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, 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 /= sizeof(pixel);
346 for (k = 0; k < 4; k++)
347 offset_table[(k + sao_left_class) & 31] = sao_offset_val[k + 1];
348 for (y = 0; y < height; y++) {
349 for (x = 0; x < width; x++)
350 dst[x] = av_clip_pixel(src[x] + offset_table[av_clip_pixel(src[x] >> shift)]);
356 #define CMP(a, b) ((a) > (b) ? 1 : ((a) == (b) ? 0 : -1))
358 static void FUNC(sao_edge_filter)(uint8_t *_dst, uint8_t *_src,
359 ptrdiff_t stride, SAOParams *sao,
360 int width, int height,
361 int c_idx, int init_x, int init_y) {
363 static const uint8_t edge_idx[] = { 1, 2, 0, 3, 4 };
364 static const int8_t pos[4][2][2] = {
365 { { -1, 0 }, { 1, 0 } }, // horizontal
366 { { 0, -1 }, { 0, 1 } }, // vertical
367 { { -1, -1 }, { 1, 1 } }, // 45 degree
368 { { 1, -1 }, { -1, 1 } }, // 135 degree
370 int *sao_offset_val = sao->offset_val[c_idx];
371 int sao_eo_class = sao->eo_class[c_idx];
372 pixel *dst = (pixel *)_dst;
373 pixel *src = (pixel *)_src;
375 int y_stride = init_y * stride;
376 int pos_0_0 = pos[sao_eo_class][0][0];
377 int pos_0_1 = pos[sao_eo_class][0][1];
378 int pos_1_0 = pos[sao_eo_class][1][0];
379 int pos_1_1 = pos[sao_eo_class][1][1];
382 int y_stride_0_1 = (init_y + pos_0_1) * stride;
383 int y_stride_1_1 = (init_y + pos_1_1) * stride;
384 for (y = init_y; y < height; y++) {
385 for (x = init_x; x < width; x++) {
386 int diff0 = CMP(src[x + y_stride], src[x + pos_0_0 + y_stride_0_1]);
387 int diff1 = CMP(src[x + y_stride], src[x + pos_1_0 + y_stride_1_1]);
388 int offset_val = edge_idx[2 + diff0 + diff1];
389 dst[x + y_stride] = av_clip_pixel(src[x + y_stride] + sao_offset_val[offset_val]);
392 y_stride_0_1 += stride;
393 y_stride_1_1 += stride;
397 static void FUNC(sao_edge_filter_0)(uint8_t *_dst, uint8_t *_src,
398 ptrdiff_t stride, SAOParams *sao,
399 int *borders, int _width, int _height,
400 int c_idx, uint8_t *vert_edge,
401 uint8_t *horiz_edge, uint8_t *diag_edge)
404 pixel *dst = (pixel *)_dst;
405 pixel *src = (pixel *)_src;
406 int *sao_offset_val = sao->offset_val[c_idx];
407 int sao_eo_class = sao->eo_class[c_idx];
408 int init_x = 0, init_y = 0, width = _width, height = _height;
410 stride /= sizeof(pixel);
412 if (sao_eo_class != SAO_EO_VERT) {
414 int offset_val = sao_offset_val[0];
416 for (y = 0; y < height; y++) {
417 dst[y_stride] = av_clip_pixel(src[y_stride] + offset_val);
423 int offset_val = sao_offset_val[0];
424 int x_stride = width - 1;
425 for (x = 0; x < height; x++) {
426 dst[x_stride] = av_clip_pixel(src[x_stride] + offset_val);
432 if (sao_eo_class != SAO_EO_HORIZ) {
434 int offset_val = sao_offset_val[0];
435 for (x = init_x; x < width; x++)
436 dst[x] = av_clip_pixel(src[x] + offset_val);
440 int offset_val = sao_offset_val[0];
441 int y_stride = stride * (height - 1);
442 for (x = init_x; x < width; x++)
443 dst[x + y_stride] = av_clip_pixel(src[x + y_stride] + offset_val);
448 FUNC(sao_edge_filter)((uint8_t *)dst, (uint8_t *)src, stride, sao, width, height, c_idx, init_x, init_y);
451 static void FUNC(sao_edge_filter_1)(uint8_t *_dst, uint8_t *_src,
452 ptrdiff_t stride, SAOParams *sao,
453 int *borders, int _width, int _height,
454 int c_idx, uint8_t *vert_edge,
455 uint8_t *horiz_edge, uint8_t *diag_edge)
458 pixel *dst = (pixel *)_dst;
459 pixel *src = (pixel *)_src;
460 int *sao_offset_val = sao->offset_val[c_idx];
461 int sao_eo_class = sao->eo_class[c_idx];
462 int init_x = 0, init_y = 0, width = _width, height = _height;
464 stride /= sizeof(pixel);
466 if (sao_eo_class != SAO_EO_VERT) {
468 int offset_val = sao_offset_val[0];
470 for (y = 0; y < height; y++) {
471 dst[y_stride] = av_clip_pixel(src[y_stride] + offset_val);
477 int offset_val = sao_offset_val[0];
478 int x_stride = width - 1;
479 for (x = 0; x < height; x++) {
480 dst[x_stride] = av_clip_pixel(src[x_stride] + 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 = stride * (height - 1);
496 for (x = init_x; x < width; x++)
497 dst[x + y_stride] = av_clip_pixel(src[x + y_stride] + offset_val);
502 FUNC(sao_edge_filter)((uint8_t *)dst, (uint8_t *)src, stride, sao, width, height, c_idx, init_x, init_y);
505 int save_upper_left = !diag_edge[0] && sao_eo_class == SAO_EO_135D && !borders[0] && !borders[1];
506 int save_upper_right = !diag_edge[1] && sao_eo_class == SAO_EO_45D && !borders[1] && !borders[2];
507 int save_lower_right = !diag_edge[2] && sao_eo_class == SAO_EO_135D && !borders[2] && !borders[3];
508 int save_lower_left = !diag_edge[3] && sao_eo_class == SAO_EO_45D && !borders[0] && !borders[3];
510 // Restore pixels that can't be modified
511 if(vert_edge[0] && sao_eo_class != SAO_EO_VERT) {
512 for(y = init_y+save_upper_left; y< height-save_lower_left; y++)
513 dst[y*stride] = src[y*stride];
515 if(vert_edge[1] && sao_eo_class != SAO_EO_VERT) {
516 for(y = init_y+save_upper_right; y< height-save_lower_right; y++)
517 dst[y*stride+width-1] = src[y*stride+width-1];
520 if(horiz_edge[0] && sao_eo_class != SAO_EO_HORIZ) {
521 for(x = init_x+save_upper_left; x < width-save_upper_right; x++)
524 if(horiz_edge[1] && sao_eo_class != SAO_EO_HORIZ) {
525 for(x = init_x+save_lower_left; x < width-save_lower_right; x++)
526 dst[(height-1)*stride+x] = src[(height-1)*stride+x];
528 if(diag_edge[0] && sao_eo_class == SAO_EO_135D)
530 if(diag_edge[1] && sao_eo_class == SAO_EO_45D)
531 dst[width-1] = src[width-1];
532 if(diag_edge[2] && sao_eo_class == SAO_EO_135D)
533 dst[stride*(height-1)+width-1] = src[stride*(height-1)+width-1];
534 if(diag_edge[3] && sao_eo_class == SAO_EO_45D)
535 dst[stride*(height-1)] = src[stride*(height-1)];
542 ////////////////////////////////////////////////////////////////////////////////
544 ////////////////////////////////////////////////////////////////////////////////
545 static void FUNC(put_hevc_pel_pixels)(int16_t *dst, ptrdiff_t dststride,
546 uint8_t *_src, ptrdiff_t _srcstride,
547 int height, intptr_t mx, intptr_t my, int width)
550 pixel *src = (pixel *)_src;
551 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
553 for (y = 0; y < height; y++) {
554 for (x = 0; x < width; x++)
555 dst[x] = src[x] << (14 - BIT_DEPTH);
561 static void FUNC(put_hevc_pel_uni_pixels)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
562 int height, intptr_t mx, intptr_t my, int width)
565 pixel *src = (pixel *)_src;
566 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
567 pixel *dst = (pixel *)_dst;
568 ptrdiff_t dststride = _dststride / sizeof(pixel);
570 for (y = 0; y < height; y++) {
571 memcpy(dst, src, width * sizeof(pixel));
577 static void FUNC(put_hevc_pel_bi_pixels)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
578 int16_t *src2, ptrdiff_t src2stride,
579 int height, intptr_t mx, intptr_t my, int width)
582 pixel *src = (pixel *)_src;
583 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
584 pixel *dst = (pixel *)_dst;
585 ptrdiff_t dststride = _dststride / sizeof(pixel);
587 int shift = 14 + 1 - BIT_DEPTH;
589 int offset = 1 << (shift - 1);
594 for (y = 0; y < height; y++) {
595 for (x = 0; x < width; x++)
596 dst[x] = av_clip_pixel(((src[x] << (14 - BIT_DEPTH)) + src2[x] + offset) >> shift);
603 static void FUNC(put_hevc_pel_uni_w_pixels)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
604 int height, int denom, int wx, int ox, intptr_t mx, intptr_t my, int width)
607 pixel *src = (pixel *)_src;
608 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
609 pixel *dst = (pixel *)_dst;
610 ptrdiff_t dststride = _dststride / sizeof(pixel);
611 int shift = denom + 14 - BIT_DEPTH;
613 int offset = 1 << (shift - 1);
618 ox = ox * (1 << (BIT_DEPTH - 8));
619 for (y = 0; y < height; y++) {
620 for (x = 0; x < width; x++)
621 dst[x] = av_clip_pixel((((src[x] << (14 - BIT_DEPTH)) * wx + offset) >> shift) + ox);
627 static void FUNC(put_hevc_pel_bi_w_pixels)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
628 int16_t *src2, ptrdiff_t src2stride,
629 int height, int denom, int wx0, int wx1,
630 int ox0, int ox1, intptr_t mx, intptr_t my, int width)
633 pixel *src = (pixel *)_src;
634 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
635 pixel *dst = (pixel *)_dst;
636 ptrdiff_t dststride = _dststride / sizeof(pixel);
638 int shift = 14 + 1 - BIT_DEPTH;
639 int log2Wd = denom + shift - 1;
641 ox0 = ox0 * (1 << (BIT_DEPTH - 8));
642 ox1 = ox1 * (1 << (BIT_DEPTH - 8));
643 for (y = 0; y < height; y++) {
644 for (x = 0; x < width; x++) {
645 dst[x] = av_clip_pixel(( (src[x] << (14 - BIT_DEPTH)) * wx1 + src2[x] * wx0 + ((ox0 + ox1 + 1) << log2Wd)) >> (log2Wd + 1));
653 ////////////////////////////////////////////////////////////////////////////////
655 ////////////////////////////////////////////////////////////////////////////////
656 #define QPEL_FILTER(src, stride) \
657 (filter[0] * src[x - 3 * stride] + \
658 filter[1] * src[x - 2 * stride] + \
659 filter[2] * src[x - stride] + \
660 filter[3] * src[x ] + \
661 filter[4] * src[x + stride] + \
662 filter[5] * src[x + 2 * stride] + \
663 filter[6] * src[x + 3 * stride] + \
664 filter[7] * src[x + 4 * stride])
666 static void FUNC(put_hevc_qpel_h)(int16_t *dst, ptrdiff_t dststride,
667 uint8_t *_src, ptrdiff_t _srcstride,
668 int height, intptr_t mx, intptr_t my, int width)
671 pixel *src = (pixel*)_src;
672 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
673 const int8_t *filter = ff_hevc_qpel_filters[mx - 1];
674 for (y = 0; y < height; y++) {
675 for (x = 0; x < width; x++)
676 dst[x] = QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8);
682 static void FUNC(put_hevc_qpel_v)(int16_t *dst, ptrdiff_t dststride,
683 uint8_t *_src, ptrdiff_t _srcstride,
684 int height, intptr_t mx, intptr_t my, int width)
687 pixel *src = (pixel*)_src;
688 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
689 const int8_t *filter = ff_hevc_qpel_filters[my - 1];
690 for (y = 0; y < height; y++) {
691 for (x = 0; x < width; x++)
692 dst[x] = QPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8);
698 static void FUNC(put_hevc_qpel_hv)(int16_t *dst,
701 ptrdiff_t _srcstride,
702 int height, intptr_t mx,
703 intptr_t my, int width)
706 const int8_t *filter;
707 pixel *src = (pixel*)_src;
708 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
709 int16_t tmp_array[(MAX_PB_SIZE + QPEL_EXTRA) * MAX_PB_SIZE];
710 int16_t *tmp = tmp_array;
712 src -= QPEL_EXTRA_BEFORE * srcstride;
713 filter = ff_hevc_qpel_filters[mx - 1];
714 for (y = 0; y < height + QPEL_EXTRA; y++) {
715 for (x = 0; x < width; x++)
716 tmp[x] = QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8);
721 tmp = tmp_array + QPEL_EXTRA_BEFORE * MAX_PB_SIZE;
722 filter = ff_hevc_qpel_filters[my - 1];
723 for (y = 0; y < height; y++) {
724 for (x = 0; x < width; x++)
725 dst[x] = QPEL_FILTER(tmp, MAX_PB_SIZE) >> 6;
731 static void FUNC(put_hevc_qpel_uni_h)(uint8_t *_dst, ptrdiff_t _dststride,
732 uint8_t *_src, ptrdiff_t _srcstride,
733 int height, intptr_t mx, intptr_t my, int width)
736 pixel *src = (pixel*)_src;
737 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
738 pixel *dst = (pixel *)_dst;
739 ptrdiff_t dststride = _dststride / sizeof(pixel);
740 const int8_t *filter = ff_hevc_qpel_filters[mx - 1];
741 int shift = 14 - BIT_DEPTH;
744 int offset = 1 << (shift - 1);
749 for (y = 0; y < height; y++) {
750 for (x = 0; x < width; x++)
751 dst[x] = av_clip_pixel(((QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) + offset) >> shift);
757 static void FUNC(put_hevc_qpel_bi_h)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
758 int16_t *src2, ptrdiff_t src2stride,
759 int height, intptr_t mx, intptr_t my, int width)
762 pixel *src = (pixel*)_src;
763 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
764 pixel *dst = (pixel *)_dst;
765 ptrdiff_t dststride = _dststride / sizeof(pixel);
767 const int8_t *filter = ff_hevc_qpel_filters[mx - 1];
769 int shift = 14 + 1 - BIT_DEPTH;
771 int offset = 1 << (shift - 1);
776 for (y = 0; y < height; y++) {
777 for (x = 0; x < width; x++)
778 dst[x] = av_clip_pixel(((QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) + src2[x] + offset) >> shift);
785 static void FUNC(put_hevc_qpel_uni_v)(uint8_t *_dst, ptrdiff_t _dststride,
786 uint8_t *_src, ptrdiff_t _srcstride,
787 int height, intptr_t mx, intptr_t my, int width)
790 pixel *src = (pixel*)_src;
791 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
792 pixel *dst = (pixel *)_dst;
793 ptrdiff_t dststride = _dststride / sizeof(pixel);
794 const int8_t *filter = ff_hevc_qpel_filters[my - 1];
795 int shift = 14 - BIT_DEPTH;
798 int offset = 1 << (shift - 1);
803 for (y = 0; y < height; y++) {
804 for (x = 0; x < width; x++)
805 dst[x] = av_clip_pixel(((QPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) + offset) >> shift);
812 static void FUNC(put_hevc_qpel_bi_v)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
813 int16_t *src2, ptrdiff_t src2stride,
814 int height, intptr_t mx, intptr_t my, int width)
817 pixel *src = (pixel*)_src;
818 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
819 pixel *dst = (pixel *)_dst;
820 ptrdiff_t dststride = _dststride / sizeof(pixel);
822 const int8_t *filter = ff_hevc_qpel_filters[my - 1];
824 int shift = 14 + 1 - BIT_DEPTH;
826 int offset = 1 << (shift - 1);
831 for (y = 0; y < height; y++) {
832 for (x = 0; x < width; x++)
833 dst[x] = av_clip_pixel(((QPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) + src2[x] + offset) >> shift);
840 static void FUNC(put_hevc_qpel_uni_hv)(uint8_t *_dst, ptrdiff_t _dststride,
841 uint8_t *_src, ptrdiff_t _srcstride,
842 int height, intptr_t mx, intptr_t my, int width)
845 const int8_t *filter;
846 pixel *src = (pixel*)_src;
847 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
848 pixel *dst = (pixel *)_dst;
849 ptrdiff_t dststride = _dststride / sizeof(pixel);
850 int16_t tmp_array[(MAX_PB_SIZE + QPEL_EXTRA) * MAX_PB_SIZE];
851 int16_t *tmp = tmp_array;
852 int shift = 14 - BIT_DEPTH;
855 int offset = 1 << (shift - 1);
860 src -= QPEL_EXTRA_BEFORE * srcstride;
861 filter = ff_hevc_qpel_filters[mx - 1];
862 for (y = 0; y < height + QPEL_EXTRA; y++) {
863 for (x = 0; x < width; x++)
864 tmp[x] = QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8);
869 tmp = tmp_array + QPEL_EXTRA_BEFORE * MAX_PB_SIZE;
870 filter = ff_hevc_qpel_filters[my - 1];
872 for (y = 0; y < height; y++) {
873 for (x = 0; x < width; x++)
874 dst[x] = av_clip_pixel(((QPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) + offset) >> shift);
880 static void FUNC(put_hevc_qpel_bi_hv)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
881 int16_t *src2, ptrdiff_t src2stride,
882 int height, intptr_t mx, intptr_t my, int width)
885 const int8_t *filter;
886 pixel *src = (pixel*)_src;
887 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
888 pixel *dst = (pixel *)_dst;
889 ptrdiff_t dststride = _dststride / sizeof(pixel);
890 int16_t tmp_array[(MAX_PB_SIZE + QPEL_EXTRA) * MAX_PB_SIZE];
891 int16_t *tmp = tmp_array;
892 int shift = 14 + 1 - BIT_DEPTH;
894 int offset = 1 << (shift - 1);
899 src -= QPEL_EXTRA_BEFORE * srcstride;
900 filter = ff_hevc_qpel_filters[mx - 1];
901 for (y = 0; y < height + QPEL_EXTRA; y++) {
902 for (x = 0; x < width; x++)
903 tmp[x] = QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8);
908 tmp = tmp_array + QPEL_EXTRA_BEFORE * MAX_PB_SIZE;
909 filter = ff_hevc_qpel_filters[my - 1];
911 for (y = 0; y < height; y++) {
912 for (x = 0; x < width; x++)
913 dst[x] = av_clip_pixel(((QPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) + src2[x] + offset) >> shift);
920 static void FUNC(put_hevc_qpel_uni_w_h)(uint8_t *_dst, ptrdiff_t _dststride,
921 uint8_t *_src, ptrdiff_t _srcstride,
922 int height, int denom, int wx, int ox,
923 intptr_t mx, intptr_t my, int width)
926 pixel *src = (pixel*)_src;
927 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
928 pixel *dst = (pixel *)_dst;
929 ptrdiff_t dststride = _dststride / sizeof(pixel);
930 const int8_t *filter = ff_hevc_qpel_filters[mx - 1];
931 int shift = denom + 14 - BIT_DEPTH;
933 int offset = 1 << (shift - 1);
938 ox = ox * (1 << (BIT_DEPTH - 8));
939 for (y = 0; y < height; y++) {
940 for (x = 0; x < width; x++)
941 dst[x] = av_clip_pixel((((QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) * wx + offset) >> shift) + ox);
947 static void FUNC(put_hevc_qpel_bi_w_h)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
948 int16_t *src2, ptrdiff_t src2stride,
949 int height, int denom, int wx0, int wx1,
950 int ox0, int ox1, intptr_t mx, intptr_t my, int width)
953 pixel *src = (pixel*)_src;
954 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
955 pixel *dst = (pixel *)_dst;
956 ptrdiff_t dststride = _dststride / sizeof(pixel);
958 const int8_t *filter = ff_hevc_qpel_filters[mx - 1];
960 int shift = 14 + 1 - BIT_DEPTH;
961 int log2Wd = denom + shift - 1;
963 ox0 = ox0 * (1 << (BIT_DEPTH - 8));
964 ox1 = ox1 * (1 << (BIT_DEPTH - 8));
965 for (y = 0; y < height; y++) {
966 for (x = 0; x < width; x++)
967 dst[x] = av_clip_pixel(((QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) * wx1 + src2[x] * wx0 +
968 ((ox0 + ox1 + 1) << log2Wd)) >> (log2Wd + 1));
975 static void FUNC(put_hevc_qpel_uni_w_v)(uint8_t *_dst, ptrdiff_t _dststride,
976 uint8_t *_src, ptrdiff_t _srcstride,
977 int height, int denom, int wx, int ox,
978 intptr_t mx, intptr_t my, int width)
981 pixel *src = (pixel*)_src;
982 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
983 pixel *dst = (pixel *)_dst;
984 ptrdiff_t dststride = _dststride / sizeof(pixel);
985 const int8_t *filter = ff_hevc_qpel_filters[my - 1];
986 int shift = denom + 14 - BIT_DEPTH;
988 int offset = 1 << (shift - 1);
993 ox = ox * (1 << (BIT_DEPTH - 8));
994 for (y = 0; y < height; y++) {
995 for (x = 0; x < width; x++)
996 dst[x] = av_clip_pixel((((QPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) * wx + offset) >> shift) + ox);
1002 static void FUNC(put_hevc_qpel_bi_w_v)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1003 int16_t *src2, ptrdiff_t src2stride,
1004 int height, int denom, int wx0, int wx1,
1005 int ox0, int ox1, intptr_t mx, intptr_t my, int width)
1008 pixel *src = (pixel*)_src;
1009 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1010 pixel *dst = (pixel *)_dst;
1011 ptrdiff_t dststride = _dststride / sizeof(pixel);
1013 const int8_t *filter = ff_hevc_qpel_filters[my - 1];
1015 int shift = 14 + 1 - BIT_DEPTH;
1016 int log2Wd = denom + shift - 1;
1018 ox0 = ox0 * (1 << (BIT_DEPTH - 8));
1019 ox1 = ox1 * (1 << (BIT_DEPTH - 8));
1020 for (y = 0; y < height; y++) {
1021 for (x = 0; x < width; x++)
1022 dst[x] = av_clip_pixel(((QPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) * wx1 + src2[x] * wx0 +
1023 ((ox0 + ox1 + 1) << log2Wd)) >> (log2Wd + 1));
1030 static void FUNC(put_hevc_qpel_uni_w_hv)(uint8_t *_dst, ptrdiff_t _dststride,
1031 uint8_t *_src, ptrdiff_t _srcstride,
1032 int height, int denom, int wx, int ox,
1033 intptr_t mx, intptr_t my, int width)
1036 const int8_t *filter;
1037 pixel *src = (pixel*)_src;
1038 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1039 pixel *dst = (pixel *)_dst;
1040 ptrdiff_t dststride = _dststride / sizeof(pixel);
1041 int16_t tmp_array[(MAX_PB_SIZE + QPEL_EXTRA) * MAX_PB_SIZE];
1042 int16_t *tmp = tmp_array;
1043 int shift = denom + 14 - BIT_DEPTH;
1045 int offset = 1 << (shift - 1);
1050 src -= QPEL_EXTRA_BEFORE * srcstride;
1051 filter = ff_hevc_qpel_filters[mx - 1];
1052 for (y = 0; y < height + QPEL_EXTRA; y++) {
1053 for (x = 0; x < width; x++)
1054 tmp[x] = QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8);
1059 tmp = tmp_array + QPEL_EXTRA_BEFORE * MAX_PB_SIZE;
1060 filter = ff_hevc_qpel_filters[my - 1];
1062 ox = ox * (1 << (BIT_DEPTH - 8));
1063 for (y = 0; y < height; y++) {
1064 for (x = 0; x < width; x++)
1065 dst[x] = av_clip_pixel((((QPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) * wx + offset) >> shift) + ox);
1071 static void FUNC(put_hevc_qpel_bi_w_hv)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1072 int16_t *src2, ptrdiff_t src2stride,
1073 int height, int denom, int wx0, int wx1,
1074 int ox0, int ox1, intptr_t mx, intptr_t my, int width)
1077 const int8_t *filter;
1078 pixel *src = (pixel*)_src;
1079 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1080 pixel *dst = (pixel *)_dst;
1081 ptrdiff_t dststride = _dststride / sizeof(pixel);
1082 int16_t tmp_array[(MAX_PB_SIZE + QPEL_EXTRA) * MAX_PB_SIZE];
1083 int16_t *tmp = tmp_array;
1084 int shift = 14 + 1 - BIT_DEPTH;
1085 int log2Wd = denom + shift - 1;
1087 src -= QPEL_EXTRA_BEFORE * srcstride;
1088 filter = ff_hevc_qpel_filters[mx - 1];
1089 for (y = 0; y < height + QPEL_EXTRA; y++) {
1090 for (x = 0; x < width; x++)
1091 tmp[x] = QPEL_FILTER(src, 1) >> (BIT_DEPTH - 8);
1096 tmp = tmp_array + QPEL_EXTRA_BEFORE * MAX_PB_SIZE;
1097 filter = ff_hevc_qpel_filters[my - 1];
1099 ox0 = ox0 * (1 << (BIT_DEPTH - 8));
1100 ox1 = ox1 * (1 << (BIT_DEPTH - 8));
1101 for (y = 0; y < height; y++) {
1102 for (x = 0; x < width; x++)
1103 dst[x] = av_clip_pixel(((QPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) * wx1 + src2[x] * wx0 +
1104 ((ox0 + ox1 + 1) << log2Wd)) >> (log2Wd + 1));
1111 ////////////////////////////////////////////////////////////////////////////////
1113 ////////////////////////////////////////////////////////////////////////////////
1114 #define EPEL_FILTER(src, stride) \
1115 (filter[0] * src[x - stride] + \
1116 filter[1] * src[x] + \
1117 filter[2] * src[x + stride] + \
1118 filter[3] * src[x + 2 * stride])
1120 static void FUNC(put_hevc_epel_h)(int16_t *dst, ptrdiff_t dststride,
1121 uint8_t *_src, ptrdiff_t _srcstride,
1122 int height, intptr_t mx, intptr_t my, int width)
1125 pixel *src = (pixel *)_src;
1126 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1127 const int8_t *filter = ff_hevc_epel_filters[mx - 1];
1128 for (y = 0; y < height; y++) {
1129 for (x = 0; x < width; x++)
1130 dst[x] = EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8);
1136 static void FUNC(put_hevc_epel_v)(int16_t *dst, ptrdiff_t dststride,
1137 uint8_t *_src, ptrdiff_t _srcstride,
1138 int height, intptr_t mx, intptr_t my, int width)
1141 pixel *src = (pixel *)_src;
1142 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1143 const int8_t *filter = ff_hevc_epel_filters[my - 1];
1145 for (y = 0; y < height; y++) {
1146 for (x = 0; x < width; x++)
1147 dst[x] = EPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8);
1153 static void FUNC(put_hevc_epel_hv)(int16_t *dst, ptrdiff_t dststride,
1154 uint8_t *_src, ptrdiff_t _srcstride,
1155 int height, intptr_t mx, intptr_t my, int width)
1158 pixel *src = (pixel *)_src;
1159 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1160 const int8_t *filter = ff_hevc_epel_filters[mx - 1];
1161 int16_t tmp_array[(MAX_PB_SIZE + EPEL_EXTRA) * MAX_PB_SIZE];
1162 int16_t *tmp = tmp_array;
1164 src -= EPEL_EXTRA_BEFORE * srcstride;
1166 for (y = 0; y < height + EPEL_EXTRA; y++) {
1167 for (x = 0; x < width; x++)
1168 tmp[x] = EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8);
1173 tmp = tmp_array + EPEL_EXTRA_BEFORE * MAX_PB_SIZE;
1174 filter = ff_hevc_epel_filters[my - 1];
1176 for (y = 0; y < height; y++) {
1177 for (x = 0; x < width; x++)
1178 dst[x] = EPEL_FILTER(tmp, MAX_PB_SIZE) >> 6;
1184 static void FUNC(put_hevc_epel_uni_h)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1185 int height, intptr_t mx, intptr_t my, int width)
1188 pixel *src = (pixel *)_src;
1189 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1190 pixel *dst = (pixel *)_dst;
1191 ptrdiff_t dststride = _dststride / sizeof(pixel);
1192 const int8_t *filter = ff_hevc_epel_filters[mx - 1];
1193 int shift = 14 - BIT_DEPTH;
1195 int offset = 1 << (shift - 1);
1200 for (y = 0; y < height; y++) {
1201 for (x = 0; x < width; x++)
1202 dst[x] = av_clip_pixel(((EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) + offset) >> shift);
1208 static void FUNC(put_hevc_epel_bi_h)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1209 int16_t *src2, ptrdiff_t src2stride,
1210 int height, intptr_t mx, intptr_t my, int width)
1213 pixel *src = (pixel *)_src;
1214 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1215 pixel *dst = (pixel *)_dst;
1216 ptrdiff_t dststride = _dststride / sizeof(pixel);
1217 const int8_t *filter = ff_hevc_epel_filters[mx - 1];
1218 int shift = 14 + 1 - BIT_DEPTH;
1220 int offset = 1 << (shift - 1);
1225 for (y = 0; y < height; y++) {
1226 for (x = 0; x < width; x++) {
1227 dst[x] = av_clip_pixel(((EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) + src2[x] + offset) >> shift);
1235 static void FUNC(put_hevc_epel_uni_v)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1236 int height, intptr_t mx, intptr_t my, int width)
1239 pixel *src = (pixel *)_src;
1240 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1241 pixel *dst = (pixel *)_dst;
1242 ptrdiff_t dststride = _dststride / sizeof(pixel);
1243 const int8_t *filter = ff_hevc_epel_filters[my - 1];
1244 int shift = 14 - BIT_DEPTH;
1246 int offset = 1 << (shift - 1);
1251 for (y = 0; y < height; y++) {
1252 for (x = 0; x < width; x++)
1253 dst[x] = av_clip_pixel(((EPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) + offset) >> shift);
1259 static void FUNC(put_hevc_epel_bi_v)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1260 int16_t *src2, ptrdiff_t src2stride,
1261 int height, intptr_t mx, intptr_t my, int width)
1264 pixel *src = (pixel *)_src;
1265 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1266 const int8_t *filter = ff_hevc_epel_filters[my - 1];
1267 pixel *dst = (pixel *)_dst;
1268 ptrdiff_t dststride = _dststride / sizeof(pixel);
1269 int shift = 14 + 1 - BIT_DEPTH;
1271 int offset = 1 << (shift - 1);
1276 for (y = 0; y < height; y++) {
1277 for (x = 0; x < width; x++)
1278 dst[x] = av_clip_pixel(((EPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) + src2[x] + offset) >> shift);
1285 static void FUNC(put_hevc_epel_uni_hv)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1286 int height, intptr_t mx, intptr_t my, int width)
1289 pixel *src = (pixel *)_src;
1290 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1291 pixel *dst = (pixel *)_dst;
1292 ptrdiff_t dststride = _dststride / sizeof(pixel);
1293 const int8_t *filter = ff_hevc_epel_filters[mx - 1];
1294 int16_t tmp_array[(MAX_PB_SIZE + EPEL_EXTRA) * MAX_PB_SIZE];
1295 int16_t *tmp = tmp_array;
1296 int shift = 14 - BIT_DEPTH;
1298 int offset = 1 << (shift - 1);
1303 src -= EPEL_EXTRA_BEFORE * srcstride;
1305 for (y = 0; y < height + EPEL_EXTRA; y++) {
1306 for (x = 0; x < width; x++)
1307 tmp[x] = EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8);
1312 tmp = tmp_array + EPEL_EXTRA_BEFORE * MAX_PB_SIZE;
1313 filter = ff_hevc_epel_filters[my - 1];
1315 for (y = 0; y < height; y++) {
1316 for (x = 0; x < width; x++)
1317 dst[x] = av_clip_pixel(((EPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) + offset) >> shift);
1323 static void FUNC(put_hevc_epel_bi_hv)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1324 int16_t *src2, ptrdiff_t src2stride,
1325 int height, intptr_t mx, intptr_t my, int width)
1328 pixel *src = (pixel *)_src;
1329 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1330 pixel *dst = (pixel *)_dst;
1331 ptrdiff_t dststride = _dststride / sizeof(pixel);
1332 const int8_t *filter = ff_hevc_epel_filters[mx - 1];
1333 int16_t tmp_array[(MAX_PB_SIZE + EPEL_EXTRA) * MAX_PB_SIZE];
1334 int16_t *tmp = tmp_array;
1335 int shift = 14 + 1 - BIT_DEPTH;
1337 int offset = 1 << (shift - 1);
1342 src -= EPEL_EXTRA_BEFORE * srcstride;
1344 for (y = 0; y < height + EPEL_EXTRA; y++) {
1345 for (x = 0; x < width; x++)
1346 tmp[x] = EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8);
1351 tmp = tmp_array + EPEL_EXTRA_BEFORE * MAX_PB_SIZE;
1352 filter = ff_hevc_epel_filters[my - 1];
1354 for (y = 0; y < height; y++) {
1355 for (x = 0; x < width; x++)
1356 dst[x] = av_clip_pixel(((EPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) + src2[x] + offset) >> shift);
1363 static void FUNC(put_hevc_epel_uni_w_h)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1364 int height, int denom, int wx, int ox, intptr_t mx, intptr_t my, int width)
1367 pixel *src = (pixel *)_src;
1368 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1369 pixel *dst = (pixel *)_dst;
1370 ptrdiff_t dststride = _dststride / sizeof(pixel);
1371 const int8_t *filter = ff_hevc_epel_filters[mx - 1];
1372 int shift = denom + 14 - BIT_DEPTH;
1374 int offset = 1 << (shift - 1);
1379 ox = ox * (1 << (BIT_DEPTH - 8));
1380 for (y = 0; y < height; y++) {
1381 for (x = 0; x < width; x++) {
1382 dst[x] = av_clip_pixel((((EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) * wx + offset) >> shift) + ox);
1389 static void FUNC(put_hevc_epel_bi_w_h)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1390 int16_t *src2, ptrdiff_t src2stride,
1391 int height, int denom, int wx0, int wx1,
1392 int ox0, int ox1, intptr_t mx, intptr_t my, int width)
1395 pixel *src = (pixel *)_src;
1396 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1397 pixel *dst = (pixel *)_dst;
1398 ptrdiff_t dststride = _dststride / sizeof(pixel);
1399 const int8_t *filter = ff_hevc_epel_filters[mx - 1];
1400 int shift = 14 + 1 - BIT_DEPTH;
1401 int log2Wd = denom + shift - 1;
1403 ox0 = ox0 * (1 << (BIT_DEPTH - 8));
1404 ox1 = ox1 * (1 << (BIT_DEPTH - 8));
1405 for (y = 0; y < height; y++) {
1406 for (x = 0; x < width; x++)
1407 dst[x] = av_clip_pixel(((EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8)) * wx1 + src2[x] * wx0 +
1408 ((ox0 + ox1 + 1) << log2Wd)) >> (log2Wd + 1));
1415 static void FUNC(put_hevc_epel_uni_w_v)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1416 int height, int denom, int wx, int ox, intptr_t mx, intptr_t my, int width)
1419 pixel *src = (pixel *)_src;
1420 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1421 pixel *dst = (pixel *)_dst;
1422 ptrdiff_t dststride = _dststride / sizeof(pixel);
1423 const int8_t *filter = ff_hevc_epel_filters[my - 1];
1424 int shift = denom + 14 - BIT_DEPTH;
1426 int offset = 1 << (shift - 1);
1431 ox = ox * (1 << (BIT_DEPTH - 8));
1432 for (y = 0; y < height; y++) {
1433 for (x = 0; x < width; x++) {
1434 dst[x] = av_clip_pixel((((EPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) * wx + offset) >> shift) + ox);
1441 static void FUNC(put_hevc_epel_bi_w_v)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1442 int16_t *src2, ptrdiff_t src2stride,
1443 int height, int denom, int wx0, int wx1,
1444 int ox0, int ox1, intptr_t mx, intptr_t my, int width)
1447 pixel *src = (pixel *)_src;
1448 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1449 const int8_t *filter = ff_hevc_epel_filters[my - 1];
1450 pixel *dst = (pixel *)_dst;
1451 ptrdiff_t dststride = _dststride / sizeof(pixel);
1452 int shift = 14 + 1 - BIT_DEPTH;
1453 int log2Wd = denom + shift - 1;
1455 ox0 = ox0 * (1 << (BIT_DEPTH - 8));
1456 ox1 = ox1 * (1 << (BIT_DEPTH - 8));
1457 for (y = 0; y < height; y++) {
1458 for (x = 0; x < width; x++)
1459 dst[x] = av_clip_pixel(((EPEL_FILTER(src, srcstride) >> (BIT_DEPTH - 8)) * wx1 + src2[x] * wx0 +
1460 ((ox0 + ox1 + 1) << log2Wd)) >> (log2Wd + 1));
1467 static void FUNC(put_hevc_epel_uni_w_hv)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1468 int height, int denom, int wx, int ox, intptr_t mx, intptr_t my, int width)
1471 pixel *src = (pixel *)_src;
1472 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1473 pixel *dst = (pixel *)_dst;
1474 ptrdiff_t dststride = _dststride / sizeof(pixel);
1475 const int8_t *filter = ff_hevc_epel_filters[mx - 1];
1476 int16_t tmp_array[(MAX_PB_SIZE + EPEL_EXTRA) * MAX_PB_SIZE];
1477 int16_t *tmp = tmp_array;
1478 int shift = denom + 14 - BIT_DEPTH;
1480 int offset = 1 << (shift - 1);
1485 src -= EPEL_EXTRA_BEFORE * srcstride;
1487 for (y = 0; y < height + EPEL_EXTRA; y++) {
1488 for (x = 0; x < width; x++)
1489 tmp[x] = EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8);
1494 tmp = tmp_array + EPEL_EXTRA_BEFORE * MAX_PB_SIZE;
1495 filter = ff_hevc_epel_filters[my - 1];
1497 ox = ox * (1 << (BIT_DEPTH - 8));
1498 for (y = 0; y < height; y++) {
1499 for (x = 0; x < width; x++)
1500 dst[x] = av_clip_pixel((((EPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) * wx + offset) >> shift) + ox);
1506 static void FUNC(put_hevc_epel_bi_w_hv)(uint8_t *_dst, ptrdiff_t _dststride, uint8_t *_src, ptrdiff_t _srcstride,
1507 int16_t *src2, ptrdiff_t src2stride,
1508 int height, int denom, int wx0, int wx1,
1509 int ox0, int ox1, intptr_t mx, intptr_t my, int width)
1512 pixel *src = (pixel *)_src;
1513 ptrdiff_t srcstride = _srcstride / sizeof(pixel);
1514 pixel *dst = (pixel *)_dst;
1515 ptrdiff_t dststride = _dststride / sizeof(pixel);
1516 const int8_t *filter = ff_hevc_epel_filters[mx - 1];
1517 int16_t tmp_array[(MAX_PB_SIZE + EPEL_EXTRA) * MAX_PB_SIZE];
1518 int16_t *tmp = tmp_array;
1519 int shift = 14 + 1 - BIT_DEPTH;
1520 int log2Wd = denom + shift - 1;
1522 src -= EPEL_EXTRA_BEFORE * srcstride;
1524 for (y = 0; y < height + EPEL_EXTRA; y++) {
1525 for (x = 0; x < width; x++)
1526 tmp[x] = EPEL_FILTER(src, 1) >> (BIT_DEPTH - 8);
1531 tmp = tmp_array + EPEL_EXTRA_BEFORE * MAX_PB_SIZE;
1532 filter = ff_hevc_epel_filters[my - 1];
1534 ox0 = ox0 * (1 << (BIT_DEPTH - 8));
1535 ox1 = ox1 * (1 << (BIT_DEPTH - 8));
1536 for (y = 0; y < height; y++) {
1537 for (x = 0; x < width; x++)
1538 dst[x] = av_clip_pixel(((EPEL_FILTER(tmp, MAX_PB_SIZE) >> 6) * wx1 + src2[x] * wx0 +
1539 ((ox0 + ox1 + 1) << log2Wd)) >> (log2Wd + 1));
1545 #define P3 pix[-4 * xstride]
1546 #define P2 pix[-3 * xstride]
1547 #define P1 pix[-2 * xstride]
1548 #define P0 pix[-1 * xstride]
1549 #define Q0 pix[0 * xstride]
1550 #define Q1 pix[1 * xstride]
1551 #define Q2 pix[2 * xstride]
1552 #define Q3 pix[3 * xstride]
1554 // line three. used only for deblocking decision
1555 #define TP3 pix[-4 * xstride + 3 * ystride]
1556 #define TP2 pix[-3 * xstride + 3 * ystride]
1557 #define TP1 pix[-2 * xstride + 3 * ystride]
1558 #define TP0 pix[-1 * xstride + 3 * ystride]
1559 #define TQ0 pix[0 * xstride + 3 * ystride]
1560 #define TQ1 pix[1 * xstride + 3 * ystride]
1561 #define TQ2 pix[2 * xstride + 3 * ystride]
1562 #define TQ3 pix[3 * xstride + 3 * ystride]
1564 static void FUNC(hevc_loop_filter_luma)(uint8_t *_pix,
1565 ptrdiff_t _xstride, ptrdiff_t _ystride,
1566 int *_beta, int *_tc,
1567 uint8_t *_no_p, uint8_t *_no_q)
1570 pixel *pix = (pixel *)_pix;
1571 ptrdiff_t xstride = _xstride / sizeof(pixel);
1572 ptrdiff_t ystride = _ystride / sizeof(pixel);
1574 for (j = 0; j < 2; j++) {
1575 const int dp0 = abs(P2 - 2 * P1 + P0);
1576 const int dq0 = abs(Q2 - 2 * Q1 + Q0);
1577 const int dp3 = abs(TP2 - 2 * TP1 + TP0);
1578 const int dq3 = abs(TQ2 - 2 * TQ1 + TQ0);
1579 const int d0 = dp0 + dq0;
1580 const int d3 = dp3 + dq3;
1581 const int beta = _beta[j] << (BIT_DEPTH - 8);
1582 const int tc = _tc[j] << (BIT_DEPTH - 8);
1583 const int no_p = _no_p[j];
1584 const int no_q = _no_q[j];
1586 if (d0 + d3 >= beta) {
1590 const int beta_3 = beta >> 3;
1591 const int beta_2 = beta >> 2;
1592 const int tc25 = ((tc * 5 + 1) >> 1);
1594 if (abs(P3 - P0) + abs(Q3 - Q0) < beta_3 && abs(P0 - Q0) < tc25 &&
1595 abs(TP3 - TP0) + abs(TQ3 - TQ0) < beta_3 && abs(TP0 - TQ0) < tc25 &&
1596 (d0 << 1) < beta_2 && (d3 << 1) < beta_2) {
1598 const int tc2 = tc << 1;
1599 for (d = 0; d < 4; d++) {
1609 P0 = p0 + av_clip(((p2 + 2 * p1 + 2 * p0 + 2 * q0 + q1 + 4) >> 3) - p0, -tc2, tc2);
1610 P1 = p1 + av_clip(((p2 + p1 + p0 + q0 + 2) >> 2) - p1, -tc2, tc2);
1611 P2 = p2 + av_clip(((2 * p3 + 3 * p2 + p1 + p0 + q0 + 4) >> 3) - p2, -tc2, tc2);
1614 Q0 = q0 + av_clip(((p1 + 2 * p0 + 2 * q0 + 2 * q1 + q2 + 4) >> 3) - q0, -tc2, tc2);
1615 Q1 = q1 + av_clip(((p0 + q0 + q1 + q2 + 2) >> 2) - q1, -tc2, tc2);
1616 Q2 = q2 + av_clip(((2 * q3 + 3 * q2 + q1 + q0 + p0 + 4) >> 3) - q2, -tc2, tc2);
1620 } else { // normal filtering
1623 const int tc_2 = tc >> 1;
1624 if (dp0 + dp3 < ((beta + (beta >> 1)) >> 3))
1626 if (dq0 + dq3 < ((beta + (beta >> 1)) >> 3))
1629 for (d = 0; d < 4; d++) {
1636 int delta0 = (9 * (q0 - p0) - 3 * (q1 - p1) + 8) >> 4;
1637 if (abs(delta0) < 10 * tc) {
1638 delta0 = av_clip(delta0, -tc, tc);
1640 P0 = av_clip_pixel(p0 + delta0);
1642 Q0 = av_clip_pixel(q0 - delta0);
1643 if (!no_p && nd_p > 1) {
1644 const int deltap1 = av_clip((((p2 + p0 + 1) >> 1) - p1 + delta0) >> 1, -tc_2, tc_2);
1645 P1 = av_clip_pixel(p1 + deltap1);
1647 if (!no_q && nd_q > 1) {
1648 const int deltaq1 = av_clip((((q2 + q0 + 1) >> 1) - q1 - delta0) >> 1, -tc_2, tc_2);
1649 Q1 = av_clip_pixel(q1 + deltaq1);
1659 static void FUNC(hevc_loop_filter_chroma)(uint8_t *_pix, ptrdiff_t _xstride,
1660 ptrdiff_t _ystride, int *_tc,
1661 uint8_t *_no_p, uint8_t *_no_q)
1663 int d, j, no_p, no_q;
1664 pixel *pix = (pixel *)_pix;
1665 ptrdiff_t xstride = _xstride / sizeof(pixel);
1666 ptrdiff_t ystride = _ystride / sizeof(pixel);
1668 for (j = 0; j < 2; j++) {
1669 const int tc = _tc[j] << (BIT_DEPTH - 8);
1677 for (d = 0; d < 4; d++) {
1683 delta0 = av_clip((((q0 - p0) * 4) + p1 - q1 + 4) >> 3, -tc, tc);
1685 P0 = av_clip_pixel(p0 + delta0);
1687 Q0 = av_clip_pixel(q0 - delta0);
1693 static void FUNC(hevc_h_loop_filter_chroma)(uint8_t *pix, ptrdiff_t stride,
1694 int *tc, uint8_t *no_p,
1697 FUNC(hevc_loop_filter_chroma)(pix, stride, sizeof(pixel), tc, no_p, no_q);
1700 static void FUNC(hevc_v_loop_filter_chroma)(uint8_t *pix, ptrdiff_t stride,
1701 int *tc, uint8_t *no_p,
1704 FUNC(hevc_loop_filter_chroma)(pix, sizeof(pixel), stride, tc, no_p, no_q);
1707 static void FUNC(hevc_h_loop_filter_luma)(uint8_t *pix, ptrdiff_t stride,
1708 int *beta, int *tc, uint8_t *no_p,
1711 FUNC(hevc_loop_filter_luma)(pix, stride, sizeof(pixel),
1712 beta, tc, no_p, no_q);
1715 static void FUNC(hevc_v_loop_filter_luma)(uint8_t *pix, ptrdiff_t stride,
1716 int *beta, int *tc, uint8_t *no_p,
1719 FUNC(hevc_loop_filter_luma)(pix, sizeof(pixel), stride,
1720 beta, tc, no_p, no_q);