32 |
* |
* |
33 |
* History: |
* History: |
34 |
* |
* |
35 |
|
* 22.04.2002 add some B-frame decode support; chenm001 <chenm001@163.com> |
36 |
|
* 29.03.2002 interlacing fix - compensated block wasn't being used when |
37 |
|
* reconstructing blocks, thus artifacts |
38 |
|
* interlacing speedup - used transfers to re-interlace |
39 |
|
* interlaced decoding should be as fast as progressive now |
40 |
|
* 26.03.2002 interlacing support - moved transfers outside decode loop |
41 |
* 26.12.2001 decoder_mbinter: dequant/idct moved within if(coded) block |
* 26.12.2001 decoder_mbinter: dequant/idct moved within if(coded) block |
42 |
* 22.12.2001 block based interpolation |
* 22.12.2001 block based interpolation |
43 |
* 01.12.2001 inital version; (c)2001 peter ross <pross@cs.rmit.edu.au> |
* 01.12.2001 inital version; (c)2001 peter ross <pross@cs.rmit.edu.au> |
74 |
{ |
{ |
75 |
DECODER * dec; |
DECODER * dec; |
76 |
|
|
77 |
dec = xvid_malloc(sizeof(DECODER), 16); |
dec = xvid_malloc(sizeof(DECODER), CACHE_LINE); |
78 |
if (dec == NULL) |
if (dec == NULL) |
79 |
{ |
{ |
80 |
return XVID_ERR_MEMORY; |
return XVID_ERR_MEMORY; |
96 |
return XVID_ERR_MEMORY; |
return XVID_ERR_MEMORY; |
97 |
} |
} |
98 |
|
|
99 |
if (image_create(&dec->refn, dec->edged_width, dec->edged_height)) |
if (image_create(&dec->refn[0], dec->edged_width, dec->edged_height)) |
100 |
{ |
{ |
101 |
image_destroy(&dec->cur, dec->edged_width, dec->edged_height); |
image_destroy(&dec->cur, dec->edged_width, dec->edged_height); |
102 |
xvid_free(dec); |
xvid_free(dec); |
103 |
return XVID_ERR_MEMORY; |
return XVID_ERR_MEMORY; |
104 |
} |
} |
105 |
|
// add by chenm001 <chenm001@163.com> |
106 |
|
// for support B-frame to reference last 2 frame |
107 |
|
if (image_create(&dec->refn[1], dec->edged_width, dec->edged_height)) |
108 |
|
{ |
109 |
|
image_destroy(&dec->cur, dec->edged_width, dec->edged_height); |
110 |
|
image_destroy(&dec->refn[0], dec->edged_width, dec->edged_height); |
111 |
|
xvid_free(dec); |
112 |
|
return XVID_ERR_MEMORY; |
113 |
|
} |
114 |
|
|
115 |
dec->mbs = xvid_malloc(sizeof(MACROBLOCK) * dec->mb_width * dec->mb_height, 16); |
dec->mbs = xvid_malloc(sizeof(MACROBLOCK) * dec->mb_width * dec->mb_height, CACHE_LINE); |
116 |
if (dec->mbs == NULL) |
if (dec->mbs == NULL) |
117 |
{ |
{ |
118 |
image_destroy(&dec->cur, dec->edged_width, dec->edged_height); |
image_destroy(&dec->cur, dec->edged_width, dec->edged_height); |
121 |
} |
} |
122 |
|
|
123 |
init_timer(); |
init_timer(); |
|
create_vlc_tables(); |
|
124 |
|
|
125 |
return XVID_ERR_OK; |
return XVID_ERR_OK; |
126 |
} |
} |
129 |
int decoder_destroy(DECODER * dec) |
int decoder_destroy(DECODER * dec) |
130 |
{ |
{ |
131 |
xvid_free(dec->mbs); |
xvid_free(dec->mbs); |
132 |
image_destroy(&dec->refn, dec->edged_width, dec->edged_height); |
image_destroy(&dec->refn[0], dec->edged_width, dec->edged_height); |
133 |
image_destroy(&dec->cur, dec->edged_width, dec->edged_height); |
image_destroy(&dec->cur, dec->edged_width, dec->edged_height); |
134 |
xvid_free(dec); |
xvid_free(dec); |
135 |
|
|
|
destroy_vlc_tables(); |
|
|
|
|
136 |
write_timer(); |
write_timer(); |
137 |
return XVID_ERR_OK; |
return XVID_ERR_OK; |
138 |
} |
} |
147 |
|
|
148 |
// decode an intra macroblock |
// decode an intra macroblock |
149 |
|
|
150 |
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) |
void decoder_mbintra(DECODER * dec, |
151 |
{ |
MACROBLOCK * pMB, |
152 |
uint32_t k; |
const uint32_t x_pos, |
153 |
|
const uint32_t y_pos, |
154 |
|
const uint32_t acpred_flag, |
155 |
|
const uint32_t cbp, |
156 |
|
Bitstream * bs, |
157 |
|
const uint32_t quant, |
158 |
|
const uint32_t intra_dc_threshold) |
159 |
|
{ |
160 |
|
|
161 |
|
DECLARE_ALIGNED_MATRIX(block, 6, 64, int16_t, CACHE_LINE); |
162 |
|
DECLARE_ALIGNED_MATRIX(data, 6, 64, int16_t, CACHE_LINE); |
163 |
|
|
164 |
|
uint32_t stride = dec->edged_width; |
165 |
|
uint32_t stride2 = stride / 2; |
166 |
|
uint32_t next_block = stride * 8; |
167 |
|
uint32_t i; |
168 |
|
uint32_t iQuant = pMB->quant; |
169 |
|
uint8_t *pY_Cur, *pU_Cur, *pV_Cur; |
170 |
|
|
171 |
|
pY_Cur = dec->cur.y + (y_pos << 4) * stride + (x_pos << 4); |
172 |
|
pU_Cur = dec->cur.u + (y_pos << 3) * stride2 + (x_pos << 3); |
173 |
|
pV_Cur = dec->cur.v + (y_pos << 3) * stride2 + (x_pos << 3); |
174 |
|
|
175 |
|
memset(block, 0, 6*64*sizeof(int16_t)); // clear |
176 |
|
|
177 |
for (k = 0; k < 6; k++) |
for (i = 0; i < 6; i++) |
178 |
{ |
{ |
179 |
uint32_t dcscalar; |
uint32_t iDcScaler = get_dc_scaler(iQuant, i < 4); |
|
int16_t block[64]; |
|
|
int16_t data[64]; |
|
180 |
int16_t predictors[8]; |
int16_t predictors[8]; |
181 |
int start_coeff; |
int start_coeff; |
182 |
|
|
|
dcscalar = get_dc_scaler(mb->quant, k < 4); |
|
|
|
|
183 |
start_timer(); |
start_timer(); |
184 |
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], iQuant, iDcScaler, predictors); |
185 |
if (!acpred_flag) |
if (!acpred_flag) |
186 |
{ |
{ |
187 |
mb->acpred_directions[k] = 0; |
pMB->acpred_directions[i] = 0; |
188 |
} |
} |
189 |
stop_prediction_timer(); |
stop_prediction_timer(); |
190 |
|
|
|
memset(block, 0, 64*sizeof(int16_t)); // clear |
|
|
|
|
191 |
if (quant < intra_dc_threshold) |
if (quant < intra_dc_threshold) |
192 |
{ |
{ |
193 |
int dc_size; |
int dc_size; |
194 |
int dc_dif; |
int dc_dif; |
195 |
|
|
196 |
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); |
197 |
dc_dif = dc_size ? get_dc_dif(bs, dc_size) : 0 ; |
dc_dif = dc_size ? get_dc_dif(bs, dc_size) : 0 ; |
198 |
|
|
199 |
if (dc_size > 8) |
if (dc_size > 8) |
201 |
BitstreamSkip(bs, 1); // marker |
BitstreamSkip(bs, 1); // marker |
202 |
} |
} |
203 |
|
|
204 |
block[0] = dc_dif; |
block[i*64 + 0] = dc_dif; |
205 |
start_coeff = 1; |
start_coeff = 1; |
206 |
} |
} |
207 |
else |
else |
210 |
} |
} |
211 |
|
|
212 |
start_timer(); |
start_timer(); |
213 |
if (cbp & (1 << (5-k))) // coded |
if (cbp & (1 << (5-i))) // coded |
214 |
{ |
{ |
215 |
get_intra_block(bs, block, mb->acpred_directions[k], start_coeff); |
get_intra_block(bs, &block[i*64], pMB->acpred_directions[i], start_coeff); |
216 |
} |
} |
217 |
stop_coding_timer(); |
stop_coding_timer(); |
218 |
|
|
219 |
start_timer(); |
start_timer(); |
220 |
add_acdc(mb, k, block, dcscalar, predictors); |
add_acdc(pMB, i, &block[i*64], iDcScaler, predictors); |
221 |
stop_prediction_timer(); |
stop_prediction_timer(); |
222 |
|
|
223 |
start_timer(); |
start_timer(); |
224 |
if (dec->quant_type == 0) |
if (dec->quant_type == 0) |
225 |
{ |
{ |
226 |
dequant_intra(data, block, mb->quant, dcscalar); |
dequant_intra(&data[i*64], &block[i*64], iQuant, iDcScaler); |
227 |
} |
} |
228 |
else |
else |
229 |
{ |
{ |
230 |
dequant4_intra(data, block, mb->quant, dcscalar); |
dequant4_intra(&data[i*64], &block[i*64], iQuant, iDcScaler); |
231 |
} |
} |
232 |
stop_iquant_timer(); |
stop_iquant_timer(); |
233 |
|
|
234 |
start_timer(); |
start_timer(); |
235 |
idct(data); |
idct(&data[i*64]); |
236 |
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); |
|
|
} |
|
|
else if (k == 4) |
|
|
{ |
|
|
transfer_16to8copy(dec->cur.u+ 8*y*(dec->edged_width/2) + 8*x, data, (dec->edged_width/2)); |
|
237 |
} |
} |
238 |
else // if (k == 5) |
|
239 |
|
if (dec->interlacing && pMB->field_dct) |
240 |
{ |
{ |
241 |
transfer_16to8copy(dec->cur.v + 8*y*(dec->edged_width/2) + 8*x, data, (dec->edged_width/2)); |
next_block = stride; |
242 |
|
stride *= 2; |
243 |
} |
} |
244 |
|
|
245 |
|
start_timer(); |
246 |
|
transfer_16to8copy(pY_Cur, &data[0*64], stride); |
247 |
|
transfer_16to8copy(pY_Cur + 8, &data[1*64], stride); |
248 |
|
transfer_16to8copy(pY_Cur + next_block, &data[2*64], stride); |
249 |
|
transfer_16to8copy(pY_Cur + 8 + next_block, &data[3*64], stride); |
250 |
|
transfer_16to8copy(pU_Cur, &data[4*64], stride2); |
251 |
|
transfer_16to8copy(pV_Cur, &data[5*64], stride2); |
252 |
stop_transfer_timer(); |
stop_transfer_timer(); |
253 |
} |
} |
|
} |
|
254 |
|
|
255 |
|
|
256 |
|
|
264 |
|
|
265 |
// decode an inter macroblock |
// decode an inter macroblock |
266 |
|
|
267 |
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 decoder_mbinter(DECODER * dec, |
268 |
{ |
const MACROBLOCK * pMB, |
269 |
const uint32_t stride = dec->edged_width; |
const uint32_t x_pos, |
270 |
const uint32_t stride2 = dec->edged_width / 2; |
const uint32_t y_pos, |
271 |
|
const uint32_t acpred_flag, |
272 |
|
const uint32_t cbp, |
273 |
|
Bitstream * bs, |
274 |
|
const uint32_t quant, |
275 |
|
const uint32_t rounding) |
276 |
|
{ |
277 |
|
|
278 |
|
DECLARE_ALIGNED_MATRIX(block,6, 64, int16_t, CACHE_LINE); |
279 |
|
DECLARE_ALIGNED_MATRIX(data, 6, 64, int16_t, CACHE_LINE); |
280 |
|
|
281 |
|
uint32_t stride = dec->edged_width; |
282 |
|
uint32_t stride2 = stride / 2; |
283 |
|
uint32_t next_block = stride * 8; |
284 |
|
uint32_t i; |
285 |
|
uint32_t iQuant = pMB->quant; |
286 |
|
uint8_t *pY_Cur, *pU_Cur, *pV_Cur; |
287 |
int uv_dx, uv_dy; |
int uv_dx, uv_dy; |
|
uint32_t k; |
|
288 |
|
|
289 |
if (mb->mode == MODE_INTER || mb->mode == MODE_INTER_Q) |
pY_Cur = dec->cur.y + (y_pos << 4) * stride + (x_pos << 4); |
290 |
|
pU_Cur = dec->cur.u + (y_pos << 3) * stride2 + (x_pos << 3); |
291 |
|
pV_Cur = dec->cur.v + (y_pos << 3) * stride2 + (x_pos << 3); |
292 |
|
|
293 |
|
if (pMB->mode == MODE_INTER || pMB->mode == MODE_INTER_Q) |
294 |
{ |
{ |
295 |
uv_dx = mb->mvs[0].x; |
uv_dx = pMB->mvs[0].x; |
296 |
uv_dy = mb->mvs[0].y; |
uv_dy = pMB->mvs[0].y; |
297 |
|
|
298 |
uv_dx = (uv_dx & 3) ? (uv_dx >> 1) | 1 : uv_dx / 2; |
uv_dx = (uv_dx & 3) ? (uv_dx >> 1) | 1 : uv_dx / 2; |
299 |
uv_dy = (uv_dy & 3) ? (uv_dy >> 1) | 1 : uv_dy / 2; |
uv_dy = (uv_dy & 3) ? (uv_dy >> 1) | 1 : uv_dy / 2; |
301 |
else |
else |
302 |
{ |
{ |
303 |
int sum; |
int sum; |
304 |
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; |
305 |
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) ); |
306 |
|
|
307 |
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; |
308 |
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) ); |
309 |
} |
} |
310 |
|
|
311 |
start_timer(); |
start_timer(); |
312 |
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 , pMB->mvs[0].x, pMB->mvs[0].y, stride, rounding); |
313 |
interpolate8x8_switch(dec->cur.y, dec->refn.y, 16*x + 8, 16*y , mb->mvs[1].x, mb->mvs[1].y, stride, rounding); |
interpolate8x8_switch(dec->cur.y, dec->refn[0].y, 16*x_pos + 8, 16*y_pos , pMB->mvs[1].x, pMB->mvs[1].y, stride, rounding); |
314 |
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, 16*y_pos + 8, pMB->mvs[2].x, pMB->mvs[2].y, stride, rounding); |
315 |
interpolate8x8_switch(dec->cur.y, dec->refn.y, 16*x + 8, 16*y + 8, mb->mvs[3].x, mb->mvs[3].y, stride, rounding); |
interpolate8x8_switch(dec->cur.y, dec->refn[0].y, 16*x_pos + 8, 16*y_pos + 8, pMB->mvs[3].x, pMB->mvs[3].y, stride, rounding); |
316 |
interpolate8x8_switch(dec->cur.u, dec->refn.u, 8*x, 8*y, uv_dx, uv_dy, stride2, rounding); |
interpolate8x8_switch(dec->cur.u, dec->refn[0].u, 8*x_pos, 8*y_pos, uv_dx, uv_dy, stride2, rounding); |
317 |
interpolate8x8_switch(dec->cur.v, dec->refn.v, 8*x, 8*y, uv_dx, uv_dy, stride2, rounding); |
interpolate8x8_switch(dec->cur.v, dec->refn[0].v, 8*x_pos, 8*y_pos, uv_dx, uv_dy, stride2, rounding); |
318 |
stop_comp_timer(); |
stop_comp_timer(); |
319 |
|
|
320 |
|
for (i = 0; i < 6; i++) |
|
for (k = 0; k < 6; k++) |
|
321 |
{ |
{ |
322 |
int16_t block[64]; |
if (cbp & (1 << (5-i))) // coded |
|
int16_t data[64]; |
|
|
|
|
|
if (cbp & (1 << (5-k))) // coded |
|
323 |
{ |
{ |
324 |
memset(block, 0, 64 * sizeof(int16_t)); // clear |
memset(&block[i*64], 0, 64 * sizeof(int16_t)); // clear |
325 |
|
|
326 |
start_timer(); |
start_timer(); |
327 |
get_inter_block(bs, block); |
get_inter_block(bs, &block[i*64]); |
328 |
stop_coding_timer(); |
stop_coding_timer(); |
329 |
|
|
330 |
start_timer(); |
start_timer(); |
331 |
if (dec->quant_type == 0) |
if (dec->quant_type == 0) |
332 |
{ |
{ |
333 |
dequant_inter(data, block, mb->quant); |
dequant_inter(&data[i*64], &block[i*64], iQuant); |
334 |
} |
} |
335 |
else |
else |
336 |
{ |
{ |
337 |
dequant4_inter(data, block, mb->quant); |
dequant4_inter(&data[i*64], &block[i*64], iQuant); |
338 |
} |
} |
339 |
stop_iquant_timer(); |
stop_iquant_timer(); |
340 |
|
|
341 |
start_timer(); |
start_timer(); |
342 |
idct(data); |
idct(&data[i*64]); |
343 |
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); |
|
344 |
} |
} |
|
else if (k == 4) |
|
|
{ |
|
|
transfer_16to8add(dec->cur.u + 8*y*stride2 + 8*x, data, stride2); |
|
345 |
} |
} |
346 |
else // k == 5 |
|
347 |
|
if (dec->interlacing && pMB->field_dct) |
348 |
{ |
{ |
349 |
transfer_16to8add(dec->cur.v + 8*y*stride2 + 8*x, data, stride2); |
next_block = stride; |
350 |
|
stride *= 2; |
351 |
} |
} |
352 |
|
|
353 |
|
start_timer(); |
354 |
|
if (cbp & 32) |
355 |
|
transfer_16to8add(pY_Cur, &data[0*64], stride); |
356 |
|
if (cbp & 16) |
357 |
|
transfer_16to8add(pY_Cur + 8, &data[1*64], stride); |
358 |
|
if (cbp & 8) |
359 |
|
transfer_16to8add(pY_Cur + next_block, &data[2*64], stride); |
360 |
|
if (cbp & 4) |
361 |
|
transfer_16to8add(pY_Cur + 8 + next_block, &data[3*64], stride); |
362 |
|
if (cbp & 2) |
363 |
|
transfer_16to8add(pU_Cur, &data[4*64], stride2); |
364 |
|
if (cbp & 1) |
365 |
|
transfer_16to8add(pV_Cur, &data[5*64], stride2); |
366 |
stop_transfer_timer(); |
stop_transfer_timer(); |
367 |
} |
} |
|
} |
|
|
} |
|
|
|
|
368 |
|
|
369 |
|
|
370 |
void decoder_iframe(DECODER * dec, Bitstream * bs, int quant, int intra_dc_threshold) |
void decoder_iframe(DECODER * dec, Bitstream * bs, int quant, int intra_dc_threshold) |
371 |
{ |
{ |
372 |
|
|
373 |
uint32_t x, y; |
uint32_t x, y; |
374 |
|
|
375 |
for (y = 0; y < dec->mb_height; y++) |
for (y = 0; y < dec->mb_height; y++) |
413 |
} |
} |
414 |
mb->quant = quant; |
mb->quant = quant; |
415 |
|
|
416 |
|
if (dec->interlacing) |
417 |
|
{ |
418 |
|
mb->field_dct = BitstreamGetBit(bs); |
419 |
|
DEBUG1("deci: field_dct: ", mb->field_dct); |
420 |
|
} |
421 |
|
|
422 |
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, intra_dc_threshold); |
423 |
} |
} |
424 |
} |
} |
425 |
|
|
426 |
} |
} |
427 |
|
|
428 |
|
|
429 |
void get_motion_vector(DECODER *dec, Bitstream *bs, int x, int y, int k, VECTOR * mv, int fcode) |
void get_motion_vector(DECODER *dec, Bitstream *bs, int x, int y, int k, VECTOR * mv, int fcode) |
430 |
{ |
{ |
431 |
|
|
432 |
int scale_fac = 1 << (fcode - 1); |
int scale_fac = 1 << (fcode - 1); |
433 |
int high = (32 * scale_fac) - 1; |
int high = (32 * scale_fac) - 1; |
434 |
int low = ((-32) * scale_fac); |
int low = ((-32) * scale_fac); |
478 |
|
|
479 |
void decoder_pframe(DECODER * dec, Bitstream * bs, int rounding, int quant, int fcode, int intra_dc_threshold) |
void decoder_pframe(DECODER * dec, Bitstream * bs, int rounding, int quant, int fcode, int intra_dc_threshold) |
480 |
{ |
{ |
|
uint32_t x, y; |
|
481 |
|
|
482 |
image_swap(&dec->cur, &dec->refn); |
uint32_t x, y; |
483 |
|
|
484 |
start_timer(); |
start_timer(); |
485 |
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, dec->width, dec->height, dec->interlacing); |
486 |
stop_edges_timer(); |
stop_edges_timer(); |
487 |
|
|
488 |
for (y = 0; y < dec->mb_height; y++) |
for (y = 0; y < dec->mb_height; y++) |
535 |
} |
} |
536 |
mb->quant = quant; |
mb->quant = quant; |
537 |
|
|
538 |
|
if (dec->interlacing) |
539 |
|
{ |
540 |
|
mb->field_dct = BitstreamGetBit(bs); |
541 |
|
DEBUG1("decp: field_dct: ", mb->field_dct); |
542 |
|
|
543 |
if (mb->mode == MODE_INTER || mb->mode == MODE_INTER_Q) |
if (mb->mode == MODE_INTER || mb->mode == MODE_INTER_Q) |
544 |
{ |
{ |
545 |
|
mb->field_pred = BitstreamGetBit(bs); |
546 |
|
DEBUG1("decp: field_pred: ", mb->field_pred); |
547 |
|
|
548 |
|
if (mb->field_pred) |
549 |
|
{ |
550 |
|
mb->field_for_top = BitstreamGetBit(bs); |
551 |
|
DEBUG1("decp: field_for_top: ", mb->field_for_top); |
552 |
|
mb->field_for_bot = BitstreamGetBit(bs); |
553 |
|
DEBUG1("decp: field_for_bot: ", mb->field_for_bot); |
554 |
|
} |
555 |
|
} |
556 |
|
} |
557 |
|
|
558 |
|
if (mb->mode == MODE_INTER || mb->mode == MODE_INTER_Q) |
559 |
|
{ |
560 |
|
if (dec->interlacing && mb->field_pred) |
561 |
|
{ |
562 |
|
get_motion_vector(dec, bs, x, y, 0, &mb->mvs[0], fcode); |
563 |
|
get_motion_vector(dec, bs, x, y, 0, &mb->mvs[1], fcode); |
564 |
|
} |
565 |
|
else |
566 |
|
{ |
567 |
get_motion_vector(dec, bs, x, y, 0, &mb->mvs[0], fcode); |
get_motion_vector(dec, bs, x, y, 0, &mb->mvs[0], fcode); |
568 |
mb->mvs[1].x = mb->mvs[2].x = mb->mvs[3].x = mb->mvs[0].x; |
mb->mvs[1].x = mb->mvs[2].x = mb->mvs[3].x = mb->mvs[0].x; |
569 |
mb->mvs[1].y = mb->mvs[2].y = mb->mvs[3].y = mb->mvs[0].y; |
mb->mvs[1].y = mb->mvs[2].y = mb->mvs[3].y = mb->mvs[0].y; |
570 |
} |
} |
571 |
|
} |
572 |
else if (mb->mode == MODE_INTER4V /* || mb->mode == MODE_INTER4V_Q */) |
else if (mb->mode == MODE_INTER4V /* || mb->mode == MODE_INTER4V_Q */) |
573 |
{ |
{ |
574 |
get_motion_vector(dec, bs, x, y, 0, &mb->mvs[0], fcode); |
get_motion_vector(dec, bs, x, y, 0, &mb->mvs[0], fcode); |
598 |
start_timer(); |
start_timer(); |
599 |
|
|
600 |
transfer8x8_copy(dec->cur.y + (16*y)*dec->edged_width + (16*x), |
transfer8x8_copy(dec->cur.y + (16*y)*dec->edged_width + (16*x), |
601 |
dec->refn.y + (16*y)*dec->edged_width + (16*x), |
dec->refn[0].y + (16*y)*dec->edged_width + (16*x), |
602 |
dec->edged_width); |
dec->edged_width); |
603 |
|
|
604 |
transfer8x8_copy(dec->cur.y + (16*y)*dec->edged_width + (16*x+8), |
transfer8x8_copy(dec->cur.y + (16*y)*dec->edged_width + (16*x+8), |
605 |
dec->refn.y + (16*y)*dec->edged_width + (16*x+8), |
dec->refn[0].y + (16*y)*dec->edged_width + (16*x+8), |
606 |
dec->edged_width); |
dec->edged_width); |
607 |
|
|
608 |
transfer8x8_copy(dec->cur.y + (16*y+8)*dec->edged_width + (16*x), |
transfer8x8_copy(dec->cur.y + (16*y+8)*dec->edged_width + (16*x), |
609 |
dec->refn.y + (16*y+8)*dec->edged_width + (16*x), |
dec->refn[0].y + (16*y+8)*dec->edged_width + (16*x), |
610 |
dec->edged_width); |
dec->edged_width); |
611 |
|
|
612 |
transfer8x8_copy(dec->cur.y + (16*y+8)*dec->edged_width + (16*x+8), |
transfer8x8_copy(dec->cur.y + (16*y+8)*dec->edged_width + (16*x+8), |
613 |
dec->refn.y + (16*y+8)*dec->edged_width + (16*x+8), |
dec->refn[0].y + (16*y+8)*dec->edged_width + (16*x+8), |
614 |
dec->edged_width); |
dec->edged_width); |
615 |
|
|
616 |
transfer8x8_copy(dec->cur.u + (8*y)*dec->edged_width/2 + (8*x), |
transfer8x8_copy(dec->cur.u + (8*y)*dec->edged_width/2 + (8*x), |
617 |
dec->refn.u + (8*y)*dec->edged_width/2 + (8*x), |
dec->refn[0].u + (8*y)*dec->edged_width/2 + (8*x), |
618 |
dec->edged_width/2); |
dec->edged_width/2); |
619 |
|
|
620 |
transfer8x8_copy(dec->cur.v + (8*y)*dec->edged_width/2 + (8*x), |
transfer8x8_copy(dec->cur.v + (8*y)*dec->edged_width/2 + (8*x), |
621 |
dec->refn.v + (8*y)*dec->edged_width/2 + (8*x), |
dec->refn[0].v + (8*y)*dec->edged_width/2 + (8*x), |
622 |
dec->edged_width/2); |
dec->edged_width/2); |
623 |
|
|
624 |
stop_transfer_timer(); |
stop_transfer_timer(); |
629 |
|
|
630 |
int decoder_decode(DECODER * dec, XVID_DEC_FRAME * frame) |
int decoder_decode(DECODER * dec, XVID_DEC_FRAME * frame) |
631 |
{ |
{ |
632 |
|
|
633 |
Bitstream bs; |
Bitstream bs; |
634 |
uint32_t rounding; |
uint32_t rounding; |
635 |
uint32_t quant; |
uint32_t quant; |
636 |
uint32_t fcode; |
uint32_t fcode; |
637 |
uint32_t intra_dc_threshold; |
uint32_t intra_dc_threshold; |
638 |
|
uint32_t vop_type; |
639 |
|
|
640 |
start_global_timer(); |
start_global_timer(); |
641 |
|
|
642 |
BitstreamInit(&bs, frame->bitstream, frame->length); |
BitstreamInit(&bs, frame->bitstream, frame->length); |
643 |
|
|
644 |
switch (BitstreamReadHeaders(&bs, dec, &rounding, &quant, &fcode, &intra_dc_threshold)) |
// add by chenm001 <chenm001@163.com> |
645 |
|
// for support B-frame to reference last 2 frame |
646 |
|
vop_type=BitstreamReadHeaders(&bs, dec, &rounding, &quant, &fcode, &intra_dc_threshold); |
647 |
|
|
648 |
|
if (vop_type==I_VOP || vop_type==P_VOP){ |
649 |
|
image_swap(&dec->refn[0], &dec->refn[1]); |
650 |
|
image_swap(&dec->cur, &dec->refn[0]); |
651 |
|
} |
652 |
|
|
653 |
|
switch (vop_type) |
654 |
{ |
{ |
655 |
case P_VOP : |
case P_VOP : |
656 |
decoder_pframe(dec, &bs, rounding, quant, fcode, intra_dc_threshold); |
decoder_pframe(dec, &bs, rounding, quant, fcode, intra_dc_threshold); |
683 |
stop_global_timer(); |
stop_global_timer(); |
684 |
|
|
685 |
return XVID_ERR_OK; |
return XVID_ERR_OK; |
686 |
|
|
687 |
} |
} |