Artifacts on Export From Darktable

I’ve had a couple of quite heavily processed exports when using the negadoctor module with large blocky artifacts, in case it’s useful to know. Only appears sporadically and it’s not persistent.

Re-exported after some additional processing and this one is now fine:

1 Like

More giant pixel artefacts (in the sky)

You’re going to need to provide more than a catalog of pictures if you’re hoping for a resolution. Help us help you.


Yep, thanks. I figured so but have other priorities to deal with so just an fyi/placeholder for now.

1 Like

It certainly looks like a “tiling issue” with internal correction data not calculated from whole image.

If you want some investigation we would need debug log with -d common plus a raw and corresponding xmp.

20230813_0082.CR2 (64.3 MB)
20230813_0082_01.CR2.xmp (34.2 KB)

In this example, I can see the tile in the darktable but not in the exported jpeg, previous examples it’s been the other way around. The tile artifact appears when I use both profiled noise reduction and highlight reconstruction in this example. Can you be more specific on generating the debug log? I see the terminal command but do I need to specify Darktable or something. Thanks

This shows an artifact in the jpeg when I turn on raw chromatic aberations

Tiling happens at runtime for every module that doesn’t fit into memory with all module specific requirements. The requirements while exporting depend on settings especially the “high quality resampling” , memory requirements for darkroom are different.

So to analyse what might introduce the artefacts you should call dt from a terminal like
darktable -d pipe -d opencl -d tiling

Well, i tried your files, can’t reproduce any tiling problem even on my pretty small intel notebook.