[argyllcms] negative primaries

One of the issues Graeme mentions in “What’s wrong with the ICC
profile format anyway ?" is the fact that ICC v2 specifications
disallow negative tristimulus values.

Recently I was comparing the gamut shape and size of several working
space profiles with the gamut of raw-rendered tiffs of highly
saturated flower images. I came to the same conclusion Ben Goren did
in the argyllcms archive
http://www.freelists.org/post/argyllcms/Optimal-color-space-for-an-image,4,
that none of the spaces smaller than ProPhotoRGB were big enough to
hold even moderately saturated images.

But really even ProPhoto isn't big enough for all images. The blue
channel of one particular flower image while still in camera space had
lots and lots of detail. After conversion to ProPhotoRGB, the details
in the blue channel had turned to a solid black blob. Once I started
looking, a lot of images turned out to have this problem of blue
channel detail going to black upon conversion to ProPhoto. I wrote up
my findings here:
http://www.dustystones.com/2010/photo-essays/intro/open-source-digital-imaging-intro.html
and here: 
http://www.dustystones.com/2010/photo-essays/negative-primaries/negative-primaries.html.

I'm pretty sure the custom working space I created to use in place of
ProPhoto could be a lot smaller and still hold all the image details,
but I haven't yet experimented to see what the primaries would need to
be. FWIW, the idea of a working space with negative primaries is not a
totally new thing as Joseph Holmes sells a working space profile with
negative primaries. See
http://www.josephholmes.com/propages/gamuts/DCam5.jpg.

Also, the ICC is coming around to realizing that negative primaries
are a good thing. Section 1.1.3, “PCS encoding range is limited to
[0,2) [sic]” of ICC Votable Proposal Submission, Floating-Point Device
Encoding Range, dated June 16, 2006, says "It is also possible that
some device values may have corresponding XYZ values that are
negative. Such values can result from digital camera color analysis
matrices, or chromatic adaptation transforms applied to extremely
saturated blue colors. In most cases, it is acceptable to clip
negative XYZ values to zero as such values do not correspond to real
colors. However in some cases this may be unacceptable, for example if
perfect round-tripping is desired."

I'd like to know, if what the ICC says is true, is the blue channel
detail in camera space imaginary, because it looks real, even if it
gets clipped to solid black upon conversion to an icc-compliant
working space.

Kind Regards,
Elle Stone

Other related posts: