4 * Copyright (C) 2012 - 2013 Guillaume Martres
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
26 #include "bit_depth_template.c"
30 static void FUNC(put_pcm)(uint8_t *_dst, ptrdiff_t stride, int width, int height,
31 GetBitContext *gb, int pcm_bit_depth)
34 pixel *dst = (pixel *)_dst;
36 stride /= sizeof(pixel);
38 for (y = 0; y < height; y++) {
39 for (x = 0; x < width; x++)
40 dst[x] = get_bits(gb, pcm_bit_depth) << (BIT_DEPTH - pcm_bit_depth);
45 static void FUNC(transform_add4x4)(uint8_t *_dst, int16_t *coeffs,
49 pixel *dst = (pixel *)_dst;
51 stride /= sizeof(pixel);
53 for (y = 0; y < 4; y++) {
54 for (x = 0; x < 4; x++) {
55 dst[x] = av_clip_pixel(dst[x] + *coeffs);
62 static void FUNC(transform_add8x8)(uint8_t *_dst, int16_t *coeffs,
66 pixel *dst = (pixel *)_dst;
68 stride /= sizeof(pixel);
70 for (y = 0; y < 8; y++) {
71 for (x = 0; x < 8; x++) {
72 dst[x] = av_clip_pixel(dst[x] + *coeffs);
79 static void FUNC(transform_add16x16)(uint8_t *_dst, int16_t *coeffs,
83 pixel *dst = (pixel *)_dst;
85 stride /= sizeof(pixel);
87 for (y = 0; y < 16; y++) {
88 for (x = 0; x < 16; x++) {
89 dst[x] = av_clip_pixel(dst[x] + *coeffs);
96 static void FUNC(transform_add32x32)(uint8_t *_dst, int16_t *coeffs,
100 pixel *dst = (pixel *)_dst;
102 stride /= sizeof(pixel);
104 for (y = 0; y < 32; y++) {
105 for (x = 0; x < 32; x++) {
106 dst[x] = av_clip_pixel(dst[x] + *coeffs);
114 static void FUNC(transform_rdpcm)(int16_t *_coeffs, int16_t log2_size, int mode)
116 int16_t *coeffs = (int16_t *) _coeffs;
118 int size = 1 << log2_size;
122 for (y = 0; y < size - 1; y++) {
123 for (x = 0; x < size; x++)
124 coeffs[x] += coeffs[x - size];
128 for (y = 0; y < size; y++) {
129 for (x = 1; x < size; x++)
130 coeffs[x] += coeffs[x - 1];
136 static void FUNC(transform_skip)(int16_t *_coeffs, int16_t log2_size)
138 int shift = 15 - BIT_DEPTH - log2_size;
140 int size = 1 << log2_size;
141 int16_t *coeffs = _coeffs;
145 int offset = 1 << (shift - 1);
146 for (y = 0; y < size; y++) {
147 for (x = 0; x < size; x++) {
148 *coeffs = (*coeffs + offset) >> shift;
153 for (y = 0; y < size; y++) {
154 for (x = 0; x < size; x++) {
155 *coeffs = *coeffs << -shift;
162 #define SET(dst, x) (dst) = (x)
163 #define SCALE(dst, x) (dst) = av_clip_int16(((x) + add) >> shift)
164 #define ADD_AND_SCALE(dst, x) \
165 (dst) = av_clip_pixel((dst) + av_clip_int16(((x) + add) >> shift))
167 #define TR_4x4_LUMA(dst, src, step, assign) \
169 int c0 = src[0 * step] + src[2 * step]; \
170 int c1 = src[2 * step] + src[3 * step]; \
171 int c2 = src[0 * step] - src[3 * step]; \
172 int c3 = 74 * src[1 * step]; \
174 assign(dst[2 * step], 74 * (src[0 * step] - \
177 assign(dst[0 * step], 29 * c0 + 55 * c1 + c3); \
178 assign(dst[1 * step], 55 * c2 - 29 * c1 + c3); \
179 assign(dst[3 * step], 55 * c0 + 29 * c2 - c3); \
182 static void FUNC(transform_4x4_luma)(int16_t *coeffs)
186 int add = 1 << (shift - 1);
187 int16_t *src = coeffs;
189 for (i = 0; i < 4; i++) {
190 TR_4x4_LUMA(src, src, 4, SCALE);
194 shift = 20 - BIT_DEPTH;
195 add = 1 << (shift - 1);
196 for (i = 0; i < 4; i++) {
197 TR_4x4_LUMA(coeffs, coeffs, 1, SCALE);
204 #define TR_4(dst, src, dstep, sstep, assign, end) \
206 const int e0 = 64 * src[0 * sstep] + 64 * src[2 * sstep]; \
207 const int e1 = 64 * src[0 * sstep] - 64 * src[2 * sstep]; \
208 const int o0 = 83 * src[1 * sstep] + 36 * src[3 * sstep]; \
209 const int o1 = 36 * src[1 * sstep] - 83 * src[3 * sstep]; \
211 assign(dst[0 * dstep], e0 + o0); \
212 assign(dst[1 * dstep], e1 + o1); \
213 assign(dst[2 * dstep], e1 - o1); \
214 assign(dst[3 * dstep], e0 - o0); \
217 #define TR_8(dst, src, dstep, sstep, assign, end) \
221 int o_8[4] = { 0 }; \
222 for (i = 0; i < 4; i++) \
223 for (j = 1; j < end; j += 2) \
224 o_8[i] += transform[4 * j][i] * src[j * sstep]; \
225 TR_4(e_8, src, 1, 2 * sstep, SET, 4); \
227 for (i = 0; i < 4; i++) { \
228 assign(dst[i * dstep], e_8[i] + o_8[i]); \
229 assign(dst[(7 - i) * dstep], e_8[i] - o_8[i]); \
233 #define TR_16(dst, src, dstep, sstep, assign, end) \
237 int o_16[8] = { 0 }; \
238 for (i = 0; i < 8; i++) \
239 for (j = 1; j < end; j += 2) \
240 o_16[i] += transform[2 * j][i] * src[j * sstep]; \
241 TR_8(e_16, src, 1, 2 * sstep, SET, 8); \
243 for (i = 0; i < 8; i++) { \
244 assign(dst[i * dstep], e_16[i] + o_16[i]); \
245 assign(dst[(15 - i) * dstep], e_16[i] - o_16[i]); \
249 #define TR_32(dst, src, dstep, sstep, assign, end) \
253 int o_32[16] = { 0 }; \
254 for (i = 0; i < 16; i++) \
255 for (j = 1; j < end; j += 2) \
256 o_32[i] += transform[j][i] * src[j * sstep]; \
257 TR_16(e_32, src, 1, 2 * sstep, SET, end/2); \
259 for (i = 0; i < 16; i++) { \
260 assign(dst[i * dstep], e_32[i] + o_32[i]); \
261 assign(dst[(31 - i) * dstep], e_32[i] - o_32[i]); \
265 #define IDCT_VAR4(H) \
266 int limit2 = FFMIN(col_limit + 4, H)
267 #define IDCT_VAR8(H) \
268 int limit = FFMIN(col_limit, H); \
269 int limit2 = FFMIN(col_limit + 4, H)
270 #define IDCT_VAR16(H) IDCT_VAR8(H)
271 #define IDCT_VAR32(H) IDCT_VAR8(H)
274 static void FUNC(idct_##H ##x ##H )( \
275 int16_t *coeffs, int col_limit) { \
278 int add = 1 << (shift - 1); \
279 int16_t *src = coeffs; \
282 for (i = 0; i < H; i++) { \
283 TR_ ## H(src, src, H, H, SCALE, limit2); \
284 if (limit2 < H && i%4 == 0 && !!i) \
289 shift = 20 - BIT_DEPTH; \
290 add = 1 << (shift - 1); \
291 for (i = 0; i < H; i++) { \
292 TR_ ## H(coeffs, coeffs, 1, 1, SCALE, limit); \
298 static void FUNC(idct_##H ##x ##H ##_dc)( \
301 int shift = 14 - BIT_DEPTH; \
302 int add = 1 << (shift - 1); \
303 int coeff = (((coeffs[0] + 1) >> 1) + add) >> shift; \
305 for (j = 0; j < H; j++) { \
306 for (i = 0; i < H; i++) { \
307 coeffs[i+j*H] = coeff; \
331 static void FUNC(sao_band_filter_0)(uint8_t *_dst, uint8_t *_src,
332 ptrdiff_t stride_dst, ptrdiff_t stride_src, SAOParams *sao,
333 int *borders, int width, int height,
336 pixel *dst = (pixel *)_dst;
337 pixel *src = (pixel *)_src;
338 int offset_table[32] = { 0 };
340 int shift = BIT_DEPTH - 5;
341 int *sao_offset_val = sao->offset_val[c_idx];
342 int sao_left_class = sao->band_position[c_idx];
344 stride_dst /= sizeof(pixel);
345 stride_src /= sizeof(pixel);
347 for (k = 0; k < 4; k++)
348 offset_table[(k + sao_left_class) & 31] = sao_offset_val[k + 1];
349 for (y = 0; y < height; y++) {
350 for (x = 0; x < width; x++)
351 dst[x] = av_clip_pixel(src[x] + offset_table[src[x] >> shift]);
357 #define CMP(a, b) ((a) > (b) ? 1 : ((a) == (b) ? 0 : -1))
359 static void FUNC(sao_edge_filter)(uint8_t *_dst, uint8_t *_src,
360 ptrdiff_t stride_dst, ptrdiff_t stride_src, SAOParams *sao,
361 int width, int height,
362 int c_idx, int init_x, int init_y) {
364 static const uint8_t edge_idx[] = { 1, 2, 0, 3, 4 };
365 static const int8_t pos[4][2][2] = {
366 { { -1, 0 }, { 1, 0 } }, // horizontal
367 { { 0, -1 }, { 0, 1 } }, // vertical
368 { { -1, -1 }, { 1, 1 } }, // 45 degree
369 { { 1, -1 }, { -1, 1 } }, // 135 degree
371 int *sao_offset_val = sao->offset_val[c_idx];
372 int sao_eo_class = sao->eo_class[c_idx];
373 pixel *dst = (pixel *)_dst;
374 pixel *src = (pixel *)_src;
376 int y_stride_src = init_y * stride_src;
377 int y_stride_dst = init_y * stride_dst;
378 int pos_0_0 = pos[sao_eo_class][0][0];
379 int pos_0_1 = pos[sao_eo_class][0][1];
380 int pos_1_0 = pos[sao_eo_class][1][0];
381 int pos_1_1 = pos[sao_eo_class][1][1];
384 int y_stride_0_1 = (init_y + pos_0_1) * stride_src;
385 int y_stride_1_1 = (init_y + pos_1_1) * stride_src;
386 for (y = init_y; y < height; y++) {
387 for (x = init_x; x < width; x++) {
388 int diff0 = CMP(src[x + y_stride_src], src[x + pos_0_0 + y_stride_0_1]);
389 int diff1 = CMP(src[x + y_stride_src], src[x + pos_1_0 + y_stride_1_1]);
390 int offset_val = edge_idx[2 + diff0 + diff1];
391 dst[x + y_stride_dst] = av_clip_pixel(src[x + y_stride_src] + sao_offset_val[offset_val]);
393 y_stride_src += stride_src;
394 y_stride_dst += stride_dst;
395 y_stride_0_1 += stride_src;
396 y_stride_1_1 += stride_src;
400 static void FUNC(sao_edge_filter_0)(uint8_t *_dst, uint8_t *_src,
401 ptrdiff_t stride_dst, ptrdiff_t stride_src, SAOParams *sao,
402 int *borders, int _width, int _height,
403 int c_idx, uint8_t *vert_edge,
404 uint8_t *horiz_edge, uint8_t *diag_edge)
407 pixel *dst = (pixel *)_dst;
408 pixel *src = (pixel *)_src;
409 int *sao_offset_val = sao->offset_val[c_idx];
410 int sao_eo_class = sao->eo_class[c_idx];
411 int init_x = 0, init_y = 0, width = _width, height = _height;
413 stride_dst /= sizeof(pixel);
414 stride_src /= sizeof(pixel);
416 if (sao_eo_class != SAO_EO_VERT) {
418 int offset_val = sao_offset_val[0];
419 for (y = 0; y < height; y++) {
420 dst[y * stride_dst] = av_clip_pixel(src[y * stride_src] + offset_val);
425 int offset_val = sao_offset_val[0];
426 int offset = width - 1;
427 for (x = 0; x < height; x++) {
428 dst[x * stride_dst + offset] = av_clip_pixel(src[x * stride_src + offset] + offset_val);
433 if (sao_eo_class != SAO_EO_HORIZ) {
435 int offset_val = sao_offset_val[0];
436 for (x = init_x; x < width; x++)
437 dst[x] = av_clip_pixel(src[x] + offset_val);
441 int offset_val = sao_offset_val[0];
442 int y_stride_dst = stride_dst * (height - 1);
443 int y_stride_src = stride_src * (height - 1);
444 for (x = init_x; x < width; x++)
445 dst[x + y_stride_dst] = av_clip_pixel(src[x + y_stride_src] + offset_val);
450 FUNC(sao_edge_filter)((uint8_t *)dst, (uint8_t *)src, stride_dst, stride_src, sao, width, height, c_idx, init_x, init_y);
453 static void FUNC(sao_edge_filter_1)(uint8_t *_dst, uint8_t *_src,
454 ptrdiff_t stride_dst, ptrdiff_t stride_src, SAOParams *sao,
455 int *borders, int _width, int _height,
456 int c_idx, uint8_t *vert_edge,
457 uint8_t *horiz_edge, uint8_t *diag_edge)
460 pixel *dst = (pixel *)_dst;
461 pixel *src = (pixel *)_src;
462 int *sao_offset_val = sao->offset_val[c_idx];
463 int sao_eo_class = sao->eo_class[c_idx];
464 int init_x = 0, init_y = 0, width = _width, height = _height;
466 stride_dst /= sizeof(pixel);
467 stride_src /= sizeof(pixel);
469 if (sao_eo_class != SAO_EO_VERT) {
471 int offset_val = sao_offset_val[0];
472 for (y = 0; y < height; y++) {
473 dst[y * stride_dst] = av_clip_pixel(src[y * stride_src] + offset_val);
478 int offset_val = sao_offset_val[0];
479 int offset = width - 1;
480 for (x = 0; x < height; x++) {
481 dst[x * stride_dst + offset] = av_clip_pixel(src[x * stride_src + offset] + offset_val);
486 if (sao_eo_class != SAO_EO_HORIZ) {
488 int offset_val = sao_offset_val[0];
489 for (x = init_x; x < width; x++)
490 dst[x] = av_clip_pixel(src[x] + offset_val);
494 int offset_val = sao_offset_val[0];
495 int y_stride_dst = stride_dst * (height - 1);
496 int y_stride_src = stride_src * (height - 1);
497 for (x = init_x; x < width; x++)
498 dst[x + y_stride_dst] = av_clip_pixel(src[x + y_stride_src] + offset_val);
503 FUNC(sao_edge_filter)((uint8_t *)dst, (uint8_t *)src, stride_dst, stride_src, sao, width, height, c_idx, init_x, init_y);
506 int save_upper_left = !diag_edge[0] && sao_eo_class == SAO_EO_135D && !borders[0] && !borders[1];
507 int save_upper_right = !diag_edge[1] && sao_eo_class == SAO_EO_45D && !borders[1] && !borders[2];
508 int save_lower_right = !diag_edge[2] && sao_eo_class == SAO_EO_135D && !borders[2] && !borders[3];
509 int save_lower_left = !diag_edge[3] && sao_eo_class == SAO_EO_45D && !borders[0] && !borders[3];
511 // Restore pixels that can't be modified
512 if(vert_edge[0] && sao_eo_class != SAO_EO_VERT) {
513 for(y = init_y+save_upper_left; y< height-save_lower_left; y++)
514 dst[y*stride_dst] = src[y*stride_src];
516 if(vert_edge[1] && sao_eo_class != SAO_EO_VERT) {
517 for(y = init_y+save_upper_right; y< height-save_lower_right; y++)
518 dst[y*stride_dst+width-1] = src[y*stride_src+width-1];
521 if(horiz_edge[0] && sao_eo_class != SAO_EO_HORIZ) {
522 for(x = init_x+save_upper_left; x < width-save_upper_right; x++)
525 if(horiz_edge[1] && sao_eo_class != SAO_EO_HORIZ) {
526 for(x = init_x+save_lower_left; x < width-save_lower_right; x++)
527 dst[(height-1)*stride_dst+x] = src[(height-1)*stride_src+x];
529 if(diag_edge[0] && sao_eo_class == SAO_EO_135D)
531 if(diag_edge[1] && sao_eo_class == SAO_EO_45D)
532 dst[width-1] = src[width-1];
533 if(diag_edge[2] && sao_eo_class == SAO_EO_135D)
534 dst[stride_dst*(height-1)+width-1] = src[stride_src*(height-1)+width-1];
535 if(diag_edge[3] && sao_eo_class == SAO_EO_45D)
536 dst[stride_dst*(height-1)] = src[stride_src*(height-1)];
543 ////////////////////////////////////////////////////////////////////////////////
545 ////////////////////////////////////////////////////////////////////////////////
546 static void FUNC(put_hevc_pel_pixels)(int16_t *dst, ptrdiff_t dststride,
547 uint8_t *_src, ptrdiff_t _srcstride,
548 int height, intptr_t mx, intptr_t my, int width)
551 pixel *src = (pixel *)_src;
552 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
554 for (y = 0; y < height; y++) {
555 for (x = 0; x < width; x++)
556 dst[x] = src[x] << (14 - BIT_DEPTH);
562 static void FUNC(put_hevc_pel_uni_pixels)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
563 int height, intptr_t mx, intptr_t my, int width)
566 pixel *src = (pixel *)_src;
567 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
568 pixel *dst = (pixel *)_dst;
569 ptrdiff_t dststride = _dststride / sizeof(pixel);
571 for (y = 0; y < height; y++) {
572 memcpy(dst, src, width * sizeof(pixel));
578 static void FUNC(put_hevc_pel_bi_pixels)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
579 int16_t *src2, ptrdiff_t src2stride,
580 int height, intptr_t mx, intptr_t my, int width)
583 pixel *src = (pixel *)_src;
584 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
585 pixel *dst = (pixel *)_dst;
586 ptrdiff_t dststride = _dststride / sizeof(pixel);
588 int shift = 14 + 1 - BIT_DEPTH;
590 int offset = 1 << (shift - 1);
595 for (y = 0; y < height; y++) {
596 for (x = 0; x < width; x++)
597 dst[x] = av_clip_pixel(((src[x] << (14 - BIT_DEPTH)) + src2[x] + offset) >> shift);
604 static void FUNC(put_hevc_pel_uni_w_pixels)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
605 int height, int denom, int wx, int ox, intptr_t mx, intptr_t my, int width)
608 pixel *src = (pixel *)_src;
609 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
610 pixel *dst = (pixel *)_dst;
611 ptrdiff_t dststride = _dststride / sizeof(pixel);
612 int shift = denom + 14 - BIT_DEPTH;
614 int offset = 1 << (shift - 1);
619 ox = ox * (1 << (BIT_DEPTH - 8));
620 for (y = 0; y < height; y++) {
621 for (x = 0; x < width; x++)
622 dst[x] = av_clip_pixel((((src[x] << (14 - BIT_DEPTH)) * wx + offset) >> shift) + ox);
628 static void FUNC(put_hevc_pel_bi_w_pixels)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
629 int16_t *src2, ptrdiff_t src2stride,
630 int height, int denom, int wx0, int wx1,
631 int ox0, int ox1, intptr_t mx, intptr_t my, int width)
634 pixel *src = (pixel *)_src;
635 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
636 pixel *dst = (pixel *)_dst;
637 ptrdiff_t dststride = _dststride / sizeof(pixel);
639 int shift = 14 + 1 - BIT_DEPTH;
640 int log2Wd = denom + shift - 1;
642 ox0 = ox0 * (1 << (BIT_DEPTH - 8));
643 ox1 = ox1 * (1 << (BIT_DEPTH - 8));
644 for (y = 0; y < height; y++) {
645 for (x = 0; x < width; x++) {
646 dst[x] = av_clip_pixel(( (src[x] << (14 - BIT_DEPTH)) * wx1 + src2[x] * wx0 + ((ox0 + ox1 + 1) << log2Wd)) >> (log2Wd + 1));
654 ////////////////////////////////////////////////////////////////////////////////
656 ////////////////////////////////////////////////////////////////////////////////
657 #define QPEL_FILTER(src, stride) \
658 (filter[0] * src[x - 3 * stride] + \
659 filter[1] * src[x - 2 * stride] + \
660 filter[2] * src[x - stride] + \
661 filter[3] * src[x ] + \
662 filter[4] * src[x + stride] + \
663 filter[5] * src[x + 2 * stride] + \
664 filter[6] * src[x + 3 * stride] + \
665 filter[7] * src[x + 4 * stride])
667 static void FUNC(put_hevc_qpel_h)(int16_t *dst, ptrdiff_t dststride,
668 uint8_t *_src, ptrdiff_t _srcstride,
669 int height, intptr_t mx, intptr_t my, int width)
672 pixel *src = (pixel*)_src;
673 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
674 const int8_t *filter = ff_hevc_qpel_filters[mx - 1];
675 for (y = 0; y < height; y++) {
676 for (x = 0; x < width; x++)
677 dst[x] = QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8);
683 static void FUNC(put_hevc_qpel_v)(int16_t *dst, ptrdiff_t dststride,
684 uint8_t *_src, ptrdiff_t _srcstride,
685 int height, intptr_t mx, intptr_t my, int width)
688 pixel *src = (pixel*)_src;
689 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
690 const int8_t *filter = ff_hevc_qpel_filters[my - 1];
691 for (y = 0; y < height; y++) {
692 for (x = 0; x < width; x++)
693 dst[x] = QPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8);
699 static void FUNC(put_hevc_qpel_hv)(int16_t *dst,
702 ptrdiff_t _srcstride,
703 int height, intptr_t mx,
704 intptr_t my, int width)
707 const int8_t *filter;
708 pixel *src = (pixel*)_src;
709 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
710 int16_t tmp_array[(MAX_PB_SIZE + QPEL_EXTRA) * MAX_PB_SIZE];
711 int16_t *tmp = tmp_array;
713 src -= QPEL_EXTRA_BEFORE * srcstride;
714 filter = ff_hevc_qpel_filters[mx - 1];
715 for (y = 0; y < height + QPEL_EXTRA; y++) {
716 for (x = 0; x < width; x++)
717 tmp[x] = QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8);
722 tmp = tmp_array + QPEL_EXTRA_BEFORE * MAX_PB_SIZE;
723 filter = ff_hevc_qpel_filters[my - 1];
724 for (y = 0; y < height; y++) {
725 for (x = 0; x < width; x++)
726 dst[x] = QPEL_FILTER(tmp, MAX_PB_SIZE) >> 6;
732 static void FUNC(put_hevc_qpel_uni_h)(uint8_t *_dst, ptrdiff_t _dststride,
733 uint8_t *_src, ptrdiff_t _srcstride,
734 int height, intptr_t mx, intptr_t my, int width)
737 pixel *src = (pixel*)_src;
738 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
739 pixel *dst = (pixel *)_dst;
740 ptrdiff_t dststride = _dststride / sizeof(pixel);
741 const int8_t *filter = ff_hevc_qpel_filters[mx - 1];
742 int shift = 14 - BIT_DEPTH;
745 int offset = 1 << (shift - 1);
750 for (y = 0; y < height; y++) {
751 for (x = 0; x < width; x++)
752 dst[x] = av_clip_pixel(((QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) + offset) >> shift);
758 static void FUNC(put_hevc_qpel_bi_h)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
759 int16_t *src2, ptrdiff_t src2stride,
760 int height, intptr_t mx, intptr_t my, int width)
763 pixel *src = (pixel*)_src;
764 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
765 pixel *dst = (pixel *)_dst;
766 ptrdiff_t dststride = _dststride / sizeof(pixel);
768 const int8_t *filter = ff_hevc_qpel_filters[mx - 1];
770 int shift = 14 + 1 - BIT_DEPTH;
772 int offset = 1 << (shift - 1);
777 for (y = 0; y < height; y++) {
778 for (x = 0; x < width; x++)
779 dst[x] = av_clip_pixel(((QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) + src2[x] + offset) >> shift);
786 static void FUNC(put_hevc_qpel_uni_v)(uint8_t *_dst, ptrdiff_t _dststride,
787 uint8_t *_src, ptrdiff_t _srcstride,
788 int height, intptr_t mx, intptr_t my, int width)
791 pixel *src = (pixel*)_src;
792 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
793 pixel *dst = (pixel *)_dst;
794 ptrdiff_t dststride = _dststride / sizeof(pixel);
795 const int8_t *filter = ff_hevc_qpel_filters[my - 1];
796 int shift = 14 - BIT_DEPTH;
799 int offset = 1 << (shift - 1);
804 for (y = 0; y < height; y++) {
805 for (x = 0; x < width; x++)
806 dst[x] = av_clip_pixel(((QPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) + offset) >> shift);
813 static void FUNC(put_hevc_qpel_bi_v)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
814 int16_t *src2, ptrdiff_t src2stride,
815 int height, intptr_t mx, intptr_t my, int width)
818 pixel *src = (pixel*)_src;
819 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
820 pixel *dst = (pixel *)_dst;
821 ptrdiff_t dststride = _dststride / sizeof(pixel);
823 const int8_t *filter = ff_hevc_qpel_filters[my - 1];
825 int shift = 14 + 1 - BIT_DEPTH;
827 int offset = 1 << (shift - 1);
832 for (y = 0; y < height; y++) {
833 for (x = 0; x < width; x++)
834 dst[x] = av_clip_pixel(((QPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) + src2[x] + offset) >> shift);
841 static void FUNC(put_hevc_qpel_uni_hv)(uint8_t *_dst, ptrdiff_t _dststride,
842 uint8_t *_src, ptrdiff_t _srcstride,
843 int height, intptr_t mx, intptr_t my, int width)
846 const int8_t *filter;
847 pixel *src = (pixel*)_src;
848 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
849 pixel *dst = (pixel *)_dst;
850 ptrdiff_t dststride = _dststride / sizeof(pixel);
851 int16_t tmp_array[(MAX_PB_SIZE + QPEL_EXTRA) * MAX_PB_SIZE];
852 int16_t *tmp = tmp_array;
853 int shift = 14 - BIT_DEPTH;
856 int offset = 1 << (shift - 1);
861 src -= QPEL_EXTRA_BEFORE * srcstride;
862 filter = ff_hevc_qpel_filters[mx - 1];
863 for (y = 0; y < height + QPEL_EXTRA; y++) {
864 for (x = 0; x < width; x++)
865 tmp[x] = QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8);
870 tmp = tmp_array + QPEL_EXTRA_BEFORE * MAX_PB_SIZE;
871 filter = ff_hevc_qpel_filters[my - 1];
873 for (y = 0; y < height; y++) {
874 for (x = 0; x < width; x++)
875 dst[x] = av_clip_pixel(((QPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) + offset) >> shift);
881 static void FUNC(put_hevc_qpel_bi_hv)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
882 int16_t *src2, ptrdiff_t src2stride,
883 int height, intptr_t mx, intptr_t my, int width)
886 const int8_t *filter;
887 pixel *src = (pixel*)_src;
888 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
889 pixel *dst = (pixel *)_dst;
890 ptrdiff_t dststride = _dststride / sizeof(pixel);
891 int16_t tmp_array[(MAX_PB_SIZE + QPEL_EXTRA) * MAX_PB_SIZE];
892 int16_t *tmp = tmp_array;
893 int shift = 14 + 1 - BIT_DEPTH;
895 int offset = 1 << (shift - 1);
900 src -= QPEL_EXTRA_BEFORE * srcstride;
901 filter = ff_hevc_qpel_filters[mx - 1];
902 for (y = 0; y < height + QPEL_EXTRA; y++) {
903 for (x = 0; x < width; x++)
904 tmp[x] = QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8);
909 tmp = tmp_array + QPEL_EXTRA_BEFORE * MAX_PB_SIZE;
910 filter = ff_hevc_qpel_filters[my - 1];
912 for (y = 0; y < height; y++) {
913 for (x = 0; x < width; x++)
914 dst[x] = av_clip_pixel(((QPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) + src2[x] + offset) >> shift);
921 static void FUNC(put_hevc_qpel_uni_w_h)(uint8_t *_dst, ptrdiff_t _dststride,
922 uint8_t *_src, ptrdiff_t _srcstride,
923 int height, int denom, int wx, int ox,
924 intptr_t mx, intptr_t my, int width)
927 pixel *src = (pixel*)_src;
928 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
929 pixel *dst = (pixel *)_dst;
930 ptrdiff_t dststride = _dststride / sizeof(pixel);
931 const int8_t *filter = ff_hevc_qpel_filters[mx - 1];
932 int shift = denom + 14 - BIT_DEPTH;
934 int offset = 1 << (shift - 1);
939 ox = ox * (1 << (BIT_DEPTH - 8));
940 for (y = 0; y < height; y++) {
941 for (x = 0; x < width; x++)
942 dst[x] = av_clip_pixel((((QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) * wx + offset) >> shift) + ox);
948 static void FUNC(put_hevc_qpel_bi_w_h)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
949 int16_t *src2, ptrdiff_t src2stride,
950 int height, int denom, int wx0, int wx1,
951 int ox0, int ox1, intptr_t mx, intptr_t my, int width)
954 pixel *src = (pixel*)_src;
955 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
956 pixel *dst = (pixel *)_dst;
957 ptrdiff_t dststride = _dststride / sizeof(pixel);
959 const int8_t *filter = ff_hevc_qpel_filters[mx - 1];
961 int shift = 14 + 1 - BIT_DEPTH;
962 int log2Wd = denom + shift - 1;
964 ox0 = ox0 * (1 << (BIT_DEPTH - 8));
965 ox1 = ox1 * (1 << (BIT_DEPTH - 8));
966 for (y = 0; y < height; y++) {
967 for (x = 0; x < width; x++)
968 dst[x] = av_clip_pixel(((QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) * wx1 + src2[x] * wx0 +
969 ((ox0 + ox1 + 1) << log2Wd)) >> (log2Wd + 1));
976 static void FUNC(put_hevc_qpel_uni_w_v)(uint8_t *_dst, ptrdiff_t _dststride,
977 uint8_t *_src, ptrdiff_t _srcstride,
978 int height, int denom, int wx, int ox,
979 intptr_t mx, intptr_t my, int width)
982 pixel *src = (pixel*)_src;
983 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
984 pixel *dst = (pixel *)_dst;
985 ptrdiff_t dststride = _dststride / sizeof(pixel);
986 const int8_t *filter = ff_hevc_qpel_filters[my - 1];
987 int shift = denom + 14 - BIT_DEPTH;
989 int offset = 1 << (shift - 1);
994 ox = ox * (1 << (BIT_DEPTH - 8));
995 for (y = 0; y < height; y++) {
996 for (x = 0; x < width; x++)
997 dst[x] = av_clip_pixel((((QPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) * wx + offset) >> shift) + ox);
1003 static void FUNC(put_hevc_qpel_bi_w_v)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1004 int16_t *src2, ptrdiff_t src2stride,
1005 int height, int denom, int wx0, int wx1,
1006 int ox0, int ox1, intptr_t mx, intptr_t my, int width)
1009 pixel *src = (pixel*)_src;
1010 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1011 pixel *dst = (pixel *)_dst;
1012 ptrdiff_t dststride = _dststride / sizeof(pixel);
1014 const int8_t *filter = ff_hevc_qpel_filters[my - 1];
1016 int shift = 14 + 1 - BIT_DEPTH;
1017 int log2Wd = denom + shift - 1;
1019 ox0 = ox0 * (1 << (BIT_DEPTH - 8));
1020 ox1 = ox1 * (1 << (BIT_DEPTH - 8));
1021 for (y = 0; y < height; y++) {
1022 for (x = 0; x < width; x++)
1023 dst[x] = av_clip_pixel(((QPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) * wx1 + src2[x] * wx0 +
1024 ((ox0 + ox1 + 1) << log2Wd)) >> (log2Wd + 1));
1031 static void FUNC(put_hevc_qpel_uni_w_hv)(uint8_t *_dst, ptrdiff_t _dststride,
1032 uint8_t *_src, ptrdiff_t _srcstride,
1033 int height, int denom, int wx, int ox,
1034 intptr_t mx, intptr_t my, int width)
1037 const int8_t *filter;
1038 pixel *src = (pixel*)_src;
1039 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1040 pixel *dst = (pixel *)_dst;
1041 ptrdiff_t dststride = _dststride / sizeof(pixel);
1042 int16_t tmp_array[(MAX_PB_SIZE + QPEL_EXTRA) * MAX_PB_SIZE];
1043 int16_t *tmp = tmp_array;
1044 int shift = denom + 14 - BIT_DEPTH;
1046 int offset = 1 << (shift - 1);
1051 src -= QPEL_EXTRA_BEFORE * srcstride;
1052 filter = ff_hevc_qpel_filters[mx - 1];
1053 for (y = 0; y < height + QPEL_EXTRA; y++) {
1054 for (x = 0; x < width; x++)
1055 tmp[x] = QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8);
1060 tmp = tmp_array + QPEL_EXTRA_BEFORE * MAX_PB_SIZE;
1061 filter = ff_hevc_qpel_filters[my - 1];
1063 ox = ox * (1 << (BIT_DEPTH - 8));
1064 for (y = 0; y < height; y++) {
1065 for (x = 0; x < width; x++)
1066 dst[x] = av_clip_pixel((((QPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) * wx + offset) >> shift) + ox);
1072 static void FUNC(put_hevc_qpel_bi_w_hv)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1073 int16_t *src2, ptrdiff_t src2stride,
1074 int height, int denom, int wx0, int wx1,
1075 int ox0, int ox1, intptr_t mx, intptr_t my, int width)
1078 const int8_t *filter;
1079 pixel *src = (pixel*)_src;
1080 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1081 pixel *dst = (pixel *)_dst;
1082 ptrdiff_t dststride = _dststride / sizeof(pixel);
1083 int16_t tmp_array[(MAX_PB_SIZE + QPEL_EXTRA) * MAX_PB_SIZE];
1084 int16_t *tmp = tmp_array;
1085 int shift = 14 + 1 - BIT_DEPTH;
1086 int log2Wd = denom + shift - 1;
1088 src -= QPEL_EXTRA_BEFORE * srcstride;
1089 filter = ff_hevc_qpel_filters[mx - 1];
1090 for (y = 0; y < height + QPEL_EXTRA; y++) {
1091 for (x = 0; x < width; x++)
1092 tmp[x] = QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8);
1097 tmp = tmp_array + QPEL_EXTRA_BEFORE * MAX_PB_SIZE;
1098 filter = ff_hevc_qpel_filters[my - 1];
1100 ox0 = ox0 * (1 << (BIT_DEPTH - 8));
1101 ox1 = ox1 * (1 << (BIT_DEPTH - 8));
1102 for (y = 0; y < height; y++) {
1103 for (x = 0; x < width; x++)
1104 dst[x] = av_clip_pixel(((QPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) * wx1 + src2[x] * wx0 +
1105 ((ox0 + ox1 + 1) << log2Wd)) >> (log2Wd + 1));
1112 ////////////////////////////////////////////////////////////////////////////////
1114 ////////////////////////////////////////////////////////////////////////////////
1115 #define EPEL_FILTER(src, stride) \
1116 (filter[0] * src[x - stride] + \
1117 filter[1] * src[x] + \
1118 filter[2] * src[x + stride] + \
1119 filter[3] * src[x + 2 * stride])
1121 static void FUNC(put_hevc_epel_h)(int16_t *dst, ptrdiff_t dststride,
1122 uint8_t *_src, ptrdiff_t _srcstride,
1123 int height, intptr_t mx, intptr_t my, int width)
1126 pixel *src = (pixel *)_src;
1127 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1128 const int8_t *filter = ff_hevc_epel_filters[mx - 1];
1129 for (y = 0; y < height; y++) {
1130 for (x = 0; x < width; x++)
1131 dst[x] = EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8);
1137 static void FUNC(put_hevc_epel_v)(int16_t *dst, ptrdiff_t dststride,
1138 uint8_t *_src, ptrdiff_t _srcstride,
1139 int height, intptr_t mx, intptr_t my, int width)
1142 pixel *src = (pixel *)_src;
1143 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1144 const int8_t *filter = ff_hevc_epel_filters[my - 1];
1146 for (y = 0; y < height; y++) {
1147 for (x = 0; x < width; x++)
1148 dst[x] = EPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8);
1154 static void FUNC(put_hevc_epel_hv)(int16_t *dst, ptrdiff_t dststride,
1155 uint8_t *_src, ptrdiff_t _srcstride,
1156 int height, intptr_t mx, intptr_t my, int width)
1159 pixel *src = (pixel *)_src;
1160 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1161 const int8_t *filter = ff_hevc_epel_filters[mx - 1];
1162 int16_t tmp_array[(MAX_PB_SIZE + EPEL_EXTRA) * MAX_PB_SIZE];
1163 int16_t *tmp = tmp_array;
1165 src -= EPEL_EXTRA_BEFORE * srcstride;
1167 for (y = 0; y < height + EPEL_EXTRA; y++) {
1168 for (x = 0; x < width; x++)
1169 tmp[x] = EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8);
1174 tmp = tmp_array + EPEL_EXTRA_BEFORE * MAX_PB_SIZE;
1175 filter = ff_hevc_epel_filters[my - 1];
1177 for (y = 0; y < height; y++) {
1178 for (x = 0; x < width; x++)
1179 dst[x] = EPEL_FILTER(tmp, MAX_PB_SIZE) >> 6;
1185 static void FUNC(put_hevc_epel_uni_h)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1186 int height, intptr_t mx, intptr_t my, int width)
1189 pixel *src = (pixel *)_src;
1190 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1191 pixel *dst = (pixel *)_dst;
1192 ptrdiff_t dststride = _dststride / sizeof(pixel);
1193 const int8_t *filter = ff_hevc_epel_filters[mx - 1];
1194 int shift = 14 - BIT_DEPTH;
1196 int offset = 1 << (shift - 1);
1201 for (y = 0; y < height; y++) {
1202 for (x = 0; x < width; x++)
1203 dst[x] = av_clip_pixel(((EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) + offset) >> shift);
1209 static void FUNC(put_hevc_epel_bi_h)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1210 int16_t *src2, ptrdiff_t src2stride,
1211 int height, intptr_t mx, intptr_t my, int width)
1214 pixel *src = (pixel *)_src;
1215 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1216 pixel *dst = (pixel *)_dst;
1217 ptrdiff_t dststride = _dststride / sizeof(pixel);
1218 const int8_t *filter = ff_hevc_epel_filters[mx - 1];
1219 int shift = 14 + 1 - BIT_DEPTH;
1221 int offset = 1 << (shift - 1);
1226 for (y = 0; y < height; y++) {
1227 for (x = 0; x < width; x++) {
1228 dst[x] = av_clip_pixel(((EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) + src2[x] + offset) >> shift);
1236 static void FUNC(put_hevc_epel_uni_v)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1237 int height, intptr_t mx, intptr_t my, int width)
1240 pixel *src = (pixel *)_src;
1241 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1242 pixel *dst = (pixel *)_dst;
1243 ptrdiff_t dststride = _dststride / sizeof(pixel);
1244 const int8_t *filter = ff_hevc_epel_filters[my - 1];
1245 int shift = 14 - BIT_DEPTH;
1247 int offset = 1 << (shift - 1);
1252 for (y = 0; y < height; y++) {
1253 for (x = 0; x < width; x++)
1254 dst[x] = av_clip_pixel(((EPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) + offset) >> shift);
1260 static void FUNC(put_hevc_epel_bi_v)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1261 int16_t *src2, ptrdiff_t src2stride,
1262 int height, intptr_t mx, intptr_t my, int width)
1265 pixel *src = (pixel *)_src;
1266 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1267 const int8_t *filter = ff_hevc_epel_filters[my - 1];
1268 pixel *dst = (pixel *)_dst;
1269 ptrdiff_t dststride = _dststride / sizeof(pixel);
1270 int shift = 14 + 1 - BIT_DEPTH;
1272 int offset = 1 << (shift - 1);
1277 for (y = 0; y < height; y++) {
1278 for (x = 0; x < width; x++)
1279 dst[x] = av_clip_pixel(((EPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) + src2[x] + offset) >> shift);
1286 static void FUNC(put_hevc_epel_uni_hv)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1287 int height, intptr_t mx, intptr_t my, int width)
1290 pixel *src = (pixel *)_src;
1291 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1292 pixel *dst = (pixel *)_dst;
1293 ptrdiff_t dststride = _dststride / sizeof(pixel);
1294 const int8_t *filter = ff_hevc_epel_filters[mx - 1];
1295 int16_t tmp_array[(MAX_PB_SIZE + EPEL_EXTRA) * MAX_PB_SIZE];
1296 int16_t *tmp = tmp_array;
1297 int shift = 14 - BIT_DEPTH;
1299 int offset = 1 << (shift - 1);
1304 src -= EPEL_EXTRA_BEFORE * srcstride;
1306 for (y = 0; y < height + EPEL_EXTRA; y++) {
1307 for (x = 0; x < width; x++)
1308 tmp[x] = EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8);
1313 tmp = tmp_array + EPEL_EXTRA_BEFORE * MAX_PB_SIZE;
1314 filter = ff_hevc_epel_filters[my - 1];
1316 for (y = 0; y < height; y++) {
1317 for (x = 0; x < width; x++)
1318 dst[x] = av_clip_pixel(((EPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) + offset) >> shift);
1324 static void FUNC(put_hevc_epel_bi_hv)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1325 int16_t *src2, ptrdiff_t src2stride,
1326 int height, intptr_t mx, intptr_t my, int width)
1329 pixel *src = (pixel *)_src;
1330 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1331 pixel *dst = (pixel *)_dst;
1332 ptrdiff_t dststride = _dststride / sizeof(pixel);
1333 const int8_t *filter = ff_hevc_epel_filters[mx - 1];
1334 int16_t tmp_array[(MAX_PB_SIZE + EPEL_EXTRA) * MAX_PB_SIZE];
1335 int16_t *tmp = tmp_array;
1336 int shift = 14 + 1 - BIT_DEPTH;
1338 int offset = 1 << (shift - 1);
1343 src -= EPEL_EXTRA_BEFORE * srcstride;
1345 for (y = 0; y < height + EPEL_EXTRA; y++) {
1346 for (x = 0; x < width; x++)
1347 tmp[x] = EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8);
1352 tmp = tmp_array + EPEL_EXTRA_BEFORE * MAX_PB_SIZE;
1353 filter = ff_hevc_epel_filters[my - 1];
1355 for (y = 0; y < height; y++) {
1356 for (x = 0; x < width; x++)
1357 dst[x] = av_clip_pixel(((EPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) + src2[x] + offset) >> shift);
1364 static void FUNC(put_hevc_epel_uni_w_h)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1365 int height, int denom, int wx, int ox, intptr_t mx, intptr_t my, int width)
1368 pixel *src = (pixel *)_src;
1369 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1370 pixel *dst = (pixel *)_dst;
1371 ptrdiff_t dststride = _dststride / sizeof(pixel);
1372 const int8_t *filter = ff_hevc_epel_filters[mx - 1];
1373 int shift = denom + 14 - BIT_DEPTH;
1375 int offset = 1 << (shift - 1);
1380 ox = ox * (1 << (BIT_DEPTH - 8));
1381 for (y = 0; y < height; y++) {
1382 for (x = 0; x < width; x++) {
1383 dst[x] = av_clip_pixel((((EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) * wx + offset) >> shift) + ox);
1390 static void FUNC(put_hevc_epel_bi_w_h)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1391 int16_t *src2, ptrdiff_t src2stride,
1392 int height, int denom, int wx0, int wx1,
1393 int ox0, int ox1, intptr_t mx, intptr_t my, int width)
1396 pixel *src = (pixel *)_src;
1397 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1398 pixel *dst = (pixel *)_dst;
1399 ptrdiff_t dststride = _dststride / sizeof(pixel);
1400 const int8_t *filter = ff_hevc_epel_filters[mx - 1];
1401 int shift = 14 + 1 - BIT_DEPTH;
1402 int log2Wd = denom + shift - 1;
1404 ox0 = ox0 * (1 << (BIT_DEPTH - 8));
1405 ox1 = ox1 * (1 << (BIT_DEPTH - 8));
1406 for (y = 0; y < height; y++) {
1407 for (x = 0; x < width; x++)
1408 dst[x] = av_clip_pixel(((EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) * wx1 + src2[x] * wx0 +
1409 ((ox0 + ox1 + 1) << log2Wd)) >> (log2Wd + 1));
1416 static void FUNC(put_hevc_epel_uni_w_v)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1417 int height, int denom, int wx, int ox, intptr_t mx, intptr_t my, int width)
1420 pixel *src = (pixel *)_src;
1421 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1422 pixel *dst = (pixel *)_dst;
1423 ptrdiff_t dststride = _dststride / sizeof(pixel);
1424 const int8_t *filter = ff_hevc_epel_filters[my - 1];
1425 int shift = denom + 14 - BIT_DEPTH;
1427 int offset = 1 << (shift - 1);
1432 ox = ox * (1 << (BIT_DEPTH - 8));
1433 for (y = 0; y < height; y++) {
1434 for (x = 0; x < width; x++) {
1435 dst[x] = av_clip_pixel((((EPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) * wx + offset) >> shift) + ox);
1442 static void FUNC(put_hevc_epel_bi_w_v)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1443 int16_t *src2, ptrdiff_t src2stride,
1444 int height, int denom, int wx0, int wx1,
1445 int ox0, int ox1, intptr_t mx, intptr_t my, int width)
1448 pixel *src = (pixel *)_src;
1449 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1450 const int8_t *filter = ff_hevc_epel_filters[my - 1];
1451 pixel *dst = (pixel *)_dst;
1452 ptrdiff_t dststride = _dststride / sizeof(pixel);
1453 int shift = 14 + 1 - BIT_DEPTH;
1454 int log2Wd = denom + shift - 1;
1456 ox0 = ox0 * (1 << (BIT_DEPTH - 8));
1457 ox1 = ox1 * (1 << (BIT_DEPTH - 8));
1458 for (y = 0; y < height; y++) {
1459 for (x = 0; x < width; x++)
1460 dst[x] = av_clip_pixel(((EPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) * wx1 + src2[x] * wx0 +
1461 ((ox0 + ox1 + 1) << log2Wd)) >> (log2Wd + 1));
1468 static void FUNC(put_hevc_epel_uni_w_hv)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1469 int height, int denom, int wx, int ox, intptr_t mx, intptr_t my, int width)
1472 pixel *src = (pixel *)_src;
1473 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1474 pixel *dst = (pixel *)_dst;
1475 ptrdiff_t dststride = _dststride / sizeof(pixel);
1476 const int8_t *filter = ff_hevc_epel_filters[mx - 1];
1477 int16_t tmp_array[(MAX_PB_SIZE + EPEL_EXTRA) * MAX_PB_SIZE];
1478 int16_t *tmp = tmp_array;
1479 int shift = denom + 14 - BIT_DEPTH;
1481 int offset = 1 << (shift - 1);
1486 src -= EPEL_EXTRA_BEFORE * srcstride;
1488 for (y = 0; y < height + EPEL_EXTRA; y++) {
1489 for (x = 0; x < width; x++)
1490 tmp[x] = EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8);
1495 tmp = tmp_array + EPEL_EXTRA_BEFORE * MAX_PB_SIZE;
1496 filter = ff_hevc_epel_filters[my - 1];
1498 ox = ox * (1 << (BIT_DEPTH - 8));
1499 for (y = 0; y < height; y++) {
1500 for (x = 0; x < width; x++)
1501 dst[x] = av_clip_pixel((((EPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) * wx + offset) >> shift) + ox);
1507 static void FUNC(put_hevc_epel_bi_w_hv)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1508 int16_t *src2, ptrdiff_t src2stride,
1509 int height, int denom, int wx0, int wx1,
1510 int ox0, int ox1, intptr_t mx, intptr_t my, int width)
1513 pixel *src = (pixel *)_src;
1514 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1515 pixel *dst = (pixel *)_dst;
1516 ptrdiff_t dststride = _dststride / sizeof(pixel);
1517 const int8_t *filter = ff_hevc_epel_filters[mx - 1];
1518 int16_t tmp_array[(MAX_PB_SIZE + EPEL_EXTRA) * MAX_PB_SIZE];
1519 int16_t *tmp = tmp_array;
1520 int shift = 14 + 1 - BIT_DEPTH;
1521 int log2Wd = denom + shift - 1;
1523 src -= EPEL_EXTRA_BEFORE * srcstride;
1525 for (y = 0; y < height + EPEL_EXTRA; y++) {
1526 for (x = 0; x < width; x++)
1527 tmp[x] = EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8);
1532 tmp = tmp_array + EPEL_EXTRA_BEFORE * MAX_PB_SIZE;
1533 filter = ff_hevc_epel_filters[my - 1];
1535 ox0 = ox0 * (1 << (BIT_DEPTH - 8));
1536 ox1 = ox1 * (1 << (BIT_DEPTH - 8));
1537 for (y = 0; y < height; y++) {
1538 for (x = 0; x < width; x++)
1539 dst[x] = av_clip_pixel(((EPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) * wx1 + src2[x] * wx0 +
1540 ((ox0 + ox1 + 1) << log2Wd)) >> (log2Wd + 1));
1546 #define P3 pix[-4 * xstride]
1547 #define P2 pix[-3 * xstride]
1548 #define P1 pix[-2 * xstride]
1549 #define P0 pix[-1 * xstride]
1550 #define Q0 pix[0 * xstride]
1551 #define Q1 pix[1 * xstride]
1552 #define Q2 pix[2 * xstride]
1553 #define Q3 pix[3 * xstride]
1555 // line three. used only for deblocking decision
1556 #define TP3 pix[-4 * xstride + 3 * ystride]
1557 #define TP2 pix[-3 * xstride + 3 * ystride]
1558 #define TP1 pix[-2 * xstride + 3 * ystride]
1559 #define TP0 pix[-1 * xstride + 3 * ystride]
1560 #define TQ0 pix[0 * xstride + 3 * ystride]
1561 #define TQ1 pix[1 * xstride + 3 * ystride]
1562 #define TQ2 pix[2 * xstride + 3 * ystride]
1563 #define TQ3 pix[3 * xstride + 3 * ystride]
1565 static void FUNC(hevc_loop_filter_luma)(uint8_t *_pix,
1566 ptrdiff_t _xstride, ptrdiff_t _ystride,
1567 int *_beta, int *_tc,
1568 uint8_t *_no_p, uint8_t *_no_q)
1571 pixel *pix = (pixel *)_pix;
1572 ptrdiff_t xstride = _xstride / sizeof(pixel);
1573 ptrdiff_t ystride = _ystride / sizeof(pixel);
1575 for (j = 0; j < 2; j++) {
1576 const int dp0 = abs(P2 - 2 * P1 + P0);
1577 const int dq0 = abs(Q2 - 2 * Q1 + Q0);
1578 const int dp3 = abs(TP2 - 2 * TP1 + TP0);
1579 const int dq3 = abs(TQ2 - 2 * TQ1 + TQ0);
1580 const int d0 = dp0 + dq0;
1581 const int d3 = dp3 + dq3;
1582 const int beta = _beta[j] << (BIT_DEPTH - 8);
1583 const int tc = _tc[j] << (BIT_DEPTH - 8);
1584 const int no_p = _no_p[j];
1585 const int no_q = _no_q[j];
1587 if (d0 + d3 >= beta) {
1591 const int beta_3 = beta >> 3;
1592 const int beta_2 = beta >> 2;
1593 const int tc25 = ((tc * 5 + 1) >> 1);
1595 if (abs(P3 - P0) + abs(Q3 - Q0) < beta_3 && abs(P0 - Q0) < tc25 &&
1596 abs(TP3 - TP0) + abs(TQ3 - TQ0) < beta_3 && abs(TP0 - TQ0) < tc25 &&
1597 (d0 << 1) < beta_2 && (d3 << 1) < beta_2) {
1599 const int tc2 = tc << 1;
1600 for (d = 0; d < 4; d++) {
1610 P0 = p0 + av_clip(((p2 + 2 * p1 + 2 * p0 + 2 * q0 + q1 + 4) >> 3) - p0, -tc2, tc2);
1611 P1 = p1 + av_clip(((p2 + p1 + p0 + q0 + 2) >> 2) - p1, -tc2, tc2);
1612 P2 = p2 + av_clip(((2 * p3 + 3 * p2 + p1 + p0 + q0 + 4) >> 3) - p2, -tc2, tc2);
1615 Q0 = q0 + av_clip(((p1 + 2 * p0 + 2 * q0 + 2 * q1 + q2 + 4) >> 3) - q0, -tc2, tc2);
1616 Q1 = q1 + av_clip(((p0 + q0 + q1 + q2 + 2) >> 2) - q1, -tc2, tc2);
1617 Q2 = q2 + av_clip(((2 * q3 + 3 * q2 + q1 + q0 + p0 + 4) >> 3) - q2, -tc2, tc2);
1621 } else { // normal filtering
1624 const int tc_2 = tc >> 1;
1625 if (dp0 + dp3 < ((beta + (beta >> 1)) >> 3))
1627 if (dq0 + dq3 < ((beta + (beta >> 1)) >> 3))
1630 for (d = 0; d < 4; d++) {
1637 int delta0 = (9 * (q0 - p0) - 3 * (q1 - p1) + 8) >> 4;
1638 if (abs(delta0) < 10 * tc) {
1639 delta0 = av_clip(delta0, -tc, tc);
1641 P0 = av_clip_pixel(p0 + delta0);
1643 Q0 = av_clip_pixel(q0 - delta0);
1644 if (!no_p && nd_p > 1) {
1645 const int deltap1 = av_clip((((p2 + p0 + 1) >> 1) - p1 + delta0) >> 1, -tc_2, tc_2);
1646 P1 = av_clip_pixel(p1 + deltap1);
1648 if (!no_q && nd_q > 1) {
1649 const int deltaq1 = av_clip((((q2 + q0 + 1) >> 1) - q1 - delta0) >> 1, -tc_2, tc_2);
1650 Q1 = av_clip_pixel(q1 + deltaq1);
1660 static void FUNC(hevc_loop_filter_chroma)(uint8_t *_pix, ptrdiff_t _xstride,
1661 ptrdiff_t _ystride, int *_tc,
1662 uint8_t *_no_p, uint8_t *_no_q)
1664 int d, j, no_p, no_q;
1665 pixel *pix = (pixel *)_pix;
1666 ptrdiff_t xstride = _xstride / sizeof(pixel);
1667 ptrdiff_t ystride = _ystride / sizeof(pixel);
1669 for (j = 0; j < 2; j++) {
1670 const int tc = _tc[j] << (BIT_DEPTH - 8);
1678 for (d = 0; d < 4; d++) {
1684 delta0 = av_clip((((q0 - p0) * 4) + p1 - q1 + 4) >> 3, -tc, tc);
1686 P0 = av_clip_pixel(p0 + delta0);
1688 Q0 = av_clip_pixel(q0 - delta0);
1694 static void FUNC(hevc_h_loop_filter_chroma)(uint8_t *pix, ptrdiff_t stride,
1695 int *tc, uint8_t *no_p,
1698 FUNC(hevc_loop_filter_chroma)(pix, stride, sizeof(pixel), tc, no_p, no_q);
1701 static void FUNC(hevc_v_loop_filter_chroma)(uint8_t *pix, ptrdiff_t stride,
1702 int *tc, uint8_t *no_p,
1705 FUNC(hevc_loop_filter_chroma)(pix, sizeof(pixel), stride, tc, no_p, no_q);
1708 static void FUNC(hevc_h_loop_filter_luma)(uint8_t *pix, ptrdiff_t stride,
1709 int *beta, int *tc, uint8_t *no_p,
1712 FUNC(hevc_loop_filter_luma)(pix, stride, sizeof(pixel),
1713 beta, tc, no_p, no_q);
1716 static void FUNC(hevc_v_loop_filter_luma)(uint8_t *pix, ptrdiff_t stride,
1717 int *beta, int *tc, uint8_t *no_p,
1720 FUNC(hevc_loop_filter_luma)(pix, sizeof(pixel), stride,
1721 beta, tc, no_p, no_q);