1 |
/************************************************************************** |
/************************************************************************** |
2 |
* |
* |
3 |
* XVID MPEG-4 VIDEO CODEC |
* XVID MPEG-4 VIDEO CODEC |
4 |
* image stuff |
* - Image management functions - |
5 |
* |
* |
6 |
* This program is an implementation of a part of one or more MPEG-4 |
* Copyright(C) 2001-2004 Peter Ross <pross@xvid.org> |
|
* Video tools as specified in ISO/IEC 14496-2 standard. Those intending |
|
|
* to use this software module in hardware or software products are |
|
|
* advised that its use may infringe existing patents or copyrights, and |
|
|
* any such use would be at such party's own risk. The original |
|
|
* developer of this software module and his/her company, and subsequent |
|
|
* editors and their companies, will have no liability for use of this |
|
|
* software or modifications or derivatives thereof. |
|
7 |
* |
* |
8 |
* This program is free software; you can redistribute it and/or modify |
* This program is free software; you can redistribute it and/or modify |
9 |
* 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 |
* |
* |
18 |
* You should have received a copy of the GNU General Public License |
* You should have received a copy of the GNU General Public License |
19 |
* along with this program; if not, write to the Free Software |
* along with this program; if not, write to the Free Software |
20 |
* Foundation, Inc., 675 Mass Ave, Cambridge, MA 02139, USA. |
* Foundation, Inc., 59 Temple Place, Suite 330, Boston, MA 02111-1307 USA |
21 |
* |
* |
22 |
*************************************************************************/ |
* $Id$ |
|
|
|
|
/************************************************************************** |
|
|
* |
|
|
* History: |
|
23 |
* |
* |
24 |
* 05.10.2002 support for interpolated images in qpel mode - Isibaar |
****************************************************************************/ |
|
* 01.05.2002 BFRAME image-based u,v interpolation |
|
|
* 22.04.2002 added some B-frame support |
|
|
* 14.04.2002 added image_dump_yuvpgm(), added image_mad() |
|
|
* XVID_CSP_USER input support |
|
|
* 09.04.2002 PSNR calculations - Isibaar |
|
|
* 06.04.2002 removed interlaced edging from U,V blocks (as per spec) |
|
|
* 26.03.2002 interlacing support (field-based edging in set_edges) |
|
|
* 26.01.2002 rgb555, rgb565 |
|
|
* 07.01.2001 commented u,v interpolation (not required for uv-block-based) |
|
|
* 23.12.2001 removed #ifdefs, added function pointers + init_common() |
|
|
* 22.12.2001 cpu #ifdefs |
|
|
* 19.12.2001 image_dump(); useful for debugging |
|
|
* 6.12.2001 inital version; (c)2001 peter ross <pross@cs.rmit.edu.au> |
|
|
* |
|
|
*************************************************************************/ |
|
25 |
|
|
26 |
#include <stdlib.h> |
#include <stdlib.h> |
27 |
#include <string.h> /* memcpy, memset */ |
#include <string.h> /* memcpy, memset */ |
28 |
#include <math.h> |
#include <math.h> |
|
|
|
29 |
#include "../portab.h" |
#include "../portab.h" |
30 |
#include "../global.h" /* XVID_CSP_XXX's */ |
#include "../global.h" /* XVID_CSP_XXX's */ |
31 |
#include "../xvid.h" /* XVID_CSP_XXX's */ |
#include "../xvid.h" /* XVID_CSP_XXX's */ |
32 |
#include "image.h" |
#include "image.h" |
33 |
#include "colorspace.h" |
#include "colorspace.h" |
34 |
#include "interpolate8x8.h" |
#include "interpolate8x8.h" |
|
#include "reduced.h" |
|
35 |
#include "../utils/mem_align.h" |
#include "../utils/mem_align.h" |
36 |
|
#include "../motion/sad.h" |
37 |
|
|
38 |
#include "font.h" /* XXX: remove later */ |
#include "font.h" /* XXX: remove later */ |
39 |
|
|
48 |
{ |
{ |
49 |
const uint32_t edged_width2 = edged_width / 2; |
const uint32_t edged_width2 = edged_width / 2; |
50 |
const uint32_t edged_height2 = edged_height / 2; |
const uint32_t edged_height2 = edged_height / 2; |
|
uint32_t i; |
|
51 |
|
|
52 |
image->y = |
image->y = |
53 |
xvid_malloc(edged_width * (edged_height + 1) + SAFETY, CACHE_LINE); |
xvid_malloc(edged_width * (edged_height + 1) + SAFETY, CACHE_LINE); |
54 |
if (image->y == NULL) { |
if (image->y == NULL) { |
55 |
return -1; |
return -1; |
56 |
} |
} |
57 |
|
memset(image->y, 0, edged_width * (edged_height + 1) + SAFETY); |
|
for (i = 0; i < edged_width * edged_height + SAFETY; i++) { |
|
|
image->y[i] = 0; |
|
|
} |
|
58 |
|
|
59 |
image->u = xvid_malloc(edged_width2 * edged_height2 + SAFETY, CACHE_LINE); |
image->u = xvid_malloc(edged_width2 * edged_height2 + SAFETY, CACHE_LINE); |
60 |
if (image->u == NULL) { |
if (image->u == NULL) { |
61 |
xvid_free(image->y); |
xvid_free(image->y); |
62 |
|
image->y = NULL; |
63 |
return -1; |
return -1; |
64 |
} |
} |
65 |
|
memset(image->u, 0, edged_width2 * edged_height2 + SAFETY); |
66 |
|
|
67 |
image->v = xvid_malloc(edged_width2 * edged_height2 + SAFETY, CACHE_LINE); |
image->v = xvid_malloc(edged_width2 * edged_height2 + SAFETY, CACHE_LINE); |
68 |
if (image->v == NULL) { |
if (image->v == NULL) { |
69 |
xvid_free(image->u); |
xvid_free(image->u); |
70 |
|
image->u = NULL; |
71 |
xvid_free(image->y); |
xvid_free(image->y); |
72 |
|
image->y = NULL; |
73 |
return -1; |
return -1; |
74 |
} |
} |
75 |
|
memset(image->v, 0, edged_width2 * edged_height2 + SAFETY); |
76 |
|
|
77 |
image->y += EDGE_SIZE * edged_width + EDGE_SIZE; |
image->y += EDGE_SIZE * edged_width + EDGE_SIZE; |
78 |
image->u += EDGE_SIZE2 * edged_width2 + EDGE_SIZE2; |
image->u += EDGE_SIZE2 * edged_width2 + EDGE_SIZE2; |
92 |
|
|
93 |
if (image->y) { |
if (image->y) { |
94 |
xvid_free(image->y - (EDGE_SIZE * edged_width + EDGE_SIZE)); |
xvid_free(image->y - (EDGE_SIZE * edged_width + EDGE_SIZE)); |
95 |
|
image->y = NULL; |
96 |
} |
} |
97 |
if (image->u) { |
if (image->u) { |
98 |
xvid_free(image->u - (EDGE_SIZE2 * edged_width2 + EDGE_SIZE2)); |
xvid_free(image->u - (EDGE_SIZE2 * edged_width2 + EDGE_SIZE2)); |
99 |
|
image->u = NULL; |
100 |
} |
} |
101 |
if (image->v) { |
if (image->v) { |
102 |
xvid_free(image->v - (EDGE_SIZE2 * edged_width2 + EDGE_SIZE2)); |
xvid_free(image->v - (EDGE_SIZE2 * edged_width2 + EDGE_SIZE2)); |
103 |
|
image->v = NULL; |
104 |
} |
} |
105 |
} |
} |
106 |
|
|
126 |
memcpy(image1->v, image2->v, edged_width * height / 4); |
memcpy(image1->v, image2->v, edged_width * height / 4); |
127 |
} |
} |
128 |
|
|
129 |
|
/* setedges bug was fixed in this BS version */ |
130 |
|
#define SETEDGES_BUG_BEFORE 18 |
131 |
|
|
132 |
void |
void |
133 |
image_setedges(IMAGE * image, |
image_setedges(IMAGE * image, |
134 |
uint32_t edged_width, |
uint32_t edged_width, |
135 |
uint32_t edged_height, |
uint32_t edged_height, |
136 |
uint32_t width, |
uint32_t width, |
137 |
uint32_t height) |
uint32_t height, |
138 |
|
int bs_version) |
139 |
{ |
{ |
140 |
const uint32_t edged_width2 = edged_width / 2; |
const uint32_t edged_width2 = edged_width / 2; |
141 |
const uint32_t width2 = width / 2; |
uint32_t width2; |
142 |
uint32_t i; |
uint32_t i; |
143 |
uint8_t *dst; |
uint8_t *dst; |
144 |
uint8_t *src; |
uint8_t *src; |
145 |
|
|
|
|
|
146 |
dst = image->y - (EDGE_SIZE + EDGE_SIZE * edged_width); |
dst = image->y - (EDGE_SIZE + EDGE_SIZE * edged_width); |
147 |
src = image->y; |
src = image->y; |
148 |
|
|
149 |
|
/* According to the Standard Clause 7.6.4, padding is done starting at 16 |
150 |
|
* pixel width and height multiples. This was not respected in old xvids */ |
151 |
|
if (bs_version == 0 || bs_version >= SETEDGES_BUG_BEFORE) { |
152 |
|
width = (width+15)&~15; |
153 |
|
height = (height+15)&~15; |
154 |
|
} |
155 |
|
|
156 |
|
width2 = width/2; |
157 |
|
|
158 |
for (i = 0; i < EDGE_SIZE; i++) { |
for (i = 0; i < EDGE_SIZE; i++) { |
159 |
memset(dst, *src, EDGE_SIZE); |
memset(dst, *src, EDGE_SIZE); |
160 |
memcpy(dst + EDGE_SIZE, src, width); |
memcpy(dst + EDGE_SIZE, src, width); |
262 |
n_ptr = refn->y; |
n_ptr = refn->y; |
263 |
h_ptr = refh->y; |
h_ptr = refh->y; |
264 |
v_ptr = refv->y; |
v_ptr = refv->y; |
|
hv_ptr = refhv->y; |
|
265 |
|
|
266 |
n_ptr -= offset; |
n_ptr -= offset; |
267 |
h_ptr -= offset; |
h_ptr -= offset; |
268 |
v_ptr -= offset; |
v_ptr -= offset; |
|
hv_ptr -= offset; |
|
269 |
|
|
270 |
|
/* Note we initialize the hv pointer later, as we can optimize code a bit |
271 |
|
* doing it down to up in quarterpel and up to down in halfpel */ |
272 |
if(quarterpel) { |
if(quarterpel) { |
273 |
|
|
274 |
for (y = 0; y < (edged_height - EDGE_SIZE); y += 8) { |
for (y = 0; y < (edged_height - EDGE_SIZE); y += 8) { |
290 |
n_ptr += stride_add; |
n_ptr += stride_add; |
291 |
} |
} |
292 |
|
|
293 |
h_ptr = refh->y; |
h_ptr = refh->y + (edged_height - EDGE_SIZE - EDGE_SIZE2)*edged_width - EDGE_SIZE2; |
294 |
h_ptr -= offset; |
hv_ptr = refhv->y + (edged_height - EDGE_SIZE - EDGE_SIZE2)*edged_width - EDGE_SIZE2; |
295 |
|
|
296 |
for (y = 0; y < (edged_height - EDGE_SIZE); y = y + 8) { |
for (y = 0; y < (edged_height - EDGE_SIZE); y = y + 8) { |
297 |
|
hv_ptr -= stride_add; |
298 |
|
h_ptr -= stride_add; |
299 |
|
hv_ptr -= EDGE_SIZE; |
300 |
|
h_ptr -= EDGE_SIZE; |
301 |
|
|
302 |
for (x = 0; x < (edged_width - EDGE_SIZE); x = x + 8) { |
for (x = 0; x < (edged_width - EDGE_SIZE); x = x + 8) { |
303 |
|
hv_ptr -= 8; |
304 |
|
h_ptr -= 8; |
305 |
interpolate8x8_6tap_lowpass_v(hv_ptr, h_ptr, edged_width, rounding); |
interpolate8x8_6tap_lowpass_v(hv_ptr, h_ptr, edged_width, rounding); |
|
hv_ptr += 8; |
|
|
h_ptr += 8; |
|
|
} |
|
|
|
|
|
hv_ptr += EDGE_SIZE; |
|
|
h_ptr += EDGE_SIZE; |
|
|
|
|
|
hv_ptr += stride_add; |
|
|
h_ptr += stride_add; |
|
306 |
} |
} |
307 |
} |
} |
308 |
else { |
} else { |
309 |
|
|
310 |
|
hv_ptr = refhv->y; |
311 |
|
hv_ptr -= offset; |
312 |
|
|
313 |
for (y = 0; y < (edged_height - EDGE_SIZE); y += 8) { |
for (y = 0; y < (edged_height - EDGE_SIZE); y += 8) { |
314 |
for (x = 0; x < (edged_width - EDGE_SIZE); x += 8) { |
for (x = 0; x < (edged_width - EDGE_SIZE); x += 8) { |
603 |
interlacing?rgbai_to_yv12_c:rgba_to_yv12_c, 4); |
interlacing?rgbai_to_yv12_c:rgba_to_yv12_c, 4); |
604 |
break; |
break; |
605 |
|
|
606 |
|
case XVID_CSP_ARGB: |
607 |
|
safe_packed_conv( |
608 |
|
src[0], src_stride[0], image->y, image->u, image->v, |
609 |
|
edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP), |
610 |
|
interlacing?argbi_to_yv12 : argb_to_yv12, |
611 |
|
interlacing?argbi_to_yv12_c: argb_to_yv12_c, 4); |
612 |
|
break; |
613 |
|
|
614 |
case XVID_CSP_YUY2: |
case XVID_CSP_YUY2: |
615 |
safe_packed_conv( |
safe_packed_conv( |
616 |
src[0], src_stride[0], image->y, image->u, image->v, |
src[0], src_stride[0], image->y, image->u, image->v, |
621 |
|
|
622 |
case XVID_CSP_YVYU: /* u/v swapped */ |
case XVID_CSP_YVYU: /* u/v swapped */ |
623 |
safe_packed_conv( |
safe_packed_conv( |
624 |
src[0], src_stride[0], image->y, image->v, image->y, |
src[0], src_stride[0], image->y, image->v, image->u, |
625 |
edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP), |
edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP), |
626 |
interlacing?yuyvi_to_yv12 :yuyv_to_yv12, |
interlacing?yuyvi_to_yv12 :yuyv_to_yv12, |
627 |
interlacing?yuyvi_to_yv12_c:yuyv_to_yv12_c, 2); |
interlacing?yuyvi_to_yv12_c:yuyv_to_yv12_c, 2); |
635 |
interlacing?uyvyi_to_yv12_c:uyvy_to_yv12_c, 2); |
interlacing?uyvyi_to_yv12_c:uyvy_to_yv12_c, 2); |
636 |
break; |
break; |
637 |
|
|
638 |
case XVID_CSP_I420: |
case XVID_CSP_I420: /* YCbCr == YUV == internal colorspace for MPEG */ |
639 |
yv12_to_yv12(image->y, image->u, image->v, edged_width, edged_width2, |
yv12_to_yv12(image->y, image->u, image->v, edged_width, edged_width2, |
640 |
src[0], src[0] + src_stride[0]*height, src[0] + src_stride[0]*height + (src_stride[0]/2)*height2, |
src[0], src[0] + src_stride[0]*height, src[0] + src_stride[0]*height + (src_stride[0]/2)*height2, |
641 |
src_stride[0], src_stride[0]/2, width, height, (csp & XVID_CSP_VFLIP)); |
src_stride[0], src_stride[0]/2, width, height, (csp & XVID_CSP_VFLIP)); |
642 |
break |
break; |
643 |
; |
|
644 |
case XVID_CSP_YV12: /* u/v swapped */ |
case XVID_CSP_YV12: /* YCrCb == YVA == U and V plane swapped */ |
645 |
yv12_to_yv12(image->y, image->v, image->u, edged_width, edged_width2, |
yv12_to_yv12(image->y, image->v, image->u, edged_width, edged_width2, |
646 |
src[0], src[0] + src_stride[0]*height, src[0] + src_stride[0]*height + (src_stride[0]/2)*height2, |
src[0], src[0] + src_stride[0]*height, src[0] + src_stride[0]*height + (src_stride[0]/2)*height2, |
647 |
src_stride[0], src_stride[0]/2, width, height, (csp & XVID_CSP_VFLIP)); |
src_stride[0], src_stride[0]/2, width, height, (csp & XVID_CSP_VFLIP)); |
648 |
break; |
break; |
649 |
|
|
650 |
case XVID_CSP_USER: |
case XVID_CSP_PLANAR: /* YCbCr with arbitrary pointers and different strides for Y and UV */ |
|
/*XXX: support for different u & v strides */ |
|
651 |
yv12_to_yv12(image->y, image->u, image->v, edged_width, edged_width2, |
yv12_to_yv12(image->y, image->u, image->v, edged_width, edged_width2, |
652 |
src[0], src[1], src[2], src_stride[0], src_stride[1], |
src[0], src[1], src[2], src_stride[0], src_stride[1], /* v: dst_stride[2] not yet supported */ |
653 |
width, height, (csp & XVID_CSP_VFLIP)); |
width, height, (csp & XVID_CSP_VFLIP)); |
654 |
break; |
break; |
655 |
|
|
717 |
int height, |
int height, |
718 |
uint32_t edged_width, |
uint32_t edged_width, |
719 |
uint8_t * dst[4], |
uint8_t * dst[4], |
720 |
uint32_t dst_stride[4], |
int dst_stride[4], |
721 |
int csp, |
int csp, |
722 |
int interlacing) |
int interlacing) |
723 |
{ |
{ |
779 |
interlacing?yv12_to_rgbai_c:yv12_to_rgba_c, 4); |
interlacing?yv12_to_rgbai_c:yv12_to_rgba_c, 4); |
780 |
return 0; |
return 0; |
781 |
|
|
782 |
|
case XVID_CSP_ARGB: |
783 |
|
safe_packed_conv( |
784 |
|
dst[0], dst_stride[0], image->y, image->u, image->v, |
785 |
|
edged_width, edged_width2, width, height, (csp & XVID_CSP_VFLIP), |
786 |
|
interlacing?yv12_to_argbi :yv12_to_argb, |
787 |
|
interlacing?yv12_to_argbi_c:yv12_to_argb_c, 4); |
788 |
|
return 0; |
789 |
|
|
790 |
case XVID_CSP_YUY2: |
case XVID_CSP_YUY2: |
791 |
safe_packed_conv( |
safe_packed_conv( |
792 |
dst[0], dst_stride[0], image->y, image->u, image->v, |
dst[0], dst_stride[0], image->y, image->u, image->v, |
811 |
interlacing?yv12_to_uyvyi_c:yv12_to_uyvy_c, 2); |
interlacing?yv12_to_uyvyi_c:yv12_to_uyvy_c, 2); |
812 |
return 0; |
return 0; |
813 |
|
|
814 |
case XVID_CSP_I420: |
case XVID_CSP_I420: /* YCbCr == YUV == internal colorspace for MPEG */ |
815 |
yv12_to_yv12(dst[0], dst[0] + dst_stride[0]*height, dst[0] + dst_stride[0]*height + (dst_stride[0]/2)*height2, |
yv12_to_yv12(dst[0], dst[0] + dst_stride[0]*height, dst[0] + dst_stride[0]*height + (dst_stride[0]/2)*height2, |
816 |
dst_stride[0], dst_stride[0]/2, |
dst_stride[0], dst_stride[0]/2, |
817 |
image->y, image->u, image->v, edged_width, edged_width2, |
image->y, image->u, image->v, edged_width, edged_width2, |
818 |
width, height, (csp & XVID_CSP_VFLIP)); |
width, height, (csp & XVID_CSP_VFLIP)); |
819 |
return 0; |
return 0; |
820 |
|
|
821 |
case XVID_CSP_YV12: /* u,v swapped */ |
case XVID_CSP_YV12: /* YCrCb == YVU == U and V plane swapped */ |
822 |
yv12_to_yv12(dst[0], dst[0] + dst_stride[0]*height, dst[0] + dst_stride[0]*height + (dst_stride[0]/2)*height2, |
yv12_to_yv12(dst[0], dst[0] + dst_stride[0]*height, dst[0] + dst_stride[0]*height + (dst_stride[0]/2)*height2, |
823 |
dst_stride[0], dst_stride[0]/2, |
dst_stride[0], dst_stride[0]/2, |
824 |
image->y, image->v, image->u, edged_width, edged_width2, |
image->y, image->v, image->u, edged_width, edged_width2, |
825 |
width, height, (csp & XVID_CSP_VFLIP)); |
width, height, (csp & XVID_CSP_VFLIP)); |
826 |
return 0; |
return 0; |
827 |
|
|
828 |
case XVID_CSP_USER : /* u,v swapped */ |
case XVID_CSP_PLANAR: /* YCbCr with arbitrary pointers and different strides for Y and UV */ |
829 |
yv12_to_yv12(dst[0], dst[1], dst[2], |
yv12_to_yv12(dst[0], dst[1], dst[2], |
830 |
dst_stride[0], dst_stride[1], /* v: dst_stride[2] */ |
dst_stride[0], dst_stride[1], /* v: dst_stride[2] not yet supported */ |
831 |
image->y, image->v, image->u, edged_width, edged_width2, |
image->y, image->u, image->v, edged_width, edged_width2, |
832 |
width, height, (csp & XVID_CSP_VFLIP)); |
width, height, (csp & XVID_CSP_VFLIP)); |
833 |
return 0; |
return 0; |
834 |
|
|
898 |
uint16_t width, |
uint16_t width, |
899 |
uint16_t height) |
uint16_t height) |
900 |
{ |
{ |
901 |
int diff, x, y; |
int y, bwidth, bheight; |
902 |
long sse=0; |
long sse=0; |
903 |
|
|
904 |
for (y = 0; y < height; y++) { |
bwidth = width & (~0x07); |
905 |
|
bheight = height & (~0x07); |
906 |
|
|
907 |
|
/* Compute the 8x8 integer part */ |
908 |
|
for (y = 0; y<bheight; y += 8) { |
909 |
|
int x; |
910 |
|
|
911 |
|
/* Compute sse for the band */ |
912 |
|
for (x = 0; x<bwidth; x += 8) |
913 |
|
sse += sse8_8bit(orig + x, recon + x, stride); |
914 |
|
|
915 |
|
/* remaining pixels of the 8 pixels high band */ |
916 |
|
for (x = bwidth; x < width; x++) { |
917 |
|
int diff; |
918 |
|
diff = *(orig + 0*stride + x) - *(recon + 0*stride + x); |
919 |
|
sse += diff * diff; |
920 |
|
diff = *(orig + 1*stride + x) - *(recon + 1*stride + x); |
921 |
|
sse += diff * diff; |
922 |
|
diff = *(orig + 2*stride + x) - *(recon + 2*stride + x); |
923 |
|
sse += diff * diff; |
924 |
|
diff = *(orig + 3*stride + x) - *(recon + 3*stride + x); |
925 |
|
sse += diff * diff; |
926 |
|
diff = *(orig + 4*stride + x) - *(recon + 4*stride + x); |
927 |
|
sse += diff * diff; |
928 |
|
diff = *(orig + 5*stride + x) - *(recon + 5*stride + x); |
929 |
|
sse += diff * diff; |
930 |
|
diff = *(orig + 6*stride + x) - *(recon + 6*stride + x); |
931 |
|
sse += diff * diff; |
932 |
|
diff = *(orig + 7*stride + x) - *(recon + 7*stride + x); |
933 |
|
sse += diff * diff; |
934 |
|
} |
935 |
|
|
936 |
|
orig += 8*stride; |
937 |
|
recon += 8*stride; |
938 |
|
} |
939 |
|
|
940 |
|
/* Compute the down rectangle sse */ |
941 |
|
for (y = bheight; y < height; y++) { |
942 |
|
int x; |
943 |
for (x = 0; x < width; x++) { |
for (x = 0; x < width; x++) { |
944 |
|
int diff; |
945 |
diff = *(orig + x) - *(recon + x); |
diff = *(orig + x) - *(recon + x); |
946 |
sse += diff * diff; |
sse += diff * diff; |
947 |
} |
} |
948 |
orig += stride; |
orig += stride; |
949 |
recon += stride; |
recon += stride; |
950 |
} |
} |
951 |
return sse; |
|
952 |
|
return (sse); |
953 |
} |
} |
954 |
|
|
955 |
#if 0 |
#if 0 |
1078 |
} |
} |
1079 |
|
|
1080 |
void |
void |
1081 |
output_slice(IMAGE * cur, int std, int width, xvid_image_t* out_frm, int mbx, int mby,int mbl) { |
output_slice(IMAGE * cur, int stride, int width, xvid_image_t* out_frm, int mbx, int mby,int mbl) { |
1082 |
uint8_t *dY,*dU,*dV,*sY,*sU,*sV; |
uint8_t *dY,*dU,*dV,*sY,*sU,*sV; |
1083 |
int std2 = std >> 1; |
int stride2 = stride >> 1; |
1084 |
int w = mbl << 4, w2,i; |
int w = mbl << 4, w2,i; |
1085 |
|
|
1086 |
if(w > width) |
if(w > width) |
1090 |
dY = (uint8_t*)out_frm->plane[0] + (mby << 4) * out_frm->stride[0] + (mbx << 4); |
dY = (uint8_t*)out_frm->plane[0] + (mby << 4) * out_frm->stride[0] + (mbx << 4); |
1091 |
dU = (uint8_t*)out_frm->plane[1] + (mby << 3) * out_frm->stride[1] + (mbx << 3); |
dU = (uint8_t*)out_frm->plane[1] + (mby << 3) * out_frm->stride[1] + (mbx << 3); |
1092 |
dV = (uint8_t*)out_frm->plane[2] + (mby << 3) * out_frm->stride[2] + (mbx << 3); |
dV = (uint8_t*)out_frm->plane[2] + (mby << 3) * out_frm->stride[2] + (mbx << 3); |
1093 |
sY = cur->y + (mby << 4) * std + (mbx << 4); |
sY = cur->y + (mby << 4) * stride + (mbx << 4); |
1094 |
sU = cur->u + (mby << 3) * std2 + (mbx << 3); |
sU = cur->u + (mby << 3) * stride2 + (mbx << 3); |
1095 |
sV = cur->v + (mby << 3) * std2 + (mbx << 3); |
sV = cur->v + (mby << 3) * stride2 + (mbx << 3); |
1096 |
|
|
1097 |
for(i = 0 ; i < 16 ; i++) { |
for(i = 0 ; i < 16 ; i++) { |
1098 |
memcpy(dY,sY,w); |
memcpy(dY,sY,w); |
1099 |
dY += out_frm->stride[0]; |
dY += out_frm->stride[0]; |
1100 |
sY += std; |
sY += stride; |
1101 |
} |
} |
1102 |
for(i = 0 ; i < 8 ; i++) { |
for(i = 0 ; i < 8 ; i++) { |
1103 |
memcpy(dU,sU,w2); |
memcpy(dU,sU,w2); |
1104 |
dU += out_frm->stride[1]; |
dU += out_frm->stride[1]; |
1105 |
sU += std2; |
sU += stride2; |
1106 |
} |
} |
1107 |
for(i = 0 ; i < 8 ; i++) { |
for(i = 0 ; i < 8 ; i++) { |
1108 |
memcpy(dV,sV,w2); |
memcpy(dV,sV,w2); |
1109 |
dV += out_frm->stride[2]; |
dV += out_frm->stride[2]; |
1110 |
sV += std2; |
sV += stride2; |
1111 |
} |
} |
1112 |
} |
} |
1113 |
|
|
1137 |
p += edged_width/2; |
p += edged_width/2; |
1138 |
} |
} |
1139 |
} |
} |
|
|
|
|
|
|
|
/* reduced resolution deblocking filter |
|
|
block = block size (16=rrv, 8=full resolution) |
|
|
flags = XVID_DEC_YDEBLOCK|XVID_DEC_UVDEBLOCK |
|
|
*/ |
|
|
void |
|
|
image_deblock_rrv(IMAGE * img, int edged_width, |
|
|
const MACROBLOCK * mbs, int mb_width, int mb_height, int mb_stride, |
|
|
int block, int flags) |
|
|
{ |
|
|
const int edged_width2 = edged_width /2; |
|
|
const int nblocks = block / 8; /* skals code uses 8pixel block uints */ |
|
|
int i,j; |
|
|
|
|
|
/* luma: j,i in block units */ |
|
|
|
|
|
for (j = 1; j < mb_height*2; j++) /* horizontal deblocking */ |
|
|
for (i = 0; i < mb_width*2; i++) |
|
|
{ |
|
|
if (mbs[(j-1)/2*mb_stride + (i/2)].mode != MODE_NOT_CODED || |
|
|
mbs[(j+0)/2*mb_stride + (i/2)].mode != MODE_NOT_CODED) |
|
|
{ |
|
|
hfilter_31(img->y + (j*block - 1)*edged_width + i*block, |
|
|
img->y + (j*block + 0)*edged_width + i*block, nblocks); |
|
|
} |
|
|
} |
|
|
|
|
|
for (j = 0; j < mb_height*2; j++) /* vertical deblocking */ |
|
|
for (i = 1; i < mb_width*2; i++) |
|
|
{ |
|
|
if (mbs[(j/2)*mb_stride + (i-1)/2].mode != MODE_NOT_CODED || |
|
|
mbs[(j/2)*mb_stride + (i+0)/2].mode != MODE_NOT_CODED) |
|
|
{ |
|
|
vfilter_31(img->y + (j*block)*edged_width + i*block - 1, |
|
|
img->y + (j*block)*edged_width + i*block + 0, |
|
|
edged_width, nblocks); |
|
|
} |
|
|
} |
|
|
|
|
|
|
|
|
|
|
|
/* chroma */ |
|
|
|
|
|
for (j = 1; j < mb_height; j++) /* horizontal deblocking */ |
|
|
for (i = 0; i < mb_width; i++) |
|
|
{ |
|
|
if (mbs[(j-1)*mb_stride + i].mode != MODE_NOT_CODED || |
|
|
mbs[(j+0)*mb_stride + i].mode != MODE_NOT_CODED) |
|
|
{ |
|
|
hfilter_31(img->u + (j*block - 1)*edged_width2 + i*block, |
|
|
img->u + (j*block + 0)*edged_width2 + i*block, nblocks); |
|
|
hfilter_31(img->v + (j*block - 1)*edged_width2 + i*block, |
|
|
img->v + (j*block + 0)*edged_width2 + i*block, nblocks); |
|
|
} |
|
|
} |
|
|
|
|
|
for (j = 0; j < mb_height; j++) /* vertical deblocking */ |
|
|
for (i = 1; i < mb_width; i++) |
|
|
{ |
|
|
if (mbs[j*mb_stride + i - 1].mode != MODE_NOT_CODED || |
|
|
mbs[j*mb_stride + i + 0].mode != MODE_NOT_CODED) |
|
|
{ |
|
|
vfilter_31(img->u + (j*block)*edged_width2 + i*block - 1, |
|
|
img->u + (j*block)*edged_width2 + i*block + 0, |
|
|
edged_width2, nblocks); |
|
|
vfilter_31(img->v + (j*block)*edged_width2 + i*block - 1, |
|
|
img->v + (j*block)*edged_width2 + i*block + 0, |
|
|
edged_width2, nblocks); |
|
|
} |
|
|
} |
|
|
|
|
|
|
|
|
} |
|
|
|
|