FFmpeg
hevcpred_template.c
Go to the documentation of this file.
1 /*
2  * HEVC video decoder
3  *
4  * Copyright (C) 2012 - 2013 Guillaume Martres
5  *
6  * This file is part of FFmpeg.
7  *
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.
12  *
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.
17  *
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
21  */
22 
23 #include "libavutil/pixdesc.h"
24 
25 #include "bit_depth_template.c"
26 #include "hevcpred.h"
27 
28 #define POS(x, y) src[(x) + stride * (y)]
29 
30 static av_always_inline void FUNC(intra_pred)(HEVCLocalContext *lc, int x0, int y0,
31  int log2_size, int c_idx)
32 {
33 #define PU(x) \
34  ((x) >> s->ps.sps->log2_min_pu_size)
35 #define MVF(x, y) \
36  (s->ref->tab_mvf[(x) + (y) * min_pu_width])
37 #define MVF_PU(x, y) \
38  MVF(PU(x0 + ((x) * (1 << hshift))), PU(y0 + ((y) * (1 << vshift))))
39 #define IS_INTRA(x, y) \
40  (MVF_PU(x, y).pred_flag == PF_INTRA)
41 #define MIN_TB_ADDR_ZS(x, y) \
42  s->ps.pps->min_tb_addr_zs[(y) * (s->ps.sps->tb_mask+2) + (x)]
43 #define EXTEND(ptr, val, len) \
44 do { \
45  pixel4 pix = PIXEL_SPLAT_X4(val); \
46  for (i = 0; i < (len); i += 4) \
47  AV_WN4P(ptr + i, pix); \
48 } while (0)
49 
50 #define EXTEND_RIGHT_CIP(ptr, start, length) \
51  for (i = start; i < (start) + (length); i += 4) \
52  if (!IS_INTRA(i, -1)) \
53  AV_WN4P(&ptr[i], a); \
54  else \
55  a = PIXEL_SPLAT_X4(ptr[i+3])
56 #define EXTEND_LEFT_CIP(ptr, start, length) \
57  for (i = start; i > (start) - (length); i--) \
58  if (!IS_INTRA(i - 1, -1)) \
59  ptr[i - 1] = ptr[i]
60 #define EXTEND_UP_CIP(ptr, start, length) \
61  for (i = (start); i > (start) - (length); i -= 4) \
62  if (!IS_INTRA(-1, i - 3)) \
63  AV_WN4P(&ptr[i - 3], a); \
64  else \
65  a = PIXEL_SPLAT_X4(ptr[i - 3])
66 #define EXTEND_DOWN_CIP(ptr, start, length) \
67  for (i = start; i < (start) + (length); i += 4) \
68  if (!IS_INTRA(-1, i)) \
69  AV_WN4P(&ptr[i], a); \
70  else \
71  a = PIXEL_SPLAT_X4(ptr[i + 3])
72 
73  const HEVCContext *const s = lc->parent;
74  int i;
75  int hshift = s->ps.sps->hshift[c_idx];
76  int vshift = s->ps.sps->vshift[c_idx];
77  int size = (1 << log2_size);
78  int size_in_luma_h = size << hshift;
79  int size_in_tbs_h = size_in_luma_h >> s->ps.sps->log2_min_tb_size;
80  int size_in_luma_v = size << vshift;
81  int size_in_tbs_v = size_in_luma_v >> s->ps.sps->log2_min_tb_size;
82  int x = x0 >> hshift;
83  int y = y0 >> vshift;
84  int x_tb = (x0 >> s->ps.sps->log2_min_tb_size) & s->ps.sps->tb_mask;
85  int y_tb = (y0 >> s->ps.sps->log2_min_tb_size) & s->ps.sps->tb_mask;
86  int spin = c_idx && !size_in_tbs_v && ((2 * y0) & (1 << s->ps.sps->log2_min_tb_size));
87 
88  int cur_tb_addr = MIN_TB_ADDR_ZS(x_tb, y_tb);
89 
90  ptrdiff_t stride = s->frame->linesize[c_idx] / sizeof(pixel);
91  pixel *src = (pixel*)s->frame->data[c_idx] + x + y * stride;
92 
93  int min_pu_width = s->ps.sps->min_pu_width;
94 
95  enum IntraPredMode mode = c_idx ? lc->tu.intra_pred_mode_c :
96  lc->tu.intra_pred_mode;
97  pixel4 a;
98  pixel left_array[2 * MAX_TB_SIZE + 1];
99  pixel filtered_left_array[2 * MAX_TB_SIZE + 1];
100  pixel top_array[2 * MAX_TB_SIZE + 1];
101  pixel filtered_top_array[2 * MAX_TB_SIZE + 1];
102 
103  pixel *left = left_array + 1;
104  pixel *top = top_array + 1;
105  pixel *filtered_left = filtered_left_array + 1;
106  pixel *filtered_top = filtered_top_array + 1;
107  int cand_bottom_left = lc->na.cand_bottom_left && cur_tb_addr > MIN_TB_ADDR_ZS( x_tb - 1, (y_tb + size_in_tbs_v + spin) & s->ps.sps->tb_mask);
108  int cand_left = lc->na.cand_left;
109  int cand_up_left = lc->na.cand_up_left;
110  int cand_up = lc->na.cand_up;
111  int cand_up_right = lc->na.cand_up_right && !spin && cur_tb_addr > MIN_TB_ADDR_ZS((x_tb + size_in_tbs_h) & s->ps.sps->tb_mask, y_tb - 1);
112 
113  int bottom_left_size = (FFMIN(y0 + 2 * size_in_luma_v, s->ps.sps->height) -
114  (y0 + size_in_luma_v)) >> vshift;
115  int top_right_size = (FFMIN(x0 + 2 * size_in_luma_h, s->ps.sps->width) -
116  (x0 + size_in_luma_h)) >> hshift;
117 
118  if (s->ps.pps->constrained_intra_pred_flag == 1) {
119  int size_in_luma_pu_v = PU(size_in_luma_v);
120  int size_in_luma_pu_h = PU(size_in_luma_h);
121  int on_pu_edge_x = !av_mod_uintp2(x0, s->ps.sps->log2_min_pu_size);
122  int on_pu_edge_y = !av_mod_uintp2(y0, s->ps.sps->log2_min_pu_size);
123  if (!size_in_luma_pu_h)
124  size_in_luma_pu_h++;
125  if (cand_bottom_left == 1 && on_pu_edge_x) {
126  int x_left_pu = PU(x0 - 1);
127  int y_bottom_pu = PU(y0 + size_in_luma_v);
128  int max = FFMIN(size_in_luma_pu_v, s->ps.sps->min_pu_height - y_bottom_pu);
129  cand_bottom_left = 0;
130  for (i = 0; i < max; i += 2)
131  cand_bottom_left |= (MVF(x_left_pu, y_bottom_pu + i).pred_flag == PF_INTRA);
132  }
133  if (cand_left == 1 && on_pu_edge_x) {
134  int x_left_pu = PU(x0 - 1);
135  int y_left_pu = PU(y0);
136  int max = FFMIN(size_in_luma_pu_v, s->ps.sps->min_pu_height - y_left_pu);
137  cand_left = 0;
138  for (i = 0; i < max; i += 2)
139  cand_left |= (MVF(x_left_pu, y_left_pu + i).pred_flag == PF_INTRA);
140  }
141  if (cand_up_left == 1) {
142  int x_left_pu = PU(x0 - 1);
143  int y_top_pu = PU(y0 - 1);
144  cand_up_left = MVF(x_left_pu, y_top_pu).pred_flag == PF_INTRA;
145  }
146  if (cand_up == 1 && on_pu_edge_y) {
147  int x_top_pu = PU(x0);
148  int y_top_pu = PU(y0 - 1);
149  int max = FFMIN(size_in_luma_pu_h, s->ps.sps->min_pu_width - x_top_pu);
150  cand_up = 0;
151  for (i = 0; i < max; i += 2)
152  cand_up |= (MVF(x_top_pu + i, y_top_pu).pred_flag == PF_INTRA);
153  }
154  if (cand_up_right == 1 && on_pu_edge_y) {
155  int y_top_pu = PU(y0 - 1);
156  int x_right_pu = PU(x0 + size_in_luma_h);
157  int max = FFMIN(size_in_luma_pu_h, s->ps.sps->min_pu_width - x_right_pu);
158  cand_up_right = 0;
159  for (i = 0; i < max; i += 2)
160  cand_up_right |= (MVF(x_right_pu + i, y_top_pu).pred_flag == PF_INTRA);
161  }
162  memset(left, 128, 2 * MAX_TB_SIZE*sizeof(pixel));
163  memset(top , 128, 2 * MAX_TB_SIZE*sizeof(pixel));
164  top[-1] = 128;
165  }
166  if (cand_up_left) {
167  left[-1] = POS(-1, -1);
168  top[-1] = left[-1];
169  }
170  if (cand_up)
171  memcpy(top, src - stride, size * sizeof(pixel));
172  if (cand_up_right) {
173  memcpy(top + size, src - stride + size, size * sizeof(pixel));
174  EXTEND(top + size + top_right_size, POS(size + top_right_size - 1, -1),
175  size - top_right_size);
176  }
177  if (cand_left)
178  for (i = 0; i < size; i++)
179  left[i] = POS(-1, i);
180  if (cand_bottom_left) {
181  for (i = size; i < size + bottom_left_size; i++)
182  left[i] = POS(-1, i);
183  EXTEND(left + size + bottom_left_size, POS(-1, size + bottom_left_size - 1),
184  size - bottom_left_size);
185  }
186 
187  if (s->ps.pps->constrained_intra_pred_flag == 1) {
188  if (cand_bottom_left || cand_left || cand_up_left || cand_up || cand_up_right) {
189  int size_max_x = x0 + ((2 * size) << hshift) < s->ps.sps->width ?
190  2 * size : (s->ps.sps->width - x0) >> hshift;
191  int size_max_y = y0 + ((2 * size) << vshift) < s->ps.sps->height ?
192  2 * size : (s->ps.sps->height - y0) >> vshift;
193  int j = size + (cand_bottom_left? bottom_left_size: 0) -1;
194  if (!cand_up_right) {
195  size_max_x = x0 + ((size) << hshift) < s->ps.sps->width ?
196  size : (s->ps.sps->width - x0) >> hshift;
197  }
198  if (!cand_bottom_left) {
199  size_max_y = y0 + (( size) << vshift) < s->ps.sps->height ?
200  size : (s->ps.sps->height - y0) >> vshift;
201  }
202  if (cand_bottom_left || cand_left || cand_up_left) {
203  while (j > -1 && !IS_INTRA(-1, j))
204  j--;
205  if (!IS_INTRA(-1, j)) {
206  j = 0;
207  while (j < size_max_x && !IS_INTRA(j, -1))
208  j++;
209  EXTEND_LEFT_CIP(top, j, j + 1);
210  left[-1] = top[-1];
211  }
212  } else {
213  j = 0;
214  while (j < size_max_x && !IS_INTRA(j, -1))
215  j++;
216  if (j > 0)
217  if (cand_up_left) {
218  EXTEND_LEFT_CIP(top, j, j + 1);
219  } else {
220  EXTEND_LEFT_CIP(top, j, j);
221  top[-1] = top[0];
222  }
223  left[-1] = top[-1];
224  }
225  left[-1] = top[-1];
226  if (cand_bottom_left || cand_left) {
227  a = PIXEL_SPLAT_X4(left[-1]);
228  EXTEND_DOWN_CIP(left, 0, size_max_y);
229  }
230  if (!cand_left)
231  EXTEND(left, left[-1], size);
232  if (!cand_bottom_left)
233  EXTEND(left + size, left[size - 1], size);
234  if (x0 != 0 && y0 != 0) {
235  a = PIXEL_SPLAT_X4(left[size_max_y - 1]);
236  EXTEND_UP_CIP(left, size_max_y - 1, size_max_y);
237  if (!IS_INTRA(-1, - 1))
238  left[-1] = left[0];
239  } else if (x0 == 0) {
240  EXTEND(left, 0, size_max_y);
241  } else {
242  a = PIXEL_SPLAT_X4(left[size_max_y - 1]);
243  EXTEND_UP_CIP(left, size_max_y - 1, size_max_y);
244  }
245  top[-1] = left[-1];
246  if (y0 != 0) {
247  a = PIXEL_SPLAT_X4(left[-1]);
248  EXTEND_RIGHT_CIP(top, 0, size_max_x);
249  }
250  }
251  }
252  // Infer the unavailable samples
253  if (!cand_bottom_left) {
254  if (cand_left) {
255  EXTEND(left + size, left[size - 1], size);
256  } else if (cand_up_left) {
257  EXTEND(left, left[-1], 2 * size);
258  cand_left = 1;
259  } else if (cand_up) {
260  left[-1] = top[0];
261  EXTEND(left, left[-1], 2 * size);
262  cand_up_left = 1;
263  cand_left = 1;
264  } else if (cand_up_right) {
265  EXTEND(top, top[size], size);
266  left[-1] = top[size];
267  EXTEND(left, left[-1], 2 * size);
268  cand_up = 1;
269  cand_up_left = 1;
270  cand_left = 1;
271  } else { // No samples available
272  left[-1] = (1 << (BIT_DEPTH - 1));
273  EXTEND(top, left[-1], 2 * size);
274  EXTEND(left, left[-1], 2 * size);
275  }
276  }
277 
278  if (!cand_left)
279  EXTEND(left, left[size], size);
280  if (!cand_up_left) {
281  left[-1] = left[0];
282  }
283  if (!cand_up)
284  EXTEND(top, left[-1], size);
285  if (!cand_up_right)
286  EXTEND(top + size, top[size - 1], size);
287 
288  top[-1] = left[-1];
289 
290  // Filtering process
291  if (!s->ps.sps->intra_smoothing_disabled_flag && (c_idx == 0 || s->ps.sps->chroma_format_idc == 3)) {
292  if (mode != INTRA_DC && size != 4){
293  int intra_hor_ver_dist_thresh[] = { 7, 1, 0 };
294  int min_dist_vert_hor = FFMIN(FFABS((int)(mode - 26U)),
295  FFABS((int)(mode - 10U)));
296  if (min_dist_vert_hor > intra_hor_ver_dist_thresh[log2_size - 3]) {
297  int threshold = 1 << (BIT_DEPTH - 5);
298  if (s->ps.sps->sps_strong_intra_smoothing_enable_flag && c_idx == 0 &&
299  log2_size == 5 &&
300  FFABS(top[-1] + top[63] - 2 * top[31]) < threshold &&
301  FFABS(left[-1] + left[63] - 2 * left[31]) < threshold) {
302  // We can't just overwrite values in top because it could be
303  // a pointer into src
304  filtered_top[-1] = top[-1];
305  filtered_top[63] = top[63];
306  for (i = 0; i < 63; i++)
307  filtered_top[i] = ((64 - (i + 1)) * top[-1] +
308  (i + 1) * top[63] + 32) >> 6;
309  for (i = 0; i < 63; i++)
310  left[i] = ((64 - (i + 1)) * left[-1] +
311  (i + 1) * left[63] + 32) >> 6;
312  top = filtered_top;
313  } else {
314  filtered_left[2 * size - 1] = left[2 * size - 1];
315  filtered_top[2 * size - 1] = top[2 * size - 1];
316  for (i = 2 * size - 2; i >= 0; i--)
317  filtered_left[i] = (left[i + 1] + 2 * left[i] +
318  left[i - 1] + 2) >> 2;
319  filtered_top[-1] =
320  filtered_left[-1] = (left[0] + 2 * left[-1] + top[0] + 2) >> 2;
321  for (i = 2 * size - 2; i >= 0; i--)
322  filtered_top[i] = (top[i + 1] + 2 * top[i] +
323  top[i - 1] + 2) >> 2;
324  left = filtered_left;
325  top = filtered_top;
326  }
327  }
328  }
329  }
330 
331  switch (mode) {
332  case INTRA_PLANAR:
333  s->hpc.pred_planar[log2_size - 2]((uint8_t *)src, (uint8_t *)top,
334  (uint8_t *)left, stride);
335  break;
336  case INTRA_DC:
337  s->hpc.pred_dc((uint8_t *)src, (uint8_t *)top,
338  (uint8_t *)left, stride, log2_size, c_idx);
339  break;
340  default:
341  s->hpc.pred_angular[log2_size - 2]((uint8_t *)src, (uint8_t *)top,
342  (uint8_t *)left, stride, c_idx,
343  mode);
344  break;
345  }
346 }
347 
348 #define INTRA_PRED(size) \
349 static void FUNC(intra_pred_ ## size)(HEVCLocalContext *lc, int x0, int y0, int c_idx) \
350 { \
351  FUNC(intra_pred)(lc, x0, y0, size, c_idx); \
352 }
353 
354 INTRA_PRED(2)
355 INTRA_PRED(3)
356 INTRA_PRED(4)
357 INTRA_PRED(5)
358 
359 #undef INTRA_PRED
360 
361 static av_always_inline void FUNC(pred_planar)(uint8_t *_src, const uint8_t *_top,
362  const uint8_t *_left, ptrdiff_t stride,
363  int trafo_size)
364 {
365  int x, y;
366  pixel *src = (pixel *)_src;
367  const pixel *top = (const pixel *)_top;
368  const pixel *left = (const pixel *)_left;
369  int size = 1 << trafo_size;
370  for (y = 0; y < size; y++)
371  for (x = 0; x < size; x++)
372  POS(x, y) = ((size - 1 - x) * left[y] + (x + 1) * top[size] +
373  (size - 1 - y) * top[x] + (y + 1) * left[size] + size) >> (trafo_size + 1);
374 }
375 
376 #define PRED_PLANAR(size)\
377 static void FUNC(pred_planar_ ## size)(uint8_t *src, const uint8_t *top, \
378  const uint8_t *left, ptrdiff_t stride) \
379 { \
380  FUNC(pred_planar)(src, top, left, stride, size + 2); \
381 }
382 
383 PRED_PLANAR(0)
384 PRED_PLANAR(1)
385 PRED_PLANAR(2)
386 PRED_PLANAR(3)
387 
388 #undef PRED_PLANAR
389 
390 static void FUNC(pred_dc)(uint8_t *_src, const uint8_t *_top,
391  const uint8_t *_left,
392  ptrdiff_t stride, int log2_size, int c_idx)
393 {
394  int i, j, x, y;
395  int size = (1 << log2_size);
396  pixel *src = (pixel *)_src;
397  const pixel *top = (const pixel *)_top;
398  const pixel *left = (const pixel *)_left;
399  int dc = size;
400  pixel4 a;
401  for (i = 0; i < size; i++)
402  dc += left[i] + top[i];
403 
404  dc >>= log2_size + 1;
405 
406  a = PIXEL_SPLAT_X4(dc);
407 
408  for (i = 0; i < size; i++)
409  for (j = 0; j < size; j+=4)
410  AV_WN4P(&POS(j, i), a);
411 
412  if (c_idx == 0 && size < 32) {
413  POS(0, 0) = (left[0] + 2 * dc + top[0] + 2) >> 2;
414  for (x = 1; x < size; x++)
415  POS(x, 0) = (top[x] + 3 * dc + 2) >> 2;
416  for (y = 1; y < size; y++)
417  POS(0, y) = (left[y] + 3 * dc + 2) >> 2;
418  }
419 }
420 
421 static av_always_inline void FUNC(pred_angular)(uint8_t *_src,
422  const uint8_t *_top,
423  const uint8_t *_left,
424  ptrdiff_t stride, int c_idx,
425  int mode, int size)
426 {
427  int x, y;
428  pixel *src = (pixel *)_src;
429  const pixel *top = (const pixel *)_top;
430  const pixel *left = (const pixel *)_left;
431 
432  static const int intra_pred_angle[] = {
433  32, 26, 21, 17, 13, 9, 5, 2, 0, -2, -5, -9, -13, -17, -21, -26, -32,
434  -26, -21, -17, -13, -9, -5, -2, 0, 2, 5, 9, 13, 17, 21, 26, 32
435  };
436  static const int inv_angle[] = {
437  -4096, -1638, -910, -630, -482, -390, -315, -256, -315, -390, -482,
438  -630, -910, -1638, -4096
439  };
440 
441  int angle = intra_pred_angle[mode - 2];
442  pixel ref_array[3 * MAX_TB_SIZE + 4];
443  pixel *ref_tmp = ref_array + size;
444  const pixel *ref;
445  int last = (size * angle) >> 5;
446 
447  if (mode >= 18) {
448  ref = top - 1;
449  if (angle < 0 && last < -1) {
450  for (x = 0; x <= size; x += 4)
451  AV_WN4P(&ref_tmp[x], AV_RN4P(&top[x - 1]));
452  for (x = last; x <= -1; x++)
453  ref_tmp[x] = left[-1 + ((x * inv_angle[mode - 11] + 128) >> 8)];
454  ref = ref_tmp;
455  }
456 
457  for (y = 0; y < size; y++) {
458  int idx = ((y + 1) * angle) >> 5;
459  int fact = ((y + 1) * angle) & 31;
460  if (fact) {
461  for (x = 0; x < size; x += 4) {
462  POS(x , y) = ((32 - fact) * ref[x + idx + 1] +
463  fact * ref[x + idx + 2] + 16) >> 5;
464  POS(x + 1, y) = ((32 - fact) * ref[x + 1 + idx + 1] +
465  fact * ref[x + 1 + idx + 2] + 16) >> 5;
466  POS(x + 2, y) = ((32 - fact) * ref[x + 2 + idx + 1] +
467  fact * ref[x + 2 + idx + 2] + 16) >> 5;
468  POS(x + 3, y) = ((32 - fact) * ref[x + 3 + idx + 1] +
469  fact * ref[x + 3 + idx + 2] + 16) >> 5;
470  }
471  } else {
472  for (x = 0; x < size; x += 4)
473  AV_WN4P(&POS(x, y), AV_RN4P(&ref[x + idx + 1]));
474  }
475  }
476  if (mode == 26 && c_idx == 0 && size < 32) {
477  for (y = 0; y < size; y++)
478  POS(0, y) = av_clip_pixel(top[0] + ((left[y] - left[-1]) >> 1));
479  }
480  } else {
481  ref = left - 1;
482  if (angle < 0 && last < -1) {
483  for (x = 0; x <= size; x += 4)
484  AV_WN4P(&ref_tmp[x], AV_RN4P(&left[x - 1]));
485  for (x = last; x <= -1; x++)
486  ref_tmp[x] = top[-1 + ((x * inv_angle[mode - 11] + 128) >> 8)];
487  ref = ref_tmp;
488  }
489 
490  for (x = 0; x < size; x++) {
491  int idx = ((x + 1) * angle) >> 5;
492  int fact = ((x + 1) * angle) & 31;
493  if (fact) {
494  for (y = 0; y < size; y++) {
495  POS(x, y) = ((32 - fact) * ref[y + idx + 1] +
496  fact * ref[y + idx + 2] + 16) >> 5;
497  }
498  } else {
499  for (y = 0; y < size; y++)
500  POS(x, y) = ref[y + idx + 1];
501  }
502  }
503  if (mode == 10 && c_idx == 0 && size < 32) {
504  for (x = 0; x < size; x += 4) {
505  POS(x, 0) = av_clip_pixel(left[0] + ((top[x ] - top[-1]) >> 1));
506  POS(x + 1, 0) = av_clip_pixel(left[0] + ((top[x + 1] - top[-1]) >> 1));
507  POS(x + 2, 0) = av_clip_pixel(left[0] + ((top[x + 2] - top[-1]) >> 1));
508  POS(x + 3, 0) = av_clip_pixel(left[0] + ((top[x + 3] - top[-1]) >> 1));
509  }
510  }
511  }
512 }
513 
514 static void FUNC(pred_angular_0)(uint8_t *src, const uint8_t *top,
515  const uint8_t *left,
516  ptrdiff_t stride, int c_idx, int mode)
517 {
518  FUNC(pred_angular)(src, top, left, stride, c_idx, mode, 1 << 2);
519 }
520 
521 static void FUNC(pred_angular_1)(uint8_t *src, const uint8_t *top,
522  const uint8_t *left,
523  ptrdiff_t stride, int c_idx, int mode)
524 {
525  FUNC(pred_angular)(src, top, left, stride, c_idx, mode, 1 << 3);
526 }
527 
528 static void FUNC(pred_angular_2)(uint8_t *src, const uint8_t *top,
529  const uint8_t *left,
530  ptrdiff_t stride, int c_idx, int mode)
531 {
532  FUNC(pred_angular)(src, top, left, stride, c_idx, mode, 1 << 4);
533 }
534 
535 static void FUNC(pred_angular_3)(uint8_t *src, const uint8_t *top,
536  const uint8_t *left,
537  ptrdiff_t stride, int c_idx, int mode)
538 {
539  FUNC(pred_angular)(src, top, left, stride, c_idx, mode, 1 << 5);
540 }
541 
542 #undef EXTEND_LEFT_CIP
543 #undef EXTEND_RIGHT_CIP
544 #undef EXTEND_UP_CIP
545 #undef EXTEND_DOWN_CIP
546 #undef IS_INTRA
547 #undef MVF_PU
548 #undef MVF
549 #undef PU
550 #undef EXTEND
551 #undef MIN_TB_ADDR_ZS
552 #undef POS
HEVCLocalContext
Definition: hevcdec.h:432
BIT_DEPTH
#define BIT_DEPTH
Definition: bit_depth_template.c:24
POS
#define POS(x, y)
Definition: hevcpred_template.c:28
PU
#define PU(x)
AV_RN4P
#define AV_RN4P
Definition: bit_depth_template.c:91
av_mod_uintp2
#define av_mod_uintp2
Definition: common.h:122
pixdesc.h
INTRA_DC
@ INTRA_DC
Definition: hevcdec.h:173
max
#define max(a, b)
Definition: cuda_runtime.h:33
EXTEND_UP_CIP
#define EXTEND_UP_CIP(ptr, start, length)
PF_INTRA
@ PF_INTRA
Definition: hevcdec.h:165
pixel4
#define pixel4
Definition: bit_depth_template.c:83
pred_angular
static av_always_inline void FUNC() pred_angular(uint8_t *_src, const uint8_t *_top, const uint8_t *_left, ptrdiff_t stride, int c_idx, int mode, int size)
Definition: hevcpred_template.c:421
hevcpred.h
s
#define s(width, name)
Definition: cbs_vp9.c:256
EXTEND_LEFT_CIP
#define EXTEND_LEFT_CIP(ptr, start, length)
pred_planar
static av_always_inline void FUNC() pred_planar(uint8_t *_src, const uint8_t *_top, const uint8_t *_left, ptrdiff_t stride, int trafo_size)
Definition: hevcpred_template.c:361
IS_INTRA
#define IS_INTRA(x, y)
PIXEL_SPLAT_X4
#define PIXEL_SPLAT_X4(x)
Definition: bit_depth_template.c:96
FFABS
#define FFABS(a)
Absolute value, Note, INT_MIN / INT64_MIN result in undefined behavior as they are not representable ...
Definition: common.h:64
MVF
#define MVF(x, y)
pixel
uint8_t pixel
Definition: tiny_ssim.c:41
pred_angular_1
static void FUNC() pred_angular_1(uint8_t *src, const uint8_t *top, const uint8_t *left, ptrdiff_t stride, int c_idx, int mode)
Definition: hevcpred_template.c:521
bit_depth_template.c
dc
Tag MUST be and< 10hcoeff half pel interpolation filter coefficients, hcoeff[0] are the 2 middle coefficients[1] are the next outer ones and so on, resulting in a filter like:...eff[2], hcoeff[1], hcoeff[0], hcoeff[0], hcoeff[1], hcoeff[2] ... the sign of the coefficients is not explicitly stored but alternates after each coeff and coeff[0] is positive, so ...,+,-,+,-,+,+,-,+,-,+,... hcoeff[0] is not explicitly stored but found by subtracting the sum of all stored coefficients with signs from 32 hcoeff[0]=32 - hcoeff[1] - hcoeff[2] - ... a good choice for hcoeff and htaps is htaps=6 hcoeff={40,-10, 2} an alternative which requires more computations at both encoder and decoder side and may or may not be better is htaps=8 hcoeff={42,-14, 6,-2}ref_frames minimum of the number of available reference frames and max_ref_frames for example the first frame after a key frame always has ref_frames=1spatial_decomposition_type wavelet type 0 is a 9/7 symmetric compact integer wavelet 1 is a 5/3 symmetric compact integer wavelet others are reserved stored as delta from last, last is reset to 0 if always_reset||keyframeqlog quality(logarithmic quantizer scale) stored as delta from last, last is reset to 0 if always_reset||keyframemv_scale stored as delta from last, last is reset to 0 if always_reset||keyframe FIXME check that everything works fine if this changes between framesqbias dequantization bias stored as delta from last, last is reset to 0 if always_reset||keyframeblock_max_depth maximum depth of the block tree stored as delta from last, last is reset to 0 if always_reset||keyframequant_table quantization tableHighlevel bitstream structure:==============================--------------------------------------------|Header|--------------------------------------------|------------------------------------|||Block0||||split?||||yes no||||......... intra?||||:Block01 :yes no||||:Block02 :....... ..........||||:Block03 ::y DC ::ref index:||||:Block04 ::cb DC ::motion x :||||......... :cr DC ::motion y :||||....... ..........|||------------------------------------||------------------------------------|||Block1|||...|--------------------------------------------|------------ ------------ ------------|||Y subbands||Cb subbands||Cr subbands||||--- ---||--- ---||--- ---|||||LL0||HL0||||LL0||HL0||||LL0||HL0|||||--- ---||--- ---||--- ---||||--- ---||--- ---||--- ---|||||LH0||HH0||||LH0||HH0||||LH0||HH0|||||--- ---||--- ---||--- ---||||--- ---||--- ---||--- ---|||||HL1||LH1||||HL1||LH1||||HL1||LH1|||||--- ---||--- ---||--- ---||||--- ---||--- ---||--- ---|||||HH1||HL2||||HH1||HL2||||HH1||HL2|||||...||...||...|||------------ ------------ ------------|--------------------------------------------Decoding process:=================------------|||Subbands|------------||||------------|Intra DC||||LL0 subband prediction ------------|\ Dequantization ------------------- \||Reference frames|\ IDWT|------- -------|Motion \|||Frame 0||Frame 1||Compensation . OBMC v -------|------- -------|--------------. \------> Frame n output Frame Frame<----------------------------------/|...|------------------- Range Coder:============Binary Range Coder:------------------- The implemented range coder is an adapted version based upon "Range encoding: an algorithm for removing redundancy from a digitised message." by G. N. N. Martin. The symbols encoded by the Snow range coder are bits(0|1). The associated probabilities are not fix but change depending on the symbol mix seen so far. bit seen|new state ---------+----------------------------------------------- 0|256 - state_transition_table[256 - old_state];1|state_transition_table[old_state];state_transition_table={ 0, 0, 0, 0, 0, 0, 0, 0, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 190, 191, 192, 194, 194, 195, 196, 197, 198, 199, 200, 201, 202, 202, 204, 205, 206, 207, 208, 209, 209, 210, 211, 212, 213, 215, 215, 216, 217, 218, 219, 220, 220, 222, 223, 224, 225, 226, 227, 227, 229, 229, 230, 231, 232, 234, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 248, 0, 0, 0, 0, 0, 0, 0};FIXME Range Coding of integers:------------------------- FIXME Neighboring Blocks:===================left and top are set to the respective blocks unless they are outside of the image in which case they are set to the Null block top-left is set to the top left block unless it is outside of the image in which case it is set to the left block if this block has no larger parent block or it is at the left side of its parent block and the top right block is not outside of the image then the top right block is used for top-right else the top-left block is used Null block y, cb, cr are 128 level, ref, mx and my are 0 Motion Vector Prediction:=========================1. the motion vectors of all the neighboring blocks are scaled to compensate for the difference of reference frames scaled_mv=(mv *(256 *(current_reference+1)/(mv.reference+1))+128)> the median of the scaled top and top right vectors is used as motion vector prediction the used motion vector is the sum of the predictor and(mvx_diff, mvy_diff) *mv_scale Intra DC Prediction block[y][x] dc[1]
Definition: snow.txt:400
INTRA_PLANAR
@ INTRA_PLANAR
Definition: hevcdec.h:172
size
int size
Definition: twinvq_data.h:10344
pred_angular_0
static void FUNC() pred_angular_0(uint8_t *src, const uint8_t *top, const uint8_t *left, ptrdiff_t stride, int c_idx, int mode)
Definition: hevcpred_template.c:514
a
The reader does not expect b to be semantically here and if the code is changed by maybe adding a a division or other the signedness will almost certainly be mistaken To avoid this confusion a new type was SUINT is the C unsigned type but it holds a signed int to use the same example SUINT a
Definition: undefined.txt:41
fact
static double fact(double i)
Definition: af_aiir.c:941
i
#define i(width, name, range_min, range_max)
Definition: cbs_h2645.c:269
MIN_TB_ADDR_ZS
#define MIN_TB_ADDR_ZS(x, y)
av_always_inline
#define av_always_inline
Definition: attributes.h:49
FFMIN
#define FFMIN(a, b)
Definition: macros.h:49
stride
#define stride
Definition: h264pred_template.c:537
IntraPredMode
IntraPredMode
Definition: hevcdec.h:171
av_clip_pixel
#define av_clip_pixel(a)
Definition: bit_depth_template.c:98
EXTEND_RIGHT_CIP
#define EXTEND_RIGHT_CIP(ptr, start, length)
INTRA_PRED
#define INTRA_PRED(size)
Definition: hevcpred_template.c:348
left
Tag MUST be and< 10hcoeff half pel interpolation filter coefficients, hcoeff[0] are the 2 middle coefficients[1] are the next outer ones and so on, resulting in a filter like:...eff[2], hcoeff[1], hcoeff[0], hcoeff[0], hcoeff[1], hcoeff[2] ... the sign of the coefficients is not explicitly stored but alternates after each coeff and coeff[0] is positive, so ...,+,-,+,-,+,+,-,+,-,+,... hcoeff[0] is not explicitly stored but found by subtracting the sum of all stored coefficients with signs from 32 hcoeff[0]=32 - hcoeff[1] - hcoeff[2] - ... a good choice for hcoeff and htaps is htaps=6 hcoeff={40,-10, 2} an alternative which requires more computations at both encoder and decoder side and may or may not be better is htaps=8 hcoeff={42,-14, 6,-2}ref_frames minimum of the number of available reference frames and max_ref_frames for example the first frame after a key frame always has ref_frames=1spatial_decomposition_type wavelet type 0 is a 9/7 symmetric compact integer wavelet 1 is a 5/3 symmetric compact integer wavelet others are reserved stored as delta from last, last is reset to 0 if always_reset||keyframeqlog quality(logarithmic quantizer scale) stored as delta from last, last is reset to 0 if always_reset||keyframemv_scale stored as delta from last, last is reset to 0 if always_reset||keyframe FIXME check that everything works fine if this changes between framesqbias dequantization bias stored as delta from last, last is reset to 0 if always_reset||keyframeblock_max_depth maximum depth of the block tree stored as delta from last, last is reset to 0 if always_reset||keyframequant_table quantization tableHighlevel bitstream structure:==============================--------------------------------------------|Header|--------------------------------------------|------------------------------------|||Block0||||split?||||yes no||||......... intra?||||:Block01 :yes no||||:Block02 :....... ..........||||:Block03 ::y DC ::ref index:||||:Block04 ::cb DC ::motion x :||||......... :cr DC ::motion y :||||....... ..........|||------------------------------------||------------------------------------|||Block1|||...|--------------------------------------------|------------ ------------ ------------|||Y subbands||Cb subbands||Cr subbands||||--- ---||--- ---||--- ---|||||LL0||HL0||||LL0||HL0||||LL0||HL0|||||--- ---||--- ---||--- ---||||--- ---||--- ---||--- ---|||||LH0||HH0||||LH0||HH0||||LH0||HH0|||||--- ---||--- ---||--- ---||||--- ---||--- ---||--- ---|||||HL1||LH1||||HL1||LH1||||HL1||LH1|||||--- ---||--- ---||--- ---||||--- ---||--- ---||--- ---|||||HH1||HL2||||HH1||HL2||||HH1||HL2|||||...||...||...|||------------ ------------ ------------|--------------------------------------------Decoding process:=================------------|||Subbands|------------||||------------|Intra DC||||LL0 subband prediction ------------|\ Dequantization ------------------- \||Reference frames|\ IDWT|------- -------|Motion \|||Frame 0||Frame 1||Compensation . OBMC v -------|------- -------|--------------. \------> Frame n output Frame Frame<----------------------------------/|...|------------------- Range Coder:============Binary Range Coder:------------------- The implemented range coder is an adapted version based upon "Range encoding: an algorithm for removing redundancy from a digitised message." by G. N. N. Martin. The symbols encoded by the Snow range coder are bits(0|1). The associated probabilities are not fix but change depending on the symbol mix seen so far. bit seen|new state ---------+----------------------------------------------- 0|256 - state_transition_table[256 - old_state];1|state_transition_table[old_state];state_transition_table={ 0, 0, 0, 0, 0, 0, 0, 0, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, 91, 92, 93, 94, 94, 95, 96, 97, 98, 99, 100, 101, 102, 103, 104, 105, 106, 107, 108, 109, 110, 111, 112, 113, 114, 114, 115, 116, 117, 118, 119, 120, 121, 122, 123, 124, 125, 126, 127, 128, 129, 130, 131, 132, 133, 133, 134, 135, 136, 137, 138, 139, 140, 141, 142, 143, 144, 145, 146, 147, 148, 149, 150, 151, 152, 152, 153, 154, 155, 156, 157, 158, 159, 160, 161, 162, 163, 164, 165, 166, 167, 168, 169, 170, 171, 171, 172, 173, 174, 175, 176, 177, 178, 179, 180, 181, 182, 183, 184, 185, 186, 187, 188, 189, 190, 190, 191, 192, 194, 194, 195, 196, 197, 198, 199, 200, 201, 202, 202, 204, 205, 206, 207, 208, 209, 209, 210, 211, 212, 213, 215, 215, 216, 217, 218, 219, 220, 220, 222, 223, 224, 225, 226, 227, 227, 229, 229, 230, 231, 232, 234, 234, 235, 236, 237, 238, 239, 240, 241, 242, 243, 244, 245, 246, 247, 248, 248, 0, 0, 0, 0, 0, 0, 0};FIXME Range Coding of integers:------------------------- FIXME Neighboring Blocks:===================left and top are set to the respective blocks unless they are outside of the image in which case they are set to the Null block top-left is set to the top left block unless it is outside of the image in which case it is set to the left block if this block has no larger parent block or it is at the left side of its parent block and the top right block is not outside of the image then the top right block is used for top-right else the top-left block is used Null block y, cb, cr are 128 level, ref, mx and my are 0 Motion Vector Prediction:=========================1. the motion vectors of all the neighboring blocks are scaled to compensate for the difference of reference frames scaled_mv=(mv *(256 *(current_reference+1)/(mv.reference+1))+128)> the median of the scaled left
Definition: snow.txt:386
U
#define U(x)
Definition: vpx_arith.h:37
intra_pred
static av_always_inline void FUNC() intra_pred(HEVCLocalContext *lc, int x0, int y0, int log2_size, int c_idx)
Definition: hevcpred_template.c:30
pred_angular_3
static void FUNC() pred_angular_3(uint8_t *src, const uint8_t *top, const uint8_t *left, ptrdiff_t stride, int c_idx, int mode)
Definition: hevcpred_template.c:535
FUNC
#define FUNC(a)
Definition: bit_depth_template.c:104
mode
mode
Definition: ebur128.h:83
HEVCContext
Definition: hevcdec.h:490
AV_WN4P
#define AV_WN4P
Definition: bit_depth_template.c:94
ref
static int ref[MAX_W *MAX_W]
Definition: jpeg2000dwt.c:112
PRED_PLANAR
#define PRED_PLANAR(size)
Definition: hevcpred_template.c:376
EXTEND
#define EXTEND(ptr, val, len)
pred_angular_2
static void FUNC() pred_angular_2(uint8_t *src, const uint8_t *top, const uint8_t *left, ptrdiff_t stride, int c_idx, int mode)
Definition: hevcpred_template.c:528
pred_dc
static void FUNC() pred_dc(uint8_t *_src, const uint8_t *_top, const uint8_t *_left, ptrdiff_t stride, int log2_size, int c_idx)
Definition: hevcpred_template.c:390
src
INIT_CLIP pixel * src
Definition: h264pred_template.c:418
MAX_TB_SIZE
#define MAX_TB_SIZE
Definition: hevcdec.h:48
EXTEND_DOWN_CIP
#define EXTEND_DOWN_CIP(ptr, start, length)