You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
If you use RasterDataset with a raster that has non-square pixels (i.e. where x-resolution is different than y-resolution), then RasterDataset will record the resolution of the entire raster as that of the first dimension (probably y-resolution). This is a problem in at least two places:
Every __getitem__ call to the RasterDataset will result in a resampling, which may be unexpected, and will definitely be slow
When using Samplers with a patch size in pixel units the bounding boxes that index into the RasterDataset will be generated assuming a square resolution so what you get back won't match up with what you expect.
Steps to reproduce
Use a RasterDataset with a raster that has non-square pixels.
Version
latest
The text was updated successfully, but these errors were encountered:
Description
If you use RasterDataset with a raster that has non-square pixels (i.e. where x-resolution is different than y-resolution), then RasterDataset will record the resolution of the entire raster as that of the first dimension (probably y-resolution). This is a problem in at least two places:
__getitem__
call to the RasterDataset will result in a resampling, which may be unexpected, and will definitely be slowSteps to reproduce
Use a RasterDataset with a raster that has non-square pixels.
Version
latest
The text was updated successfully, but these errors were encountered: