ProPhoto RGB: gamma decomposition?

I exported some TIFFs from Lightroom with ProPhoto RGB profile. But the embedded v2.1 ICC profile (as visible with Mac ColorSync or node.js exifr ICC) only contains the fixed gamma value 1.8 and the color matrix. I miss a hint about the gamma decomposition (similar to sRGB’s) as given in the following gamma formulas for ProPhoto RGB:

c <= 0.001953 ? 16 * c : Math.pow(c, 1/1.8)

c <= 16 * .001953 ? c / 16 : Math.pow(c, 1.8)

It seems that the check for the lower values belongs to a regular ProPhoto RGB interpretation, but I don’t find any hint about it in the embedded profile (as provided by Lightroom 5.6). Besides this, I searched for “gamma” in the ICC.1:2001-04 specification. The only result is in chapter 6.5.3 (“curveType”):

when count is 1, then the data entry is interpreted as a simple gamma value encoded as a u8Fixed8Number. Gamma is interpreted canonically and not as an inverse.

Can you explain to me what this means? Can I somehow derive the c <= 0.001953 ? 16 * c part from the ICC profile and this sentence about the “canonical interpretation”?

The Tone Response Curve (TRC) in an ICC profile can be one of a number of types of successive “definition resolution”, for lack of a better term. Various polynomial expressions can be defined, but if the TRC contains just a single value, the TRC is assumed to be a simple power function.

I didn’t know ROMM RGB (the “proper” name for ProPhoto, I think) had anything other than a simple power specification (1.8) for the TRC:

I think you’ll find the “standardization” of color profiles to have a bit of a “shotgun” approach…

1 Like

Thanks for your quick reply, @ggbutcher

(Lightroom shows an installed profile so I guess it won’t add any further gamma decomposition when encoding.

Anyway, here you see an alternative definition: ProPhoto RGB color space - Wikipedia )

Edit: The PDF you linked to, contains the decomposition, too? So there is no way to find it inside the ICC profile? It seems that the ProPhoto ICC profile I have installed is incomplete.

This is the profile Adobe Lightroom provides:

ProPhoto.ICC from adobe lightroom was removed by @paperdigits. Please only upload things you’re legally allowed to share.

I’d rather have no tone response curve in a profile I intend to use as a working profile; I just want the color transform. Accordingly, I use the profiles from @Elle Stone’s well-behaved ICC profile set; she built profiles for the various colorspaces with various gamma TRCs. I use her gamma=1.0 Rec2020 when I need a working profile; she also has a set of ROMM profiles:

Look in the profiles subdirectory for it.

Regarding the gamma 1.8 decomposition, I’ve noticed that a few profile builders just rely in the power TRC, I guess because they don’t feel like the refinement is worth the work, space, ???

Thinking even more about it, I don’t see any benefit in having such a profile try to accommodate display perception. I’d rather leave that to whatever calibrated profile I use for the rendition medium.

Quick check for a 16 bit channel.

What are the smallest linear values the first gamma values without decomposition >= 0 map to?
Math.pow(0…/65536, 1.8):

And with decomposition:


The float32 (usually my work space) epsilon is 6e-8, so my objection is pointless.

1 Like

I removed the ProPhoto ICC. Please only upload things you can legally share.