Jon Sneyers

Thanks, @psykauze . Now we should probably preprocess raw input to be in RG GB format.

Maybe one image per camera model is not enough to really draw conclusions, but I wonder what is going on here. The differences in compression when changing the CFA pattern...

Converting to R, (G1+G2)/2, B, (G1-G2) is not lossless: one bit is lost in the division by two and it cannot be recovered. But we could try to directly apply...

What is Y in that case distinction?

Ah, right, the parity of the sum is equal to the parity of the difference so there would indeed be enough information to restore the exact sum. I'll try that...

I'm storing it as 0x4000 + G1 - G2, which is always > 0 if the numbers are at most 0x3fff. But unfortunately, my preliminary results (just testing on a...

I pushed it as commented-out code (only in the rggb loading part, not in the decode/save part),

You could implement somethingl like that, but why? It's a lossless conversion, so not like YCbCr. And the way it is implemented in FLIF, you need 9 bit Co/Cg if...

The current metadata does not contain this information. There have been proposals to add a metadata chunk specifically for this purpose (see https://github.com/FLIF-hub/FLIF/blob/master/doc/metadata#L36), but unless we make it obligatory, it...

For games decode speed probably doesn't matter that much, since you will probably only decode when loading the game / loading a level. It should indeed not be hard to...