videolan/vlc
1
Fork 0
mirror of https://code.videolan.org/videolan/vlc synced 2024-07-25 09:41:30 +02:00
Code Releases Activity

Fix transform plugin 90-degree rotations on I422.

Browse Source 
I422 is a planar chroma with non-square pixels. It requires some pixel
play when rotating the image at a 90-degree angle. For other transformations
we just fall back to the original function.
This commit is contained in:
Sam Hocevar 2008-04-07 16:47:49 +02:00
parent 19ac4fcc29
commit cd079331b9
2 changed files with 172 additions and 23 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

13
modules/video_filter/filter_picture.h
View File

@ -21,7 +21,8 @@
* Foundation, Inc., 51 Franklin Street, Fifth Floor, Boston MA 02110-1301, USA.
*****************************************************************************/
#define CASE_PLANAR_YUV \
/* FIXME: do all of these really have square pixels? */
#define CASE_PLANAR_YUV_SQUARE \
case VLC_FOURCC('I','4','2','0'): \
case VLC_FOURCC('I','Y','U','V'): \
case VLC_FOURCC('J','4','2','0'): \
@ -29,12 +30,18 @@
case VLC_FOURCC('I','4','1','1'): \
case VLC_FOURCC('I','4','1','0'): \
case VLC_FOURCC('Y','V','U','9'): \
case VLC_FOURCC('I','4','2','2'): \
case VLC_FOURCC('J','4','2','2'): \
case VLC_FOURCC('I','4','4','4'): \
case VLC_FOURCC('J','4','4','4'): \
case VLC_FOURCC('Y','U','V','A'):
#define CASE_PLANAR_YUV_NONSQUARE \
case VLC_FOURCC('I','4','2','2'): \
case VLC_FOURCC('J','4','2','2'):
#define CASE_PLANAR_YUV \
CASE_PLANAR_YUV_SQUARE \
CASE_PLANAR_YUV_NONSQUARE \
#define CASE_PACKED_YUV_422 \
case VLC_FOURCC('U','Y','V','Y'): \
case VLC_FOURCC('U','Y','N','V'): \

182
modules/video_filter/transform.c
View File

@ -52,6 +52,7 @@ static void End ( vout_thread_t * );
static void Render ( vout_thread_t *, picture_t * );
static void FilterPlanar( vout_thread_t *, const picture_t *, picture_t * );
static void FilterI422( vout_thread_t *, const picture_t *, picture_t * );
static void FilterYUYV( vout_thread_t *, const picture_t *, picture_t * );
static int SendEvents( vlc_object_t *, char const *,
@ -145,11 +146,16 @@ static int Create( vlc_object_t *p_this )
switch( p_vout->fmt_in.i_chroma )
{
CASE_PLANAR_YUV
CASE_PLANAR_YUV_SQUARE
case VLC_FOURCC('G','R','E','Y'):
p_vout->p_sys->pf_filter = FilterPlanar;
break;
case VLC_FOURCC('I','4','2','2'):
case VLC_FOURCC('J','4','2','2'):
p_vout->p_sys->pf_filter = FilterI422;
break;
CASE_PACKED_YUV_422
p_vout->p_sys->pf_filter = FilterYUYV;
break;
@ -551,6 +557,156 @@ static void FilterPlanar( vout_thread_t *p_vout,
}
}
static void FilterI422( vout_thread_t *p_vout,
const picture_t *p_pic, picture_t *p_outpic )
{
int i_index;
switch( p_vout->p_sys->i_mode )
{
case TRANSFORM_MODE_180:
case TRANSFORM_MODE_HFLIP:
case TRANSFORM_MODE_VFLIP:
/* Fall back on the default implementation */
FilterPlanar( p_vout, p_pic, p_outpic );
return;
case TRANSFORM_MODE_90:
for( i_index = 0 ; i_index < p_pic->i_planes ; i_index++ )
{
int i_pitch = p_pic->p[i_index].i_pitch;
uint8_t *p_in = p_pic->p[i_index].p_pixels;
uint8_t *p_out = p_outpic->p[i_index].p_pixels;
uint8_t *p_out_end = p_out +
p_outpic->p[i_index].i_visible_lines *
p_outpic->p[i_index].i_pitch;
if( i_index == 0 )
{
for( ; p_out < p_out_end ; )
{
uint8_t *p_line_end;
p_out_end -= p_outpic->p[i_index].i_pitch
- p_outpic->p[i_index].i_visible_pitch;
p_line_end = p_in + p_pic->p[i_index].i_visible_lines *
i_pitch;
for( ; p_in < p_line_end ; )
{
p_line_end -= i_pitch;
*(--p_out_end) = *p_line_end;
}
p_in++;
}
}
else /* i_index == 1 or 2 */
{
for( ; p_out < p_out_end ; )
{
uint8_t *p_line_end, *p_out_end2;
p_out_end -= p_outpic->p[i_index].i_pitch
- p_outpic->p[i_index].i_visible_pitch;
p_out_end2 = p_out_end - p_outpic->p[i_index].i_pitch;
p_line_end = p_in + p_pic->p[i_index].i_visible_lines *
i_pitch;
for( ; p_in < p_line_end ; )
{
uint8_t p1, p2;
p_line_end -= i_pitch;
p1 = *p_line_end;
p_line_end -= i_pitch;
p2 = *p_line_end;
/* Trick for (x+y)/2 without overflow, based on
* x + y == (x ^ y) + 2 * (x & y) */
*(--p_out_end) = (p1 & p2) + ((p1 ^ p2) / 2);
*(--p_out_end2) = (p1 & p2) + ((p1 ^ p2) / 2);
}
p_out_end = p_out_end2;
p_in++;
}
}
}
break;
case TRANSFORM_MODE_270:
for( i_index = 0 ; i_index < p_pic->i_planes ; i_index++ )
{
int i_pitch = p_pic->p[i_index].i_pitch;
uint8_t *p_in = p_pic->p[i_index].p_pixels;
uint8_t *p_out = p_outpic->p[i_index].p_pixels;
uint8_t *p_out_end = p_out +
p_outpic->p[i_index].i_visible_lines *
p_outpic->p[i_index].i_pitch;
if( i_index == 0 )
{
for( ; p_out < p_out_end ; )
{
uint8_t *p_in_end;
p_in_end = p_in + p_pic->p[i_index].i_visible_lines *
i_pitch;
for( ; p_in < p_in_end ; )
{
p_in_end -= i_pitch;
*p_out++ = *p_in_end;
}
p_out += p_outpic->p[i_index].i_pitch
- p_outpic->p[i_index].i_visible_pitch;
p_in++;
}
}
else /* i_index == 1 or 2 */
{
for( ; p_out < p_out_end ; )
{
uint8_t *p_in_end, *p_out2;
p_in_end = p_in + p_pic->p[i_index].i_visible_lines *
i_pitch;
p_out2 = p_out + p_outpic->p[i_index].i_pitch;
for( ; p_in < p_in_end ; )
{
uint8_t p1, p2;
p_in_end -= i_pitch;
p1 = *p_in_end;
p_in_end -= i_pitch;
p2 = *p_in_end;
/* Trick for (x+y)/2 without overflow, based on
* x + y == (x ^ y) + 2 * (x & y) */
*p_out++ = (p1 & p2) + ((p1 ^ p2) / 2);
*p_out2++ = (p1 & p2) + ((p1 ^ p2) / 2);
}
p_out2 += p_outpic->p[i_index].i_pitch
- p_outpic->p[i_index].i_visible_pitch;
p_out = p_out2;
p_in++;
}
}
}
break;
default:
break;
}
}
static void FilterYUYV( vout_thread_t *p_vout,
const picture_t *p_pic, picture_t *p_outpic )
{
@ -562,6 +718,11 @@ static void FilterYUYV( vout_thread_t *p_vout,
switch( p_vout->p_sys->i_mode )
{
case TRANSFORM_MODE_HFLIP:
/* Fall back on the default implementation */
FilterPlanar( p_vout, p_pic, p_outpic );
return;
case TRANSFORM_MODE_90:
for( i_index = 0 ; i_index < p_pic->i_planes ; i_index++ )
{
@ -684,25 +845,6 @@ static void FilterYUYV( vout_thread_t *p_vout,
}
break;
case TRANSFORM_MODE_HFLIP:
for( i_index = 0 ; i_index < p_pic->i_planes ; i_index++ )
{
uint8_t *p_in = p_pic->p[i_index].p_pixels;
uint8_t *p_in_end = p_in + p_pic->p[i_index].i_visible_lines
* p_pic->p[i_index].i_pitch;
uint8_t *p_out = p_outpic->p[i_index].p_pixels;
for( ; p_in < p_in_end ; )
{
p_in_end -= p_pic->p[i_index].i_pitch;
p_vout->p_libvlc->pf_memcpy( p_out, p_in_end,
p_pic->p[i_index].i_visible_pitch );
p_out += p_pic->p[i_index].i_pitch;
}
}
break;
case TRANSFORM_MODE_VFLIP:
for( i_index = 0 ; i_index < p_pic->i_planes ; i_index++ )
{