Re: RGB colorspace question

On Wed, 2024-01-10 at 16:15 -0600, Bob Friesenhahn wrote:
> I should mention that when one is performing gamma or colorspace 
> conversions, there will always be a loss of accuracy.  The stair-step 
> representation at one gamma will not render to the same number of 
> stair-steps at a different gamma since two steps might be transformed 
> to the same number.  Due to this, it is recommended to use at least a 
> Q16 build of GraphicsMagick (numeric range 0-65535) and only save to 8 
> bit formats when required to do so. Use '-depth 8' to force saving at 
> a lower bit depth.
> 

Thanks for your continued help with this. But honestly, I think I have
to give up on it for now. The code in question [0] is tested, and the
test case is looking for the specific set of numbers that ImageMagick
produces, without regard for correctness or, even -- as I understand it
-- the human interpretation of the input image.

That is annoying to me because I'd rather have GraphicsMagick installed
than ImageMagick. But it's a minor annoyance, because I don't actually
use hap -- I'm merely packaging it. So it's relatively easy for me to
swap out GM for IM while doing that work.


[0] https://github.com/gap-packages/hap/blob/master/lib/PolyComplexes/pureCubicalComplexes.gi#L310