53 |
* 26.03.2002 interlacing support - moved transfers outside decode loop |
* 26.03.2002 interlacing support - moved transfers outside decode loop |
54 |
* 26.12.2001 decoder_mbinter: dequant/idct moved within if(coded) block |
* 26.12.2001 decoder_mbinter: dequant/idct moved within if(coded) block |
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@xvid.org> |
57 |
* |
* |
58 |
* $Id$ |
* $Id$ |
59 |
* |
* |
136 |
return XVID_ERR_MEMORY; |
return XVID_ERR_MEMORY; |
137 |
} |
} |
138 |
|
|
|
if (image_create(&dec->refh, dec->edged_width, dec->edged_height)) { |
|
|
image_destroy(&dec->cur, dec->edged_width, dec->edged_height); |
|
|
image_destroy(&dec->refn[0], dec->edged_width, dec->edged_height); |
|
|
image_destroy(&dec->refn[1], dec->edged_width, dec->edged_height); |
|
|
image_destroy(&dec->refn[2], dec->edged_width, dec->edged_height); |
|
|
xvid_free(dec); |
|
|
return XVID_ERR_MEMORY; |
|
|
} |
|
|
|
|
139 |
dec->mbs = |
dec->mbs = |
140 |
xvid_malloc(sizeof(MACROBLOCK) * dec->mb_width * dec->mb_height, |
xvid_malloc(sizeof(MACROBLOCK) * dec->mb_width * dec->mb_height, |
141 |
CACHE_LINE); |
CACHE_LINE); |
144 |
image_destroy(&dec->refn[0], dec->edged_width, dec->edged_height); |
image_destroy(&dec->refn[0], dec->edged_width, dec->edged_height); |
145 |
image_destroy(&dec->refn[1], dec->edged_width, dec->edged_height); |
image_destroy(&dec->refn[1], dec->edged_width, dec->edged_height); |
146 |
image_destroy(&dec->refn[2], dec->edged_width, dec->edged_height); |
image_destroy(&dec->refn[2], dec->edged_width, dec->edged_height); |
|
image_destroy(&dec->refh, dec->edged_width, dec->edged_height); |
|
147 |
xvid_free(dec); |
xvid_free(dec); |
148 |
return XVID_ERR_MEMORY; |
return XVID_ERR_MEMORY; |
149 |
} |
} |
161 |
image_destroy(&dec->refn[0], dec->edged_width, dec->edged_height); |
image_destroy(&dec->refn[0], dec->edged_width, dec->edged_height); |
162 |
image_destroy(&dec->refn[1], dec->edged_width, dec->edged_height); |
image_destroy(&dec->refn[1], dec->edged_width, dec->edged_height); |
163 |
image_destroy(&dec->refn[2], dec->edged_width, dec->edged_height); |
image_destroy(&dec->refn[2], dec->edged_width, dec->edged_height); |
|
image_destroy(&dec->refh, dec->edged_width, dec->edged_height); |
|
164 |
xvid_free(dec); |
xvid_free(dec); |
165 |
return XVID_ERR_MEMORY; |
return XVID_ERR_MEMORY; |
166 |
} |
} |
186 |
image_destroy(&dec->refn[0], dec->edged_width, dec->edged_height); |
image_destroy(&dec->refn[0], dec->edged_width, dec->edged_height); |
187 |
image_destroy(&dec->refn[1], dec->edged_width, dec->edged_height); |
image_destroy(&dec->refn[1], dec->edged_width, dec->edged_height); |
188 |
image_destroy(&dec->refn[2], dec->edged_width, dec->edged_height); |
image_destroy(&dec->refn[2], dec->edged_width, dec->edged_height); |
|
image_destroy(&dec->refh, dec->edged_width, dec->edged_height); |
|
189 |
image_destroy(&dec->cur, dec->edged_width, dec->edged_height); |
image_destroy(&dec->cur, dec->edged_width, dec->edged_height); |
190 |
xvid_free(dec); |
xvid_free(dec); |
191 |
|
|
268 |
start_timer(); |
start_timer(); |
269 |
if (cbp & (1 << (5 - i))) // coded |
if (cbp & (1 << (5 - i))) // coded |
270 |
{ |
{ |
271 |
int direction = dec->alternate_vertical_scan ? |
get_intra_block(bs, &block[i * 64], pMB->acpred_directions[i], |
272 |
2 : pMB->acpred_directions[i]; |
start_coeff); |
|
|
|
|
get_intra_block(bs, &block[i * 64], direction, start_coeff); |
|
273 |
} |
} |
274 |
stop_coding_timer(); |
stop_coding_timer(); |
275 |
|
|
356 |
|
|
357 |
uv_dx = (uv_dx & 3) ? (uv_dx >> 1) | 1 : uv_dx / 2; |
uv_dx = (uv_dx & 3) ? (uv_dx >> 1) | 1 : uv_dx / 2; |
358 |
uv_dy = (uv_dy & 3) ? (uv_dy >> 1) | 1 : uv_dy / 2; |
uv_dy = (uv_dy & 3) ? (uv_dy >> 1) | 1 : uv_dy / 2; |
|
|
|
|
start_timer(); |
|
|
if(dec->quarterpel) { |
|
|
interpolate16x16_quarterpel(dec->cur.y, dec->refn[0].y, dec->refh.y, dec->refh.y + 64, |
|
|
dec->refh.y + 128, 16*x_pos, 16*y_pos, |
|
|
pMB->mvs[0].x, pMB->mvs[0].y, stride, rounding); |
|
|
} |
|
|
else { |
|
|
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); |
|
|
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); |
|
|
} |
|
|
|
|
|
interpolate8x8_switch(dec->cur.u, dec->refn[0].u, 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); |
|
|
stop_comp_timer(); |
|
|
|
|
359 |
} else { |
} else { |
360 |
int sum; |
int sum; |
361 |
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; |
375 |
} |
} |
376 |
|
|
377 |
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)); |
378 |
|
} |
379 |
|
|
380 |
start_timer(); |
start_timer(); |
381 |
if(dec->quarterpel) { |
if(dec->quarterpel) { |
382 |
interpolate8x8_quarterpel(dec->cur.y, dec->refn[0].y, dec->refh.y, dec->refh.y + 64, |
DPRINTF(DPRINTF_DEBUG, "QUARTERPEL\n"); |
383 |
dec->refh.y + 128, 16*x_pos, 16*y_pos, |
interpolate8x8_quarterpel(dec->cur.y, dec->refn[0].y, 16*x_pos, 16*y_pos, |
384 |
pMB->mvs[0].x, pMB->mvs[0].y, stride, rounding); |
pMB->mvs[0].x, pMB->mvs[0].y, stride, rounding); |
385 |
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, 16*x_pos + 8, 16*y_pos, |
|
dec->refh.y + 128, 16*x_pos + 8, 16*y_pos, |
|
386 |
pMB->mvs[1].x, pMB->mvs[1].y, stride, rounding); |
pMB->mvs[1].x, pMB->mvs[1].y, stride, rounding); |
387 |
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, 16*x_pos, 16*y_pos + 8, |
|
dec->refh.y + 128, 16*x_pos, 16*y_pos + 8, |
|
388 |
pMB->mvs[2].x, pMB->mvs[2].y, stride, rounding); |
pMB->mvs[2].x, pMB->mvs[2].y, stride, rounding); |
389 |
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, 16*x_pos + 8, 16*y_pos + 8, |
|
dec->refh.y + 128, 16*x_pos + 8, 16*y_pos + 8, |
|
390 |
pMB->mvs[3].x, pMB->mvs[3].y, stride, rounding); |
pMB->mvs[3].x, pMB->mvs[3].y, stride, rounding); |
391 |
} |
} |
392 |
else { |
else { |
405 |
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, |
406 |
uv_dx, uv_dy, stride2, rounding); |
uv_dx, uv_dy, stride2, rounding); |
407 |
stop_comp_timer(); |
stop_comp_timer(); |
|
} |
|
408 |
|
|
409 |
for (i = 0; i < 6; i++) { |
for (i = 0; i < 6; i++) { |
|
int direction = dec->alternate_vertical_scan ? 2 : 0; |
|
|
|
|
410 |
if (cbp & (1 << (5 - i))) // coded |
if (cbp & (1 << (5 - i))) // coded |
411 |
{ |
{ |
412 |
memset(&block[i * 64], 0, 64 * sizeof(int16_t)); // clear |
memset(&block[i * 64], 0, 64 * sizeof(int16_t)); // clear |
413 |
|
|
414 |
start_timer(); |
start_timer(); |
415 |
get_inter_block(bs, &block[i * 64], direction); |
get_inter_block(bs, &block[i * 64]); |
416 |
stop_coding_timer(); |
stop_coding_timer(); |
417 |
|
|
418 |
start_timer(); |
start_timer(); |
509 |
|
|
510 |
if (dec->interlacing) { |
if (dec->interlacing) { |
511 |
mb->field_dct = BitstreamGetBit(bs); |
mb->field_dct = BitstreamGetBit(bs); |
512 |
DEBUG1("deci: field_dct: ", mb->field_dct); |
DPRINTF(DPRINTF_DEBUG, "deci: field_dct: %d", mb->field_dct); |
513 |
} |
} |
514 |
|
|
515 |
decoder_mbintra(dec, mb, x, y, acpred_flag, cbp, bs, quant, |
decoder_mbintra(dec, mb, x, y, acpred_flag, cbp, bs, quant, |
585 |
|
|
586 |
start_timer(); |
start_timer(); |
587 |
image_setedges(&dec->refn[0], dec->edged_width, dec->edged_height, |
image_setedges(&dec->refn[0], dec->edged_width, dec->edged_height, |
588 |
dec->width, dec->height); |
dec->width, dec->height, dec->interlacing); |
589 |
stop_edges_timer(); |
stop_edges_timer(); |
590 |
|
|
591 |
bound = 0; |
bound = 0; |
655 |
if (dec->interlacing) { |
if (dec->interlacing) { |
656 |
if (cbp || intra) { |
if (cbp || intra) { |
657 |
mb->field_dct = BitstreamGetBit(bs); |
mb->field_dct = BitstreamGetBit(bs); |
658 |
DEBUG1("decp: field_dct: ", mb->field_dct); |
DPRINTF(DPRINTF_DEBUG, "decp: field_dct: %d", mb->field_dct); |
659 |
} |
} |
660 |
|
|
661 |
if (mb->mode == MODE_INTER || mb->mode == MODE_INTER_Q) { |
if (mb->mode == MODE_INTER || mb->mode == MODE_INTER_Q) { |
662 |
mb->field_pred = BitstreamGetBit(bs); |
mb->field_pred = BitstreamGetBit(bs); |
663 |
DEBUG1("decp: field_pred: ", mb->field_pred); |
DPRINTF(DPRINTF_DEBUG, "decp: field_pred: %d", mb->field_pred); |
664 |
|
|
665 |
if (mb->field_pred) { |
if (mb->field_pred) { |
666 |
mb->field_for_top = BitstreamGetBit(bs); |
mb->field_for_top = BitstreamGetBit(bs); |
667 |
DEBUG1("decp: field_for_top: ", mb->field_for_top); |
DPRINTF(DPRINTF_DEBUG, "decp: field_for_top: %d", mb->field_for_top); |
668 |
mb->field_for_bot = BitstreamGetBit(bs); |
mb->field_for_bot = BitstreamGetBit(bs); |
669 |
DEBUG1("decp: field_for_bot: ", mb->field_for_bot); |
DPRINTF(DPRINTF_DEBUG, "decp: field_for_bot: %d", mb->field_for_bot); |
670 |
} |
} |
671 |
} |
} |
672 |
} |
} |
706 |
rounding); |
rounding); |
707 |
} else // not coded |
} else // not coded |
708 |
{ |
{ |
709 |
DEBUG2("P-frame MB at (X,Y)=",x,y); |
DPRINTF(DPRINTF_DEBUG, "P-frame MB at (X,Y)=(%d,%d)", x, y); |
710 |
|
|
711 |
mb->mode = MODE_NOT_CODED; |
mb->mode = MODE_NOT_CODED; |
712 |
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; |
713 |
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; |
877 |
stop_comp_timer(); |
stop_comp_timer(); |
878 |
|
|
879 |
for (i = 0; i < 6; i++) { |
for (i = 0; i < 6; i++) { |
|
int direction = dec->alternate_vertical_scan ? 2 : 0; |
|
|
|
|
880 |
if (cbp & (1 << (5 - i))) // coded |
if (cbp & (1 << (5 - i))) // coded |
881 |
{ |
{ |
882 |
memset(&block[i * 64], 0, 64 * sizeof(int16_t)); // clear |
memset(&block[i * 64], 0, 64 * sizeof(int16_t)); // clear |
883 |
|
|
884 |
start_timer(); |
start_timer(); |
885 |
get_inter_block(bs, &block[i * 64], direction); |
get_inter_block(bs, &block[i * 64]); |
886 |
stop_coding_timer(); |
stop_coding_timer(); |
887 |
|
|
888 |
start_timer(); |
start_timer(); |
1028 |
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, |
1029 |
b_uv_dx, b_uv_dy, stride2, 0); |
b_uv_dx, b_uv_dy, stride2, 0); |
1030 |
|
|
1031 |
interpolate8x8_avg2(dec->cur.y + (16 * y_pos * stride) + 16 * x_pos, |
interpolate8x8_c(dec->cur.y, dec->refn[2].y, 16 * x_pos, 16 * y_pos, |
1032 |
dec->cur.y + (16 * y_pos * stride) + 16 * x_pos, |
stride); |
1033 |
dec->refn[2].y + (16 * y_pos * stride) + 16 * x_pos, |
interpolate8x8_c(dec->cur.y, dec->refn[2].y, 16 * x_pos + 8, 16 * y_pos, |
1034 |
stride, 0); |
stride); |
1035 |
|
interpolate8x8_c(dec->cur.y, dec->refn[2].y, 16 * x_pos, 16 * y_pos + 8, |
1036 |
interpolate8x8_avg2(dec->cur.y + (16 * y_pos * stride) + 16 * x_pos + 8, |
stride); |
1037 |
dec->cur.y + (16 * y_pos * stride) + 16 * x_pos + 8, |
interpolate8x8_c(dec->cur.y, dec->refn[2].y, 16 * x_pos + 8, |
1038 |
dec->refn[2].y + (16 * y_pos * stride) + 16 * x_pos + 8, |
16 * y_pos + 8, stride); |
1039 |
stride, 0); |
interpolate8x8_c(dec->cur.u, dec->refn[2].u, 8 * x_pos, 8 * y_pos, |
1040 |
|
stride2); |
1041 |
interpolate8x8_avg2(dec->cur.y + (16 * (y_pos + 8) * stride) + 16 * x_pos, |
interpolate8x8_c(dec->cur.v, dec->refn[2].v, 8 * x_pos, 8 * y_pos, |
1042 |
dec->cur.y + (16 * (y_pos + 8) * stride) + 16 * x_pos, |
stride2); |
|
dec->refn[2].y + (16 * (y_pos + 8) * stride) + 16 * x_pos, |
|
|
stride, 0); |
|
|
|
|
|
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->refn[2].y + (16 * (y_pos + 8) * stride) + 16 * x_pos + 8, |
|
|
stride, 0); |
|
|
|
|
|
interpolate8x8_avg2(dec->cur.u + (8 * y_pos * stride) + 8 * x_pos, |
|
|
dec->cur.u + (8 * y_pos * stride) + 8 * x_pos, |
|
|
dec->refn[2].u + (8 * y_pos * stride) + 8 * x_pos, |
|
|
stride2, 0); |
|
|
|
|
|
interpolate8x8_avg2(dec->cur.v + (8 * y_pos * stride) + 8 * x_pos, |
|
|
dec->cur.v + (8 * y_pos * stride) + 8 * x_pos, |
|
|
dec->refn[2].v + (8 * y_pos * stride) + 8 * x_pos, |
|
|
stride2, 0); |
|
|
|
|
1043 |
stop_comp_timer(); |
stop_comp_timer(); |
1044 |
|
|
1045 |
for (i = 0; i < 6; i++) { |
for (i = 0; i < 6; i++) { |
|
int direction = dec->alternate_vertical_scan ? 2 : 0; |
|
|
|
|
1046 |
if (cbp & (1 << (5 - i))) // coded |
if (cbp & (1 << (5 - i))) // coded |
1047 |
{ |
{ |
1048 |
memset(&block[i * 64], 0, 64 * sizeof(int16_t)); // clear |
memset(&block[i * 64], 0, 64 * sizeof(int16_t)); // clear |
1049 |
|
|
1050 |
start_timer(); |
start_timer(); |
1051 |
get_inter_block(bs, &block[i * 64], direction); |
get_inter_block(bs, &block[i * 64]); |
1052 |
stop_coding_timer(); |
stop_coding_timer(); |
1053 |
|
|
1054 |
start_timer(); |
start_timer(); |
1143 |
|
|
1144 |
start_timer(); |
start_timer(); |
1145 |
image_setedges(&dec->refn[0], dec->edged_width, dec->edged_height, |
image_setedges(&dec->refn[0], dec->edged_width, dec->edged_height, |
1146 |
dec->width, dec->height); |
dec->width, dec->height, dec->interlacing); |
1147 |
image_setedges(&dec->refn[1], dec->edged_width, dec->edged_height, |
image_setedges(&dec->refn[1], dec->edged_width, dec->edged_height, |
1148 |
dec->width, dec->height); |
dec->width, dec->height, dec->interlacing); |
1149 |
stop_edges_timer(); |
stop_edges_timer(); |
1150 |
|
|
1151 |
#ifdef BFRAMES_DEC_DEBUG |
#ifdef BFRAMES_DEC_DEBUG |
1279 |
break; |
break; |
1280 |
|
|
1281 |
default: |
default: |
1282 |
DEBUG1("Not support B-frame mb_type =", mb->mb_type); |
DPRINTF(DPRINTF_ERROR, "Not support B-frame mb_type = %d", mb->mb_type); |
1283 |
} |
} |
1284 |
|
|
1285 |
} // end of FOR |
} // end of FOR |