sergiotarxz
/
libvips
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

turn on smalltile 

2D locality helps a lot with many subimages
This commit is contained in:
John Cupitt 2018-12-07 16:43:44 +00:00
parent 2a32a95be9
commit 6bc564a1e1
2 changed files with 7 additions and 3 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
ChangeLog
View File

@ -9,8 +9,9 @@
- add bandand()/or()/eor() to cplusplus binding [clcaalu] - add bandand()/or()/eor() to cplusplus binding [clcaalu]
- implement shrink-on-load for tiff pyramids [jcupitt] - implement shrink-on-load for tiff pyramids [jcupitt]
- added vips_image_set_blob_copy() [jcupitt] - added vips_image_set_blob_copy() [jcupitt]
- don't stop composite on first non-transparent image [felixbuenemann] - don't stop composite on first non-transparent image [felixbuenemann, GDmac]
- add vips_rect_overlapsrect() - add vips_rect_overlapsrect()
- composite is much faster at positioning subimages
21/11/18 started 8.7.3 21/11/18 started 8.7.3
- fix infinite loop for autofit with non-scaleable font - fix infinite loop for autofit with non-scaleable font

7
libvips/conversion/composite.cpp
View File

@ -10,7 +10,7 @@
* 11/8/18 [medakk] * 11/8/18 [medakk]
* - x/y params let you position images * - x/y params let you position images
* 27/11/18 * 27/11/18
* - don't stop on first non-transparent image [felixbuenemann] * - don't stop on first non-transparent image [felixbuenemann, GDmac]
* 6/12/18 * 6/12/18
* - do our own subimage positioning * - do our own subimage positioning
*/ */
@ -1356,8 +1356,11 @@ vips_composite_base_build( VipsObject *object )
return( -1 ); return( -1 );
in = format; in = format;
/* We want locality, so that we only prepare a few subimages each
* time.
*/
if( vips_image_pipeline_array( conversion->out, if( vips_image_pipeline_array( conversion->out,
VIPS_DEMAND_STYLE_THINSTRIP, in ) ) VIPS_DEMAND_STYLE_SMALLTILE, in ) )
return( -1 ); return( -1 );
/* The output image is always the size of the base image. /* The output image is always the size of the base image.