/src/vlc/contrib/contrib-build/bpg/libavcodec/hevcpred_template.c
Line | Count | Source |
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 | | #ifndef USE_FUNC_PTR |
29 | | static HEVC_INLINE void FUNC(pred_planar)(uint8_t *_src, const uint8_t *_top, |
30 | | const uint8_t *_left, ptrdiff_t stride, |
31 | | int trafo_size); |
32 | | static void FUNC(pred_dc)(uint8_t *_src, const uint8_t *_top, |
33 | | const uint8_t *_left, |
34 | | ptrdiff_t stride, int log2_size, int c_idx); |
35 | | static HEVC_INLINE void FUNC(pred_angular)(uint8_t *_src, |
36 | | const uint8_t *_top, |
37 | | const uint8_t *_left, |
38 | | ptrdiff_t stride, int c_idx, |
39 | | int mode, int size, |
40 | | int disable_intra_boundary_filter BIT_DEPTH_PARAM); |
41 | | #endif |
42 | | |
43 | 0 | #define POS(x, y) src[(x) + stride * (y)] |
44 | | |
45 | | #ifdef USE_FUNC_PTR |
46 | | static HEVC_INLINE void FUNC(intra_pred) |
47 | | #else |
48 | | void intra_pred |
49 | | #endif |
50 | | (HEVCContext *s, int x0, int y0, int log2_size, int c_idx) |
51 | 0 | { |
52 | 0 | #define PU(x) \ |
53 | 0 | ((x) >> s->sps->log2_min_pu_size) |
54 | 0 | #ifdef USE_PRED |
55 | 0 | #define MVF(x, y) \ |
56 | 0 | (s->ref->tab_mvf[(x) + (y) * min_pu_width].pred_flag) |
57 | | #else |
58 | | #define MVF(x, y) PF_INTRA |
59 | | #endif |
60 | 0 | #define MVF_PU(x, y) \ |
61 | 0 | MVF(PU(x0 + ((x) << hshift)), PU(y0 + ((y) << vshift))) |
62 | 0 | #define IS_INTRA(x, y) \ |
63 | 0 | (MVF_PU(x, y) == PF_INTRA) |
64 | 0 | #define MIN_TB_ADDR_ZS(x, y) \ |
65 | 0 | s->pps->min_tb_addr_zs[(y) * (s->sps->tb_mask+2) + (x)] |
66 | 0 | #define EXTEND(ptr, val, len) \ |
67 | 0 | do { \ |
68 | 0 | pixel4 pix = PIXEL_SPLAT_X4(val); \ |
69 | 0 | for (i = 0; i < (len); i += 4) \ |
70 | 0 | AV_WN4P(ptr + i, pix); \ |
71 | 0 | } while (0) |
72 | |
|
73 | 0 | #define EXTEND_RIGHT_CIP(ptr, start, length) \ |
74 | 0 | for (i = start; i < (start) + (length); i += 4) \ |
75 | 0 | if (!IS_INTRA(i, -1)) \ |
76 | 0 | AV_WN4P(&ptr[i], a); \ |
77 | 0 | else \ |
78 | 0 | a = PIXEL_SPLAT_X4(ptr[i+3]) |
79 | 0 | #define EXTEND_LEFT_CIP(ptr, start, length) \ |
80 | 0 | for (i = start; i > (start) - (length); i--) \ |
81 | 0 | if (!IS_INTRA(i - 1, -1)) \ |
82 | 0 | ptr[i - 1] = ptr[i] |
83 | 0 | #define EXTEND_UP_CIP(ptr, start, length) \ |
84 | 0 | for (i = (start); i > (start) - (length); i -= 4) \ |
85 | 0 | if (!IS_INTRA(-1, i - 3)) \ |
86 | 0 | AV_WN4P(&ptr[i - 3], a); \ |
87 | 0 | else \ |
88 | 0 | a = PIXEL_SPLAT_X4(ptr[i - 3]) |
89 | 0 | #define EXTEND_DOWN_CIP(ptr, start, length) \ |
90 | 0 | for (i = start; i < (start) + (length); i += 4) \ |
91 | 0 | if (!IS_INTRA(-1, i)) \ |
92 | 0 | AV_WN4P(&ptr[i], a); \ |
93 | 0 | else \ |
94 | 0 | a = PIXEL_SPLAT_X4(ptr[i + 3]) |
95 | |
|
96 | 0 | HEVCLocalContext *lc = s->HEVClc; |
97 | 0 | #ifdef USE_VAR_BIT_DEPTH |
98 | 0 | int bit_depth = s->sps->bit_depth; |
99 | 0 | #endif |
100 | 0 | int i; |
101 | 0 | int hshift = s->sps->hshift[c_idx]; |
102 | 0 | int vshift = s->sps->vshift[c_idx]; |
103 | 0 | int size = (1 << log2_size); |
104 | 0 | int size_in_luma_h = size << hshift; |
105 | 0 | int size_in_tbs_h = size_in_luma_h >> s->sps->log2_min_tb_size; |
106 | 0 | int size_in_luma_v = size << vshift; |
107 | 0 | int size_in_tbs_v = size_in_luma_v >> s->sps->log2_min_tb_size; |
108 | 0 | int x = x0 >> hshift; |
109 | 0 | int y = y0 >> vshift; |
110 | 0 | int x_tb = (x0 >> s->sps->log2_min_tb_size) & s->sps->tb_mask; |
111 | 0 | int y_tb = (y0 >> s->sps->log2_min_tb_size) & s->sps->tb_mask; |
112 | |
|
113 | 0 | int cur_tb_addr = MIN_TB_ADDR_ZS(x_tb, y_tb); |
114 | |
|
115 | 0 | ptrdiff_t stride = s->frame->linesize[c_idx] / sizeof(pixel); |
116 | 0 | pixel *src = (pixel*)s->frame->data[c_idx] + x + y * stride; |
117 | |
|
118 | 0 | int min_pu_width = s->sps->min_pu_width; |
119 | |
|
120 | 0 | enum IntraPredMode mode = c_idx ? lc->tu.intra_pred_mode_c : |
121 | 0 | lc->tu.intra_pred_mode; |
122 | 0 | pixel4 a; |
123 | 0 | pixel left_array[2 * MAX_TB_SIZE + 1]; |
124 | 0 | pixel filtered_left_array[2 * MAX_TB_SIZE + 1]; |
125 | 0 | pixel top_array[2 * MAX_TB_SIZE + 1]; |
126 | 0 | pixel filtered_top_array[2 * MAX_TB_SIZE + 1]; |
127 | |
|
128 | 0 | pixel *left = left_array + 1; |
129 | 0 | pixel *top = top_array + 1; |
130 | 0 | pixel *filtered_left = filtered_left_array + 1; |
131 | 0 | pixel *filtered_top = filtered_top_array + 1; |
132 | 0 | 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) & s->sps->tb_mask); |
133 | 0 | int cand_left = lc->na.cand_left; |
134 | 0 | int cand_up_left = lc->na.cand_up_left; |
135 | 0 | int cand_up = lc->na.cand_up; |
136 | 0 | int cand_up_right = lc->na.cand_up_right && cur_tb_addr > MIN_TB_ADDR_ZS((x_tb + size_in_tbs_h) & s->sps->tb_mask, y_tb - 1); |
137 | |
|
138 | 0 | int bottom_left_size = (FFMIN(y0 + 2 * size_in_luma_v, s->sps->height) - |
139 | 0 | (y0 + size_in_luma_v)) >> vshift; |
140 | 0 | int top_right_size = (FFMIN(x0 + 2 * size_in_luma_h, s->sps->width) - |
141 | 0 | (x0 + size_in_luma_h)) >> hshift; |
142 | 0 | int disable_intra_boundary_filter; |
143 | |
|
144 | 0 | if (s->pps->constrained_intra_pred_flag == 1) { |
145 | 0 | int size_in_luma_pu_v = PU(size_in_luma_v); |
146 | 0 | int size_in_luma_pu_h = PU(size_in_luma_h); |
147 | 0 | int on_pu_edge_x = !(x0 & ((1 << s->sps->log2_min_pu_size) - 1)); |
148 | 0 | int on_pu_edge_y = !(y0 & ((1 << s->sps->log2_min_pu_size) - 1)); |
149 | 0 | if (!size_in_luma_pu_h) |
150 | 0 | size_in_luma_pu_h++; |
151 | 0 | if (cand_bottom_left == 1 && on_pu_edge_x) { |
152 | 0 | int x_left_pu = PU(x0 - 1); |
153 | 0 | int y_bottom_pu = PU(y0 + size_in_luma_v); |
154 | 0 | int max = FFMIN(size_in_luma_pu_v, s->sps->min_pu_height - y_bottom_pu); |
155 | 0 | cand_bottom_left = 0; |
156 | 0 | for (i = 0; i < max; i += 2) |
157 | 0 | cand_bottom_left |= (MVF(x_left_pu, y_bottom_pu + i) == PF_INTRA); |
158 | 0 | } |
159 | 0 | if (cand_left == 1 && on_pu_edge_x) { |
160 | 0 | int x_left_pu = PU(x0 - 1); |
161 | 0 | int y_left_pu = PU(y0); |
162 | 0 | int max = FFMIN(size_in_luma_pu_v, s->sps->min_pu_height - y_left_pu); |
163 | 0 | cand_left = 0; |
164 | 0 | for (i = 0; i < max; i += 2) |
165 | 0 | cand_left |= (MVF(x_left_pu, y_left_pu + i) == PF_INTRA); |
166 | 0 | } |
167 | 0 | if (cand_up_left == 1) { |
168 | 0 | int x_left_pu = PU(x0 - 1); |
169 | 0 | int y_top_pu = PU(y0 - 1); |
170 | 0 | cand_up_left = MVF(x_left_pu, y_top_pu) == PF_INTRA; |
171 | 0 | } |
172 | 0 | if (cand_up == 1 && on_pu_edge_y) { |
173 | 0 | int x_top_pu = PU(x0); |
174 | 0 | int y_top_pu = PU(y0 - 1); |
175 | 0 | int max = FFMIN(size_in_luma_pu_h, s->sps->min_pu_width - x_top_pu); |
176 | 0 | cand_up = 0; |
177 | 0 | for (i = 0; i < max; i += 2) |
178 | 0 | cand_up |= (MVF(x_top_pu + i, y_top_pu) == PF_INTRA); |
179 | 0 | } |
180 | 0 | if (cand_up_right == 1 && on_pu_edge_y) { |
181 | 0 | int y_top_pu = PU(y0 - 1); |
182 | 0 | int x_right_pu = PU(x0 + size_in_luma_h); |
183 | 0 | int max = FFMIN(size_in_luma_pu_h, s->sps->min_pu_width - x_right_pu); |
184 | 0 | cand_up_right = 0; |
185 | 0 | for (i = 0; i < max; i += 2) |
186 | 0 | cand_up_right |= (MVF(x_right_pu + i, y_top_pu) == PF_INTRA); |
187 | 0 | } |
188 | 0 | memset(left, 128, 2 * MAX_TB_SIZE*sizeof(pixel)); |
189 | 0 | memset(top , 128, 2 * MAX_TB_SIZE*sizeof(pixel)); |
190 | 0 | top[-1] = 128; |
191 | 0 | } |
192 | 0 | if (cand_up_left) { |
193 | 0 | left[-1] = POS(-1, -1); |
194 | 0 | top[-1] = left[-1]; |
195 | 0 | } |
196 | 0 | if (cand_up) |
197 | 0 | memcpy(top, src - stride, size * sizeof(pixel)); |
198 | 0 | if (cand_up_right) { |
199 | 0 | memcpy(top + size, src - stride + size, size * sizeof(pixel)); |
200 | 0 | EXTEND(top + size + top_right_size, POS(size + top_right_size - 1, -1), |
201 | 0 | size - top_right_size); |
202 | 0 | } |
203 | 0 | if (cand_left) |
204 | 0 | for (i = 0; i < size; i++) |
205 | 0 | left[i] = POS(-1, i); |
206 | 0 | if (cand_bottom_left) { |
207 | 0 | for (i = size; i < size + bottom_left_size; i++) |
208 | 0 | left[i] = POS(-1, i); |
209 | 0 | EXTEND(left + size + bottom_left_size, POS(-1, size + bottom_left_size - 1), |
210 | 0 | size - bottom_left_size); |
211 | 0 | } |
212 | |
|
213 | 0 | if (s->pps->constrained_intra_pred_flag == 1) { |
214 | 0 | if (cand_bottom_left || cand_left || cand_up_left || cand_up || cand_up_right) { |
215 | 0 | int size_max_x = x0 + ((2 * size) << hshift) < s->sps->width ? |
216 | 0 | 2 * size : (s->sps->width - x0) >> hshift; |
217 | 0 | int size_max_y = y0 + ((2 * size) << vshift) < s->sps->height ? |
218 | 0 | 2 * size : (s->sps->height - y0) >> vshift; |
219 | 0 | int j = size + (cand_bottom_left? bottom_left_size: 0) -1; |
220 | 0 | if (!cand_up_right) { |
221 | 0 | size_max_x = x0 + ((size) << hshift) < s->sps->width ? |
222 | 0 | size : (s->sps->width - x0) >> hshift; |
223 | 0 | } |
224 | 0 | if (!cand_bottom_left) { |
225 | 0 | size_max_y = y0 + (( size) << vshift) < s->sps->height ? |
226 | 0 | size : (s->sps->height - y0) >> vshift; |
227 | 0 | } |
228 | 0 | if (cand_bottom_left || cand_left || cand_up_left) { |
229 | 0 | while (j > -1 && !IS_INTRA(-1, j)) |
230 | 0 | j--; |
231 | 0 | if (!IS_INTRA(-1, j)) { |
232 | 0 | j = 0; |
233 | 0 | while (j < size_max_x && !IS_INTRA(j, -1)) |
234 | 0 | j++; |
235 | 0 | EXTEND_LEFT_CIP(top, j, j + 1); |
236 | 0 | left[-1] = top[-1]; |
237 | 0 | } |
238 | 0 | } else { |
239 | 0 | j = 0; |
240 | 0 | while (j < size_max_x && !IS_INTRA(j, -1)) |
241 | 0 | j++; |
242 | 0 | if (j > 0) |
243 | 0 | if (x0 > 0) { |
244 | 0 | EXTEND_LEFT_CIP(top, j, j + 1); |
245 | 0 | } else { |
246 | 0 | EXTEND_LEFT_CIP(top, j, j); |
247 | 0 | top[-1] = top[0]; |
248 | 0 | } |
249 | 0 | left[-1] = top[-1]; |
250 | 0 | } |
251 | 0 | left[-1] = top[-1]; |
252 | 0 | if (cand_bottom_left || cand_left) { |
253 | 0 | a = PIXEL_SPLAT_X4(left[-1]); |
254 | 0 | EXTEND_DOWN_CIP(left, 0, size_max_y); |
255 | 0 | } |
256 | 0 | if (!cand_left) |
257 | 0 | EXTEND(left, left[-1], size); |
258 | 0 | if (!cand_bottom_left) |
259 | 0 | EXTEND(left + size, left[size - 1], size); |
260 | 0 | if (x0 != 0 && y0 != 0) { |
261 | 0 | a = PIXEL_SPLAT_X4(left[size_max_y - 1]); |
262 | 0 | EXTEND_UP_CIP(left, size_max_y - 1, size_max_y); |
263 | 0 | if (!IS_INTRA(-1, - 1)) |
264 | 0 | left[-1] = left[0]; |
265 | 0 | } else if (x0 == 0) { |
266 | 0 | EXTEND(left, 0, size_max_y); |
267 | 0 | } else { |
268 | 0 | a = PIXEL_SPLAT_X4(left[size_max_y - 1]); |
269 | 0 | EXTEND_UP_CIP(left, size_max_y - 1, size_max_y); |
270 | 0 | } |
271 | 0 | top[-1] = left[-1]; |
272 | 0 | if (y0 != 0) { |
273 | 0 | a = PIXEL_SPLAT_X4(left[-1]); |
274 | 0 | EXTEND_RIGHT_CIP(top, 0, size_max_x); |
275 | 0 | } |
276 | 0 | } |
277 | 0 | } |
278 | | // Infer the unavailable samples |
279 | 0 | if (!cand_bottom_left) { |
280 | 0 | if (cand_left) { |
281 | 0 | EXTEND(left + size, left[size - 1], size); |
282 | 0 | } else if (cand_up_left) { |
283 | 0 | EXTEND(left, left[-1], 2 * size); |
284 | 0 | cand_left = 1; |
285 | 0 | } else if (cand_up) { |
286 | 0 | left[-1] = top[0]; |
287 | 0 | EXTEND(left, left[-1], 2 * size); |
288 | 0 | cand_up_left = 1; |
289 | 0 | cand_left = 1; |
290 | 0 | } else if (cand_up_right) { |
291 | 0 | EXTEND(top, top[size], size); |
292 | 0 | left[-1] = top[size]; |
293 | 0 | EXTEND(left, left[-1], 2 * size); |
294 | 0 | cand_up = 1; |
295 | 0 | cand_up_left = 1; |
296 | 0 | cand_left = 1; |
297 | 0 | } else { // No samples available |
298 | 0 | left[-1] = (1 << (BIT_DEPTH - 1)); |
299 | 0 | EXTEND(top, left[-1], 2 * size); |
300 | 0 | EXTEND(left, left[-1], 2 * size); |
301 | 0 | } |
302 | 0 | } |
303 | |
|
304 | 0 | if (!cand_left) |
305 | 0 | EXTEND(left, left[size], size); |
306 | 0 | if (!cand_up_left) { |
307 | 0 | left[-1] = left[0]; |
308 | 0 | } |
309 | 0 | if (!cand_up) |
310 | 0 | EXTEND(top, left[-1], size); |
311 | 0 | if (!cand_up_right) |
312 | 0 | EXTEND(top + size, top[size - 1], size); |
313 | |
|
314 | 0 | top[-1] = left[-1]; |
315 | | |
316 | | // Filtering process |
317 | 0 | if (!s->sps->intra_smoothing_disabled_flag && (c_idx == 0 || s->sps->chroma_format_idc == 3)) { |
318 | 0 | if (mode != INTRA_DC && size != 4){ |
319 | 0 | int intra_hor_ver_dist_thresh[] = { 7, 1, 0 }; |
320 | 0 | int min_dist_vert_hor = FFMIN(FFABS((int)(mode - 26U)), |
321 | 0 | FFABS((int)(mode - 10U))); |
322 | 0 | if (min_dist_vert_hor > intra_hor_ver_dist_thresh[log2_size - 3]) { |
323 | 0 | int threshold = 1 << (BIT_DEPTH - 5); |
324 | 0 | if (s->sps->sps_strong_intra_smoothing_enable_flag && c_idx == 0 && |
325 | 0 | log2_size == 5 && |
326 | 0 | FFABS(top[-1] + top[63] - 2 * top[31]) < threshold && |
327 | 0 | FFABS(left[-1] + left[63] - 2 * left[31]) < threshold) { |
328 | | // We can't just overwrite values in top because it could be |
329 | | // a pointer into src |
330 | 0 | filtered_top[-1] = top[-1]; |
331 | 0 | filtered_top[63] = top[63]; |
332 | 0 | for (i = 0; i < 63; i++) |
333 | 0 | filtered_top[i] = ((64 - (i + 1)) * top[-1] + |
334 | 0 | (i + 1) * top[63] + 32) >> 6; |
335 | 0 | for (i = 0; i < 63; i++) |
336 | 0 | left[i] = ((64 - (i + 1)) * left[-1] + |
337 | 0 | (i + 1) * left[63] + 32) >> 6; |
338 | 0 | top = filtered_top; |
339 | 0 | } else { |
340 | 0 | filtered_left[2 * size - 1] = left[2 * size - 1]; |
341 | 0 | filtered_top[2 * size - 1] = top[2 * size - 1]; |
342 | 0 | for (i = 2 * size - 2; i >= 0; i--) |
343 | 0 | filtered_left[i] = (left[i + 1] + 2 * left[i] + |
344 | 0 | left[i - 1] + 2) >> 2; |
345 | 0 | filtered_top[-1] = |
346 | 0 | filtered_left[-1] = (left[0] + 2 * left[-1] + top[0] + 2) >> 2; |
347 | 0 | for (i = 2 * size - 2; i >= 0; i--) |
348 | 0 | filtered_top[i] = (top[i + 1] + 2 * top[i] + |
349 | 0 | top[i - 1] + 2) >> 2; |
350 | 0 | left = filtered_left; |
351 | 0 | top = filtered_top; |
352 | 0 | } |
353 | 0 | } |
354 | 0 | } |
355 | 0 | } |
356 | |
|
357 | 0 | switch (mode) { |
358 | 0 | case INTRA_PLANAR: |
359 | | #ifdef USE_FUNC_PTR |
360 | | s->hpc.pred_planar[log2_size - 2]((uint8_t *)src, (uint8_t *)top, |
361 | | (uint8_t *)left, stride); |
362 | | #else |
363 | 0 | FUNC(pred_planar)((uint8_t *)src, (uint8_t *)top, |
364 | 0 | (uint8_t *)left, stride, log2_size); |
365 | 0 | #endif |
366 | 0 | break; |
367 | 0 | case INTRA_DC: |
368 | | #ifdef USE_FUNC_PTR |
369 | | s->hpc.pred_dc((uint8_t *)src, (uint8_t *)top, |
370 | | (uint8_t *)left, stride, log2_size, c_idx); |
371 | | #else |
372 | 0 | FUNC(pred_dc)((uint8_t *)src, (uint8_t *)top, |
373 | 0 | (uint8_t *)left, stride, log2_size, c_idx); |
374 | 0 | #endif |
375 | 0 | break; |
376 | 0 | default: |
377 | 0 | disable_intra_boundary_filter = (s->sps->implicit_rdpcm_enabled_flag && |
378 | 0 | lc->cu.cu_transquant_bypass_flag); |
379 | | #ifdef USE_FUNC_PTR |
380 | | s->hpc.pred_angular[log2_size - 2]((uint8_t *)src, (uint8_t *)top, |
381 | | (uint8_t *)left, stride, c_idx, |
382 | | mode, |
383 | | disable_intra_boundary_filter BIT_DEPTH_ARG); |
384 | | #else |
385 | 0 | FUNC(pred_angular)((uint8_t *)src, (uint8_t *)top, |
386 | 0 | (uint8_t *)left, stride, c_idx, |
387 | 0 | mode, 1 << log2_size, |
388 | 0 | disable_intra_boundary_filter BIT_DEPTH_ARG); |
389 | 0 | #endif |
390 | 0 | break; |
391 | 0 | } |
392 | 0 | } |
393 | | |
394 | | #ifdef USE_FUNC_PTR |
395 | | |
396 | | #define INTRA_PRED(size) \ |
397 | | static void FUNC(intra_pred_ ## size)(HEVCContext *s, int x0, int y0, int c_idx) \ |
398 | | { \ |
399 | | FUNC(intra_pred)(s, x0, y0, size, c_idx); \ |
400 | | } |
401 | | |
402 | | INTRA_PRED(2) |
403 | | INTRA_PRED(3) |
404 | | INTRA_PRED(4) |
405 | | INTRA_PRED(5) |
406 | | |
407 | | #undef INTRA_PRED |
408 | | |
409 | | #endif |
410 | | |
411 | | static HEVC_INLINE void FUNC(pred_planar)(uint8_t *_src, const uint8_t *_top, |
412 | | const uint8_t *_left, ptrdiff_t stride, |
413 | | int trafo_size) |
414 | 0 | { |
415 | 0 | int x, y; |
416 | 0 | pixel *src = (pixel *)_src; |
417 | 0 | const pixel *top = (const pixel *)_top; |
418 | 0 | const pixel *left = (const pixel *)_left; |
419 | 0 | int size = 1 << trafo_size; |
420 | 0 | for (y = 0; y < size; y++) |
421 | 0 | for (x = 0; x < size; x++) |
422 | 0 | POS(x, y) = ((size - 1 - x) * left[y] + (x + 1) * top[size] + |
423 | 0 | (size - 1 - y) * top[x] + (y + 1) * left[size] + size) >> (trafo_size + 1); |
424 | 0 | } |
425 | | |
426 | | #ifdef USE_FUNC_PTR |
427 | | |
428 | | #define PRED_PLANAR(size)\ |
429 | | static void FUNC(pred_planar_ ## size)(uint8_t *src, const uint8_t *top, \ |
430 | | const uint8_t *left, ptrdiff_t stride) \ |
431 | | { \ |
432 | | FUNC(pred_planar)(src, top, left, stride, size + 2); \ |
433 | | } |
434 | | |
435 | | PRED_PLANAR(0) |
436 | | PRED_PLANAR(1) |
437 | | PRED_PLANAR(2) |
438 | | PRED_PLANAR(3) |
439 | | |
440 | | #undef PRED_PLANAR |
441 | | |
442 | | #endif |
443 | | |
444 | | static void FUNC(pred_dc)(uint8_t *_src, const uint8_t *_top, |
445 | | const uint8_t *_left, |
446 | | ptrdiff_t stride, int log2_size, int c_idx) |
447 | 0 | { |
448 | 0 | int i, j, x, y; |
449 | 0 | int size = (1 << log2_size); |
450 | 0 | pixel *src = (pixel *)_src; |
451 | 0 | const pixel *top = (const pixel *)_top; |
452 | 0 | const pixel *left = (const pixel *)_left; |
453 | 0 | int dc = size; |
454 | 0 | pixel4 a; |
455 | 0 | for (i = 0; i < size; i++) |
456 | 0 | dc += left[i] + top[i]; |
457 | |
|
458 | 0 | dc >>= log2_size + 1; |
459 | |
|
460 | 0 | a = PIXEL_SPLAT_X4(dc); |
461 | |
|
462 | 0 | for (i = 0; i < size; i++) |
463 | 0 | for (j = 0; j < size; j+=4) |
464 | 0 | AV_WN4P(&POS(j, i), a); |
465 | |
|
466 | 0 | if (c_idx == 0 && size < 32) { |
467 | 0 | POS(0, 0) = (left[0] + 2 * dc + top[0] + 2) >> 2; |
468 | 0 | for (x = 1; x < size; x++) |
469 | 0 | POS(x, 0) = (top[x] + 3 * dc + 2) >> 2; |
470 | 0 | for (y = 1; y < size; y++) |
471 | 0 | POS(0, y) = (left[y] + 3 * dc + 2) >> 2; |
472 | 0 | } |
473 | 0 | } |
474 | | |
475 | | static HEVC_INLINE void FUNC(pred_angular)(uint8_t *_src, |
476 | | const uint8_t *_top, |
477 | | const uint8_t *_left, |
478 | | ptrdiff_t stride, int c_idx, |
479 | | int mode, int size, |
480 | | int disable_intra_boundary_filter BIT_DEPTH_PARAM) |
481 | 0 | { |
482 | 0 | int x, y; |
483 | 0 | pixel *src = (pixel *)_src; |
484 | 0 | const pixel *top = (const pixel *)_top; |
485 | 0 | const pixel *left = (const pixel *)_left; |
486 | |
|
487 | 0 | static const int8_t intra_pred_angle[] = { |
488 | 0 | 32, 26, 21, 17, 13, 9, 5, 2, 0, -2, -5, -9, -13, -17, -21, -26, -32, |
489 | 0 | -26, -21, -17, -13, -9, -5, -2, 0, 2, 5, 9, 13, 17, 21, 26, 32 |
490 | 0 | }; |
491 | 0 | static const int16_t inv_angle[] = { |
492 | 0 | -4096, -1638, -910, -630, -482, -390, -315, -256, -315, -390, -482, |
493 | 0 | -630, -910, -1638, -4096 |
494 | 0 | }; |
495 | |
|
496 | 0 | int angle = intra_pred_angle[mode - 2]; |
497 | 0 | pixel ref_array[3 * MAX_TB_SIZE + 4]; |
498 | 0 | pixel *ref_tmp = ref_array + size; |
499 | 0 | const pixel *ref; |
500 | 0 | int last = (size * angle) >> 5; |
501 | |
|
502 | 0 | if (mode >= 18) { |
503 | 0 | ref = top - 1; |
504 | 0 | if (angle < 0 && last < -1) { |
505 | 0 | for (x = 0; x <= size; x += 4) |
506 | 0 | AV_WN4P(&ref_tmp[x], AV_RN4P(&top[x - 1])); |
507 | 0 | for (x = last; x <= -1; x++) |
508 | 0 | ref_tmp[x] = left[-1 + ((x * inv_angle[mode - 11] + 128) >> 8)]; |
509 | 0 | ref = ref_tmp; |
510 | 0 | } |
511 | |
|
512 | 0 | for (y = 0; y < size; y++) { |
513 | 0 | int idx = ((y + 1) * angle) >> 5; |
514 | 0 | int fact = ((y + 1) * angle) & 31; |
515 | 0 | if (fact) { |
516 | 0 | for (x = 0; x < size; x += 4) { |
517 | 0 | POS(x , y) = ((32 - fact) * ref[x + idx + 1] + |
518 | 0 | fact * ref[x + idx + 2] + 16) >> 5; |
519 | 0 | POS(x + 1, y) = ((32 - fact) * ref[x + 1 + idx + 1] + |
520 | 0 | fact * ref[x + 1 + idx + 2] + 16) >> 5; |
521 | 0 | POS(x + 2, y) = ((32 - fact) * ref[x + 2 + idx + 1] + |
522 | 0 | fact * ref[x + 2 + idx + 2] + 16) >> 5; |
523 | 0 | POS(x + 3, y) = ((32 - fact) * ref[x + 3 + idx + 1] + |
524 | 0 | fact * ref[x + 3 + idx + 2] + 16) >> 5; |
525 | 0 | } |
526 | 0 | } else { |
527 | 0 | for (x = 0; x < size; x += 4) |
528 | 0 | AV_WN4P(&POS(x, y), AV_RN4P(&ref[x + idx + 1])); |
529 | 0 | } |
530 | 0 | } |
531 | 0 | if (mode == 26 && c_idx == 0 && size < 32 && |
532 | 0 | !disable_intra_boundary_filter) { |
533 | 0 | for (y = 0; y < size; y++) |
534 | 0 | POS(0, y) = av_clip_pixel(top[0] + ((left[y] - left[-1]) >> 1)); |
535 | 0 | } |
536 | 0 | } else { |
537 | 0 | ref = left - 1; |
538 | 0 | if (angle < 0 && last < -1) { |
539 | 0 | for (x = 0; x <= size; x += 4) |
540 | 0 | AV_WN4P(&ref_tmp[x], AV_RN4P(&left[x - 1])); |
541 | 0 | for (x = last; x <= -1; x++) |
542 | 0 | ref_tmp[x] = top[-1 + ((x * inv_angle[mode - 11] + 128) >> 8)]; |
543 | 0 | ref = ref_tmp; |
544 | 0 | } |
545 | |
|
546 | 0 | for (x = 0; x < size; x++) { |
547 | 0 | int idx = ((x + 1) * angle) >> 5; |
548 | 0 | int fact = ((x + 1) * angle) & 31; |
549 | 0 | if (fact) { |
550 | 0 | for (y = 0; y < size; y++) { |
551 | 0 | POS(x, y) = ((32 - fact) * ref[y + idx + 1] + |
552 | 0 | fact * ref[y + idx + 2] + 16) >> 5; |
553 | 0 | } |
554 | 0 | } else { |
555 | 0 | for (y = 0; y < size; y++) |
556 | 0 | POS(x, y) = ref[y + idx + 1]; |
557 | 0 | } |
558 | 0 | } |
559 | 0 | if (mode == 10 && c_idx == 0 && size < 32 && |
560 | 0 | !disable_intra_boundary_filter) { |
561 | 0 | for (x = 0; x < size; x += 4) { |
562 | 0 | POS(x, 0) = av_clip_pixel(left[0] + ((top[x ] - top[-1]) >> 1)); |
563 | 0 | POS(x + 1, 0) = av_clip_pixel(left[0] + ((top[x + 1] - top[-1]) >> 1)); |
564 | 0 | POS(x + 2, 0) = av_clip_pixel(left[0] + ((top[x + 2] - top[-1]) >> 1)); |
565 | 0 | POS(x + 3, 0) = av_clip_pixel(left[0] + ((top[x + 3] - top[-1]) >> 1)); |
566 | 0 | } |
567 | 0 | } |
568 | 0 | } |
569 | 0 | } |
570 | | |
571 | | #ifdef USE_FUNC_PTR |
572 | | |
573 | | static void FUNC(pred_angular_0)(uint8_t *src, const uint8_t *top, |
574 | | const uint8_t *left, |
575 | | ptrdiff_t stride, int c_idx, int mode, |
576 | | int disable_intra_boundary_filter BIT_DEPTH_PARAM) |
577 | | { |
578 | | FUNC(pred_angular)(src, top, left, stride, c_idx, mode, 1 << 2, |
579 | | disable_intra_boundary_filter BIT_DEPTH_ARG); |
580 | | } |
581 | | |
582 | | static void FUNC(pred_angular_1)(uint8_t *src, const uint8_t *top, |
583 | | const uint8_t *left, |
584 | | ptrdiff_t stride, int c_idx, int mode, |
585 | | int disable_intra_boundary_filter BIT_DEPTH_PARAM) |
586 | | { |
587 | | FUNC(pred_angular)(src, top, left, stride, c_idx, mode, 1 << 3, |
588 | | disable_intra_boundary_filter BIT_DEPTH_ARG); |
589 | | } |
590 | | |
591 | | static void FUNC(pred_angular_2)(uint8_t *src, const uint8_t *top, |
592 | | const uint8_t *left, |
593 | | ptrdiff_t stride, int c_idx, int mode, |
594 | | int disable_intra_boundary_filter BIT_DEPTH_PARAM) |
595 | | { |
596 | | FUNC(pred_angular)(src, top, left, stride, c_idx, mode, 1 << 4, |
597 | | disable_intra_boundary_filter BIT_DEPTH_ARG); |
598 | | } |
599 | | |
600 | | static void FUNC(pred_angular_3)(uint8_t *src, const uint8_t *top, |
601 | | const uint8_t *left, |
602 | | ptrdiff_t stride, int c_idx, int mode, |
603 | | int disable_intra_boundary_filter BIT_DEPTH_PARAM) |
604 | | { |
605 | | FUNC(pred_angular)(src, top, left, stride, c_idx, mode, 1 << 5, |
606 | | disable_intra_boundary_filter BIT_DEPTH_ARG); |
607 | | } |
608 | | #endif |
609 | | |
610 | | #undef EXTEND_LEFT_CIP |
611 | | #undef EXTEND_RIGHT_CIP |
612 | | #undef EXTEND_UP_CIP |
613 | | #undef EXTEND_DOWN_CIP |
614 | | #undef IS_INTRA |
615 | | #undef MVF_PU |
616 | | #undef MVF |
617 | | #undef PU |
618 | | #undef EXTEND |
619 | | #undef MIN_TB_ADDR_ZS |
620 | | #undef POS |