You could use osr's CoordinateTransformation to evaluate the new location of
the four corners in the destination spatial reference:
From there a min/max on the four coordinates will give you the extent of the
new image space.
If the input and output CRS are very dissimilar you might want to evaluate a
regular grid instead to get a more representative extent.
Physical Sciences Specialist
Canadian Ice Service, Meteorological Service of Canada
Environment and Climate Change Canada / Government of Canada
benjamin.descha...@canada.ca / Tel: 613-943-4293
Spécialiste des sciences physiques
Service Canadien des Glaces, Service Météorologique du Canada,
Environnement et Changement climatique Canada / Gouvernement du Canada
benjamin.descha...@canada.ca / Tél. : 613-943-4293
From: Bang Pham Huu [mailto:b.pham...@jacobs-university.de]
Sent: May 17, 2018 5:34 AM
Subject: [gdal-dev] How to estimate the width/height size of output image in a
target CRS without running gdalwarp?
I have a hard problem to find the width/height of a 2D GeoTiff file which is
reprojected from a source CRS (e.g: EPSG:4326) to target CRS
(e.g: EPSG:3857) without running gdalwarp. Because, the file could be large
(GBs) and I don't need the projected output file, just need the width/height of
What could be done with gdalwarp is:
gdalwarp -s_srs EPSG:4326 -t_srs EPSG:3857 full.tif full_3857.tif
Size is 879, 811 (width/height).
I've searched a lot and what seems to be good without doing the real projection
is this gdal function: GDALSuggestedWarpOutput2
However, it doesn't help because my application developed in Java and it uses
gdal-java (GDAL JNI)
as library. Unfortunately, I cannot find this GDALSuggestedWarpOutput2() from
gdal-java http://gdal.org/java/overview-summary.html then cannot invoke this
C++ function from my application to test.
Can anyone please give me a hint to solve this problem?
gdal-dev mailing list