@lucfr this switches back to your rio-cogeo function and hopefully now creating the overviews and levels correctly.
Looked good on first try, but not checked in detail.
I still fail to reproduce what rio-cogeo does with pure gdal use, although I checked all flags the package sets. So it seems to work now (correct internal masking, no nodata value, no band artifacts), but I don't know why.
closes #22
@lucfr this switches back to your rio-cogeo function and hopefully now creating the overviews and levels correctly. Looked good on first try, but not checked in detail.
I still fail to reproduce what rio-cogeo does with pure gdal use, although I checked all flags the package sets. So it seems to work now (correct internal masking, no nodata value, no band artifacts), but I don't know why.