[argyllcms] Re: White Point

  • From: Ben Goren <ben@xxxxxxxxxxxxxxxx>
  • To: argyllcms@xxxxxxxxxxxxx
  • Date: Mon, 9 Nov 2015 16:11:19 -0700

On Nov 9, 2015, at 11:23 AM, Elle Stone <ellestone@xxxxxxxxxxxxxxxxxxxx> wrote:

But it's interesting that BetaRGB, Ben Goren's recommended editing color
space, comes out at the top of the list, and sRGB and ProPhotoRGB don't fair
so well.

Bruce Lindbloom's intent in creating BetaRGB was to create the most efficient
color space for encoding the various samples he had access to -- the various
usual suspects of test charts. As it turns out, that wound up being a proxy for
the most efficient way to encode Pointer's Gamut.

(Incidentally, I haven't yet done the math, but the irregularities in the
surface of Pointer's Gamut strongly suggest to me that it's a derived
approximation. A refinement using calculations of synthetic spectra should
result in a much smoother and more realistic gamut..but it's also going to be
within shouting distance of Pointer's Gamut, so it might not be the most urgent
task to be done.)

I'm coming to the conclusion that the concept of a Platonically ideal "best"
profile / color space is a coherent one.

As you've demonstrated so well, Elle, it's not uncommon for a digital camera to
distinguish between colors that humans can't distinguish (and, sadly,
vice-versa), and, in such cases, you can recover or create detail by working in
the camera's native space and using non-absolute means to convert those colors
to ones that lie within the human (or preferred working space's) gamut.

But simply tagging the image with the camera's native profile isn't necessarily
a good idea for general-purpose editing, since there _is,_ after all, that
disconnect between the camera's gamut and ours.

And then there're all the other concerns that have motivated all the rest of
the angst driving this discussion...those looking to maximize recordable color
gamut at the expense of volumetric efficiency; those looking to maximize
volumetric efficiency at the expense of color gamut; those looking to optimize
a workflow for this or that input or output device; and so on.

BetaRGB has turned out to be a really good compromise. I think it's probably
time to have a fresh look at the subject...I don't remember when Pointer's
Gamut came on the scene, but I'm pretty sure common awareness of it was long
after Bruce designed BetaRGB. And that Python library might very well be suited
to performing some much more sophisticated (and even multi-dimensional)
analyses than were practical when Bruce designed BetaRGB. But I have a very
strong hunch that an optimized color space with the same general goals as Bruce
had for BetaRGB will wind up within shouting distance of what he came up with.
Probably a bit bigger volume, maybe some rotation of the primaries...but not


Attachment: signature.asc
Description: Message signed with OpenPGP using GPGMail

Other related posts: