I had a conversation with a poster here yesterday and the day before about out of gamut Macbeth color checker patches when rendered in sRGB.
Turns out that’s a complicated question, and the answer depends on the illuminant and the adaptation algorithm. I’m working on a post about it.
But that’s not what this post is about. During the conversation, my reader asked me which nonlinearity Bruce Lindbloom was using for one of his presentations showing how the cyan Macbeth patch was out of sRGB gamut. Since Bruce has used the “real” sRGB nonlinearity and gamma = 2.2 with sRGB primaries elsewhere, I couldn’t answer the question.
But it got me thinking. Could the nonlinearity of an RGB color space possibly affect the gamut? I thought not. Upon further thinking I still think not, if infinite precision is allowed.
[The text below was added based upon the first comment to this post]
In order for what I wrote to be correct, there need to be some constraints on the non-linearity:
- It needs to be continuous, and everywhere defined in the input range, which I’ll assume to be [0,1] — the closed interval.
- It needs to map 0 to 0, and 1 to 1. In 4.3.1.3 of this document, Adobe constrains Adobe RGB so that zero inputs map to zero outputs..
There may be some more, and I’ll thank you all for pointing them out, but let’s start with these.
If any aspect of the gamma curve affects the gamut, it’s the black point, since that controls how much trace “contaminants” of the other color channels leak into the most extreme saturated colors.
Thanks. I was assuming that 0 -> 0, but, thanks to you, I’ve added some text that makes it explicit.
Jim