Even Rouault

That sounds like an interesting idea of optimization. It should probably only be used after verifying that the columns of the table are of a known type for which we...

> This issue needs revisiting now that upstream libjxl seems to have added some support for this in [libjxl/libjxl#1812](https://github.com/libjxl/libjxl/pull/1812) Implemented in https://github.com/OSGeo/gdal/pull/7198. I've verified that lossless mode is lossless, and...

> Here's a hardcoded diff where lossy mode on a 16bit image which only has 12bits of range works correctly (i.e. results are visually identical at distance=1): I guess this...

- There were test failures: https://github.com/OSGeo/gdal/actions/runs/3739616110/jobs/6347049360 - This needs to be rebased on top of latest HEAD "git fetch origin master; git rebase origin/master" (assuming "origin" is the name of...

your PR is in a wrong state with hundreds of files appearing modified. You probably git merge instead of git rebase. If you can't find the right git incantation to...

@jratike80 Regarding "Nearblack had problems with some concave nodata areas in the middle of the image", on your example ``nearblack 187082488-6b907631-9f82-48c5-b696-9b886512225a.png -o out.tif -color 0,0,0 -color 255,255,255 -setalpha -nb 0``...

Nearblack floodfill algorithm implemented in https://github.com/OSGeo/gdal/pull/7877

@hobu I'm not totally clear on what you've in mind if all bands are to be used, rather than selecting a single band. In the situation when there is no...

> When the images are "natural" images like RGB or RGB-NIR I do not think that a footprint per band is a good default. Every band tend to have extreme...

hum, for multi-band, https://github.com/stac-utils/stactools/blob/main/src/stactools/core/utils/raster_footprint.py seems to take the union of the validity masks: https://github.com/stac-utils/stactools/blob/main/src/stactools/core/utils/raster_footprint.py#LL263C26-L263C26 (one of the challenges compared to existing approaches, is that given the GDAL philosophy of being...