[IDX][0][1] = put_vp8_bilinear ## SIZE ## _h_c; \
486 dsp->put_vp8_bilinear_pixels_tab[IDX][0][2] = put_vp8_bilinear ## SIZE ## _h_c; \
487 dsp->put_vp8_bilinear_pixels_tab[IDX][1][0] = put_vp8_bilinear ## SIZE ## _v_c; \
488 dsp->put_vp8_bilinear_pixels_tab[IDX][1][1] = put_vp8_bilinear ## SIZE ## _hv_c; \
489 dsp->put_vp8_bilinear_pixels_tab[IDX][1][2] = put_vp8_bilinear ## SIZE ## _hv_c; \
490 dsp->put_vp8_bilinear_pixels_tab[IDX][2][0] = put_vp8_bilinear ## SIZE ## _v_c; \
491 dsp->put_vp8_bilinear_pixels_tab[IDX][2][1] = put_vp8_bilinear ## SIZE ## _hv_c; \
492 dsp->put_vp8_bilinear_pixels_tab[IDX][2][2] = put_vp8_bilinear ## SIZE ## _hv_c