1 |
/************************************************************************** |
/************************************************************************** |
2 |
* |
* |
3 |
* XVID MPEG-4 VIDEO CODEC |
* XVID MPEG-4 VIDEO CODEC |
4 |
* decoder main |
* - Decoder main module - |
5 |
* |
* |
6 |
* This program is an implementation of a part of one or more MPEG-4 |
* Copyright(C) 2002 MinChen <chenm001@163.com> |
7 |
* Video tools as specified in ISO/IEC 14496-2 standard. Those intending |
* 2002 Peter Ross <pross@xvid.org> |
|
* to use this software module in hardware or software products are |
|
|
* advised that its use may infringe existing patents or copyrights, and |
|
|
* any such use would be at such party's own risk. The original |
|
|
* developer of this software module and his/her company, and subsequent |
|
|
* editors and their companies, will have no liability for use of this |
|
|
* software or modifications or derivatives thereof. |
|
8 |
* |
* |
9 |
* This program is free software; you can redistribute it and/or modify |
* This file is part of XviD, a free MPEG-4 video encoder/decoder |
10 |
* it under the terms of the GNU General Public License as published by |
* |
11 |
|
* XviD is free software; you can redistribute it and/or modify it |
12 |
|
* under the terms of the GNU General Public License as published by |
13 |
* the Free Software Foundation; either version 2 of the License, or |
* the Free Software Foundation; either version 2 of the License, or |
14 |
* (at your option) any later version. |
* (at your option) any later version. |
15 |
* |
* |
20 |
* |
* |
21 |
* You should have received a copy of the GNU General Public License |
* You should have received a copy of the GNU General Public License |
22 |
* along with this program; if not, write to the Free Software |
* along with this program; if not, write to the Free Software |
23 |
* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. |
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
24 |
* |
* |
25 |
*************************************************************************/ |
* Under section 8 of the GNU General Public License, the copyright |
26 |
|
* holders of XVID explicitly forbid distribution in the following |
27 |
/************************************************************************** |
* countries: |
28 |
|
* |
29 |
|
* - Japan |
30 |
|
* - United States of America |
31 |
* |
* |
32 |
* History: |
* Linking XviD statically or dynamically with other modules is making a |
33 |
|
* combined work based on XviD. Thus, the terms and conditions of the |
34 |
|
* GNU General Public License cover the whole combination. |
35 |
* |
* |
36 |
* 26.12.2001 decoder_mbinter: dequant/idct moved within if(coded) block |
* As a special exception, the copyright holders of XviD give you |
37 |
* 22.12.2001 block based interpolation |
* permission to link XviD with independent modules that communicate with |
38 |
* 01.12.2001 inital version; (c)2001 peter ross <pross@cs.rmit.edu.au> |
* XviD solely through the VFW1.1 and DShow interfaces, regardless of the |
39 |
|
* license terms of these independent modules, and to copy and distribute |
40 |
|
* the resulting combined work under terms of your choice, provided that |
41 |
|
* every copy of the combined work is accompanied by a complete copy of |
42 |
|
* the source code of XviD (the version of XviD used to produce the |
43 |
|
* combined work), being distributed under the terms of the GNU General |
44 |
|
* Public License plus this exception. An independent module is a module |
45 |
|
* which is not derived from or based on XviD. |
46 |
|
* |
47 |
|
* Note that people who make modified versions of XviD are not obligated |
48 |
|
* to grant this special exception for their modified versions; it is |
49 |
|
* their choice whether to do so. The GNU General Public License gives |
50 |
|
* permission to release a modified version without this exception; this |
51 |
|
* exception also makes it possible to release a modified version which |
52 |
|
* carries forward this exception. |
53 |
|
* |
54 |
|
* $Id$ |
55 |
* |
* |
56 |
*************************************************************************/ |
*************************************************************************/ |
57 |
|
|
58 |
#include <stdlib.h> |
#include <stdlib.h> |
59 |
#include <string.h> // memset |
#include <string.h> |
60 |
|
|
61 |
|
#ifdef BFRAMES_DEC_DEBUG |
62 |
|
#define BFRAMES_DEC |
63 |
|
#endif |
64 |
|
|
65 |
#include "xvid.h" |
#include "xvid.h" |
66 |
#include "portab.h" |
#include "portab.h" |
83 |
|
|
84 |
#include "image/image.h" |
#include "image/image.h" |
85 |
#include "image/colorspace.h" |
#include "image/colorspace.h" |
86 |
|
#include "utils/mem_align.h" |
87 |
|
|
88 |
int decoder_create(XVID_DEC_PARAM * param) |
int |
89 |
|
decoder_create(XVID_DEC_PARAM * param) |
90 |
{ |
{ |
91 |
DECODER * dec; |
DECODER * dec; |
92 |
|
|
93 |
dec = malloc(sizeof(DECODER)); |
dec = xvid_malloc(sizeof(DECODER), CACHE_LINE); |
94 |
if (dec == NULL) |
if (dec == NULL) { |
|
{ |
|
95 |
return XVID_ERR_MEMORY; |
return XVID_ERR_MEMORY; |
96 |
} |
} |
97 |
param->handle = dec; |
param->handle = dec; |
104 |
|
|
105 |
dec->edged_width = 16 * dec->mb_width + 2 * EDGE_SIZE; |
dec->edged_width = 16 * dec->mb_width + 2 * EDGE_SIZE; |
106 |
dec->edged_height = 16 * dec->mb_height + 2 * EDGE_SIZE; |
dec->edged_height = 16 * dec->mb_height + 2 * EDGE_SIZE; |
107 |
|
dec->low_delay = 0; |
108 |
|
|
109 |
if (image_create(&dec->cur, dec->edged_width, dec->edged_height)) |
if (image_create(&dec->cur, dec->edged_width, dec->edged_height)) { |
110 |
{ |
xvid_free(dec); |
|
free(dec); |
|
111 |
return XVID_ERR_MEMORY; |
return XVID_ERR_MEMORY; |
112 |
} |
} |
113 |
|
|
114 |
if (image_create(&dec->refn, dec->edged_width, dec->edged_height)) |
if (image_create(&dec->refn[0], dec->edged_width, dec->edged_height)) { |
115 |
{ |
image_destroy(&dec->cur, dec->edged_width, dec->edged_height); |
116 |
|
xvid_free(dec); |
117 |
|
return XVID_ERR_MEMORY; |
118 |
|
} |
119 |
|
/* add by chenm001 <chenm001@163.com> */ |
120 |
|
/* for support B-frame to reference last 2 frame */ |
121 |
|
if (image_create(&dec->refn[1], dec->edged_width, dec->edged_height)) { |
122 |
|
image_destroy(&dec->cur, dec->edged_width, dec->edged_height); |
123 |
|
image_destroy(&dec->refn[0], dec->edged_width, dec->edged_height); |
124 |
|
xvid_free(dec); |
125 |
|
return XVID_ERR_MEMORY; |
126 |
|
} |
127 |
|
if (image_create(&dec->refn[2], dec->edged_width, dec->edged_height)) { |
128 |
image_destroy(&dec->cur, dec->edged_width, dec->edged_height); |
image_destroy(&dec->cur, dec->edged_width, dec->edged_height); |
129 |
free(dec); |
image_destroy(&dec->refn[0], dec->edged_width, dec->edged_height); |
130 |
|
image_destroy(&dec->refn[1], dec->edged_width, dec->edged_height); |
131 |
|
xvid_free(dec); |
132 |
return XVID_ERR_MEMORY; |
return XVID_ERR_MEMORY; |
133 |
} |
} |
134 |
|
|
135 |
dec->mbs = malloc(sizeof(MACROBLOCK) * dec->mb_width * dec->mb_height); |
dec->mbs = |
136 |
if (dec->mbs == NULL) |
xvid_malloc(sizeof(MACROBLOCK) * dec->mb_width * dec->mb_height, |
137 |
{ |
CACHE_LINE); |
138 |
|
if (dec->mbs == NULL) { |
139 |
image_destroy(&dec->cur, dec->edged_width, dec->edged_height); |
image_destroy(&dec->cur, dec->edged_width, dec->edged_height); |
140 |
free(dec); |
image_destroy(&dec->refn[0], dec->edged_width, dec->edged_height); |
141 |
|
image_destroy(&dec->refn[1], dec->edged_width, dec->edged_height); |
142 |
|
image_destroy(&dec->refn[2], dec->edged_width, dec->edged_height); |
143 |
|
xvid_free(dec); |
144 |
return XVID_ERR_MEMORY; |
return XVID_ERR_MEMORY; |
145 |
} |
} |
146 |
|
|
147 |
|
memset(dec->mbs, 0, sizeof(MACROBLOCK) * dec->mb_width * dec->mb_height); |
148 |
|
|
149 |
|
/* add by chenm001 <chenm001@163.com> */ |
150 |
|
/* for skip MB flag */ |
151 |
|
dec->last_mbs = |
152 |
|
xvid_malloc(sizeof(MACROBLOCK) * dec->mb_width * dec->mb_height, |
153 |
|
CACHE_LINE); |
154 |
|
if (dec->last_mbs == NULL) { |
155 |
|
xvid_free(dec->mbs); |
156 |
|
image_destroy(&dec->cur, dec->edged_width, dec->edged_height); |
157 |
|
image_destroy(&dec->refn[0], dec->edged_width, dec->edged_height); |
158 |
|
image_destroy(&dec->refn[1], dec->edged_width, dec->edged_height); |
159 |
|
image_destroy(&dec->refn[2], dec->edged_width, dec->edged_height); |
160 |
|
xvid_free(dec); |
161 |
|
return XVID_ERR_MEMORY; |
162 |
|
} |
163 |
|
|
164 |
|
memset(dec->last_mbs, 0, sizeof(MACROBLOCK) * dec->mb_width * dec->mb_height); |
165 |
|
|
166 |
init_timer(); |
init_timer(); |
167 |
create_vlc_tables(); |
|
168 |
|
/* add by chenm001 <chenm001@163.com> */ |
169 |
|
/* for support B-frame to save reference frame's time */ |
170 |
|
dec->frames = -1; |
171 |
|
dec->time = dec->time_base = dec->last_time_base = 0; |
172 |
|
|
173 |
return XVID_ERR_OK; |
return XVID_ERR_OK; |
174 |
} |
} |
175 |
|
|
176 |
|
|
177 |
int decoder_destroy(DECODER * dec) |
int |
178 |
|
decoder_destroy(DECODER * dec) |
179 |
{ |
{ |
180 |
free(dec->mbs); |
xvid_free(dec->last_mbs); |
181 |
image_destroy(&dec->refn, dec->edged_width, dec->edged_height); |
xvid_free(dec->mbs); |
182 |
|
image_destroy(&dec->refn[0], dec->edged_width, dec->edged_height); |
183 |
|
image_destroy(&dec->refn[1], dec->edged_width, dec->edged_height); |
184 |
|
image_destroy(&dec->refn[2], dec->edged_width, dec->edged_height); |
185 |
image_destroy(&dec->cur, dec->edged_width, dec->edged_height); |
image_destroy(&dec->cur, dec->edged_width, dec->edged_height); |
186 |
free(dec); |
xvid_free(dec); |
|
|
|
|
destroy_vlc_tables(); |
|
187 |
|
|
188 |
write_timer(); |
write_timer(); |
189 |
return XVID_ERR_OK; |
return XVID_ERR_OK; |
191 |
|
|
192 |
|
|
193 |
|
|
194 |
static const int32_t dquant_table[4] = |
static const int32_t dquant_table[4] = { |
|
{ |
|
195 |
-1, -2, 1, 2 |
-1, -2, 1, 2 |
196 |
}; |
}; |
197 |
|
|
198 |
|
|
|
// decode an intra macroblock |
|
199 |
|
|
|
void decoder_mbintra(DECODER * dec, MACROBLOCK * mb, int x, int y, uint32_t acpred_flag, uint32_t cbp, Bitstream * bs, int quant, int intra_dc_threshold) |
|
|
{ |
|
|
uint32_t k; |
|
200 |
|
|
201 |
for (k = 0; k < 6; k++) |
/* decode an intra macroblock */ |
202 |
{ |
|
203 |
uint32_t dcscalar; |
void |
204 |
int16_t block[64]; |
decoder_mbintra(DECODER * dec, |
205 |
int16_t data[64]; |
MACROBLOCK * pMB, |
206 |
|
const uint32_t x_pos, |
207 |
|
const uint32_t y_pos, |
208 |
|
const uint32_t acpred_flag, |
209 |
|
const uint32_t cbp, |
210 |
|
Bitstream * bs, |
211 |
|
const uint32_t quant, |
212 |
|
const uint32_t intra_dc_threshold, |
213 |
|
const unsigned int bound) |
214 |
|
{ |
215 |
|
|
216 |
|
DECLARE_ALIGNED_MATRIX(block, 6, 64, int16_t, CACHE_LINE); |
217 |
|
DECLARE_ALIGNED_MATRIX(data, 6, 64, int16_t, CACHE_LINE); |
218 |
|
|
219 |
|
uint32_t stride = dec->edged_width; |
220 |
|
uint32_t stride2 = stride / 2; |
221 |
|
uint32_t next_block = stride * 8; |
222 |
|
uint32_t i; |
223 |
|
uint32_t iQuant = pMB->quant; |
224 |
|
uint8_t *pY_Cur, *pU_Cur, *pV_Cur; |
225 |
|
|
226 |
|
pY_Cur = dec->cur.y + (y_pos << 4) * stride + (x_pos << 4); |
227 |
|
pU_Cur = dec->cur.u + (y_pos << 3) * stride2 + (x_pos << 3); |
228 |
|
pV_Cur = dec->cur.v + (y_pos << 3) * stride2 + (x_pos << 3); |
229 |
|
|
230 |
|
memset(block, 0, 6 * 64 * sizeof(int16_t)); /* clear */ |
231 |
|
|
232 |
|
for (i = 0; i < 6; i++) { |
233 |
|
uint32_t iDcScaler = get_dc_scaler(iQuant, i < 4); |
234 |
int16_t predictors[8]; |
int16_t predictors[8]; |
235 |
int start_coeff; |
int start_coeff; |
236 |
|
|
|
dcscalar = get_dc_scaler(mb->quant, k < 4); |
|
|
|
|
237 |
start_timer(); |
start_timer(); |
238 |
predict_acdc(dec->mbs, x, y, dec->mb_width, k, block, mb->quant, dcscalar, predictors); |
predict_acdc(dec->mbs, x_pos, y_pos, dec->mb_width, i, &block[i * 64], |
239 |
if (!acpred_flag) |
iQuant, iDcScaler, predictors, bound); |
240 |
{ |
if (!acpred_flag) { |
241 |
mb->acpred_directions[k] = 0; |
pMB->acpred_directions[i] = 0; |
242 |
} |
} |
243 |
stop_prediction_timer(); |
stop_prediction_timer(); |
244 |
|
|
245 |
memset(block, 0, 64*sizeof(int16_t)); // clear |
if (quant < intra_dc_threshold) { |
|
|
|
|
if (quant < intra_dc_threshold) |
|
|
{ |
|
246 |
int dc_size; |
int dc_size; |
247 |
int dc_dif; |
int dc_dif; |
248 |
|
|
249 |
dc_size = k < 4 ? get_dc_size_lum(bs) : get_dc_size_chrom(bs); |
dc_size = i < 4 ? get_dc_size_lum(bs) : get_dc_size_chrom(bs); |
250 |
dc_dif = dc_size ? get_dc_dif(bs, dc_size) : 0 ; |
dc_dif = dc_size ? get_dc_dif(bs, dc_size) : 0 ; |
251 |
|
|
252 |
if (dc_size > 8) |
if (dc_size > 8) { |
253 |
{ |
BitstreamSkip(bs, 1); /* marker */ |
|
BitstreamSkip(bs, 1); // marker |
|
254 |
} |
} |
255 |
|
|
256 |
block[0] = dc_dif; |
block[i * 64 + 0] = dc_dif; |
257 |
start_coeff = 1; |
start_coeff = 1; |
258 |
} |
|
259 |
else |
DPRINTF(DPRINTF_COEFF,"block[0] %i", dc_dif); |
260 |
{ |
} else { |
261 |
start_coeff = 0; |
start_coeff = 0; |
262 |
} |
} |
263 |
|
|
264 |
start_timer(); |
start_timer(); |
265 |
if (cbp & (1 << (5-k))) // coded |
if (cbp & (1 << (5 - i))) /* coded */ |
266 |
{ |
{ |
267 |
get_intra_block(bs, block, mb->acpred_directions[k], start_coeff); |
get_intra_block(bs, &block[i * 64], pMB->acpred_directions[i], |
268 |
|
start_coeff); |
269 |
} |
} |
270 |
stop_coding_timer(); |
stop_coding_timer(); |
271 |
|
|
272 |
start_timer(); |
start_timer(); |
273 |
add_acdc(mb, k, block, dcscalar, predictors); |
add_acdc(pMB, i, &block[i * 64], iDcScaler, predictors); |
274 |
stop_prediction_timer(); |
stop_prediction_timer(); |
275 |
|
|
276 |
start_timer(); |
start_timer(); |
277 |
if (dec->quant_type == 0) |
if (dec->quant_type == 0) { |
278 |
{ |
dequant_intra(&data[i * 64], &block[i * 64], iQuant, iDcScaler); |
279 |
dequant_intra(data, block, mb->quant, dcscalar); |
} else { |
280 |
} |
dequant4_intra(&data[i * 64], &block[i * 64], iQuant, iDcScaler); |
|
else |
|
|
{ |
|
|
dequant4_intra(data, block, mb->quant, dcscalar); |
|
281 |
} |
} |
282 |
stop_iquant_timer(); |
stop_iquant_timer(); |
283 |
|
|
284 |
start_timer(); |
start_timer(); |
285 |
idct(data); |
idct(&data[i * 64]); |
286 |
stop_idct_timer(); |
stop_idct_timer(); |
|
|
|
|
start_timer(); |
|
|
if (k < 4) |
|
|
{ |
|
|
transfer_16to8copy(dec->cur.y + (16*y*dec->edged_width) + 16*x + (4*(k&2)*dec->edged_width) + 8*(k&1), data, dec->edged_width); |
|
287 |
} |
} |
288 |
else if (k == 4) |
|
289 |
{ |
if (dec->interlacing && pMB->field_dct) { |
290 |
transfer_16to8copy(dec->cur.u+ 8*y*(dec->edged_width/2) + 8*x, data, (dec->edged_width/2)); |
next_block = stride; |
291 |
} |
stride *= 2; |
|
else // if (k == 5) |
|
|
{ |
|
|
transfer_16to8copy(dec->cur.v + 8*y*(dec->edged_width/2) + 8*x, data, (dec->edged_width/2)); |
|
292 |
} |
} |
293 |
|
|
294 |
|
start_timer(); |
295 |
|
transfer_16to8copy(pY_Cur, &data[0 * 64], stride); |
296 |
|
transfer_16to8copy(pY_Cur + 8, &data[1 * 64], stride); |
297 |
|
transfer_16to8copy(pY_Cur + next_block, &data[2 * 64], stride); |
298 |
|
transfer_16to8copy(pY_Cur + 8 + next_block, &data[3 * 64], stride); |
299 |
|
transfer_16to8copy(pU_Cur, &data[4 * 64], stride2); |
300 |
|
transfer_16to8copy(pV_Cur, &data[5 * 64], stride2); |
301 |
stop_transfer_timer(); |
stop_transfer_timer(); |
302 |
} |
} |
|
} |
|
303 |
|
|
304 |
|
|
305 |
|
|
311 |
{ 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2 }; |
{ 0, 0, 0, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 2, 2 }; |
312 |
|
|
313 |
|
|
314 |
// decode an inter macroblock |
/* decode an inter macroblock */ |
315 |
|
|
316 |
void decoder_mbinter(DECODER * dec, MACROBLOCK * mb, int x, int y, uint32_t acpred_flag, uint32_t cbp, Bitstream * bs, int quant, int rounding) |
void |
317 |
{ |
decoder_mbinter(DECODER * dec, |
318 |
const uint32_t stride = dec->edged_width; |
const MACROBLOCK * pMB, |
319 |
const uint32_t stride2 = dec->edged_width / 2; |
const uint32_t x_pos, |
320 |
|
const uint32_t y_pos, |
321 |
|
const uint32_t acpred_flag, |
322 |
|
const uint32_t cbp, |
323 |
|
Bitstream * bs, |
324 |
|
const uint32_t quant, |
325 |
|
const uint32_t rounding) |
326 |
|
{ |
327 |
|
|
328 |
|
DECLARE_ALIGNED_MATRIX(block, 6, 64, int16_t, CACHE_LINE); |
329 |
|
DECLARE_ALIGNED_MATRIX(data, 6, 64, int16_t, CACHE_LINE); |
330 |
|
|
331 |
|
uint32_t stride = dec->edged_width; |
332 |
|
uint32_t stride2 = stride / 2; |
333 |
|
uint32_t next_block = stride * 8; |
334 |
|
uint32_t i; |
335 |
|
uint32_t iQuant = pMB->quant; |
336 |
|
uint8_t *pY_Cur, *pU_Cur, *pV_Cur; |
337 |
int uv_dx, uv_dy; |
int uv_dx, uv_dy; |
|
uint32_t k; |
|
338 |
|
|
339 |
if (mb->mode == MODE_INTER || mb->mode == MODE_INTER_Q) |
pY_Cur = dec->cur.y + (y_pos << 4) * stride + (x_pos << 4); |
340 |
{ |
pU_Cur = dec->cur.u + (y_pos << 3) * stride2 + (x_pos << 3); |
341 |
uv_dx = mb->mvs[0].x; |
pV_Cur = dec->cur.v + (y_pos << 3) * stride2 + (x_pos << 3); |
342 |
uv_dy = mb->mvs[0].y; |
|
343 |
|
if (pMB->mode == MODE_INTER || pMB->mode == MODE_INTER_Q) { |
344 |
|
uv_dx = pMB->mvs[0].x; |
345 |
|
uv_dy = pMB->mvs[0].y; |
346 |
|
|
347 |
uv_dx = (uv_dx & 3) ? (uv_dx >> 1) | 1 : uv_dx / 2; |
uv_dx = (uv_dx & 3) ? (uv_dx >> 1) | 1 : uv_dx / 2; |
348 |
uv_dy = (uv_dy & 3) ? (uv_dy >> 1) | 1 : uv_dy / 2; |
uv_dy = (uv_dy & 3) ? (uv_dy >> 1) | 1 : uv_dy / 2; |
349 |
} |
} else { |
|
else |
|
|
{ |
|
350 |
int sum; |
int sum; |
351 |
sum = mb->mvs[0].x + mb->mvs[1].x + mb->mvs[2].x + mb->mvs[3].x; |
sum = pMB->mvs[0].x + pMB->mvs[1].x + pMB->mvs[2].x + pMB->mvs[3].x; |
352 |
|
|
353 |
uv_dx = (sum == 0 ? 0 : SIGN(sum) * (roundtab[ABS(sum) % 16] + (ABS(sum) / 16) * 2) ); |
uv_dx = (sum == 0 ? 0 : SIGN(sum) * (roundtab[ABS(sum) % 16] + (ABS(sum) / 16) * 2) ); |
354 |
|
|
355 |
sum = mb->mvs[0].y + mb->mvs[1].y + mb->mvs[2].y + mb->mvs[3].y; |
sum = pMB->mvs[0].y + pMB->mvs[1].y + pMB->mvs[2].y + pMB->mvs[3].y; |
356 |
|
|
357 |
uv_dy = (sum == 0 ? 0 : SIGN(sum) * (roundtab[ABS(sum) % 16] + (ABS(sum) / 16) * 2) ); |
uv_dy = (sum == 0 ? 0 : SIGN(sum) * (roundtab[ABS(sum) % 16] + (ABS(sum) / 16) * 2) ); |
358 |
} |
} |
359 |
|
|
360 |
start_timer(); |
start_timer(); |
361 |
interpolate8x8_switch(dec->cur.y, dec->refn.y, 16*x, 16*y , mb->mvs[0].x, mb->mvs[0].y, stride, rounding); |
interpolate8x8_switch(dec->cur.y, dec->refn[0].y, 16*x_pos, 16*y_pos, |
362 |
interpolate8x8_switch(dec->cur.y, dec->refn.y, 16*x + 8, 16*y , mb->mvs[1].x, mb->mvs[1].y, stride, rounding); |
pMB->mvs[0].x, pMB->mvs[0].y, stride, rounding); |
363 |
interpolate8x8_switch(dec->cur.y, dec->refn.y, 16*x, 16*y + 8, mb->mvs[2].x, mb->mvs[2].y, stride, rounding); |
interpolate8x8_switch(dec->cur.y, dec->refn[0].y, 16*x_pos + 8, 16*y_pos, |
364 |
interpolate8x8_switch(dec->cur.y, dec->refn.y, 16*x + 8, 16*y + 8, mb->mvs[3].x, mb->mvs[3].y, stride, rounding); |
pMB->mvs[1].x, pMB->mvs[1].y, stride, rounding); |
365 |
interpolate8x8_switch(dec->cur.u, dec->refn.u, 8*x, 8*y, uv_dx, uv_dy, stride2, rounding); |
interpolate8x8_switch(dec->cur.y, dec->refn[0].y, 16*x_pos, 16*y_pos + 8, |
366 |
interpolate8x8_switch(dec->cur.v, dec->refn.v, 8*x, 8*y, uv_dx, uv_dy, stride2, rounding); |
pMB->mvs[2].x, pMB->mvs[2].y, stride, rounding); |
367 |
|
interpolate8x8_switch(dec->cur.y, dec->refn[0].y, 16*x_pos + 8, 16*y_pos + 8, |
368 |
|
pMB->mvs[3].x, pMB->mvs[3].y, stride, rounding); |
369 |
|
interpolate8x8_switch(dec->cur.u, dec->refn[0].u, 8 * x_pos, 8 * y_pos, |
370 |
|
uv_dx, uv_dy, stride2, rounding); |
371 |
|
interpolate8x8_switch(dec->cur.v, dec->refn[0].v, 8 * x_pos, 8 * y_pos, |
372 |
|
uv_dx, uv_dy, stride2, rounding); |
373 |
stop_comp_timer(); |
stop_comp_timer(); |
374 |
|
|
375 |
|
for (i = 0; i < 6; i++) { |
376 |
for (k = 0; k < 6; k++) |
if (cbp & (1 << (5 - i))) /* coded */ |
377 |
{ |
{ |
378 |
int16_t block[64]; |
memset(&block[i * 64], 0, 64 * sizeof(int16_t)); /* clear */ |
|
int16_t data[64]; |
|
|
|
|
|
if (cbp & (1 << (5-k))) // coded |
|
|
{ |
|
|
memset(block, 0, 64 * sizeof(int16_t)); // clear |
|
379 |
|
|
380 |
start_timer(); |
start_timer(); |
381 |
get_inter_block(bs, block); |
get_inter_block(bs, &block[i * 64]); |
382 |
stop_coding_timer(); |
stop_coding_timer(); |
383 |
|
|
384 |
start_timer(); |
start_timer(); |
385 |
if (dec->quant_type == 0) |
if (dec->quant_type == 0) { |
386 |
{ |
dequant_inter(&data[i * 64], &block[i * 64], iQuant); |
387 |
dequant_inter(data, block, mb->quant); |
} else { |
388 |
} |
dequant4_inter(&data[i * 64], &block[i * 64], iQuant); |
|
else |
|
|
{ |
|
|
dequant4_inter(data, block, mb->quant); |
|
389 |
} |
} |
390 |
stop_iquant_timer(); |
stop_iquant_timer(); |
391 |
|
|
392 |
start_timer(); |
start_timer(); |
393 |
idct(data); |
idct(&data[i * 64]); |
394 |
stop_idct_timer(); |
stop_idct_timer(); |
|
|
|
|
start_timer(); |
|
|
if (k < 4) |
|
|
{ |
|
|
transfer_16to8add(dec->cur.y + (16*y + 4*(k&2))*stride + 16*x + 8*(k&1), data, stride); |
|
395 |
} |
} |
|
else if (k == 4) |
|
|
{ |
|
|
transfer_16to8add(dec->cur.u + 8*y*stride2 + 8*x, data, stride2); |
|
396 |
} |
} |
397 |
else // k == 5 |
|
398 |
{ |
if (dec->interlacing && pMB->field_dct) { |
399 |
transfer_16to8add(dec->cur.v + 8*y*stride2 + 8*x, data, stride2); |
next_block = stride; |
400 |
|
stride *= 2; |
401 |
} |
} |
402 |
|
|
403 |
|
start_timer(); |
404 |
|
if (cbp & 32) |
405 |
|
transfer_16to8add(pY_Cur, &data[0 * 64], stride); |
406 |
|
if (cbp & 16) |
407 |
|
transfer_16to8add(pY_Cur + 8, &data[1 * 64], stride); |
408 |
|
if (cbp & 8) |
409 |
|
transfer_16to8add(pY_Cur + next_block, &data[2 * 64], stride); |
410 |
|
if (cbp & 4) |
411 |
|
transfer_16to8add(pY_Cur + 8 + next_block, &data[3 * 64], stride); |
412 |
|
if (cbp & 2) |
413 |
|
transfer_16to8add(pU_Cur, &data[4 * 64], stride2); |
414 |
|
if (cbp & 1) |
415 |
|
transfer_16to8add(pV_Cur, &data[5 * 64], stride2); |
416 |
stop_transfer_timer(); |
stop_transfer_timer(); |
417 |
} |
} |
|
} |
|
|
} |
|
418 |
|
|
419 |
|
|
420 |
|
void |
421 |
void decoder_iframe(DECODER * dec, Bitstream * bs, int quant, int intra_dc_threshold) |
decoder_iframe(DECODER * dec, |
422 |
|
Bitstream * bs, |
423 |
|
int quant, |
424 |
|
int intra_dc_threshold) |
425 |
{ |
{ |
426 |
|
uint32_t bound; |
427 |
uint32_t x, y; |
uint32_t x, y; |
428 |
|
|
429 |
for (y = 0; y < dec->mb_height; y++) |
bound = 0; |
|
{ |
|
|
for (x = 0; x < dec->mb_width; x++) |
|
|
{ |
|
|
MACROBLOCK * mb = &dec->mbs[y*dec->mb_width + x]; |
|
430 |
|
|
431 |
|
for (y = 0; y < dec->mb_height; y++) { |
432 |
|
for (x = 0; x < dec->mb_width; x++) { |
433 |
|
MACROBLOCK *mb; |
434 |
uint32_t mcbpc; |
uint32_t mcbpc; |
435 |
uint32_t cbpc; |
uint32_t cbpc; |
436 |
uint32_t acpred_flag; |
uint32_t acpred_flag; |
437 |
uint32_t cbpy; |
uint32_t cbpy; |
438 |
uint32_t cbp; |
uint32_t cbp; |
439 |
|
|
440 |
|
while (BitstreamShowBits(bs, 9) == 1) |
441 |
|
BitstreamSkip(bs, 9); |
442 |
|
|
443 |
|
if (check_resync_marker(bs, 0)) |
444 |
|
{ |
445 |
|
bound = read_video_packet_header(bs, 0, &quant); |
446 |
|
x = bound % dec->mb_width; |
447 |
|
y = bound / dec->mb_width; |
448 |
|
} |
449 |
|
mb = &dec->mbs[y * dec->mb_width + x]; |
450 |
|
|
451 |
|
DPRINTF(DPRINTF_MB, "macroblock (%i,%i) %08x", x, y, BitstreamShowBits(bs, 32)); |
452 |
|
|
453 |
mcbpc = get_mcbpc_intra(bs); |
mcbpc = get_mcbpc_intra(bs); |
454 |
mb->mode = mcbpc & 7; |
mb->mode = mcbpc & 7; |
455 |
cbpc = (mcbpc >> 4); |
cbpc = (mcbpc >> 4); |
456 |
|
|
457 |
acpred_flag = BitstreamGetBit(bs); |
acpred_flag = BitstreamGetBit(bs); |
458 |
|
|
|
if (mb->mode == MODE_STUFFING) |
|
|
{ |
|
|
DEBUG("-- STUFFING ?"); |
|
|
continue; |
|
|
} |
|
|
|
|
459 |
cbpy = get_cbpy(bs, 1); |
cbpy = get_cbpy(bs, 1); |
460 |
cbp = (cbpy << 2) | cbpc; |
cbp = (cbpy << 2) | cbpc; |
461 |
|
|
462 |
if (mb->mode == MODE_INTRA_Q) |
if (mb->mode == MODE_INTRA_Q) { |
|
{ |
|
463 |
quant += dquant_table[BitstreamGetBits(bs,2)]; |
quant += dquant_table[BitstreamGetBits(bs,2)]; |
464 |
if (quant > 31) |
if (quant > 31) { |
|
{ |
|
465 |
quant = 31; |
quant = 31; |
466 |
} |
} else if (quant < 1) { |
|
else if (quant < 1) |
|
|
{ |
|
467 |
quant = 1; |
quant = 1; |
468 |
} |
} |
469 |
} |
} |
470 |
mb->quant = quant; |
mb->quant = quant; |
471 |
|
mb->mvs[0].x = mb->mvs[0].y = |
472 |
|
mb->mvs[1].x = mb->mvs[1].y = |
473 |
|
mb->mvs[2].x = mb->mvs[2].y = |
474 |
|
mb->mvs[3].x = mb->mvs[3].y =0; |
475 |
|
|
476 |
|
if (dec->interlacing) { |
477 |
|
mb->field_dct = BitstreamGetBit(bs); |
478 |
|
DPRINTF(DPRINTF_DEBUG, "deci: field_dct: %d", mb->field_dct); |
479 |
|
} |
480 |
|
|
481 |
decoder_mbintra(dec, mb, x, y, acpred_flag, cbp, bs, quant, intra_dc_threshold); |
decoder_mbintra(dec, mb, x, y, acpred_flag, cbp, bs, quant, |
482 |
|
intra_dc_threshold, bound); |
483 |
} |
} |
484 |
|
if(dec->out_frm) |
485 |
|
output_slice(&dec->cur, dec->edged_width,dec->width,dec->out_frm,0,y,dec->mb_width); |
486 |
|
|
487 |
} |
} |
488 |
|
|
489 |
} |
} |
490 |
|
|
491 |
|
|
492 |
void get_motion_vector(DECODER *dec, Bitstream *bs, int x, int y, int k, VECTOR * mv, int fcode) |
void |
493 |
|
get_motion_vector(DECODER * dec, |
494 |
|
Bitstream * bs, |
495 |
|
int x, |
496 |
|
int y, |
497 |
|
int k, |
498 |
|
VECTOR * mv, |
499 |
|
int fcode, |
500 |
|
const int bound) |
501 |
{ |
{ |
502 |
|
|
503 |
int scale_fac = 1 << (fcode - 1); |
int scale_fac = 1 << (fcode - 1); |
504 |
int high = (32 * scale_fac) - 1; |
int high = (32 * scale_fac) - 1; |
505 |
int low = ((-32) * scale_fac); |
int low = ((-32) * scale_fac); |
506 |
int range = (64 * scale_fac); |
int range = (64 * scale_fac); |
507 |
|
|
508 |
VECTOR pmv[4]; |
VECTOR pmv; |
|
uint32_t psad[4]; |
|
|
|
|
509 |
int mv_x, mv_y; |
int mv_x, mv_y; |
|
int pmv_x, pmv_y; |
|
|
|
|
510 |
|
|
511 |
get_pmvdata(dec->mbs, x, y, dec->mb_width, k, pmv, psad); |
pmv = get_pmv2(dec->mbs, dec->mb_width, bound, x, y, k); |
|
|
|
|
pmv_x = pmv[0].x; |
|
|
pmv_y = pmv[0].y; |
|
512 |
|
|
513 |
mv_x = get_mv(bs, fcode); |
mv_x = get_mv(bs, fcode); |
514 |
mv_y = get_mv(bs, fcode); |
mv_y = get_mv(bs, fcode); |
515 |
|
|
516 |
mv_x += pmv_x; |
DPRINTF(DPRINTF_MV,"mv_diff (%i,%i) pred (%i,%i)", mv_x, mv_y, pmv.x, pmv.y); |
|
mv_y += pmv_y; |
|
517 |
|
|
518 |
if (mv_x < low) |
mv_x += pmv.x; |
519 |
{ |
mv_y += pmv.y; |
520 |
|
|
521 |
|
if (mv_x < low) { |
522 |
mv_x += range; |
mv_x += range; |
523 |
} |
} else if (mv_x > high) { |
|
else if (mv_x > high) |
|
|
{ |
|
524 |
mv_x -= range; |
mv_x -= range; |
525 |
} |
} |
526 |
|
|
527 |
if (mv_y < low) |
if (mv_y < low) { |
|
{ |
|
528 |
mv_y += range; |
mv_y += range; |
529 |
} |
} else if (mv_y > high) { |
|
else if (mv_y > high) |
|
|
{ |
|
530 |
mv_y -= range; |
mv_y -= range; |
531 |
} |
} |
532 |
|
|
536 |
} |
} |
537 |
|
|
538 |
|
|
539 |
void decoder_pframe(DECODER * dec, Bitstream * bs, int rounding, int quant, int fcode, int intra_dc_threshold) |
void |
540 |
|
decoder_pframe(DECODER * dec, |
541 |
|
Bitstream * bs, |
542 |
|
int rounding, |
543 |
|
int quant, |
544 |
|
int fcode, |
545 |
|
int intra_dc_threshold) |
546 |
{ |
{ |
|
uint32_t x, y; |
|
547 |
|
|
548 |
image_swap(&dec->cur, &dec->refn); |
uint32_t x, y; |
549 |
|
uint32_t bound; |
550 |
|
int cp_mb, st_mb; |
551 |
|
|
552 |
start_timer(); |
start_timer(); |
553 |
image_setedges(&dec->refn, dec->edged_width, dec->edged_height, dec->width, dec->height); |
image_setedges(&dec->refn[0], dec->edged_width, dec->edged_height, |
554 |
|
dec->width, dec->height); |
555 |
stop_edges_timer(); |
stop_edges_timer(); |
556 |
|
|
557 |
for (y = 0; y < dec->mb_height; y++) |
bound = 0; |
558 |
{ |
|
559 |
for (x = 0; x < dec->mb_width; x++) |
for (y = 0; y < dec->mb_height; y++) { |
560 |
|
cp_mb = st_mb = 0; |
561 |
|
for (x = 0; x < dec->mb_width; x++) { |
562 |
|
MACROBLOCK *mb; |
563 |
|
|
564 |
|
/* skip stuffing */ |
565 |
|
while (BitstreamShowBits(bs, 10) == 1) |
566 |
|
BitstreamSkip(bs, 10); |
567 |
|
|
568 |
|
if (check_resync_marker(bs, fcode - 1)) |
569 |
{ |
{ |
570 |
MACROBLOCK * mb = &dec->mbs[y*dec->mb_width + x]; |
bound = read_video_packet_header(bs, fcode - 1, &quant); |
571 |
|
x = bound % dec->mb_width; |
572 |
|
y = bound / dec->mb_width; |
573 |
|
} |
574 |
|
mb = &dec->mbs[y * dec->mb_width + x]; |
575 |
|
|
576 |
if (!BitstreamGetBit(bs)) // not_coded |
DPRINTF(DPRINTF_MB, "macroblock (%i,%i) %08x", x, y, BitstreamShowBits(bs, 32)); |
577 |
|
|
578 |
|
/*if (!(dec->mb_skip[y*dec->mb_width + x]=BitstreamGetBit(bs))) not_coded */ |
579 |
|
if (!(BitstreamGetBit(bs))) /* not_coded */ |
580 |
{ |
{ |
581 |
uint32_t mcbpc; |
uint32_t mcbpc; |
582 |
uint32_t cbpc; |
uint32_t cbpc; |
585 |
uint32_t cbp; |
uint32_t cbp; |
586 |
uint32_t intra; |
uint32_t intra; |
587 |
|
|
588 |
|
cp_mb++; |
589 |
mcbpc = get_mcbpc_inter(bs); |
mcbpc = get_mcbpc_inter(bs); |
590 |
mb->mode = mcbpc & 7; |
mb->mode = mcbpc & 7; |
591 |
cbpc = (mcbpc >> 4); |
cbpc = (mcbpc >> 4); |
592 |
|
|
593 |
|
DPRINTF(DPRINTF_MB, "mode %i", mb->mode); |
594 |
|
DPRINTF(DPRINTF_MB, "cbpc %i", cbpc); |
595 |
acpred_flag = 0; |
acpred_flag = 0; |
596 |
|
|
597 |
intra = (mb->mode == MODE_INTRA || mb->mode == MODE_INTRA_Q); |
intra = (mb->mode == MODE_INTRA || mb->mode == MODE_INTRA_Q); |
598 |
|
|
599 |
if (intra) |
if (intra) { |
|
{ |
|
600 |
acpred_flag = BitstreamGetBit(bs); |
acpred_flag = BitstreamGetBit(bs); |
601 |
} |
} |
602 |
|
|
|
if (mb->mode == MODE_STUFFING) |
|
|
{ |
|
|
DEBUG("-- STUFFING ?"); |
|
|
continue; |
|
|
} |
|
|
|
|
603 |
cbpy = get_cbpy(bs, intra); |
cbpy = get_cbpy(bs, intra); |
604 |
|
DPRINTF(DPRINTF_MB, "cbpy %i", cbpy); |
605 |
|
|
606 |
cbp = (cbpy << 2) | cbpc; |
cbp = (cbpy << 2) | cbpc; |
607 |
|
|
608 |
if (mb->mode == MODE_INTER_Q || mb->mode == MODE_INTRA_Q) |
if (mb->mode == MODE_INTER_Q || mb->mode == MODE_INTRA_Q) { |
609 |
{ |
int dquant = dquant_table[BitstreamGetBits(bs, 2)]; |
610 |
quant += dquant_table[BitstreamGetBits(bs,2)]; |
DPRINTF(DPRINTF_MB, "dquant %i", dquant); |
611 |
if (quant > 31) |
quant += dquant; |
612 |
{ |
if (quant > 31) { |
613 |
quant = 31; |
quant = 31; |
614 |
} |
} else if (quant < 1) { |
|
else if (mb->quant < 1) |
|
|
{ |
|
615 |
quant = 1; |
quant = 1; |
616 |
} |
} |
617 |
|
DPRINTF(DPRINTF_MB, "quant %i", quant); |
618 |
} |
} |
619 |
mb->quant = quant; |
mb->quant = quant; |
620 |
|
|
621 |
if (mb->mode == MODE_INTER || mb->mode == MODE_INTER_Q) |
if (dec->interlacing) { |
622 |
{ |
if (cbp || intra) { |
623 |
|
mb->field_dct = BitstreamGetBit(bs); |
624 |
get_motion_vector(dec, bs, x, y, 0, &mb->mvs[0], fcode); |
DPRINTF(DPRINTF_DEBUG, "decp: field_dct: %d", mb->field_dct); |
625 |
mb->mvs[1].x = mb->mvs[2].x = mb->mvs[3].x = mb->mvs[0].x; |
} |
626 |
mb->mvs[1].y = mb->mvs[2].y = mb->mvs[3].y = mb->mvs[0].y; |
|
627 |
|
if (mb->mode == MODE_INTER || mb->mode == MODE_INTER_Q) { |
628 |
|
mb->field_pred = BitstreamGetBit(bs); |
629 |
|
DPRINTF(DPRINTF_DEBUG, "decp: field_pred: %d", mb->field_pred); |
630 |
|
|
631 |
|
if (mb->field_pred) { |
632 |
|
mb->field_for_top = BitstreamGetBit(bs); |
633 |
|
DPRINTF(DPRINTF_DEBUG, "decp: field_for_top: %d", mb->field_for_top); |
634 |
|
mb->field_for_bot = BitstreamGetBit(bs); |
635 |
|
DPRINTF(DPRINTF_DEBUG, "decp: field_for_bot: %d", mb->field_for_bot); |
636 |
} |
} |
|
else if (mb->mode == MODE_INTER4V /* || mb->mode == MODE_INTER4V_Q */) |
|
|
{ |
|
|
get_motion_vector(dec, bs, x, y, 0, &mb->mvs[0], fcode); |
|
|
get_motion_vector(dec, bs, x, y, 1, &mb->mvs[1], fcode); |
|
|
get_motion_vector(dec, bs, x, y, 2, &mb->mvs[2], fcode); |
|
|
get_motion_vector(dec, bs, x, y, 3, &mb->mvs[3], fcode); |
|
637 |
} |
} |
|
else // MODE_INTRA, MODE_INTRA_Q |
|
|
{ |
|
|
mb->mvs[0].x = mb->mvs[1].x = mb->mvs[2].x = mb->mvs[3].x = 0; |
|
|
mb->mvs[0].y = mb->mvs[1].y = mb->mvs[2].y = mb->mvs[3].y = 0; |
|
|
decoder_mbintra(dec, mb, x, y, acpred_flag, cbp, bs, quant, intra_dc_threshold); |
|
|
continue; |
|
638 |
} |
} |
639 |
|
|
640 |
decoder_mbinter(dec, mb, x, y, acpred_flag, cbp, bs, quant, rounding); |
if (mb->mode == MODE_INTER || mb->mode == MODE_INTER_Q) { |
641 |
|
if (dec->interlacing && mb->field_pred) { |
642 |
|
get_motion_vector(dec, bs, x, y, 0, &mb->mvs[0], |
643 |
|
fcode, bound); |
644 |
|
get_motion_vector(dec, bs, x, y, 0, &mb->mvs[1], |
645 |
|
fcode, bound); |
646 |
|
} else { |
647 |
|
get_motion_vector(dec, bs, x, y, 0, &mb->mvs[0], |
648 |
|
fcode, bound); |
649 |
|
mb->mvs[1].x = mb->mvs[2].x = mb->mvs[3].x = |
650 |
|
mb->mvs[0].x; |
651 |
|
mb->mvs[1].y = mb->mvs[2].y = mb->mvs[3].y = |
652 |
|
mb->mvs[0].y; |
653 |
|
} |
654 |
|
} else if (mb->mode == MODE_INTER4V ) { |
655 |
|
|
656 |
|
get_motion_vector(dec, bs, x, y, 0, &mb->mvs[0], fcode, bound); |
657 |
|
get_motion_vector(dec, bs, x, y, 1, &mb->mvs[1], fcode, bound); |
658 |
|
get_motion_vector(dec, bs, x, y, 2, &mb->mvs[2], fcode, bound); |
659 |
|
get_motion_vector(dec, bs, x, y, 3, &mb->mvs[3], fcode, bound); |
660 |
|
} else /* MODE_INTRA, MODE_INTRA_Q */ |
661 |
|
{ |
662 |
|
mb->mvs[0].x = mb->mvs[1].x = mb->mvs[2].x = mb->mvs[3].x = |
663 |
|
0; |
664 |
|
mb->mvs[0].y = mb->mvs[1].y = mb->mvs[2].y = mb->mvs[3].y = |
665 |
|
0; |
666 |
|
decoder_mbintra(dec, mb, x, y, acpred_flag, cbp, bs, quant, |
667 |
|
intra_dc_threshold, bound); |
668 |
|
continue; |
669 |
} |
} |
670 |
else // not coded |
|
671 |
|
decoder_mbinter(dec, mb, x, y, acpred_flag, cbp, bs, quant, |
672 |
|
rounding); |
673 |
|
} else /* not coded */ |
674 |
{ |
{ |
675 |
|
DPRINTF(DPRINTF_DEBUG, "P-frame MB at (X,Y)=(%d,%d)", x, y); |
676 |
|
|
677 |
mb->mode = MODE_NOT_CODED; |
mb->mode = MODE_NOT_CODED; |
678 |
mb->mvs[0].x = mb->mvs[1].x = mb->mvs[2].x = mb->mvs[3].x = 0; |
mb->mvs[0].x = mb->mvs[1].x = mb->mvs[2].x = mb->mvs[3].x = 0; |
679 |
mb->mvs[0].y = mb->mvs[1].y = mb->mvs[2].y = mb->mvs[3].y = 0; |
mb->mvs[0].y = mb->mvs[1].y = mb->mvs[2].y = mb->mvs[3].y = 0; |
680 |
|
|
681 |
// copy macroblock directly from ref to cur |
/* copy macroblock directly from ref to cur */ |
682 |
|
|
683 |
start_timer(); |
start_timer(); |
684 |
|
|
685 |
transfer8x8_copy(dec->cur.y + (16*y)*dec->edged_width + (16*x), |
transfer8x8_copy(dec->cur.y + (16 * y) * dec->edged_width + |
686 |
dec->refn.y + (16*y)*dec->edged_width + (16*x), |
(16 * x), |
687 |
dec->edged_width); |
dec->refn[0].y + (16 * y) * dec->edged_width + |
688 |
|
(16 * x), dec->edged_width); |
689 |
transfer8x8_copy(dec->cur.y + (16*y)*dec->edged_width + (16*x+8), |
|
690 |
dec->refn.y + (16*y)*dec->edged_width + (16*x+8), |
transfer8x8_copy(dec->cur.y + (16 * y) * dec->edged_width + |
691 |
dec->edged_width); |
(16 * x + 8), |
692 |
|
dec->refn[0].y + (16 * y) * dec->edged_width + |
693 |
transfer8x8_copy(dec->cur.y + (16*y+8)*dec->edged_width + (16*x), |
(16 * x + 8), dec->edged_width); |
694 |
dec->refn.y + (16*y+8)*dec->edged_width + (16*x), |
|
695 |
dec->edged_width); |
transfer8x8_copy(dec->cur.y + (16 * y + 8) * dec->edged_width + |
696 |
|
(16 * x), |
697 |
transfer8x8_copy(dec->cur.y + (16*y+8)*dec->edged_width + (16*x+8), |
dec->refn[0].y + (16 * y + |
698 |
dec->refn.y + (16*y+8)*dec->edged_width + (16*x+8), |
8) * dec->edged_width + |
699 |
dec->edged_width); |
(16 * x), dec->edged_width); |
700 |
|
|
701 |
transfer8x8_copy(dec->cur.u + (8*y)*dec->edged_width/2 + (8*x), |
transfer8x8_copy(dec->cur.y + (16 * y + 8) * dec->edged_width + |
702 |
dec->refn.u + (8*y)*dec->edged_width/2 + (8*x), |
(16 * x + 8), |
703 |
|
dec->refn[0].y + (16 * y + |
704 |
|
8) * dec->edged_width + |
705 |
|
(16 * x + 8), dec->edged_width); |
706 |
|
|
707 |
|
transfer8x8_copy(dec->cur.u + (8 * y) * dec->edged_width / 2 + |
708 |
|
(8 * x), |
709 |
|
dec->refn[0].u + |
710 |
|
(8 * y) * dec->edged_width / 2 + (8 * x), |
711 |
dec->edged_width/2); |
dec->edged_width/2); |
712 |
|
|
713 |
transfer8x8_copy(dec->cur.v + (8*y)*dec->edged_width/2 + (8*x), |
transfer8x8_copy(dec->cur.v + (8 * y) * dec->edged_width / 2 + |
714 |
dec->refn.v + (8*y)*dec->edged_width/2 + (8*x), |
(8 * x), |
715 |
|
dec->refn[0].v + |
716 |
|
(8 * y) * dec->edged_width / 2 + (8 * x), |
717 |
dec->edged_width/2); |
dec->edged_width/2); |
|
|
|
718 |
stop_transfer_timer(); |
stop_transfer_timer(); |
719 |
|
if(dec->out_frm && cp_mb > 0) { |
720 |
|
output_slice(&dec->cur, dec->edged_width,dec->width,dec->out_frm,st_mb,y,cp_mb); |
721 |
|
cp_mb = 0; |
722 |
|
} |
723 |
|
st_mb = x+1; |
724 |
} |
} |
725 |
} |
} |
726 |
|
if(dec->out_frm && cp_mb > 0) |
727 |
|
output_slice(&dec->cur, dec->edged_width,dec->width,dec->out_frm,st_mb,y,cp_mb); |
728 |
} |
} |
729 |
} |
} |
730 |
|
|
731 |
int decoder_decode(DECODER * dec, XVID_DEC_FRAME * frame) |
/* swap two MACROBLOCK array */ |
732 |
|
void |
733 |
|
mb_swap(MACROBLOCK ** mb1, |
734 |
|
MACROBLOCK ** mb2) |
735 |
{ |
{ |
736 |
|
MACROBLOCK *temp = *mb1; |
737 |
|
|
738 |
|
*mb1 = *mb2; |
739 |
|
*mb2 = temp; |
740 |
|
} |
741 |
|
|
742 |
|
int |
743 |
|
decoder_decode(DECODER * dec, |
744 |
|
XVID_DEC_FRAME * frame) |
745 |
|
{ |
746 |
|
|
747 |
Bitstream bs; |
Bitstream bs; |
748 |
uint32_t rounding; |
uint32_t rounding; |
749 |
uint32_t quant; |
uint32_t quant; |
750 |
uint32_t fcode; |
uint32_t fcode_forward; |
751 |
|
uint32_t fcode_backward; |
752 |
uint32_t intra_dc_threshold; |
uint32_t intra_dc_threshold; |
753 |
|
uint32_t vop_type; |
754 |
|
|
755 |
start_global_timer(); |
start_global_timer(); |
756 |
|
|
757 |
|
dec->out_frm = (frame->colorspace == XVID_CSP_EXTERN) ? frame->image : NULL; |
758 |
|
|
759 |
BitstreamInit(&bs, frame->bitstream, frame->length); |
BitstreamInit(&bs, frame->bitstream, frame->length); |
760 |
|
|
761 |
switch (BitstreamReadHeaders(&bs, dec, &rounding, &quant, &fcode, &intra_dc_threshold)) |
/* add by chenm001 <chenm001@163.com> */ |
762 |
{ |
/* for support B-frame to reference last 2 frame */ |
763 |
|
dec->frames++; |
764 |
|
vop_type = |
765 |
|
BitstreamReadHeaders(&bs, dec, &rounding, &quant, &fcode_forward, |
766 |
|
&fcode_backward, &intra_dc_threshold); |
767 |
|
|
768 |
|
dec->p_bmv.x = dec->p_bmv.y = dec->p_fmv.y = dec->p_fmv.y = 0; /* init pred vector to 0 */ |
769 |
|
|
770 |
|
switch (vop_type) { |
771 |
case P_VOP : |
case P_VOP : |
772 |
decoder_pframe(dec, &bs, rounding, quant, fcode, intra_dc_threshold); |
decoder_pframe(dec, &bs, rounding, quant, fcode_forward, |
773 |
|
intra_dc_threshold); |
774 |
break; |
break; |
775 |
|
|
776 |
case I_VOP : |
case I_VOP : |
|
//DEBUG1("",intra_dc_threshold); |
|
777 |
decoder_iframe(dec, &bs, quant, intra_dc_threshold); |
decoder_iframe(dec, &bs, quant, intra_dc_threshold); |
778 |
break; |
break; |
779 |
|
case B_VOP: |
780 |
case B_VOP : // ignore |
image_copy(&dec->cur, &dec->refn[0], dec->edged_width, dec->height); |
781 |
break; |
break; |
782 |
|
case N_VOP: |
783 |
case N_VOP : // vop not coded |
/* when low_delay==0, N_VOP's should interpolate between the past and future frames */ |
784 |
|
image_copy(&dec->cur, &dec->refn[0], dec->edged_width, dec->height); |
785 |
break; |
break; |
786 |
|
|
787 |
default : |
default : |
790 |
|
|
791 |
frame->length = BitstreamPos(&bs) / 8; |
frame->length = BitstreamPos(&bs) / 8; |
792 |
|
|
|
start_timer(); |
|
793 |
image_output(&dec->cur, dec->width, dec->height, dec->edged_width, |
image_output(&dec->cur, dec->width, dec->height, dec->edged_width, |
794 |
frame->image, frame->stride, frame->colorspace); |
frame->image, frame->stride, frame->colorspace); |
795 |
stop_conv_timer(); |
|
796 |
|
if (vop_type == I_VOP || vop_type == P_VOP) { |
797 |
|
image_swap(&dec->refn[0], &dec->refn[1]); |
798 |
|
image_swap(&dec->cur, &dec->refn[0]); |
799 |
|
|
800 |
|
/* swap MACROBLOCK */ |
801 |
|
/* the Divx will not set the low_delay flage some times */ |
802 |
|
/* so follow code will wrong to not swap at that time */ |
803 |
|
/* this will broken bitstream! so I'm change it, */ |
804 |
|
/* But that is not the best way! can anyone tell me how */ |
805 |
|
/* to do another way? */ |
806 |
|
/* 18-07-2002 MinChen<chenm001@163.com> */ |
807 |
|
/*if (!dec->low_delay && vop_type == P_VOP) */ |
808 |
|
if (vop_type == P_VOP) |
809 |
|
mb_swap(&dec->mbs, &dec->last_mbs); |
810 |
|
} |
811 |
|
|
812 |
emms(); |
emms(); |
813 |
|
|