4 |
* - Decoder Module - |
* - Decoder Module - |
5 |
* |
* |
6 |
* Copyright(C) 2002 MinChen <chenm001@163.com> |
* Copyright(C) 2002 MinChen <chenm001@163.com> |
7 |
* 2002-2003 Peter Ross <pross@xvid.org> |
* 2002-2004 Peter Ross <pross@xvid.org> |
8 |
* |
* |
9 |
* This program is free software ; you can redistribute it and/or modify |
* This program is free software ; you can redistribute it and/or modify |
10 |
* 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 |
20 |
* along with this program ; if not, write to the Free Software |
* along with this program ; if not, write to the Free Software |
21 |
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
22 |
* |
* |
23 |
* $Id: decoder.c,v 1.51 2004-03-22 22:36:23 edgomez Exp $ |
* $Id: decoder.c,v 1.67 2004-08-16 22:38:06 edgomez Exp $ |
24 |
* |
* |
25 |
****************************************************************************/ |
****************************************************************************/ |
26 |
|
|
48 |
#include "image/interpolate8x8.h" |
#include "image/interpolate8x8.h" |
49 |
#include "image/reduced.h" |
#include "image/reduced.h" |
50 |
#include "image/font.h" |
#include "image/font.h" |
51 |
|
#include "image/qpel.h" |
52 |
|
|
53 |
#include "bitstream/mbcoding.h" |
#include "bitstream/mbcoding.h" |
54 |
#include "prediction/mbprediction.h" |
#include "prediction/mbprediction.h" |
223 |
dec->time = dec->time_base = dec->last_time_base = 0; |
dec->time = dec->time_base = dec->last_time_base = 0; |
224 |
dec->low_delay = 0; |
dec->low_delay = 0; |
225 |
dec->packed_mode = 0; |
dec->packed_mode = 0; |
226 |
|
dec->time_inc_resolution = 1; /* until VOL header says otherwise */ |
227 |
|
|
228 |
dec->fixed_dimensions = (dec->width > 0 && dec->height > 0); |
dec->fixed_dimensions = (dec->width > 0 && dec->height > 0); |
229 |
|
|
340 |
stop_coding_timer(); |
stop_coding_timer(); |
341 |
|
|
342 |
start_timer(); |
start_timer(); |
343 |
add_acdc(pMB, i, &block[i * 64], iDcScaler, predictors); |
add_acdc(pMB, i, &block[i * 64], iDcScaler, predictors, dec->bs_version); |
344 |
stop_prediction_timer(); |
stop_prediction_timer(); |
345 |
|
|
346 |
start_timer(); |
start_timer(); |
391 |
uint8_t * pY_Cur, |
uint8_t * pY_Cur, |
392 |
uint8_t * pU_Cur, |
uint8_t * pU_Cur, |
393 |
uint8_t * pV_Cur, |
uint8_t * pV_Cur, |
394 |
const int reduced_resolution, |
int reduced_resolution, |
395 |
const MACROBLOCK * pMB) |
const MACROBLOCK * pMB) |
396 |
{ |
{ |
397 |
DECLARE_ALIGNED_MATRIX(block, 1, 64, int16_t, CACHE_LINE); |
DECLARE_ALIGNED_MATRIX(data, 1, 64, int16_t, CACHE_LINE); |
|
DECLARE_ALIGNED_MATRIX(data, 6, 64, int16_t, CACHE_LINE); |
|
398 |
|
|
399 |
int stride = dec->edged_width; |
int stride = dec->edged_width; |
400 |
int next_block = stride * (reduced_resolution ? 16 : 8); |
int next_block = stride * (reduced_resolution ? 16 : 8); |
|
const int stride2 = stride/2; |
|
401 |
int i; |
int i; |
402 |
const uint32_t iQuant = pMB->quant; |
const uint32_t iQuant = pMB->quant; |
403 |
const int direction = dec->alternate_vertical_scan ? 2 : 0; |
const int direction = dec->alternate_vertical_scan ? 2 : 0; |
404 |
const quant_interFuncPtr dequant = dec->quant_type == 0 ? dequant_h263_inter : dequant_mpeg_inter; |
typedef void (*get_inter_block_function_t)( |
405 |
|
Bitstream * bs, |
406 |
|
int16_t * block, |
407 |
|
int direction, |
408 |
|
const int quant, |
409 |
|
const uint16_t *matrix); |
410 |
|
typedef void (*add_residual_function_t)( |
411 |
|
uint8_t *predicted_block, |
412 |
|
const int16_t *residual, |
413 |
|
int stride); |
414 |
|
|
415 |
|
const get_inter_block_function_t get_inter_block = (dec->quant_type == 0) |
416 |
|
? (get_inter_block_function_t)get_inter_block_h263 |
417 |
|
: (get_inter_block_function_t)get_inter_block_mpeg; |
418 |
|
|
419 |
|
const add_residual_function_t add_residual = (reduced_resolution) |
420 |
|
? (add_residual_function_t)add_upsampled_8x8_16to8 |
421 |
|
: (add_residual_function_t)transfer_16to8add; |
422 |
|
|
423 |
for (i = 0; i < 6; i++) { |
uint8_t *dst[6]; |
424 |
|
int strides[6]; |
425 |
|
|
|
if (cbp & (1 << (5 - i))) { /* coded */ |
|
426 |
|
|
427 |
memset(block, 0, 64 * sizeof(int16_t)); /* clear */ |
if (dec->interlacing && pMB->field_dct) { |
428 |
|
next_block = stride; |
429 |
|
stride *= 2; |
430 |
|
} |
431 |
|
|
432 |
|
reduced_resolution = !!reduced_resolution; |
433 |
|
dst[0] = pY_Cur; |
434 |
|
dst[2] = pY_Cur + next_block; |
435 |
|
dst[1] = dst[0] + (8<<reduced_resolution); |
436 |
|
dst[3] = dst[2] + (8<<reduced_resolution); |
437 |
|
dst[4] = pU_Cur; |
438 |
|
dst[5] = pV_Cur; |
439 |
|
strides[0] = strides[1] = strides[2] = strides[3] = stride; |
440 |
|
strides[4] = stride/2; |
441 |
|
strides[5] = stride/2; |
442 |
|
|
443 |
|
for (i = 0; i < 6; i++) { |
444 |
|
/* Process only coded blocks */ |
445 |
|
if (cbp & (1 << (5 - i))) { |
446 |
|
|
447 |
|
/* Clear the block */ |
448 |
|
memset(&data[0], 0, 64*sizeof(int16_t)); |
449 |
|
|
450 |
|
/* Decode coeffs and dequantize on the fly */ |
451 |
start_timer(); |
start_timer(); |
452 |
get_inter_block(bs, block, direction); |
get_inter_block(bs, &data[0], direction, iQuant, get_inter_matrix(dec->mpeg_quant_matrices)); |
453 |
stop_coding_timer(); |
stop_coding_timer(); |
454 |
|
|
455 |
|
/* iDCT */ |
456 |
start_timer(); |
start_timer(); |
457 |
dequant(&data[i * 64], block, iQuant, dec->mpeg_quant_matrices); |
idct(&data[0]); |
458 |
stop_iquant_timer(); |
stop_idct_timer(); |
459 |
|
|
460 |
|
/* Add this residual to the predicted block */ |
461 |
start_timer(); |
start_timer(); |
462 |
idct(&data[i * 64]); |
add_residual(dst[i], &data[0], strides[i]); |
463 |
stop_idct_timer(); |
stop_transfer_timer(); |
464 |
} |
} |
465 |
} |
} |
|
|
|
|
if (dec->interlacing && pMB->field_dct) { |
|
|
next_block = stride; |
|
|
stride *= 2; |
|
466 |
} |
} |
467 |
|
|
468 |
start_timer(); |
static void __inline |
469 |
if (reduced_resolution) { |
validate_vector(VECTOR * mv, unsigned int x_pos, unsigned int y_pos, const DECODER * dec) |
470 |
if (cbp & 32) |
{ |
471 |
add_upsampled_8x8_16to8(pY_Cur, &data[0 * 64], stride); |
/* clip a vector to valid range |
472 |
if (cbp & 16) |
prevents crashes if bitstream is broken |
473 |
add_upsampled_8x8_16to8(pY_Cur + 16, &data[1 * 64], stride); |
*/ |
474 |
if (cbp & 8) |
int shift = 5 + dec->quarterpel; |
475 |
add_upsampled_8x8_16to8(pY_Cur + next_block, &data[2 * 64], stride); |
int xborder_high = (int)(dec->mb_width - x_pos) << shift; |
476 |
if (cbp & 4) |
int xborder_low = (-(int)x_pos-1) << shift; |
477 |
add_upsampled_8x8_16to8(pY_Cur + 16 + next_block, &data[3 * 64], stride); |
int yborder_high = (int)(dec->mb_height - y_pos) << shift; |
478 |
if (cbp & 2) |
int yborder_low = (-(int)y_pos-1) << shift; |
479 |
add_upsampled_8x8_16to8(pU_Cur, &data[4 * 64], stride2); |
|
480 |
if (cbp & 1) |
#define CHECK_MV(mv) \ |
481 |
add_upsampled_8x8_16to8(pV_Cur, &data[5 * 64], stride2); |
do { \ |
482 |
} else { |
if ((mv).x > xborder_high) { \ |
483 |
if (cbp & 32) |
DPRINTF(XVID_DEBUG_MV, "mv.x > max -- %d > %d, MB %d, %d", (mv).x, xborder_high, x_pos, y_pos); \ |
484 |
transfer_16to8add(pY_Cur, &data[0 * 64], stride); |
(mv).x = xborder_high; \ |
485 |
if (cbp & 16) |
} else if ((mv).x < xborder_low) { \ |
486 |
transfer_16to8add(pY_Cur + 8, &data[1 * 64], stride); |
DPRINTF(XVID_DEBUG_MV, "mv.x < min -- %d < %d, MB %d, %d", (mv).x, xborder_low, x_pos, y_pos); \ |
487 |
if (cbp & 8) |
(mv).x = xborder_low; \ |
488 |
transfer_16to8add(pY_Cur + next_block, &data[2 * 64], stride); |
} \ |
489 |
if (cbp & 4) |
if ((mv).y > yborder_high) { \ |
490 |
transfer_16to8add(pY_Cur + 8 + next_block, &data[3 * 64], stride); |
DPRINTF(XVID_DEBUG_MV, "mv.y > max -- %d > %d, MB %d, %d", (mv).y, yborder_high, x_pos, y_pos); \ |
491 |
if (cbp & 2) |
(mv).y = yborder_high; \ |
492 |
transfer_16to8add(pU_Cur, &data[4 * 64], stride2); |
} else if ((mv).y < yborder_low) { \ |
493 |
if (cbp & 1) |
DPRINTF(XVID_DEBUG_MV, "mv.y < min -- %d < %d, MB %d, %d", (mv).y, yborder_low, x_pos, y_pos); \ |
494 |
transfer_16to8add(pV_Cur, &data[5 * 64], stride2); |
(mv).y = yborder_low; \ |
495 |
} |
} \ |
496 |
stop_transfer_timer(); |
} while (0) |
497 |
|
|
498 |
|
CHECK_MV(mv[0]); |
499 |
|
CHECK_MV(mv[1]); |
500 |
|
CHECK_MV(mv[2]); |
501 |
|
CHECK_MV(mv[3]); |
502 |
} |
} |
503 |
|
|
504 |
/* decode an inter macroblock */ |
/* decode an inter macroblock */ |
538 |
mv[i] = pMB->mvs[i]; |
mv[i] = pMB->mvs[i]; |
539 |
} |
} |
540 |
|
|
541 |
|
validate_vector(mv, x_pos, y_pos, dec); |
542 |
|
|
543 |
start_timer(); |
start_timer(); |
544 |
|
|
545 |
if (pMB->mode != MODE_INTER4V) { /* INTER, INTER_Q, NOT_CODED, FORWARD, BACKWARD */ |
if (pMB->mode != MODE_INTER4V) { /* INTER, INTER_Q, NOT_CODED, FORWARD, BACKWARD */ |
805 |
mv.y -= range; |
mv.y -= range; |
806 |
} |
} |
807 |
|
|
|
/* clip to valid range */ |
|
|
|
|
|
if (mv.x > ((int)(dec->mb_width - x) << (5 + dec->quarterpel)) ) |
|
|
mv.x = (int)(dec->mb_width - x) << (5 + dec->quarterpel); |
|
|
|
|
|
else if (mv.x < (int)(-x-1) << (5 + dec->quarterpel)) |
|
|
mv.x = (int)(-x-1) << (5 + dec->quarterpel); |
|
|
|
|
|
if (mv.y > ((int)(dec->mb_height - y) << (5 + dec->quarterpel)) ) |
|
|
mv.y = (int)(dec->mb_height - y) << (5 + dec->quarterpel); |
|
|
|
|
|
else if (mv.y < ((int)(-y-1)) << (5 + dec->quarterpel) ) |
|
|
mv.y = (int)(-y-1) << (5 + dec->quarterpel); |
|
|
|
|
808 |
ret_mv->x = mv.x; |
ret_mv->x = mv.x; |
809 |
ret_mv->y = mv.y; |
ret_mv->y = mv.y; |
810 |
} |
} |
831 |
mb_height = (dec->height + 31) / 32; |
mb_height = (dec->height + 31) / 32; |
832 |
} |
} |
833 |
|
|
834 |
|
if (!dec->is_edged[0]) { |
835 |
start_timer(); |
start_timer(); |
836 |
image_setedges(&dec->refn[0], dec->edged_width, dec->edged_height, |
image_setedges(&dec->refn[0], dec->edged_width, dec->edged_height, |
837 |
dec->width, dec->height, dec->bs_version); |
dec->width, dec->height, dec->bs_version); |
838 |
|
dec->is_edged[0] = 1; |
839 |
stop_edges_timer(); |
stop_edges_timer(); |
840 |
|
} |
841 |
|
|
842 |
if (gmc_warp) { |
if (gmc_warp) { |
843 |
/* accuracy: 0==1/2, 1=1/4, 2=1/8, 3=1/16 */ |
/* accuracy: 0==1/2, 1=1/4, 2=1/8, 3=1/16 */ |
1019 |
else if (mv_y > high) |
else if (mv_y > high) |
1020 |
mv_y -= range; |
mv_y -= range; |
1021 |
|
|
|
|
|
|
/* clip to valid range */ |
|
|
if (mv_x > ((int)(dec->mb_width - x) << (5 + dec->quarterpel)) ) |
|
|
mv_x = (int)(dec->mb_width - x) << (5 + dec->quarterpel); |
|
|
|
|
|
else if (mv_x < (int)(-x-1) << (5 + dec->quarterpel)) |
|
|
mv_x = (int)(-x-1) << (5 + dec->quarterpel); |
|
|
|
|
|
if (mv_y > ((int)(dec->mb_height - y) << (5 + dec->quarterpel)) ) |
|
|
mv_y = (int)(dec->mb_height - y) << (5 + dec->quarterpel); |
|
|
|
|
|
else if (mv_y < ((int)(-y-1)) << (5 + dec->quarterpel) ) |
|
|
mv_y = (int)(-y-1) << (5 + dec->quarterpel); |
|
|
|
|
1022 |
mv->x = mv_x; |
mv->x = mv_x; |
1023 |
mv->y = mv_y; |
mv->y = mv_y; |
1024 |
} |
} |
1028 |
decoder_bf_interpolate_mbinter(DECODER * dec, |
decoder_bf_interpolate_mbinter(DECODER * dec, |
1029 |
IMAGE forward, |
IMAGE forward, |
1030 |
IMAGE backward, |
IMAGE backward, |
1031 |
const MACROBLOCK * pMB, |
MACROBLOCK * pMB, |
1032 |
const uint32_t x_pos, |
const uint32_t x_pos, |
1033 |
const uint32_t y_pos, |
const uint32_t y_pos, |
1034 |
Bitstream * bs, |
Bitstream * bs, |
1045 |
pU_Cur = dec->cur.u + (y_pos << 3) * stride2 + (x_pos << 3); |
pU_Cur = dec->cur.u + (y_pos << 3) * stride2 + (x_pos << 3); |
1046 |
pV_Cur = dec->cur.v + (y_pos << 3) * stride2 + (x_pos << 3); |
pV_Cur = dec->cur.v + (y_pos << 3) * stride2 + (x_pos << 3); |
1047 |
|
|
1048 |
|
validate_vector(pMB->mvs, x_pos, y_pos, dec); |
1049 |
|
validate_vector(pMB->b_mvs, x_pos, y_pos, dec); |
1050 |
|
|
1051 |
if (!direct) { |
if (!direct) { |
1052 |
uv_dx = pMB->mvs[0].x; |
uv_dx = pMB->mvs[0].x; |
1053 |
uv_dy = pMB->mvs[0].y; |
uv_dy = pMB->mvs[0].y; |
|
|
|
1054 |
b_uv_dx = pMB->b_mvs[0].x; |
b_uv_dx = pMB->b_mvs[0].x; |
1055 |
b_uv_dy = pMB->b_mvs[0].y; |
b_uv_dy = pMB->b_mvs[0].y; |
1056 |
|
|
1063 |
|
|
1064 |
uv_dx = (uv_dx >> 1) + roundtab_79[uv_dx & 0x3]; |
uv_dx = (uv_dx >> 1) + roundtab_79[uv_dx & 0x3]; |
1065 |
uv_dy = (uv_dy >> 1) + roundtab_79[uv_dy & 0x3]; |
uv_dy = (uv_dy >> 1) + roundtab_79[uv_dy & 0x3]; |
|
|
|
1066 |
b_uv_dx = (b_uv_dx >> 1) + roundtab_79[b_uv_dx & 0x3]; |
b_uv_dx = (b_uv_dx >> 1) + roundtab_79[b_uv_dx & 0x3]; |
1067 |
b_uv_dy = (b_uv_dy >> 1) + roundtab_79[b_uv_dy & 0x3]; |
b_uv_dy = (b_uv_dy >> 1) + roundtab_79[b_uv_dy & 0x3]; |
1068 |
|
|
1069 |
} else { |
} else { |
|
if(dec->quarterpel) { |
|
|
uv_dx = (pMB->mvs[0].x / 2) + (pMB->mvs[1].x / 2) + (pMB->mvs[2].x / 2) + (pMB->mvs[3].x / 2); |
|
|
uv_dy = (pMB->mvs[0].y / 2) + (pMB->mvs[1].y / 2) + (pMB->mvs[2].y / 2) + (pMB->mvs[3].y / 2); |
|
|
b_uv_dx = (pMB->b_mvs[0].x / 2) + (pMB->b_mvs[1].x / 2) + (pMB->b_mvs[2].x / 2) + (pMB->b_mvs[3].x / 2); |
|
|
b_uv_dy = (pMB->b_mvs[0].y / 2) + (pMB->b_mvs[1].y / 2) + (pMB->b_mvs[2].y / 2) + (pMB->b_mvs[3].y / 2); |
|
|
} else { |
|
1070 |
uv_dx = pMB->mvs[0].x + pMB->mvs[1].x + pMB->mvs[2].x + pMB->mvs[3].x; |
uv_dx = pMB->mvs[0].x + pMB->mvs[1].x + pMB->mvs[2].x + pMB->mvs[3].x; |
1071 |
uv_dy = pMB->mvs[0].y + pMB->mvs[1].y + pMB->mvs[2].y + pMB->mvs[3].y; |
uv_dy = pMB->mvs[0].y + pMB->mvs[1].y + pMB->mvs[2].y + pMB->mvs[3].y; |
1072 |
b_uv_dx = pMB->b_mvs[0].x + pMB->b_mvs[1].x + pMB->b_mvs[2].x + pMB->b_mvs[3].x; |
b_uv_dx = pMB->b_mvs[0].x + pMB->b_mvs[1].x + pMB->b_mvs[2].x + pMB->b_mvs[3].x; |
1073 |
b_uv_dy = pMB->b_mvs[0].y + pMB->b_mvs[1].y + pMB->b_mvs[2].y + pMB->b_mvs[3].y; |
b_uv_dy = pMB->b_mvs[0].y + pMB->b_mvs[1].y + pMB->b_mvs[2].y + pMB->b_mvs[3].y; |
1074 |
|
|
1075 |
|
if (dec->quarterpel) { |
1076 |
|
uv_dx /= 2; |
1077 |
|
uv_dy /= 2; |
1078 |
|
b_uv_dx /= 2; |
1079 |
|
b_uv_dy /= 2; |
1080 |
} |
} |
1081 |
|
|
1082 |
uv_dx = (uv_dx >> 3) + roundtab_76[uv_dx & 0xf]; |
uv_dx = (uv_dx >> 3) + roundtab_76[uv_dx & 0xf]; |
1124 |
|
|
1125 |
if(dec->quarterpel) { |
if(dec->quarterpel) { |
1126 |
if(!direct) { |
if(!direct) { |
1127 |
interpolate16x16_quarterpel(dec->tmp.y, backward.y, dec->qtmp.y, dec->qtmp.y + 64, |
interpolate16x16_add_quarterpel(dec->cur.y, backward.y, dec->qtmp.y, dec->qtmp.y + 64, |
1128 |
dec->qtmp.y + 128, 16*x_pos, 16*y_pos, |
dec->qtmp.y + 128, 16*x_pos, 16*y_pos, |
1129 |
pMB->b_mvs[0].x, pMB->b_mvs[0].y, stride, 0); |
pMB->b_mvs[0].x, pMB->b_mvs[0].y, stride, 0); |
1130 |
} else { |
} else { |
1131 |
interpolate8x8_quarterpel(dec->tmp.y, backward.y, dec->qtmp.y, dec->qtmp.y + 64, |
interpolate8x8_add_quarterpel(dec->cur.y, backward.y, dec->qtmp.y, dec->qtmp.y + 64, |
1132 |
dec->qtmp.y + 128, 16*x_pos, 16*y_pos, |
dec->qtmp.y + 128, 16*x_pos, 16*y_pos, |
1133 |
pMB->b_mvs[0].x, pMB->b_mvs[0].y, stride, 0); |
pMB->b_mvs[0].x, pMB->b_mvs[0].y, stride, 0); |
1134 |
interpolate8x8_quarterpel(dec->tmp.y, backward.y, dec->qtmp.y, dec->qtmp.y + 64, |
interpolate8x8_add_quarterpel(dec->cur.y, backward.y, dec->qtmp.y, dec->qtmp.y + 64, |
1135 |
dec->qtmp.y + 128, 16*x_pos + 8, 16*y_pos, |
dec->qtmp.y + 128, 16*x_pos + 8, 16*y_pos, |
1136 |
pMB->b_mvs[1].x, pMB->b_mvs[1].y, stride, 0); |
pMB->b_mvs[1].x, pMB->b_mvs[1].y, stride, 0); |
1137 |
interpolate8x8_quarterpel(dec->tmp.y, backward.y, dec->qtmp.y, dec->qtmp.y + 64, |
interpolate8x8_add_quarterpel(dec->cur.y, backward.y, dec->qtmp.y, dec->qtmp.y + 64, |
1138 |
dec->qtmp.y + 128, 16*x_pos, 16*y_pos + 8, |
dec->qtmp.y + 128, 16*x_pos, 16*y_pos + 8, |
1139 |
pMB->b_mvs[2].x, pMB->b_mvs[2].y, stride, 0); |
pMB->b_mvs[2].x, pMB->b_mvs[2].y, stride, 0); |
1140 |
interpolate8x8_quarterpel(dec->tmp.y, backward.y, dec->qtmp.y, dec->qtmp.y + 64, |
interpolate8x8_add_quarterpel(dec->cur.y, backward.y, dec->qtmp.y, dec->qtmp.y + 64, |
1141 |
dec->qtmp.y + 128, 16*x_pos + 8, 16*y_pos + 8, |
dec->qtmp.y + 128, 16*x_pos + 8, 16*y_pos + 8, |
1142 |
pMB->b_mvs[3].x, pMB->b_mvs[3].y, stride, 0); |
pMB->b_mvs[3].x, pMB->b_mvs[3].y, stride, 0); |
1143 |
} |
} |
1144 |
} else { |
} else { |
1145 |
interpolate8x8_switch(dec->tmp.y, backward.y, 16 * x_pos, 16 * y_pos, |
interpolate8x8_add_switch(dec->cur.y, backward.y, 16 * x_pos, 16 * y_pos, |
1146 |
pMB->b_mvs[0].x, pMB->b_mvs[0].y, stride, 0); |
pMB->b_mvs[0].x, pMB->b_mvs[0].y, stride, 0); |
1147 |
interpolate8x8_switch(dec->tmp.y, backward.y, 16 * x_pos + 8, |
interpolate8x8_add_switch(dec->cur.y, backward.y, 16 * x_pos + 8, |
1148 |
16 * y_pos, pMB->b_mvs[1].x, pMB->b_mvs[1].y, stride, 0); |
16 * y_pos, pMB->b_mvs[1].x, pMB->b_mvs[1].y, stride, 0); |
1149 |
interpolate8x8_switch(dec->tmp.y, backward.y, 16 * x_pos, |
interpolate8x8_add_switch(dec->cur.y, backward.y, 16 * x_pos, |
1150 |
16 * y_pos + 8, pMB->b_mvs[2].x, pMB->b_mvs[2].y, stride, 0); |
16 * y_pos + 8, pMB->b_mvs[2].x, pMB->b_mvs[2].y, stride, 0); |
1151 |
interpolate8x8_switch(dec->tmp.y, backward.y, 16 * x_pos + 8, |
interpolate8x8_add_switch(dec->cur.y, backward.y, 16 * x_pos + 8, |
1152 |
16 * y_pos + 8, pMB->b_mvs[3].x, pMB->b_mvs[3].y, stride, 0); |
16 * y_pos + 8, pMB->b_mvs[3].x, pMB->b_mvs[3].y, stride, 0); |
1153 |
} |
} |
1154 |
|
|
1155 |
interpolate8x8_switch(dec->tmp.u, backward.u, 8 * x_pos, 8 * y_pos, |
interpolate8x8_add_switch(dec->cur.u, backward.u, 8 * x_pos, 8 * y_pos, |
1156 |
b_uv_dx, b_uv_dy, stride2, 0); |
b_uv_dx, b_uv_dy, stride2, 0); |
1157 |
interpolate8x8_switch(dec->tmp.v, backward.v, 8 * x_pos, 8 * y_pos, |
interpolate8x8_add_switch(dec->cur.v, backward.v, 8 * x_pos, 8 * y_pos, |
1158 |
b_uv_dx, b_uv_dy, stride2, 0); |
b_uv_dx, b_uv_dy, stride2, 0); |
1159 |
|
|
|
interpolate8x8_avg2(dec->cur.y + (16 * y_pos * stride) + 16 * x_pos, |
|
|
dec->cur.y + (16 * y_pos * stride) + 16 * x_pos, |
|
|
dec->tmp.y + (16 * y_pos * stride) + 16 * x_pos, |
|
|
stride, 1, 8); |
|
|
|
|
|
interpolate8x8_avg2(dec->cur.y + (16 * y_pos * stride) + 16 * x_pos + 8, |
|
|
dec->cur.y + (16 * y_pos * stride) + 16 * x_pos + 8, |
|
|
dec->tmp.y + (16 * y_pos * stride) + 16 * x_pos + 8, |
|
|
stride, 1, 8); |
|
|
|
|
|
interpolate8x8_avg2(dec->cur.y + ((16 * y_pos + 8) * stride) + 16 * x_pos, |
|
|
dec->cur.y + ((16 * y_pos + 8) * stride) + 16 * x_pos, |
|
|
dec->tmp.y + ((16 * y_pos + 8) * stride) + 16 * x_pos, |
|
|
stride, 1, 8); |
|
|
|
|
|
interpolate8x8_avg2(dec->cur.y + ((16 * y_pos + 8) * stride) + 16 * x_pos + 8, |
|
|
dec->cur.y + ((16 * y_pos + 8) * stride) + 16 * x_pos + 8, |
|
|
dec->tmp.y + ((16 * y_pos + 8) * stride) + 16 * x_pos + 8, |
|
|
stride, 1, 8); |
|
|
|
|
|
interpolate8x8_avg2(dec->cur.u + (8 * y_pos * stride2) + 8 * x_pos, |
|
|
dec->cur.u + (8 * y_pos * stride2) + 8 * x_pos, |
|
|
dec->tmp.u + (8 * y_pos * stride2) + 8 * x_pos, |
|
|
stride2, 1, 8); |
|
|
|
|
|
interpolate8x8_avg2(dec->cur.v + (8 * y_pos * stride2) + 8 * x_pos, |
|
|
dec->cur.v + (8 * y_pos * stride2) + 8 * x_pos, |
|
|
dec->tmp.v + (8 * y_pos * stride2) + 8 * x_pos, |
|
|
stride2, 1, 8); |
|
|
|
|
1160 |
stop_comp_timer(); |
stop_comp_timer(); |
1161 |
|
|
1162 |
if (cbp) |
if (cbp) |
1205 |
uint32_t x, y; |
uint32_t x, y; |
1206 |
VECTOR mv; |
VECTOR mv; |
1207 |
const VECTOR zeromv = {0,0}; |
const VECTOR zeromv = {0,0}; |
|
const int64_t TRB = dec->time_pp - dec->time_bp, TRD = dec->time_pp; |
|
1208 |
int i; |
int i; |
1209 |
|
|
1210 |
|
if (!dec->is_edged[0]) { |
1211 |
start_timer(); |
start_timer(); |
1212 |
image_setedges(&dec->refn[0], dec->edged_width, dec->edged_height, |
image_setedges(&dec->refn[0], dec->edged_width, dec->edged_height, |
1213 |
dec->width, dec->height, dec->bs_version); |
dec->width, dec->height, dec->bs_version); |
1214 |
|
dec->is_edged[0] = 1; |
1215 |
|
stop_edges_timer(); |
1216 |
|
} |
1217 |
|
|
1218 |
|
if (!dec->is_edged[1]) { |
1219 |
|
start_timer(); |
1220 |
image_setedges(&dec->refn[1], dec->edged_width, dec->edged_height, |
image_setedges(&dec->refn[1], dec->edged_width, dec->edged_height, |
1221 |
dec->width, dec->height, dec->bs_version); |
dec->width, dec->height, dec->bs_version); |
1222 |
|
dec->is_edged[1] = 1; |
1223 |
stop_edges_timer(); |
stop_edges_timer(); |
1224 |
|
} |
1225 |
|
|
1226 |
for (y = 0; y < dec->mb_height; y++) { |
for (y = 0; y < dec->mb_height; y++) { |
1227 |
/* Initialize Pred Motion Vector */ |
/* Initialize Pred Motion Vector */ |
1308 |
|
|
1309 |
case MODE_DIRECT_NONE_MV: |
case MODE_DIRECT_NONE_MV: |
1310 |
for (i = 0; i < 4; i++) { |
for (i = 0; i < 4; i++) { |
1311 |
mb->mvs[i].x = (int32_t) ((TRB * last_mb->mvs[i].x) / TRD + mv.x); |
mb->mvs[i].x = last_mb->mvs[i].x*dec->time_bp/dec->time_pp + mv.x; |
1312 |
mb->b_mvs[i].x = (int32_t) ((mv.x == 0) |
mb->mvs[i].y = last_mb->mvs[i].y*dec->time_bp/dec->time_pp + mv.y; |
1313 |
? ((TRB - TRD) * last_mb->mvs[i].x) / TRD |
|
1314 |
: mb->mvs[i].x - last_mb->mvs[i].x); |
mb->b_mvs[i].x = (mv.x) |
1315 |
mb->mvs[i].y = (int32_t) ((TRB * last_mb->mvs[i].y) / TRD + mv.y); |
? mb->mvs[i].x - last_mb->mvs[i].x |
1316 |
mb->b_mvs[i].y = (int32_t) ((mv.y == 0) |
: last_mb->mvs[i].x*(dec->time_bp - dec->time_pp)/dec->time_pp; |
1317 |
? ((TRB - TRD) * last_mb->mvs[i].y) / TRD |
mb->b_mvs[i].y = (mv.y) |
1318 |
: mb->mvs[i].y - last_mb->mvs[i].y); |
? mb->mvs[i].y - last_mb->mvs[i].y |
1319 |
|
: last_mb->mvs[i].y*(dec->time_bp - dec->time_pp)/dec->time_pp; |
1320 |
} |
} |
1321 |
|
|
1322 |
decoder_bf_interpolate_mbinter(dec, dec->refn[1], dec->refn[0], |
decoder_bf_interpolate_mbinter(dec, dec->refn[1], dec->refn[0], |
1360 |
xvid_dec_frame_t * frame, xvid_dec_stats_t * stats, |
xvid_dec_frame_t * frame, xvid_dec_stats_t * stats, |
1361 |
int coding_type, int quant) |
int coding_type, int quant) |
1362 |
{ |
{ |
1363 |
|
const int brightness = XVID_VERSION_MINOR(frame->version) >= 1 ? frame->brightness : 0; |
1364 |
|
|
1365 |
if (dec->cartoon_mode) |
if (dec->cartoon_mode) |
1366 |
frame->general &= ~XVID_FILMEFFECT; |
frame->general &= ~XVID_FILMEFFECT; |
1367 |
|
|
1368 |
if (frame->general & (XVID_DEBLOCKY|XVID_DEBLOCKUV|XVID_FILMEFFECT) && mbs != NULL) /* post process */ |
if ((frame->general & (XVID_DEBLOCKY|XVID_DEBLOCKUV|XVID_FILMEFFECT) || brightness!=0) |
1369 |
|
&& mbs != NULL) /* post process */ |
1370 |
{ |
{ |
1371 |
/* note: image is stored to tmp */ |
/* note: image is stored to tmp */ |
1372 |
image_copy(&dec->tmp, img, dec->edged_width, dec->height); |
image_copy(&dec->tmp, img, dec->edged_width, dec->height); |
1373 |
image_postproc(&dec->postproc, &dec->tmp, dec->edged_width, |
image_postproc(&dec->postproc, &dec->tmp, dec->edged_width, |
1374 |
mbs, dec->mb_width, dec->mb_height, dec->mb_width, |
mbs, dec->mb_width, dec->mb_height, dec->mb_width, |
1375 |
frame->general, dec->frames, (coding_type == B_VOP)); |
frame->general, brightness, dec->frames, (coding_type == B_VOP)); |
1376 |
img = &dec->tmp; |
img = &dec->tmp; |
1377 |
} |
} |
1378 |
|
|
1395 |
} |
} |
1396 |
} |
} |
1397 |
|
|
|
|
|
1398 |
int |
int |
1399 |
decoder_decode(DECODER * dec, |
decoder_decode(DECODER * dec, |
1400 |
xvid_dec_frame_t * frame, xvid_dec_stats_t * stats) |
xvid_dec_frame_t * frame, xvid_dec_stats_t * stats) |
1491 |
goto repeat; |
goto repeat; |
1492 |
} |
} |
1493 |
|
|
1494 |
|
if(dec->frames == 0 && coding_type != I_VOP) { |
1495 |
|
/* 1st frame is not an i-vop */ |
1496 |
|
goto repeat; |
1497 |
|
} |
1498 |
|
|
1499 |
dec->p_bmv.x = dec->p_bmv.y = dec->p_fmv.y = dec->p_fmv.y = 0; /* init pred vector to 0 */ |
dec->p_bmv.x = dec->p_bmv.y = dec->p_fmv.y = dec->p_fmv.y = 0; /* init pred vector to 0 */ |
1500 |
|
|
1501 |
/* packed_mode: special-N_VOP treament */ |
/* packed_mode: special-N_VOP treament */ |
1545 |
} |
} |
1546 |
|
|
1547 |
image_swap(&dec->refn[0], &dec->refn[1]); |
image_swap(&dec->refn[0], &dec->refn[1]); |
1548 |
|
dec->is_edged[1] = dec->is_edged[0]; |
1549 |
image_swap(&dec->cur, &dec->refn[0]); |
image_swap(&dec->cur, &dec->refn[0]); |
1550 |
|
dec->is_edged[0] = 0; |
1551 |
SWAP(MACROBLOCK *, dec->mbs, dec->last_mbs); |
SWAP(MACROBLOCK *, dec->mbs, dec->last_mbs); |
1552 |
dec->last_reduced_resolution = reduced_resolution; |
dec->last_reduced_resolution = reduced_resolution; |
1553 |
dec->last_coding_type = coding_type; |
dec->last_coding_type = coding_type; |
1559 |
|
|
1560 |
if (dec->low_delay) { |
if (dec->low_delay) { |
1561 |
DPRINTF(XVID_DEBUG_ERROR, "warning: bvop found in low_delay==1 stream\n"); |
DPRINTF(XVID_DEBUG_ERROR, "warning: bvop found in low_delay==1 stream\n"); |
1562 |
dec->low_delay = 1; |
dec->low_delay = 0; |
1563 |
} |
} |
1564 |
|
|
1565 |
if (dec->frames < 2) { |
if (dec->frames < 2) { |