darktable 2.0.7 was announced today with this:
Base Support (fixes, was broken in 2.0.6, apologies for inconvenience)

Le 07/10/2016 à 15:00, Roman Lebedev a écrit :
As you may have seen, an issue with white level for some
nikon cameras (or so i thought) was discovered by me:

Web page says:

For that reason, many nikon cameras in cameras.xml had completely
arbitrary whitelevel specified,
which resulted in wrong image => either way too bright (if 12bit wl
was specified, and image is from 14bit-uncompressed)
or too dark and with purple highlights (if 14bit wl was specified, and
image is from 12bit-uncompressed)
or with just purple highlights (all other cases, e.g. "*-compressed").

Using a D5200 for 3 years, I would consider myself savvy enough to notice such a problem if it hit me (be careful about shoot-time exposure ( careful about in-camera histogram, sometimes expose to the right) and darktable-time exposure, choose basecurve or even disable when appropriate, use highlight reconstruction). I never saw purple highlights (unless highlight reconstruction was turned off or tweaked, which is normal then).
Now using darktable 2.0.3 on Xubuntu 16.04.

I haven't noticed anything, perhaps because my raws are always 14-bit (don't know if compressed or not), never 12-bit.


* Can someone provide a description and perhaps some examples of the symptoms to be checked?
* What should affected users do? What does it break?
* What if users have worked on photos in buggy versions? Should they e.g. shift exposure by 2EV or whatever?

Thank you for your attention.

Stéphane Gourichon

darktable user mailing list
to unsubscribe send a mail to darktable-user+unsubscr...@lists.darktable.org

Reply via email to