55 |
* 22.12.2001 lock based interpolation |
* 22.12.2001 lock based interpolation |
56 |
* 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> |
57 |
* |
* |
58 |
* $Id: decoder.c,v 1.37.2.12 2002-11-19 13:21:24 suxen_drol Exp $ |
* $Id: decoder.c,v 1.37.2.17 2002-12-09 10:47:05 suxen_drol Exp $ |
59 |
* |
* |
60 |
*************************************************************************/ |
*************************************************************************/ |
61 |
|
|
79 |
#include "dct/fdct.h" |
#include "dct/fdct.h" |
80 |
#include "utils/mem_transfer.h" |
#include "utils/mem_transfer.h" |
81 |
#include "image/interpolate8x8.h" |
#include "image/interpolate8x8.h" |
82 |
|
#include "image/reduced.h" |
83 |
|
|
84 |
#include "bitstream/mbcoding.h" |
#include "bitstream/mbcoding.h" |
85 |
#include "prediction/mbprediction.h" |
#include "prediction/mbprediction.h" |
219 |
dec->frames = -1; |
dec->frames = -1; |
220 |
dec->time = dec->time_base = dec->last_time_base = 0; |
dec->time = dec->time_base = dec->last_time_base = 0; |
221 |
dec->low_delay = 0; |
dec->low_delay = 0; |
222 |
|
dec->packed_mode = 0; |
223 |
|
|
224 |
dec->fixed_dimensions = (dec->width > 0 && dec->height > 0); |
dec->fixed_dimensions = (dec->width > 0 && dec->height > 0); |
225 |
|
|
267 |
Bitstream * bs, |
Bitstream * bs, |
268 |
const uint32_t quant, |
const uint32_t quant, |
269 |
const uint32_t intra_dc_threshold, |
const uint32_t intra_dc_threshold, |
270 |
const unsigned int bound) |
const unsigned int bound, |
271 |
|
const int reduced_resolution) |
272 |
{ |
{ |
273 |
|
|
274 |
DECLARE_ALIGNED_MATRIX(block, 6, 64, int16_t, CACHE_LINE); |
DECLARE_ALIGNED_MATRIX(block, 6, 64, int16_t, CACHE_LINE); |
281 |
uint32_t iQuant = pMB->quant; |
uint32_t iQuant = pMB->quant; |
282 |
uint8_t *pY_Cur, *pU_Cur, *pV_Cur; |
uint8_t *pY_Cur, *pU_Cur, *pV_Cur; |
283 |
|
|
284 |
|
if (reduced_resolution) { |
285 |
|
pY_Cur = dec->cur.y + (y_pos << 5) * stride + (x_pos << 5); |
286 |
|
pU_Cur = dec->cur.u + (y_pos << 4) * stride2 + (x_pos << 4); |
287 |
|
pV_Cur = dec->cur.v + (y_pos << 4) * stride2 + (x_pos << 4); |
288 |
|
}else{ |
289 |
pY_Cur = dec->cur.y + (y_pos << 4) * stride + (x_pos << 4); |
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); |
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); |
pV_Cur = dec->cur.v + (y_pos << 3) * stride2 + (x_pos << 3); |
292 |
|
} |
293 |
|
|
294 |
memset(block, 0, 6 * 64 * sizeof(int16_t)); // clear |
memset(block, 0, 6 * 64 * sizeof(int16_t)); // clear |
295 |
|
|
350 |
start_timer(); |
start_timer(); |
351 |
idct(&data[i * 64]); |
idct(&data[i * 64]); |
352 |
stop_idct_timer(); |
stop_idct_timer(); |
353 |
|
|
354 |
} |
} |
355 |
|
|
356 |
if (dec->interlacing && pMB->field_dct) { |
if (dec->interlacing && pMB->field_dct) { |
359 |
} |
} |
360 |
|
|
361 |
start_timer(); |
start_timer(); |
362 |
|
|
363 |
|
if (reduced_resolution) |
364 |
|
{ |
365 |
|
next_block*=2; |
366 |
|
copy_upsampled_8x8_16to8(pY_Cur, &data[0 * 64], stride); |
367 |
|
copy_upsampled_8x8_16to8(pY_Cur + 16, &data[1 * 64], stride); |
368 |
|
copy_upsampled_8x8_16to8(pY_Cur + next_block, &data[2 * 64], stride); |
369 |
|
copy_upsampled_8x8_16to8(pY_Cur + 16 + next_block, &data[3 * 64], stride); |
370 |
|
copy_upsampled_8x8_16to8(pU_Cur, &data[4 * 64], stride2); |
371 |
|
copy_upsampled_8x8_16to8(pV_Cur, &data[5 * 64], stride2); |
372 |
|
}else{ |
373 |
transfer_16to8copy(pY_Cur, &data[0 * 64], stride); |
transfer_16to8copy(pY_Cur, &data[0 * 64], stride); |
374 |
transfer_16to8copy(pY_Cur + 8, &data[1 * 64], stride); |
transfer_16to8copy(pY_Cur + 8, &data[1 * 64], stride); |
375 |
transfer_16to8copy(pY_Cur + next_block, &data[2 * 64], stride); |
transfer_16to8copy(pY_Cur + next_block, &data[2 * 64], stride); |
376 |
transfer_16to8copy(pY_Cur + 8 + next_block, &data[3 * 64], stride); |
transfer_16to8copy(pY_Cur + 8 + next_block, &data[3 * 64], stride); |
377 |
transfer_16to8copy(pU_Cur, &data[4 * 64], stride2); |
transfer_16to8copy(pU_Cur, &data[4 * 64], stride2); |
378 |
transfer_16to8copy(pV_Cur, &data[5 * 64], stride2); |
transfer_16to8copy(pV_Cur, &data[5 * 64], stride2); |
379 |
|
} |
380 |
stop_transfer_timer(); |
stop_transfer_timer(); |
381 |
} |
} |
382 |
|
|
389 |
|
|
390 |
// decode an inter macroblock |
// decode an inter macroblock |
391 |
|
|
392 |
|
static void |
393 |
|
rrv_mv_scaleup(VECTOR * mv) |
394 |
|
{ |
395 |
|
if (mv->x > 0) { |
396 |
|
mv->x = 2*mv->x - 1; |
397 |
|
} else if (mv->x < 0) { |
398 |
|
mv->x = 2*mv->x + 1; |
399 |
|
} |
400 |
|
|
401 |
|
if (mv->y > 0) { |
402 |
|
mv->y = 2*mv->y - 1; |
403 |
|
} else if (mv->y < 0) { |
404 |
|
mv->y = 2*mv->y + 1; |
405 |
|
} |
406 |
|
} |
407 |
|
|
408 |
|
|
409 |
|
|
410 |
void |
void |
411 |
decoder_mbinter(DECODER * dec, |
decoder_mbinter(DECODER * dec, |
412 |
const MACROBLOCK * pMB, |
const MACROBLOCK * pMB, |
416 |
const uint32_t cbp, |
const uint32_t cbp, |
417 |
Bitstream * bs, |
Bitstream * bs, |
418 |
const uint32_t quant, |
const uint32_t quant, |
419 |
const uint32_t rounding) |
const uint32_t rounding, |
420 |
|
const int reduced_resolution) |
421 |
{ |
{ |
422 |
|
|
423 |
DECLARE_ALIGNED_MATRIX(block, 6, 64, int16_t, CACHE_LINE); |
DECLARE_ALIGNED_MATRIX(block, 6, 64, int16_t, CACHE_LINE); |
425 |
|
|
426 |
uint32_t stride = dec->edged_width; |
uint32_t stride = dec->edged_width; |
427 |
uint32_t stride2 = stride / 2; |
uint32_t stride2 = stride / 2; |
428 |
uint32_t next_block = stride * 8; |
uint32_t next_block = stride * (reduced_resolution ? 16 : 8); |
429 |
uint32_t i; |
uint32_t i; |
430 |
uint32_t iQuant = pMB->quant; |
uint32_t iQuant = pMB->quant; |
431 |
uint8_t *pY_Cur, *pU_Cur, *pV_Cur; |
uint8_t *pY_Cur, *pU_Cur, *pV_Cur; |
432 |
|
|
433 |
int uv_dx, uv_dy; |
int uv_dx, uv_dy; |
434 |
|
VECTOR mv[4]; |
435 |
|
|
436 |
|
for (i = 0; i < 4; i++) |
437 |
|
{ |
438 |
|
mv[i] = pMB->mvs[i]; |
439 |
|
//DPRINTF(DPRINTF_MB, "mv[%i] orig=%i,%i local=%i", i, pMB->mvs[i].x, pMB->mvs[i].y, mv[i].x, mv[i].y); |
440 |
|
} |
441 |
|
|
442 |
|
if (reduced_resolution) { |
443 |
|
pY_Cur = dec->cur.y + (y_pos << 5) * stride + (x_pos << 5); |
444 |
|
pU_Cur = dec->cur.u + (y_pos << 4) * stride2 + (x_pos << 4); |
445 |
|
pV_Cur = dec->cur.v + (y_pos << 4) * stride2 + (x_pos << 4); |
446 |
|
rrv_mv_scaleup(&mv[0]); |
447 |
|
rrv_mv_scaleup(&mv[1]); |
448 |
|
rrv_mv_scaleup(&mv[2]); |
449 |
|
rrv_mv_scaleup(&mv[3]); |
450 |
|
}else{ |
451 |
pY_Cur = dec->cur.y + (y_pos << 4) * stride + (x_pos << 4); |
pY_Cur = dec->cur.y + (y_pos << 4) * stride + (x_pos << 4); |
452 |
pU_Cur = dec->cur.u + (y_pos << 3) * stride2 + (x_pos << 3); |
pU_Cur = dec->cur.u + (y_pos << 3) * stride2 + (x_pos << 3); |
453 |
pV_Cur = dec->cur.v + (y_pos << 3) * stride2 + (x_pos << 3); |
pV_Cur = dec->cur.v + (y_pos << 3) * stride2 + (x_pos << 3); |
454 |
|
} |
455 |
|
|
456 |
if (pMB->mode == MODE_INTER || pMB->mode == MODE_INTER_Q) { |
if (pMB->mode == MODE_INTER || pMB->mode == MODE_INTER_Q) { |
457 |
uv_dx = pMB->mvs[0].x; |
uv_dx = mv[0].x; |
458 |
uv_dy = pMB->mvs[0].y; |
uv_dy = mv[0].y; |
459 |
|
|
460 |
if (dec->quarterpel) |
if (dec->quarterpel) |
461 |
{ |
{ |
467 |
uv_dy = (uv_dy >> 1) + roundtab_79[uv_dy & 0x3]; |
uv_dy = (uv_dy >> 1) + roundtab_79[uv_dy & 0x3]; |
468 |
|
|
469 |
start_timer(); |
start_timer(); |
470 |
|
if (reduced_resolution) |
471 |
|
{ |
472 |
|
interpolate32x32_switch(dec->cur.y, dec->refn[0].y, 32*x_pos, 32*y_pos, |
473 |
|
mv[0].x, mv[0].y, stride, rounding); |
474 |
|
interpolate16x16_switch(dec->cur.u, dec->refn[0].u, 16 * x_pos, 16 * y_pos, |
475 |
|
uv_dx, uv_dy, stride2, rounding); |
476 |
|
interpolate16x16_switch(dec->cur.v, dec->refn[0].v, 16 * x_pos, 16 * y_pos, |
477 |
|
uv_dx, uv_dy, stride2, rounding); |
478 |
|
|
479 |
|
} |
480 |
|
else |
481 |
|
{ |
482 |
if(dec->quarterpel) { |
if(dec->quarterpel) { |
483 |
interpolate16x16_quarterpel(dec->cur.y, dec->refn[0].y, dec->refh.y, dec->refh.y + 64, |
interpolate16x16_quarterpel(dec->cur.y, dec->refn[0].y, dec->refh.y, dec->refh.y + 64, |
484 |
dec->refh.y + 128, 16*x_pos, 16*y_pos, |
dec->refh.y + 128, 16*x_pos, 16*y_pos, |
485 |
pMB->mvs[0].x, pMB->mvs[0].y, stride, rounding); |
mv[0].x, mv[0].y, stride, rounding); |
486 |
} |
} |
487 |
else { |
else { |
488 |
interpolate8x8_switch(dec->cur.y, dec->refn[0].y, 16*x_pos, 16*y_pos, |
interpolate16x16_switch(dec->cur.y, dec->refn[0].y, 16*x_pos, 16*y_pos, |
489 |
pMB->mvs[0].x, pMB->mvs[0].y, stride, rounding); |
mv[0].x, mv[0].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); |
|
|
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); |
|
|
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); |
|
490 |
} |
} |
491 |
|
|
492 |
interpolate8x8_switch(dec->cur.u, dec->refn[0].u, 8 * x_pos, 8 * y_pos, |
interpolate8x8_switch(dec->cur.u, dec->refn[0].u, 8 * x_pos, 8 * y_pos, |
493 |
uv_dx, uv_dy, stride2, rounding); |
uv_dx, uv_dy, stride2, rounding); |
494 |
interpolate8x8_switch(dec->cur.v, dec->refn[0].v, 8 * x_pos, 8 * y_pos, |
interpolate8x8_switch(dec->cur.v, dec->refn[0].v, 8 * x_pos, 8 * y_pos, |
495 |
uv_dx, uv_dy, stride2, rounding); |
uv_dx, uv_dy, stride2, rounding); |
496 |
|
} |
497 |
stop_comp_timer(); |
stop_comp_timer(); |
498 |
|
|
499 |
} else { |
} else { /* MODE_INTER4V */ |
500 |
int sum; |
int sum; |
501 |
|
|
502 |
if(dec->quarterpel) |
if(dec->quarterpel) |
503 |
sum = (pMB->mvs[0].x / 2) + (pMB->mvs[1].x / 2) + (pMB->mvs[2].x / 2) + (pMB->mvs[3].x / 2); |
sum = (mv[0].x / 2) + (mv[1].x / 2) + (mv[2].x / 2) + (mv[3].x / 2); |
504 |
else |
else |
505 |
sum = pMB->mvs[0].x + pMB->mvs[1].x + pMB->mvs[2].x + pMB->mvs[3].x; |
sum = mv[0].x + mv[1].x + mv[2].x + mv[3].x; |
506 |
|
|
507 |
uv_dx = (sum >> 3) + roundtab_76[sum & 0xf]; |
uv_dx = (sum >> 3) + roundtab_76[sum & 0xf]; |
508 |
|
|
509 |
if(dec->quarterpel) |
if(dec->quarterpel) |
510 |
sum = (pMB->mvs[0].y / 2) + (pMB->mvs[1].y / 2) + (pMB->mvs[2].y / 2) + (pMB->mvs[3].y / 2); |
sum = (mv[0].y / 2) + (mv[1].y / 2) + (mv[2].y / 2) + (mv[3].y / 2); |
511 |
else |
else |
512 |
sum = pMB->mvs[0].y + pMB->mvs[1].y + pMB->mvs[2].y + pMB->mvs[3].y; |
sum = mv[0].y + mv[1].y + mv[2].y + mv[3].y; |
513 |
|
|
514 |
uv_dy = (sum >> 3) + roundtab_76[sum & 0xf]; |
uv_dy = (sum >> 3) + roundtab_76[sum & 0xf]; |
515 |
|
|
516 |
start_timer(); |
start_timer(); |
517 |
|
if (reduced_resolution) |
518 |
|
{ |
519 |
|
interpolate16x16_switch(dec->cur.y, dec->refn[0].y, 32*x_pos, 32*y_pos, |
520 |
|
mv[0].x, mv[0].y, stride, rounding); |
521 |
|
interpolate16x16_switch(dec->cur.y, dec->refn[0].y, 32*x_pos + 16, 32*y_pos, |
522 |
|
mv[1].x, mv[1].y, stride, rounding); |
523 |
|
interpolate16x16_switch(dec->cur.y, dec->refn[0].y, 32*x_pos, 32*y_pos + 16, |
524 |
|
mv[2].x, mv[2].y, stride, rounding); |
525 |
|
interpolate16x16_switch(dec->cur.y, dec->refn[0].y, 32*x_pos + 16, 32*y_pos + 16, |
526 |
|
mv[3].x, mv[3].y, stride, rounding); |
527 |
|
interpolate16x16_switch(dec->cur.u, dec->refn[0].u, 16 * x_pos, 16 * y_pos, |
528 |
|
uv_dx, uv_dy, stride2, rounding); |
529 |
|
interpolate16x16_switch(dec->cur.v, dec->refn[0].v, 16 * x_pos, 16 * y_pos, |
530 |
|
uv_dx, uv_dy, stride2, rounding); |
531 |
|
|
532 |
|
// set_block(pY_Cur, stride, 32, 32, 127); |
533 |
|
} |
534 |
|
else |
535 |
|
{ |
536 |
if(dec->quarterpel) { |
if(dec->quarterpel) { |
537 |
interpolate8x8_quarterpel(dec->cur.y, dec->refn[0].y, dec->refh.y, dec->refh.y + 64, |
interpolate8x8_quarterpel(dec->cur.y, dec->refn[0].y, dec->refh.y, dec->refh.y + 64, |
538 |
dec->refh.y + 128, 16*x_pos, 16*y_pos, |
dec->refh.y + 128, 16*x_pos, 16*y_pos, |
539 |
pMB->mvs[0].x, pMB->mvs[0].y, stride, rounding); |
mv[0].x, mv[0].y, stride, rounding); |
540 |
interpolate8x8_quarterpel(dec->cur.y, dec->refn[0].y, dec->refh.y, dec->refh.y + 64, |
interpolate8x8_quarterpel(dec->cur.y, dec->refn[0].y, dec->refh.y, dec->refh.y + 64, |
541 |
dec->refh.y + 128, 16*x_pos + 8, 16*y_pos, |
dec->refh.y + 128, 16*x_pos + 8, 16*y_pos, |
542 |
pMB->mvs[1].x, pMB->mvs[1].y, stride, rounding); |
mv[1].x, mv[1].y, stride, rounding); |
543 |
interpolate8x8_quarterpel(dec->cur.y, dec->refn[0].y, dec->refh.y, dec->refh.y + 64, |
interpolate8x8_quarterpel(dec->cur.y, dec->refn[0].y, dec->refh.y, dec->refh.y + 64, |
544 |
dec->refh.y + 128, 16*x_pos, 16*y_pos + 8, |
dec->refh.y + 128, 16*x_pos, 16*y_pos + 8, |
545 |
pMB->mvs[2].x, pMB->mvs[2].y, stride, rounding); |
mv[2].x, mv[2].y, stride, rounding); |
546 |
interpolate8x8_quarterpel(dec->cur.y, dec->refn[0].y, dec->refh.y, dec->refh.y + 64, |
interpolate8x8_quarterpel(dec->cur.y, dec->refn[0].y, dec->refh.y, dec->refh.y + 64, |
547 |
dec->refh.y + 128, 16*x_pos + 8, 16*y_pos + 8, |
dec->refh.y + 128, 16*x_pos + 8, 16*y_pos + 8, |
548 |
pMB->mvs[3].x, pMB->mvs[3].y, stride, rounding); |
mv[3].x, mv[3].y, stride, rounding); |
549 |
} |
} |
550 |
else { |
else { |
551 |
interpolate8x8_switch(dec->cur.y, dec->refn[0].y, 16*x_pos, 16*y_pos, |
interpolate8x8_switch(dec->cur.y, dec->refn[0].y, 16*x_pos, 16*y_pos, |
552 |
pMB->mvs[0].x, pMB->mvs[0].y, stride, rounding); |
mv[0].x, mv[0].y, stride, rounding); |
553 |
interpolate8x8_switch(dec->cur.y, dec->refn[0].y, 16*x_pos + 8, 16*y_pos, |
interpolate8x8_switch(dec->cur.y, dec->refn[0].y, 16*x_pos + 8, 16*y_pos, |
554 |
pMB->mvs[1].x, pMB->mvs[1].y, stride, rounding); |
mv[1].x, mv[1].y, stride, rounding); |
555 |
interpolate8x8_switch(dec->cur.y, dec->refn[0].y, 16*x_pos, 16*y_pos + 8, |
interpolate8x8_switch(dec->cur.y, dec->refn[0].y, 16*x_pos, 16*y_pos + 8, |
556 |
pMB->mvs[2].x, pMB->mvs[2].y, stride, rounding); |
mv[2].x, mv[2].y, stride, rounding); |
557 |
interpolate8x8_switch(dec->cur.y, dec->refn[0].y, 16*x_pos + 8, 16*y_pos + 8, |
interpolate8x8_switch(dec->cur.y, dec->refn[0].y, 16*x_pos + 8, 16*y_pos + 8, |
558 |
pMB->mvs[3].x, pMB->mvs[3].y, stride, rounding); |
mv[3].x, mv[3].y, stride, rounding); |
559 |
} |
} |
560 |
|
|
561 |
interpolate8x8_switch(dec->cur.u, dec->refn[0].u, 8 * x_pos, 8 * y_pos, |
interpolate8x8_switch(dec->cur.u, dec->refn[0].u, 8 * x_pos, 8 * y_pos, |
562 |
uv_dx, uv_dy, stride2, rounding); |
uv_dx, uv_dy, stride2, rounding); |
563 |
interpolate8x8_switch(dec->cur.v, dec->refn[0].v, 8 * x_pos, 8 * y_pos, |
interpolate8x8_switch(dec->cur.v, dec->refn[0].v, 8 * x_pos, 8 * y_pos, |
564 |
uv_dx, uv_dy, stride2, rounding); |
uv_dx, uv_dy, stride2, rounding); |
565 |
|
} |
566 |
stop_comp_timer(); |
stop_comp_timer(); |
567 |
} |
} |
568 |
|
|
597 |
} |
} |
598 |
|
|
599 |
start_timer(); |
start_timer(); |
600 |
|
if (reduced_resolution) |
601 |
|
{ |
602 |
|
if (cbp & 32) |
603 |
|
add_upsampled_8x8_16to8(pY_Cur, &data[0 * 64], stride); |
604 |
|
if (cbp & 16) |
605 |
|
add_upsampled_8x8_16to8(pY_Cur + 16, &data[1 * 64], stride); |
606 |
|
if (cbp & 8) |
607 |
|
add_upsampled_8x8_16to8(pY_Cur + next_block, &data[2 * 64], stride); |
608 |
|
if (cbp & 4) |
609 |
|
add_upsampled_8x8_16to8(pY_Cur + 16 + next_block, &data[3 * 64], stride); |
610 |
|
if (cbp & 2) |
611 |
|
add_upsampled_8x8_16to8(pU_Cur, &data[4 * 64], stride2); |
612 |
|
if (cbp & 1) |
613 |
|
add_upsampled_8x8_16to8(pV_Cur, &data[5 * 64], stride2); |
614 |
|
} |
615 |
|
else |
616 |
|
{ |
617 |
if (cbp & 32) |
if (cbp & 32) |
618 |
transfer_16to8add(pY_Cur, &data[0 * 64], stride); |
transfer_16to8add(pY_Cur, &data[0 * 64], stride); |
619 |
if (cbp & 16) |
if (cbp & 16) |
626 |
transfer_16to8add(pU_Cur, &data[4 * 64], stride2); |
transfer_16to8add(pU_Cur, &data[4 * 64], stride2); |
627 |
if (cbp & 1) |
if (cbp & 1) |
628 |
transfer_16to8add(pV_Cur, &data[5 * 64], stride2); |
transfer_16to8add(pV_Cur, &data[5 * 64], stride2); |
629 |
|
} |
630 |
stop_transfer_timer(); |
stop_transfer_timer(); |
631 |
} |
} |
632 |
|
|
645 |
|
|
646 |
if (reduced_resolution) |
if (reduced_resolution) |
647 |
{ |
{ |
648 |
mb_width /= 2; |
mb_width = (dec->width + 31) / 32; |
649 |
mb_height /= 2; |
mb_height = (dec->height + 31) / 32; |
650 |
} |
} |
651 |
|
|
652 |
bound = 0; |
bound = 0; |
667 |
{ |
{ |
668 |
bound = read_video_packet_header(bs, dec, 0, |
bound = read_video_packet_header(bs, dec, 0, |
669 |
&quant, NULL, NULL, &intra_dc_threshold); |
&quant, NULL, NULL, &intra_dc_threshold); |
670 |
x = bound % dec->mb_width; |
x = bound % mb_width; |
671 |
y = bound / dec->mb_width; |
y = bound / mb_width; |
672 |
} |
} |
673 |
mb = &dec->mbs[y * dec->mb_width + x]; |
mb = &dec->mbs[y * dec->mb_width + x]; |
674 |
|
|
703 |
} |
} |
704 |
|
|
705 |
decoder_mbintra(dec, mb, x, y, acpred_flag, cbp, bs, quant, |
decoder_mbintra(dec, mb, x, y, acpred_flag, cbp, bs, quant, |
706 |
intra_dc_threshold, bound); |
intra_dc_threshold, bound, reduced_resolution); |
707 |
|
|
708 |
} |
} |
709 |
if(dec->out_frm) |
if(dec->out_frm) |
710 |
output_slice(&dec->cur, dec->edged_width,dec->width,dec->out_frm,0,y,dec->mb_width); |
output_slice(&dec->cur, dec->edged_width,dec->width,dec->out_frm,0,y,mb_width); |
711 |
} |
} |
712 |
|
|
713 |
} |
} |
719 |
int x, |
int x, |
720 |
int y, |
int y, |
721 |
int k, |
int k, |
722 |
VECTOR * mv, |
VECTOR * ret_mv, |
723 |
int fcode, |
int fcode, |
724 |
const int bound) |
const int bound) |
725 |
{ |
{ |
730 |
int range = (64 * scale_fac); |
int range = (64 * scale_fac); |
731 |
|
|
732 |
VECTOR pmv; |
VECTOR pmv; |
733 |
int mv_x, mv_y; |
VECTOR mv; |
734 |
|
|
735 |
pmv = get_pmv2(dec->mbs, dec->mb_width, bound, x, y, k); |
pmv = get_pmv2(dec->mbs, dec->mb_width, bound, x, y, k); |
736 |
|
|
737 |
mv_x = get_mv(bs, fcode); |
mv.x = get_mv(bs, fcode); |
738 |
mv_y = get_mv(bs, fcode); |
mv.y = get_mv(bs, fcode); |
739 |
|
|
740 |
DPRINTF(DPRINTF_MV,"mv_diff (%i,%i) pred (%i,%i)", mv_x, mv_y, pmv.x, pmv.y); |
DPRINTF(DPRINTF_MV,"mv_diff (%i,%i) pred (%i,%i)", mv.x, mv.y, pmv.x, pmv.y); |
741 |
|
|
742 |
mv_x += pmv.x; |
mv.x += pmv.x; |
743 |
mv_y += pmv.y; |
mv.y += pmv.y; |
744 |
|
|
745 |
if (mv_x < low) { |
if (mv.x < low) { |
746 |
mv_x += range; |
mv.x += range; |
747 |
} else if (mv_x > high) { |
} else if (mv.x > high) { |
748 |
mv_x -= range; |
mv.x -= range; |
749 |
} |
} |
750 |
|
|
751 |
if (mv_y < low) { |
if (mv.y < low) { |
752 |
mv_y += range; |
mv.y += range; |
753 |
} else if (mv_y > high) { |
} else if (mv.y > high) { |
754 |
mv_y -= range; |
mv.y -= range; |
755 |
} |
} |
756 |
|
|
757 |
mv->x = mv_x; |
ret_mv->x = mv.x; |
758 |
mv->y = mv_y; |
ret_mv->y = mv.y; |
759 |
} |
} |
760 |
|
|
761 |
|
|
789 |
uint32_t x, y; |
uint32_t x, y; |
790 |
uint32_t bound; |
uint32_t bound; |
791 |
int cp_mb, st_mb; |
int cp_mb, st_mb; |
792 |
|
int mb_width = dec->mb_width; |
793 |
|
int mb_height = dec->mb_height; |
794 |
|
|
795 |
|
if (reduced_resolution) |
796 |
|
{ |
797 |
|
mb_width = (dec->width + 31) / 32; |
798 |
|
mb_height = (dec->height + 31) / 32; |
799 |
|
} |
800 |
|
|
801 |
start_timer(); |
start_timer(); |
802 |
image_setedges(&dec->refn[0], dec->edged_width, dec->edged_height, |
image_setedges(&dec->refn[0], dec->edged_width, dec->edged_height, |
805 |
|
|
806 |
bound = 0; |
bound = 0; |
807 |
|
|
808 |
for (y = 0; y < dec->mb_height; y++) { |
for (y = 0; y < mb_height; y++) { |
809 |
cp_mb = st_mb = 0; |
cp_mb = st_mb = 0; |
810 |
for (x = 0; x < dec->mb_width; x++) { |
for (x = 0; x < mb_width; x++) { |
811 |
MACROBLOCK *mb; |
MACROBLOCK *mb; |
812 |
|
|
813 |
// skip stuffing |
// skip stuffing |
818 |
{ |
{ |
819 |
bound = read_video_packet_header(bs, dec, fcode - 1, |
bound = read_video_packet_header(bs, dec, fcode - 1, |
820 |
&quant, &fcode, NULL, &intra_dc_threshold); |
&quant, &fcode, NULL, &intra_dc_threshold); |
821 |
x = bound % dec->mb_width; |
x = bound % mb_width; |
822 |
y = bound / dec->mb_width; |
y = bound / mb_width; |
823 |
} |
} |
824 |
mb = &dec->mbs[y * dec->mb_width + x]; |
mb = &dec->mbs[y * dec->mb_width + x]; |
825 |
|
|
926 |
mb->mvs[0].y = 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 = |
927 |
0; |
0; |
928 |
decoder_mbintra(dec, mb, x, y, acpred_flag, cbp, bs, quant, |
decoder_mbintra(dec, mb, x, y, acpred_flag, cbp, bs, quant, |
929 |
intra_dc_threshold, bound); |
intra_dc_threshold, bound, reduced_resolution); |
930 |
continue; |
continue; |
931 |
} |
} |
932 |
|
|
933 |
decoder_mbinter(dec, mb, x, y, acpred_flag, cbp, bs, quant, |
decoder_mbinter(dec, mb, x, y, acpred_flag, cbp, bs, quant, |
934 |
rounding); |
rounding, reduced_resolution); |
935 |
|
|
936 |
} |
} |
937 |
else if (gmc_mv) /* not coded S_VOP macroblock */ |
else if (gmc_mv) /* not coded S_VOP macroblock */ |
939 |
mb->mode = MODE_NOT_CODED; |
mb->mode = MODE_NOT_CODED; |
940 |
mb->mvs[0].x = mb->mvs[1].x = mb->mvs[2].x = mb->mvs[3].x = gmc_sanitize(gmc_mv[0].x, dec->quarterpel, fcode); |
mb->mvs[0].x = mb->mvs[1].x = mb->mvs[2].x = mb->mvs[3].x = gmc_sanitize(gmc_mv[0].x, dec->quarterpel, fcode); |
941 |
mb->mvs[0].y = mb->mvs[1].y = mb->mvs[2].y = mb->mvs[3].y = gmc_sanitize(gmc_mv[0].y, dec->quarterpel, fcode); |
mb->mvs[0].y = mb->mvs[1].y = mb->mvs[2].y = mb->mvs[3].y = gmc_sanitize(gmc_mv[0].y, dec->quarterpel, fcode); |
942 |
decoder_mbinter(dec, mb, x, y, 0, 0, bs, quant, rounding); |
decoder_mbinter(dec, mb, x, y, 0, 0, bs, quant, rounding, reduced_resolution); |
943 |
} |
} |
944 |
else /* not coded P_VOP macroblock */ |
else /* not coded P_VOP macroblock */ |
945 |
{ |
{ |
946 |
mb->mode = MODE_NOT_CODED; |
mb->mode = MODE_NOT_CODED; |
947 |
|
|
948 |
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; |
949 |
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; |
950 |
// copy macroblock directly from ref to cur |
// copy macroblock directly from ref to cur |
951 |
|
|
952 |
start_timer(); |
start_timer(); |
953 |
|
|
954 |
transfer8x8_copy(dec->cur.y + (16 * y) * dec->edged_width + |
if (reduced_resolution) |
955 |
(16 * x), |
{ |
956 |
dec->refn[0].y + (16 * y) * dec->edged_width + |
transfer32x32_copy(dec->cur.y + (32*y)*dec->edged_width + (32*x), |
957 |
(16 * x), dec->edged_width); |
dec->refn[0].y + (32*y)*dec->edged_width + (32*x), |
958 |
|
dec->edged_width); |
959 |
transfer8x8_copy(dec->cur.y + (16 * y) * dec->edged_width + |
|
960 |
(16 * x + 8), |
transfer16x16_copy(dec->cur.u + (16*y)*dec->edged_width/2 + (16*x), |
961 |
dec->refn[0].y + (16 * y) * dec->edged_width + |
dec->refn[0].u + (16*y)*dec->edged_width/2 + (16*x), |
962 |
(16 * x + 8), dec->edged_width); |
dec->edged_width/2); |
963 |
|
|
964 |
transfer8x8_copy(dec->cur.y + (16 * y + 8) * dec->edged_width + |
transfer16x16_copy(dec->cur.v + (16*y)*dec->edged_width/2 + (16*x), |
965 |
(16 * x), |
dec->refn[0].v + (16*y)*dec->edged_width/2 + (16*x), |
966 |
dec->refn[0].y + (16 * y + |
dec->edged_width/2); |
967 |
8) * dec->edged_width + |
} |
968 |
(16 * x), dec->edged_width); |
else |
969 |
|
{ |
970 |
transfer8x8_copy(dec->cur.y + (16 * y + 8) * dec->edged_width + |
transfer16x16_copy(dec->cur.y + (16*y)*dec->edged_width + (16*x), |
971 |
(16 * x + 8), |
dec->refn[0].y + (16*y)*dec->edged_width + (16*x), |
972 |
dec->refn[0].y + (16 * y + |
dec->edged_width); |
973 |
8) * dec->edged_width + |
|
974 |
(16 * x + 8), dec->edged_width); |
transfer8x8_copy(dec->cur.u + (8*y)*dec->edged_width/2 + (8*x), |
975 |
|
dec->refn[0].u + (8*y)*dec->edged_width/2 + (8*x), |
|
transfer8x8_copy(dec->cur.u + (8 * y) * dec->edged_width / 2 + |
|
|
(8 * x), |
|
|
dec->refn[0].u + |
|
|
(8 * y) * dec->edged_width / 2 + (8 * x), |
|
976 |
dec->edged_width / 2); |
dec->edged_width / 2); |
977 |
|
|
978 |
transfer8x8_copy(dec->cur.v + (8 * y) * dec->edged_width / 2 + |
transfer8x8_copy(dec->cur.v + (8*y)*dec->edged_width/2 + (8*x), |
979 |
(8 * x), |
dec->refn[0].v + (8*y)*dec->edged_width/2 + (8*x), |
|
dec->refn[0].v + |
|
|
(8 * y) * dec->edged_width / 2 + (8 * x), |
|
980 |
dec->edged_width / 2); |
dec->edged_width / 2); |
981 |
|
} |
982 |
|
|
983 |
stop_transfer_timer(); |
stop_transfer_timer(); |
984 |
|
|
1073 |
uv_dx = pMB->mvs[0].x; |
uv_dx = pMB->mvs[0].x; |
1074 |
uv_dy = pMB->mvs[0].y; |
uv_dy = pMB->mvs[0].y; |
1075 |
|
|
1076 |
uv_dx = (uv_dx & 3) ? (uv_dx >> 1) | 1 : uv_dx / 2; |
if (dec->quarterpel) |
1077 |
uv_dy = (uv_dy & 3) ? (uv_dy >> 1) | 1 : uv_dy / 2; |
{ |
1078 |
|
uv_dx /= 2; |
1079 |
|
uv_dy /= 2; |
1080 |
|
} |
1081 |
|
|
1082 |
|
uv_dx = (uv_dx >> 1) + roundtab_79[uv_dx & 0x3]; |
1083 |
|
uv_dy = (uv_dy >> 1) + roundtab_79[uv_dy & 0x3]; |
1084 |
} else { |
} else { |
1085 |
int sum; |
int sum; |
1086 |
|
|
1087 |
|
if(dec->quarterpel) |
1088 |
|
sum = (pMB->mvs[0].x / 2) + (pMB->mvs[1].x / 2) + (pMB->mvs[2].x / 2) + (pMB->mvs[3].x / 2); |
1089 |
|
else |
1090 |
sum = pMB->mvs[0].x + pMB->mvs[1].x + pMB->mvs[2].x + pMB->mvs[3].x; |
sum = pMB->mvs[0].x + pMB->mvs[1].x + pMB->mvs[2].x + pMB->mvs[3].x; |
|
uv_dx = |
|
|
(sum == |
|
|
0 ? 0 : SIGN(sum) * (roundtab[ABS(sum) % 16] + |
|
|
(ABS(sum) / 16) * 2)); |
|
1091 |
|
|
1092 |
|
uv_dx = (sum >> 3) + roundtab_76[sum & 0xf]; |
1093 |
|
|
1094 |
|
if(dec->quarterpel) |
1095 |
|
sum = (pMB->mvs[0].y / 2) + (pMB->mvs[1].y / 2) + (pMB->mvs[2].y / 2) + (pMB->mvs[3].y / 2); |
1096 |
|
else |
1097 |
sum = pMB->mvs[0].y + pMB->mvs[1].y + pMB->mvs[2].y + pMB->mvs[3].y; |
sum = pMB->mvs[0].y + pMB->mvs[1].y + pMB->mvs[2].y + pMB->mvs[3].y; |
1098 |
uv_dy = |
|
1099 |
(sum == |
uv_dy = (sum >> 3) + roundtab_76[sum & 0xf]; |
|
0 ? 0 : SIGN(sum) * (roundtab[ABS(sum) % 16] + |
|
|
(ABS(sum) / 16) * 2)); |
|
1100 |
} |
} |
1101 |
|
|
1102 |
start_timer(); |
start_timer(); |
1103 |
|
if(dec->quarterpel) { |
1104 |
|
interpolate16x16_quarterpel(dec->cur.y, dec->refn[ref].y, dec->refh.y, dec->refh.y + 64, |
1105 |
|
dec->refh.y + 128, 16*x_pos, 16*y_pos, |
1106 |
|
pMB->mvs[0].x, pMB->mvs[0].y, stride, 0); |
1107 |
|
} |
1108 |
|
else { |
1109 |
interpolate8x8_switch(dec->cur.y, dec->refn[ref].y, 16 * x_pos, 16 * y_pos, |
interpolate8x8_switch(dec->cur.y, dec->refn[ref].y, 16 * x_pos, 16 * y_pos, |
1110 |
pMB->mvs[0].x, pMB->mvs[0].y, stride, 0); |
pMB->mvs[0].x, pMB->mvs[0].y, stride, 0); |
1111 |
interpolate8x8_switch(dec->cur.y, dec->refn[ref].y, 16 * x_pos + 8, |
interpolate8x8_switch(dec->cur.y, dec->refn[ref].y, 16*x_pos + 8, 16*y_pos, |
1112 |
16 * y_pos, pMB->mvs[1].x, pMB->mvs[1].y, stride, 0); |
pMB->mvs[1].x, pMB->mvs[1].y, stride, 0); |
1113 |
interpolate8x8_switch(dec->cur.y, dec->refn[ref].y, 16 * x_pos, |
interpolate8x8_switch(dec->cur.y, dec->refn[ref].y, 16*x_pos, 16*y_pos + 8, |
1114 |
16 * y_pos + 8, pMB->mvs[2].x, pMB->mvs[2].y, stride, |
pMB->mvs[2].x, pMB->mvs[2].y, stride, 0); |
1115 |
0); |
interpolate8x8_switch(dec->cur.y, dec->refn[ref].y, 16*x_pos + 8, 16*y_pos + 8, |
1116 |
interpolate8x8_switch(dec->cur.y, dec->refn[ref].y, 16 * x_pos + 8, |
pMB->mvs[3].x, pMB->mvs[3].y, stride, 0); |
1117 |
16 * y_pos + 8, pMB->mvs[3].x, pMB->mvs[3].y, stride, |
} |
1118 |
0); |
|
1119 |
interpolate8x8_switch(dec->cur.u, dec->refn[ref].u, 8 * x_pos, 8 * y_pos, |
interpolate8x8_switch(dec->cur.u, dec->refn[ref].u, 8 * x_pos, 8 * y_pos, |
1120 |
uv_dx, uv_dy, stride2, 0); |
uv_dx, uv_dy, stride2, 0); |
1121 |
interpolate8x8_switch(dec->cur.v, dec->refn[ref].v, 8 * x_pos, 8 * y_pos, |
interpolate8x8_switch(dec->cur.v, dec->refn[ref].v, 8 * x_pos, 8 * y_pos, |
1202 |
uv_dx = pMB->mvs[0].x; |
uv_dx = pMB->mvs[0].x; |
1203 |
uv_dy = pMB->mvs[0].y; |
uv_dy = pMB->mvs[0].y; |
1204 |
|
|
|
uv_dx = (uv_dx & 3) ? (uv_dx >> 1) | 1 : uv_dx / 2; |
|
|
uv_dy = (uv_dy & 3) ? (uv_dy >> 1) | 1 : uv_dy / 2; |
|
|
|
|
1205 |
b_uv_dx = pMB->b_mvs[0].x; |
b_uv_dx = pMB->b_mvs[0].x; |
1206 |
b_uv_dy = pMB->b_mvs[0].y; |
b_uv_dy = pMB->b_mvs[0].y; |
1207 |
|
|
1208 |
b_uv_dx = (uv_dx & 3) ? (uv_dx >> 1) | 1 : uv_dx / 2; |
if (dec->quarterpel) |
1209 |
b_uv_dy = (uv_dy & 3) ? (uv_dy >> 1) | 1 : uv_dy / 2; |
{ |
1210 |
|
uv_dx /= 2; |
1211 |
|
uv_dy /= 2; |
1212 |
|
|
1213 |
|
b_uv_dx /= 2; |
1214 |
|
b_uv_dy /= 2; |
1215 |
|
} |
1216 |
|
|
1217 |
|
uv_dx = (uv_dx >> 1) + roundtab_79[uv_dx & 0x3]; |
1218 |
|
uv_dy = (uv_dy >> 1) + roundtab_79[uv_dy & 0x3]; |
1219 |
|
|
1220 |
|
b_uv_dx = (b_uv_dx >> 1) + roundtab_79[b_uv_dx & 0x3]; |
1221 |
|
b_uv_dy = (b_uv_dy >> 1) + roundtab_79[b_uv_dy & 0x3]; |
1222 |
} else { |
} else { |
1223 |
int sum; |
int sum; |
1224 |
|
|
1225 |
|
if(dec->quarterpel) |
1226 |
|
sum = (pMB->mvs[0].x / 2) + (pMB->mvs[1].x / 2) + (pMB->mvs[2].x / 2) + (pMB->mvs[3].x / 2); |
1227 |
|
else |
1228 |
sum = pMB->mvs[0].x + pMB->mvs[1].x + pMB->mvs[2].x + pMB->mvs[3].x; |
sum = pMB->mvs[0].x + pMB->mvs[1].x + pMB->mvs[2].x + pMB->mvs[3].x; |
|
uv_dx = |
|
|
(sum == |
|
|
0 ? 0 : SIGN(sum) * (roundtab[ABS(sum) % 16] + |
|
|
(ABS(sum) / 16) * 2)); |
|
1229 |
|
|
1230 |
|
uv_dx = (sum >> 3) + roundtab_76[sum & 0xf]; |
1231 |
|
|
1232 |
|
if(dec->quarterpel) |
1233 |
|
sum = (pMB->mvs[0].y / 2) + (pMB->mvs[1].y / 2) + (pMB->mvs[2].y / 2) + (pMB->mvs[3].y / 2); |
1234 |
|
else |
1235 |
sum = pMB->mvs[0].y + pMB->mvs[1].y + pMB->mvs[2].y + pMB->mvs[3].y; |
sum = pMB->mvs[0].y + pMB->mvs[1].y + pMB->mvs[2].y + pMB->mvs[3].y; |
1236 |
uv_dy = |
|
1237 |
(sum == |
uv_dy = (sum >> 3) + roundtab_76[sum & 0xf]; |
1238 |
0 ? 0 : SIGN(sum) * (roundtab[ABS(sum) % 16] + |
|
1239 |
(ABS(sum) / 16) * 2)); |
|
1240 |
|
if(dec->quarterpel) |
1241 |
sum = |
sum = (pMB->b_mvs[0].x / 2) + (pMB->b_mvs[1].x / 2) + (pMB->b_mvs[2].x / 2) + (pMB->b_mvs[3].x / 2); |
1242 |
pMB->b_mvs[0].x + pMB->b_mvs[1].x + pMB->b_mvs[2].x + |
else |
1243 |
pMB->b_mvs[3].x; |
sum = pMB->b_mvs[0].x + pMB->b_mvs[1].x + pMB->b_mvs[2].x + pMB->b_mvs[3].x; |
1244 |
b_uv_dx = |
|
1245 |
(sum == |
b_uv_dx = (sum >> 3) + roundtab_76[sum & 0xf]; |
1246 |
0 ? 0 : SIGN(sum) * (roundtab[ABS(sum) % 16] + |
|
1247 |
(ABS(sum) / 16) * 2)); |
if(dec->quarterpel) |
1248 |
|
sum = (pMB->b_mvs[0].y / 2) + (pMB->b_mvs[1].y / 2) + (pMB->b_mvs[2].y / 2) + (pMB->b_mvs[3].y / 2); |
1249 |
sum = |
else |
1250 |
pMB->b_mvs[0].y + pMB->b_mvs[1].y + pMB->b_mvs[2].y + |
sum = pMB->b_mvs[0].y + pMB->b_mvs[1].y + pMB->b_mvs[2].y + pMB->b_mvs[3].y; |
1251 |
pMB->b_mvs[3].y; |
|
1252 |
b_uv_dy = |
b_uv_dy = (sum >> 3) + roundtab_76[sum & 0xf]; |
|
(sum == |
|
|
0 ? 0 : SIGN(sum) * (roundtab[ABS(sum) % 16] + |
|
|
(ABS(sum) / 16) * 2)); |
|
1253 |
} |
} |
1254 |
|
|
1255 |
|
|
1256 |
start_timer(); |
start_timer(); |
1257 |
|
if(dec->quarterpel) { |
1258 |
|
if((pMB->mode == MODE_INTER || pMB->mode == MODE_INTER_Q)) |
1259 |
|
interpolate16x16_quarterpel(dec->cur.y, forward.y, dec->refh.y, dec->refh.y + 64, |
1260 |
|
dec->refh.y + 128, 16*x_pos, 16*y_pos, |
1261 |
|
pMB->mvs[0].x, pMB->mvs[0].y, stride, 0); |
1262 |
|
else { |
1263 |
|
interpolate8x8_quarterpel(dec->cur.y, forward.y, dec->refh.y, dec->refh.y + 64, |
1264 |
|
dec->refh.y + 128, 16*x_pos, 16*y_pos, |
1265 |
|
pMB->mvs[0].x, pMB->mvs[0].y, stride, 0); |
1266 |
|
interpolate8x8_quarterpel(dec->cur.y, forward.y, dec->refh.y, dec->refh.y + 64, |
1267 |
|
dec->refh.y + 128, 16*x_pos + 8, 16*y_pos, |
1268 |
|
pMB->mvs[1].x, pMB->mvs[1].y, stride, 0); |
1269 |
|
interpolate8x8_quarterpel(dec->cur.y, forward.y, dec->refh.y, dec->refh.y + 64, |
1270 |
|
dec->refh.y + 128, 16*x_pos, 16*y_pos + 8, |
1271 |
|
pMB->mvs[2].x, pMB->mvs[2].y, stride, 0); |
1272 |
|
interpolate8x8_quarterpel(dec->cur.y, forward.y, dec->refh.y, dec->refh.y + 64, |
1273 |
|
dec->refh.y + 128, 16*x_pos + 8, 16*y_pos + 8, |
1274 |
|
pMB->mvs[3].x, pMB->mvs[3].y, stride, 0); |
1275 |
|
} |
1276 |
|
} |
1277 |
|
else { |
1278 |
interpolate8x8_switch(dec->cur.y, forward.y, 16 * x_pos, 16 * y_pos, |
interpolate8x8_switch(dec->cur.y, forward.y, 16 * x_pos, 16 * y_pos, |
1279 |
pMB->mvs[0].x, pMB->mvs[0].y, stride, 0); |
pMB->mvs[0].x, pMB->mvs[0].y, stride, 0); |
1280 |
interpolate8x8_switch(dec->cur.y, forward.y, 16 * x_pos + 8, 16 * y_pos, |
interpolate8x8_switch(dec->cur.y, forward.y, 16 * x_pos + 8, 16 * y_pos, |
1284 |
interpolate8x8_switch(dec->cur.y, forward.y, 16 * x_pos + 8, |
interpolate8x8_switch(dec->cur.y, forward.y, 16 * x_pos + 8, |
1285 |
16 * y_pos + 8, pMB->mvs[3].x, pMB->mvs[3].y, stride, |
16 * y_pos + 8, pMB->mvs[3].x, pMB->mvs[3].y, stride, |
1286 |
0); |
0); |
1287 |
|
} |
1288 |
|
|
1289 |
interpolate8x8_switch(dec->cur.u, forward.u, 8 * x_pos, 8 * y_pos, uv_dx, |
interpolate8x8_switch(dec->cur.u, forward.u, 8 * x_pos, 8 * y_pos, uv_dx, |
1290 |
uv_dy, stride2, 0); |
uv_dy, stride2, 0); |
1291 |
interpolate8x8_switch(dec->cur.v, forward.v, 8 * x_pos, 8 * y_pos, uv_dx, |
interpolate8x8_switch(dec->cur.v, forward.v, 8 * x_pos, 8 * y_pos, uv_dx, |
1292 |
uv_dy, stride2, 0); |
uv_dy, stride2, 0); |
1293 |
|
|
1294 |
|
|
1295 |
|
if(dec->quarterpel) { |
1296 |
|
if((pMB->mode == MODE_INTER || pMB->mode == MODE_INTER_Q)) |
1297 |
|
interpolate16x16_quarterpel(dec->refn[2].y, backward.y, dec->refh.y, dec->refh.y + 64, |
1298 |
|
dec->refh.y + 128, 16*x_pos, 16*y_pos, |
1299 |
|
pMB->b_mvs[0].x, pMB->b_mvs[0].y, stride, 0); |
1300 |
|
else { |
1301 |
|
interpolate8x8_quarterpel(dec->refn[2].y, backward.y, dec->refh.y, dec->refh.y + 64, |
1302 |
|
dec->refh.y + 128, 16*x_pos, 16*y_pos, |
1303 |
|
pMB->b_mvs[0].x, pMB->b_mvs[0].y, stride, 0); |
1304 |
|
interpolate8x8_quarterpel(dec->refn[2].y, backward.y, dec->refh.y, dec->refh.y + 64, |
1305 |
|
dec->refh.y + 128, 16*x_pos + 8, 16*y_pos, |
1306 |
|
pMB->b_mvs[1].x, pMB->b_mvs[1].y, stride, 0); |
1307 |
|
interpolate8x8_quarterpel(dec->refn[2].y, backward.y, dec->refh.y, dec->refh.y + 64, |
1308 |
|
dec->refh.y + 128, 16*x_pos, 16*y_pos + 8, |
1309 |
|
pMB->b_mvs[2].x, pMB->b_mvs[2].y, stride, 0); |
1310 |
|
interpolate8x8_quarterpel(dec->refn[2].y, backward.y, dec->refh.y, dec->refh.y + 64, |
1311 |
|
dec->refh.y + 128, 16*x_pos + 8, 16*y_pos + 8, |
1312 |
|
pMB->b_mvs[3].x, pMB->b_mvs[3].y, stride, 0); |
1313 |
|
} |
1314 |
|
} |
1315 |
|
else { |
1316 |
interpolate8x8_switch(dec->refn[2].y, backward.y, 16 * x_pos, 16 * y_pos, |
interpolate8x8_switch(dec->refn[2].y, backward.y, 16 * x_pos, 16 * y_pos, |
1317 |
pMB->b_mvs[0].x, pMB->b_mvs[0].y, stride, 0); |
pMB->b_mvs[0].x, pMB->b_mvs[0].y, stride, 0); |
1318 |
interpolate8x8_switch(dec->refn[2].y, backward.y, 16 * x_pos + 8, |
interpolate8x8_switch(dec->refn[2].y, backward.y, 16 * x_pos + 8, |
1324 |
interpolate8x8_switch(dec->refn[2].y, backward.y, 16 * x_pos + 8, |
interpolate8x8_switch(dec->refn[2].y, backward.y, 16 * x_pos + 8, |
1325 |
16 * y_pos + 8, pMB->b_mvs[3].x, pMB->b_mvs[3].y, |
16 * y_pos + 8, pMB->b_mvs[3].x, pMB->b_mvs[3].y, |
1326 |
stride, 0); |
stride, 0); |
1327 |
|
} |
1328 |
|
|
1329 |
interpolate8x8_switch(dec->refn[2].u, backward.u, 8 * x_pos, 8 * y_pos, |
interpolate8x8_switch(dec->refn[2].u, backward.u, 8 * x_pos, 8 * y_pos, |
1330 |
b_uv_dx, b_uv_dy, stride2, 0); |
b_uv_dx, b_uv_dy, stride2, 0); |
1331 |
interpolate8x8_switch(dec->refn[2].v, backward.v, 8 * x_pos, 8 * y_pos, |
interpolate8x8_switch(dec->refn[2].v, backward.v, 8 * x_pos, 8 * y_pos, |
1334 |
interpolate8x8_avg2(dec->cur.y + (16 * y_pos * stride) + 16 * x_pos, |
interpolate8x8_avg2(dec->cur.y + (16 * y_pos * stride) + 16 * x_pos, |
1335 |
dec->cur.y + (16 * y_pos * stride) + 16 * x_pos, |
dec->cur.y + (16 * y_pos * stride) + 16 * x_pos, |
1336 |
dec->refn[2].y + (16 * y_pos * stride) + 16 * x_pos, |
dec->refn[2].y + (16 * y_pos * stride) + 16 * x_pos, |
1337 |
stride, 0); |
stride, 1, 8); |
1338 |
|
|
1339 |
interpolate8x8_avg2(dec->cur.y + (16 * y_pos * stride) + 16 * x_pos + 8, |
interpolate8x8_avg2(dec->cur.y + (16 * y_pos * stride) + 16 * x_pos + 8, |
1340 |
dec->cur.y + (16 * y_pos * stride) + 16 * x_pos + 8, |
dec->cur.y + (16 * y_pos * stride) + 16 * x_pos + 8, |
1341 |
dec->refn[2].y + (16 * y_pos * stride) + 16 * x_pos + 8, |
dec->refn[2].y + (16 * y_pos * stride) + 16 * x_pos + 8, |
1342 |
stride, 0); |
stride, 1, 8); |
1343 |
|
|
1344 |
interpolate8x8_avg2(dec->cur.y + ((16 * y_pos + 8) * stride) + 16 * x_pos, |
interpolate8x8_avg2(dec->cur.y + ((16 * y_pos + 8) * stride) + 16 * x_pos, |
1345 |
dec->cur.y + ((16 * y_pos + 8) * stride) + 16 * x_pos, |
dec->cur.y + ((16 * y_pos + 8) * stride) + 16 * x_pos, |
1346 |
dec->refn[2].y + ((16 * y_pos + 8) * stride) + 16 * x_pos, |
dec->refn[2].y + ((16 * y_pos + 8) * stride) + 16 * x_pos, |
1347 |
stride, 0); |
stride, 1, 8); |
1348 |
|
|
1349 |
interpolate8x8_avg2(dec->cur.y + ((16 * y_pos + 8) * stride) + 16 * x_pos + 8, |
interpolate8x8_avg2(dec->cur.y + ((16 * y_pos + 8) * stride) + 16 * x_pos + 8, |
1350 |
dec->cur.y + ((16 * y_pos + 8) * stride) + 16 * x_pos + 8, |
dec->cur.y + ((16 * y_pos + 8) * stride) + 16 * x_pos + 8, |
1351 |
dec->refn[2].y + ((16 * y_pos + 8) * stride) + 16 * x_pos + 8, |
dec->refn[2].y + ((16 * y_pos + 8) * stride) + 16 * x_pos + 8, |
1352 |
stride, 0); |
stride, 1, 8); |
1353 |
|
|
1354 |
interpolate8x8_avg2(dec->cur.u + (8 * y_pos * stride2) + 8 * x_pos, |
interpolate8x8_avg2(dec->cur.u + (8 * y_pos * stride2) + 8 * x_pos, |
1355 |
dec->cur.u + (8 * y_pos * stride2) + 8 * x_pos, |
dec->cur.u + (8 * y_pos * stride2) + 8 * x_pos, |
1356 |
dec->refn[2].u + (8 * y_pos * stride2) + 8 * x_pos, |
dec->refn[2].u + (8 * y_pos * stride2) + 8 * x_pos, |
1357 |
stride2, 0); |
stride2, 1, 8); |
1358 |
|
|
1359 |
interpolate8x8_avg2(dec->cur.v + (8 * y_pos * stride2) + 8 * x_pos, |
interpolate8x8_avg2(dec->cur.v + (8 * y_pos * stride2) + 8 * x_pos, |
1360 |
dec->cur.v + (8 * y_pos * stride2) + 8 * x_pos, |
dec->cur.v + (8 * y_pos * stride2) + 8 * x_pos, |
1361 |
dec->refn[2].v + (8 * y_pos * stride2) + 8 * x_pos, |
dec->refn[2].v + (8 * y_pos * stride2) + 8 * x_pos, |
1362 |
stride2, 0); |
stride2, 1, 8); |
1363 |
|
|
1364 |
stop_comp_timer(); |
stop_comp_timer(); |
1365 |
|
|
1671 |
BitstreamReadHeaders(&bs, dec, &rounding, &reduced_resolution, |
BitstreamReadHeaders(&bs, dec, &rounding, &reduced_resolution, |
1672 |
&quant, &fcode_forward, &fcode_backward, &intra_dc_threshold, gmc_mv); |
&quant, &fcode_forward, &fcode_backward, &intra_dc_threshold, gmc_mv); |
1673 |
|
|
1674 |
DPRINTF(DPRINTF_HEADER, "vop_type=%i", vop_type); |
//DPRINTF(DPRINTF_HEADER, "vop_type=%i", vop_type); |
1675 |
|
|
1676 |
if (vop_type == -1 && success) |
if (vop_type == -1 && success) |
1677 |
goto done; |
goto done; |
1750 |
return XVID_ERR_FAIL; |
return XVID_ERR_FAIL; |
1751 |
} |
} |
1752 |
|
|
1753 |
|
|
1754 |
|
/* reduced resolution deblocking filter */ |
1755 |
|
|
1756 |
|
if (reduced_resolution) |
1757 |
|
{ |
1758 |
|
const int rmb_width = (dec->width + 31) / 32; |
1759 |
|
const int rmb_height = (dec->height + 31) / 32; |
1760 |
|
const int edged_width2 = dec->edged_width /2; |
1761 |
|
int i,j; |
1762 |
|
|
1763 |
|
/* horizontal deblocking */ |
1764 |
|
|
1765 |
|
for (j = 1; j < rmb_height*2; j++) // luma: j,i in block units |
1766 |
|
for (i = 0; i < rmb_width*2; i++) |
1767 |
|
{ |
1768 |
|
if (dec->mbs[(j-1)/2*dec->mb_width + (i/2)].mode != MODE_NOT_CODED || |
1769 |
|
dec->mbs[(j+0)/2*dec->mb_width + (i/2)].mode != MODE_NOT_CODED) |
1770 |
|
{ |
1771 |
|
xvid_HFilter_31_C(dec->cur.y + (j*16 - 1)*dec->edged_width + i*16, |
1772 |
|
dec->cur.y + (j*16 + 0)*dec->edged_width + i*16, 2); |
1773 |
|
} |
1774 |
|
} |
1775 |
|
|
1776 |
|
for (j = 1; j < rmb_height; j++) // chroma |
1777 |
|
for (i = 0; i < rmb_width; i++) |
1778 |
|
{ |
1779 |
|
if (dec->mbs[(j-1)*dec->mb_width + i].mode != MODE_NOT_CODED || |
1780 |
|
dec->mbs[(j+0)*dec->mb_width + i].mode != MODE_NOT_CODED) |
1781 |
|
{ |
1782 |
|
hfilter_31(dec->cur.u + (j*16 - 1)*edged_width2 + i*16, |
1783 |
|
dec->cur.u + (j*16 + 0)*edged_width2 + i*16, 2); |
1784 |
|
hfilter_31(dec->cur.v + (j*16 - 1)*edged_width2 + i*16, |
1785 |
|
dec->cur.v + (j*16 + 0)*edged_width2 + i*16, 2); |
1786 |
|
} |
1787 |
|
} |
1788 |
|
|
1789 |
|
/* vertical deblocking */ |
1790 |
|
|
1791 |
|
for (j = 0; j < rmb_height*2; j++) // luma: i,j in block units |
1792 |
|
for (i = 1; i < rmb_width*2; i++) |
1793 |
|
{ |
1794 |
|
if (dec->mbs[(j/2)*dec->mb_width + (i-1)/2].mode != MODE_NOT_CODED || |
1795 |
|
dec->mbs[(j/2)*dec->mb_width + (i+0)/2].mode != MODE_NOT_CODED) |
1796 |
|
{ |
1797 |
|
vfilter_31(dec->cur.y + (j*16)*dec->edged_width + i*16 - 1, |
1798 |
|
dec->cur.y + (j*16)*dec->edged_width + i*16 + 0, |
1799 |
|
dec->edged_width, 2); |
1800 |
|
} |
1801 |
|
} |
1802 |
|
|
1803 |
|
for (j = 0; j < rmb_height; j++) // chroma |
1804 |
|
for (i = 1; i < rmb_width; i++) |
1805 |
|
{ |
1806 |
|
if (dec->mbs[j*dec->mb_width + i - 1].mode != MODE_NOT_CODED || |
1807 |
|
dec->mbs[j*dec->mb_width + i + 0].mode != MODE_NOT_CODED) |
1808 |
|
{ |
1809 |
|
vfilter_31(dec->cur.u + (j*16)*edged_width2 + i*16 - 1, |
1810 |
|
dec->cur.u + (j*16)*edged_width2 + i*16 + 0, |
1811 |
|
edged_width2, 2); |
1812 |
|
vfilter_31(dec->cur.v + (j*16)*edged_width2 + i*16 - 1, |
1813 |
|
dec->cur.v + (j*16)*edged_width2 + i*16 + 0, |
1814 |
|
edged_width2, 2); |
1815 |
|
} |
1816 |
|
} |
1817 |
|
} |
1818 |
|
|
1819 |
BitstreamByteAlign(&bs); |
BitstreamByteAlign(&bs); |
1820 |
|
|
1821 |
#ifdef BFRAMES_DEC |
#ifdef BFRAMES_DEC |
1859 |
mb_swap(&dec->mbs, &dec->last_mbs); |
mb_swap(&dec->mbs, &dec->last_mbs); |
1860 |
} |
} |
1861 |
|
|
|
success = 1; |
|
1862 |
|
|
1863 |
if (frame->length > BitstreamPos(&bs) / 8) // multiple vops packed together |
if (success == 0 && dec->packed_mode) |
1864 |
|
{ |
1865 |
|
success = 1; |
1866 |
|
// if (frame->length > BitstreamPos(&bs) / 8) // multiple vops packed together |
1867 |
goto start; |
goto start; |
1868 |
|
} |
1869 |
|
|
1870 |
done : |
done : |
1871 |
|
|