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 |
* This program is an implementation of a part of one or more MPEG-4 |
7 |
* Video tools as specified in ISO/IEC 14496-2 standard. Those intending |
* Video tools as specified in ISO/IEC 14496-2 standard. Those intending |
12 |
* editors and their companies, will have no liability for use of this |
* editors and their companies, will have no liability for use of this |
13 |
* software or modifications or derivatives thereof. |
* software or modifications or derivatives thereof. |
14 |
* |
* |
15 |
* This program is xvid_free software; you can redistribute it and/or modify |
* This program is free software; you can redistribute it and/or modify |
16 |
* it under the terms of the GNU General Public License as published by |
* it under the terms of the GNU General Public License as published by |
17 |
* the xvid_free Software Foundation; either version 2 of the License, or |
* the Free Software Foundation; either version 2 of the License, or |
18 |
* (at your option) any later version. |
* (at your option) any later version. |
19 |
* |
* |
20 |
* This program is distributed in the hope that it will be useful, |
* This program is distributed in the hope that it will be useful, |
23 |
* GNU General Public License for more details. |
* GNU General Public License for more details. |
24 |
* |
* |
25 |
* You should have received a copy of the GNU General Public License |
* You should have received a copy of the GNU General Public License |
26 |
* along with this program; if not, write to the xvid_free Software |
* along with this program; if not, write to the Free Software |
27 |
* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. |
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
28 |
* |
* |
29 |
*************************************************************************/ |
*************************************************************************/ |
30 |
|
|
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 |
* 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 lock 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> |
44 |
* |
* |
45 |
|
* $Id$ |
46 |
|
* |
47 |
*************************************************************************/ |
*************************************************************************/ |
48 |
|
|
49 |
#include <stdlib.h> |
#include <stdlib.h> |
50 |
#include <string.h> // memset |
#include <string.h> |
51 |
|
|
52 |
#include "xvid.h" |
#include "xvid.h" |
53 |
#include "portab.h" |
#include "portab.h" |
62 |
#include "dct/fdct.h" |
#include "dct/fdct.h" |
63 |
#include "utils/mem_transfer.h" |
#include "utils/mem_transfer.h" |
64 |
#include "image/interpolate8x8.h" |
#include "image/interpolate8x8.h" |
|
#include "utils/mbfunctions.h" |
|
65 |
|
|
66 |
#include "bitstream/mbcoding.h" |
#include "bitstream/mbcoding.h" |
67 |
#include "prediction/mbprediction.h" |
#include "prediction/mbprediction.h" |
98 |
return XVID_ERR_MEMORY; |
return XVID_ERR_MEMORY; |
99 |
} |
} |
100 |
|
|
101 |
if (image_create(&dec->refn, dec->edged_width, dec->edged_height)) |
if (image_create(&dec->refn[0], dec->edged_width, dec->edged_height)) |
102 |
{ |
{ |
103 |
image_destroy(&dec->cur, dec->edged_width, dec->edged_height); |
image_destroy(&dec->cur, dec->edged_width, dec->edged_height); |
104 |
xvid_free(dec); |
xvid_free(dec); |
105 |
return XVID_ERR_MEMORY; |
return XVID_ERR_MEMORY; |
106 |
} |
} |
107 |
|
// add by chenm001 <chenm001@163.com> |
108 |
|
// for support B-frame to reference last 2 frame |
109 |
|
if (image_create(&dec->refn[1], dec->edged_width, dec->edged_height)) |
110 |
|
{ |
111 |
|
image_destroy(&dec->cur, dec->edged_width, dec->edged_height); |
112 |
|
image_destroy(&dec->refn[0], dec->edged_width, dec->edged_height); |
113 |
|
xvid_free(dec); |
114 |
|
return XVID_ERR_MEMORY; |
115 |
|
} |
116 |
|
|
117 |
dec->mbs = xvid_malloc(sizeof(MACROBLOCK) * dec->mb_width * dec->mb_height, CACHE_LINE); |
dec->mbs = xvid_malloc(sizeof(MACROBLOCK) * dec->mb_width * dec->mb_height, CACHE_LINE); |
118 |
if (dec->mbs == NULL) |
if (dec->mbs == NULL) |
123 |
} |
} |
124 |
|
|
125 |
init_timer(); |
init_timer(); |
|
create_vlc_tables(); |
|
126 |
|
|
127 |
return XVID_ERR_OK; |
return XVID_ERR_OK; |
128 |
} |
} |
131 |
int decoder_destroy(DECODER * dec) |
int decoder_destroy(DECODER * dec) |
132 |
{ |
{ |
133 |
xvid_free(dec->mbs); |
xvid_free(dec->mbs); |
134 |
image_destroy(&dec->refn, dec->edged_width, dec->edged_height); |
image_destroy(&dec->refn[0], dec->edged_width, dec->edged_height); |
135 |
image_destroy(&dec->cur, dec->edged_width, dec->edged_height); |
image_destroy(&dec->cur, dec->edged_width, dec->edged_height); |
136 |
xvid_free(dec); |
xvid_free(dec); |
137 |
|
|
|
destroy_vlc_tables(); |
|
|
|
|
138 |
write_timer(); |
write_timer(); |
139 |
return XVID_ERR_OK; |
return XVID_ERR_OK; |
140 |
} |
} |
163 |
DECLARE_ALIGNED_MATRIX(block, 6, 64, int16_t, CACHE_LINE); |
DECLARE_ALIGNED_MATRIX(block, 6, 64, int16_t, CACHE_LINE); |
164 |
DECLARE_ALIGNED_MATRIX(data, 6, 64, int16_t, CACHE_LINE); |
DECLARE_ALIGNED_MATRIX(data, 6, 64, int16_t, CACHE_LINE); |
165 |
|
|
166 |
const uint32_t stride = dec->edged_width; |
uint32_t stride = dec->edged_width; |
167 |
|
uint32_t stride2 = stride / 2; |
168 |
|
uint32_t next_block = stride * 8; |
169 |
uint32_t i; |
uint32_t i; |
170 |
uint32_t iQuant = pMB->quant; |
uint32_t iQuant = pMB->quant; |
171 |
uint8_t *pY_Cur, *pU_Cur, *pV_Cur; |
uint8_t *pY_Cur, *pU_Cur, *pV_Cur; |
172 |
|
|
173 |
pY_Cur = dec->cur.y + (y_pos << 4) * stride + (x_pos << 4); |
pY_Cur = dec->cur.y + (y_pos << 4) * stride + (x_pos << 4); |
174 |
pU_Cur = dec->cur.u + (y_pos << 3) * (stride >> 1) + (x_pos << 3); |
pU_Cur = dec->cur.u + (y_pos << 3) * stride2 + (x_pos << 3); |
175 |
pV_Cur = dec->cur.v + (y_pos << 3) * (stride >> 1) + (x_pos << 3); |
pV_Cur = dec->cur.v + (y_pos << 3) * stride2 + (x_pos << 3); |
176 |
|
|
177 |
memset(block, 0, 6*64*sizeof(int16_t)); // clear |
memset(block, 0, 6*64*sizeof(int16_t)); // clear |
178 |
|
|
238 |
stop_idct_timer(); |
stop_idct_timer(); |
239 |
} |
} |
240 |
|
|
|
start_timer(); |
|
241 |
if (dec->interlacing && pMB->field_dct) |
if (dec->interlacing && pMB->field_dct) |
242 |
{ |
{ |
243 |
MBFieldToFrame(data); |
next_block = stride; |
244 |
|
stride *= 2; |
245 |
} |
} |
|
stop_interlacing_timer(); |
|
246 |
|
|
247 |
start_timer(); |
start_timer(); |
248 |
transfer_16to8copy(pY_Cur, &data[0*64], stride); |
transfer_16to8copy(pY_Cur, &data[0*64], stride); |
249 |
transfer_16to8copy(pY_Cur + 8, &data[1*64], stride); |
transfer_16to8copy(pY_Cur + 8, &data[1*64], stride); |
250 |
transfer_16to8copy(pY_Cur + 8 * stride, &data[2*64], stride); |
transfer_16to8copy(pY_Cur + next_block, &data[2*64], stride); |
251 |
transfer_16to8copy(pY_Cur + 8 + 8 * stride, &data[3*64], stride); |
transfer_16to8copy(pY_Cur + 8 + next_block, &data[3*64], stride); |
252 |
transfer_16to8copy(pU_Cur, &data[4*64], stride / 2); |
transfer_16to8copy(pU_Cur, &data[4*64], stride2); |
253 |
transfer_16to8copy(pV_Cur, &data[5*64], stride / 2); |
transfer_16to8copy(pV_Cur, &data[5*64], stride2); |
254 |
stop_transfer_timer(); |
stop_transfer_timer(); |
255 |
} |
} |
256 |
|
|
280 |
DECLARE_ALIGNED_MATRIX(block,6, 64, int16_t, CACHE_LINE); |
DECLARE_ALIGNED_MATRIX(block,6, 64, int16_t, CACHE_LINE); |
281 |
DECLARE_ALIGNED_MATRIX(data, 6, 64, int16_t, CACHE_LINE); |
DECLARE_ALIGNED_MATRIX(data, 6, 64, int16_t, CACHE_LINE); |
282 |
|
|
283 |
const uint32_t stride = dec->edged_width; |
uint32_t stride = dec->edged_width; |
284 |
const uint32_t stride2 = dec->edged_width / 2; |
uint32_t stride2 = stride / 2; |
285 |
|
uint32_t next_block = stride * 8; |
286 |
uint32_t i; |
uint32_t i; |
287 |
uint32_t iQuant = pMB->quant; |
uint32_t iQuant = pMB->quant; |
288 |
uint8_t *pY_Cur, *pU_Cur, *pV_Cur; |
uint8_t *pY_Cur, *pU_Cur, *pV_Cur; |
289 |
int uv_dx, uv_dy; |
int uv_dx, uv_dy; |
290 |
|
|
291 |
pY_Cur = dec->cur.y + (y_pos << 4) * stride + (x_pos << 4); |
pY_Cur = dec->cur.y + (y_pos << 4) * stride + (x_pos << 4); |
292 |
pU_Cur = dec->cur.u + (y_pos << 3) * (stride >> 1) + (x_pos << 3); |
pU_Cur = dec->cur.u + (y_pos << 3) * stride2 + (x_pos << 3); |
293 |
pV_Cur = dec->cur.v + (y_pos << 3) * (stride >> 1) + (x_pos << 3); |
pV_Cur = dec->cur.v + (y_pos << 3) * stride2 + (x_pos << 3); |
294 |
|
|
295 |
if (pMB->mode == MODE_INTER || pMB->mode == MODE_INTER_Q) |
if (pMB->mode == MODE_INTER || pMB->mode == MODE_INTER_Q) |
296 |
{ |
{ |
311 |
} |
} |
312 |
|
|
313 |
start_timer(); |
start_timer(); |
314 |
interpolate8x8_switch(dec->cur.y, dec->refn.y, 16*x_pos, 16*y_pos , pMB->mvs[0].x, pMB->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); |
315 |
interpolate8x8_switch(dec->cur.y, dec->refn.y, 16*x_pos + 8, 16*y_pos , pMB->mvs[1].x, pMB->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); |
316 |
interpolate8x8_switch(dec->cur.y, dec->refn.y, 16*x_pos, 16*y_pos + 8, pMB->mvs[2].x, pMB->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); |
317 |
interpolate8x8_switch(dec->cur.y, dec->refn.y, 16*x_pos + 8, 16*y_pos + 8, pMB->mvs[3].x, pMB->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); |
318 |
interpolate8x8_switch(dec->cur.u, dec->refn.u, 8*x_pos, 8*y_pos, 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); |
319 |
interpolate8x8_switch(dec->cur.v, dec->refn.v, 8*x_pos, 8*y_pos, 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); |
320 |
stop_comp_timer(); |
stop_comp_timer(); |
321 |
|
|
322 |
for (i = 0; i < 6; i++) |
for (i = 0; i < 6; i++) |
346 |
} |
} |
347 |
} |
} |
348 |
|
|
349 |
start_timer(); |
if (dec->interlacing && pMB->field_dct) |
|
if (pMB->field_dct) |
|
350 |
{ |
{ |
351 |
MBFieldToFrame(data); |
next_block = stride; |
352 |
|
stride *= 2; |
353 |
} |
} |
|
stop_interlacing_timer(); |
|
354 |
|
|
355 |
start_timer(); |
start_timer(); |
356 |
if (cbp & 32) |
if (cbp & 32) |
358 |
if (cbp & 16) |
if (cbp & 16) |
359 |
transfer_16to8add(pY_Cur + 8, &data[1*64], stride); |
transfer_16to8add(pY_Cur + 8, &data[1*64], stride); |
360 |
if (cbp & 8) |
if (cbp & 8) |
361 |
transfer_16to8add(pY_Cur + 8 * stride, &data[2*64], stride); |
transfer_16to8add(pY_Cur + next_block, &data[2*64], stride); |
362 |
if (cbp & 4) |
if (cbp & 4) |
363 |
transfer_16to8add(pY_Cur + 8 + 8 * stride, &data[3*64], stride); |
transfer_16to8add(pY_Cur + 8 + next_block, &data[3*64], stride); |
364 |
if (cbp & 2) |
if (cbp & 2) |
365 |
transfer_16to8add(pU_Cur, &data[4*64], stride / 2); |
transfer_16to8add(pU_Cur, &data[4*64], stride2); |
366 |
if (cbp & 1) |
if (cbp & 1) |
367 |
transfer_16to8add(pV_Cur, &data[5*64], stride / 2); |
transfer_16to8add(pV_Cur, &data[5*64], stride2); |
368 |
stop_transfer_timer(); |
stop_transfer_timer(); |
369 |
} |
} |
370 |
|
|
437 |
int range = (64 * scale_fac); |
int range = (64 * scale_fac); |
438 |
|
|
439 |
VECTOR pmv[4]; |
VECTOR pmv[4]; |
440 |
uint32_t psad[4]; |
int32_t psad[4]; |
441 |
|
|
442 |
int mv_x, mv_y; |
int mv_x, mv_y; |
443 |
int pmv_x, pmv_y; |
int pmv_x, pmv_y; |
483 |
|
|
484 |
uint32_t x, y; |
uint32_t x, y; |
485 |
|
|
|
image_swap(&dec->cur, &dec->refn); |
|
|
|
|
486 |
start_timer(); |
start_timer(); |
487 |
image_setedges(&dec->refn, dec->edged_width, dec->edged_height, dec->width, dec->height, dec->interlacing); |
image_setedges(&dec->refn[0], dec->edged_width, dec->edged_height, dec->width, dec->height, dec->interlacing); |
488 |
stop_edges_timer(); |
stop_edges_timer(); |
489 |
|
|
490 |
for (y = 0; y < dec->mb_height; y++) |
for (y = 0; y < dec->mb_height; y++) |
600 |
start_timer(); |
start_timer(); |
601 |
|
|
602 |
transfer8x8_copy(dec->cur.y + (16*y)*dec->edged_width + (16*x), |
transfer8x8_copy(dec->cur.y + (16*y)*dec->edged_width + (16*x), |
603 |
dec->refn.y + (16*y)*dec->edged_width + (16*x), |
dec->refn[0].y + (16*y)*dec->edged_width + (16*x), |
604 |
dec->edged_width); |
dec->edged_width); |
605 |
|
|
606 |
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), |
607 |
dec->refn.y + (16*y)*dec->edged_width + (16*x+8), |
dec->refn[0].y + (16*y)*dec->edged_width + (16*x+8), |
608 |
dec->edged_width); |
dec->edged_width); |
609 |
|
|
610 |
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), |
611 |
dec->refn.y + (16*y+8)*dec->edged_width + (16*x), |
dec->refn[0].y + (16*y+8)*dec->edged_width + (16*x), |
612 |
dec->edged_width); |
dec->edged_width); |
613 |
|
|
614 |
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), |
615 |
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), |
616 |
dec->edged_width); |
dec->edged_width); |
617 |
|
|
618 |
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), |
619 |
dec->refn.u + (8*y)*dec->edged_width/2 + (8*x), |
dec->refn[0].u + (8*y)*dec->edged_width/2 + (8*x), |
620 |
dec->edged_width/2); |
dec->edged_width/2); |
621 |
|
|
622 |
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), |
623 |
dec->refn.v + (8*y)*dec->edged_width/2 + (8*x), |
dec->refn[0].v + (8*y)*dec->edged_width/2 + (8*x), |
624 |
dec->edged_width/2); |
dec->edged_width/2); |
625 |
|
|
626 |
stop_transfer_timer(); |
stop_transfer_timer(); |
637 |
uint32_t quant; |
uint32_t quant; |
638 |
uint32_t fcode; |
uint32_t fcode; |
639 |
uint32_t intra_dc_threshold; |
uint32_t intra_dc_threshold; |
640 |
|
uint32_t vop_type; |
641 |
|
|
642 |
start_global_timer(); |
start_global_timer(); |
643 |
|
|
644 |
BitstreamInit(&bs, frame->bitstream, frame->length); |
BitstreamInit(&bs, frame->bitstream, frame->length); |
645 |
|
|
646 |
switch (BitstreamReadHeaders(&bs, dec, &rounding, &quant, &fcode, &intra_dc_threshold)) |
// add by chenm001 <chenm001@163.com> |
647 |
|
// for support B-frame to reference last 2 frame |
648 |
|
vop_type=BitstreamReadHeaders(&bs, dec, &rounding, &quant, &fcode, &intra_dc_threshold); |
649 |
|
|
650 |
|
if (vop_type==I_VOP || vop_type==P_VOP){ |
651 |
|
image_swap(&dec->refn[0], &dec->refn[1]); |
652 |
|
image_swap(&dec->cur, &dec->refn[0]); |
653 |
|
} |
654 |
|
|
655 |
|
switch (vop_type) |
656 |
{ |
{ |
657 |
case P_VOP : |
case P_VOP : |
658 |
decoder_pframe(dec, &bs, rounding, quant, fcode, intra_dc_threshold); |
decoder_pframe(dec, &bs, rounding, quant, fcode, intra_dc_threshold); |