Matt,
this is likely a rounding bug somewhere, but I couldn't reproduce with
what you provided (with GDAL master and 3.3.2 on Linux). I had to fix a
few hardcoded paths in the VRTs and create
SPOT7_FiftyMileCkSixtyMileR_321_150cm_19Jun2015_utm7.tif and
SPOT7_FiftyMileCkSixtyMileR_432_150cm_19Jun2015_utm7.tif with
gdal_create SPOT7_FiftyMileCkSixtyMileR_321_150cm_19Jun2015_utm7.tif
-outsize 31918 34252 -bands 4 -co tiled=yes -co sparse_ok=yes
gdaladdo SPOT7_FiftyMileCkSixtyMileR_321_150cm_19Jun2015_utm7.tif
--config SPARSE_OK_OVERVIEW YES # SPARSE_OK_OVERVIEW only taken
into account with GDAL master
I also create a .vrt.msk file with
gdal_create SPOT7_FiftyMileCkSixtyMileR_321_150cm_19Jun2015_utm7.vrt.msk
-outsize 31912 34252 -co sparse_ok=yes -mo INTERNAL_MASK_FLAGS_1=2 -of gtiff
gdaladdo SPOT7_FiftyMileCkSixtyMileR_321_150cm_19Jun2015_utm7.vrt.msk
--config SPARSE_OK_OVERVIEW YES
I may have missed something. It would help if your reproducer was self
contained.
Even
Le 08/12/2021 à 18:35, [email protected] a écrit :
Hello,
Why is gdal off by one for the access window? I’m using standard
utilities, no custom code. This is only happening (so far) on one file
out of dozens.
```
$ gdal_translate -co compress=zstd -co predictor=yes -co level=17 ^
-co bigtiff=yes -of cog INFILE.rgb-nir.vrt OUTFILE.tif
Input file size is 31918, 34252
0.ERROR 5: Access window out of range in RasterIO(). Requested
(0,0) of size 250x268 on raster of 249x267.
```
This is one step of a process where I’m taking two 3-band
jpeg-in-geotiff images, defining a nodata mask, and creating a final
4-band cloud geotiff. Preceding steps to this one are:
gdal_translate -b 1 RGB.tif b-red.vrt
gdal_translate -b 2 RGB.tif b-grn.vrt
gdal_translate -b 3 RGB.tif b-blu.vrt
gdal_translate -b 1 432.tif b-nir.vrt
gdalbuildvrt -separate INFILE.rgb-nir.vrt ^
b-red.vrt b-grn.vrt b-blu.vrt b-nir.vrt
gdal_translate INFILE.rgb-nir.vrt xx-to-mask.tif
nearblack –setmask xx-to-mask.tif
move xx-mask.tif.msk INFILE.rgb-nir.vrt.msk
del xx-mask.tif
gdal_translate INFILE.rgb-nir.vrt OUTFILE.tif
(Edited for legibility, options used are left out)
Gdalinfo reports the size of each image are the same:
$ findstr "Size" *.gdal-info
RGB.b-blu.vrt.gdal-info:Size is 31918, 34252
RGB.b-blu.vrt.gdal-info:Pixel Size =
(1.500000000000000,-1.500000000000000)
RGB.b-grn.vrt.gdal-info:Size is 31918, 34252
RGB.b-grn.vrt.gdal-info:Pixel Size =
(1.500000000000000,-1.500000000000000)
RGB.b-red.vrt.gdal-info:Size is 31918, 34252
RGB.b-red.vrt.gdal-info:Pixel Size =
(1.500000000000000,-1.500000000000000)
432.b-nir.vrt.gdal-info:Size is 31918, 34252
432.b-nir.vrt.gdal-info:Pixel Size =
(1.500000000000000,-1.500000000000000)
INFILE.rgb-nir.vrt.gdal-info:Size is 31918, 34252
INFILE.rgb-nir.vrt.gdal-info:Pixel Size =
(1.500000000000000,-1.500000000000000)
INFILE.rgb-nir.vrt.msk.gdal-info:Size is 31918, 34252
The source VRT and gdalinfo reports are attached.
GDAL 3.3.2, released 2021/09/01 via Qgis 3.22 on Windows 10.
*Matt Wilkie*
Geomatics Developer & Administrator
Environment | Technology, Innovation and Mapping
T 867-667-8133 | _Yukon.ca <http://yukon.ca/>_
/Hours: 08:30-16:30, Mon-Wed: Office, Thu: Remote, Fri: Away./
_______________________________________________
gdal-dev mailing list
[email protected]
https://lists.osgeo.org/mailman/listinfo/gdal-dev
--
http://www.spatialys.com
My software is free, but my time generally not.
_______________________________________________
gdal-dev mailing list
[email protected]
https://lists.osgeo.org/mailman/listinfo/gdal-dev
