Support for a DNG file
Oh, it's bayer format which was never tested. Don't people want to choose debayer algorithms and stuff when dealing with this kind of file?
Constructive discussion about what to do with bayer formats in general goes here.
Fast onboard debayer algorithms are kind of glitchy. Maybe we just need the subsampled planes and the 2x2 order of them.
I guess you could do that but then you've still effectively got something that needs to be fed into a debayer filter right after anyway. Do you plan to create one?
I've thought about this a bit and the resolution will simply be that I properly reject bayer formats.