Opened 15 years ago
Closed 15 years ago
#3060 closed defect (fixed)
gdalwarp very slow
Reported by: | ysiddiqui | Owned by: | warmerdam |
---|---|---|---|
Priority: | normal | Milestone: | |
Component: | Utilities | Version: | svn-trunk |
Severity: | normal | Keywords: | CUTLINE_ALL_TOUCHED gdalwarp |
Cc: |
Description
Running gdalwarp from one identical file to another with a cutline, the process bogs down before 10%. The process was left to run all weekend, and it only got to just below 30%. The following is the command that was used:
C:\apps\OSGEO4W\apps\gdal-dev\bin\gdalwarp -order 1 -wo "CUTLINE_ALL_TOUCHED=YES" -r near -cutline E:/POPS/temp/req30598_aoi.shp -cblend 0 E:\POPS\temp\piece1423858_temp0.tif E:\POPS\temp\piece1423858_temp1.tif
The AOI and gdalinfo reports for each of the files is attached. The image files can be found at the following FTP site:
Site: ftp.i3.com Login: fwarmerdam Password: coreopsis Filename: piece1423858.zip
Attachments (3)
Change History (6)
by , 15 years ago
Attachment: | req30598_aoi.zip added |
---|
comment:1 by , 15 years ago
Component: | default → Utilities |
---|---|
Keywords: | CUTLINE_ALL_TOUCHED gdalwarp added |
Status: | new → assigned |
Version: | unspecified → svn-trunk |
I have succeeded in reproducing this problem I observe that the warp runs in less than a minute if CUTLINE_ALL_TOUCHED is false and the debugger tells me things are hung up in rasterization code. So there is clearly some sort of bug in the all_touched code.
Digging.
comment:2 by , 15 years ago
The problem related to asymmetric rounding of values at zero and resulted in the line rasterization algorithm taking extremely small steps. Corrected in trunk (r17384). I'm rebuilding the gdal-dev package for OSGeo4W now.
comment:3 by , 15 years ago
Resolution: | → fixed |
---|---|
Status: | assigned → closed |
The gdal-dev package in OSGeo4W has been upgraded with the fix, though I have not actually retested it in that context. Please reopen if a problem persists.
cutline shapefile