Hi folks I have a VRT which is a mosaic of several thousand s3-stored COGs. Doing any kind of low-res downsampling from this VRT causes very slow performance and high memory usage because it causes requests to all the individual source tiles.
I'd like to add external overviews to the VRT via a `.vrt.ovr` file – and it appears I can do this with gdaladdo. However, a naive call to gdaladdo seems to ignore the overviews on all the source files, rendering the overviews by downloading every pixel from the full-size sources. This is a nonstarter in this environment where the full set of source tiles is enormous and remote. So, two questions: 1. Is there a way to force gdaladdo to use the COG overviews from the source VRT's source tiles? 2. If not, are there any other ways to generate a `.vrt.ovr` file which would use the source overviews? I have tried using gdal_translate to produce a .tif of the right pixel size and then renaming it to `.vrt.ovr`, but it doesn't seem to be used by GDAL – so I guess there's more to `ovr` files than just TIF files with the right name! -- Regards, Craig Platform Engineer Koordinates +64 21 256 9488 <+64%2021%20256%209488> / koordinates.com / @koordinates <https://twitter.com/koordinates>
_______________________________________________ gdal-dev mailing list gdal-dev@lists.osgeo.org https://lists.osgeo.org/mailman/listinfo/gdal-dev