[argyllcms] Re: Profile mismatch behaviour

Pascal de Bruijn wrote:

I'm curious as how does most color management software respond to
such a mismatch? Is an automatic XYZ-Lab conversion done? Or will it
render invalid/incorrect results?


There are two standard profile connection spaces (PCS's), XYZ and D50 L*a*b*.
All CMMs should convert between these two automatically.


Is it possible to convert the RGB-Lab profile to a RGB-XYZ profile?


In theory it is, but not necessarily without change. The device
behavior model is usually based on the chosen PCS, and the
detailed numbers will usually depend on it (ie. linear blends between two
XYZ values will yield different results to linear blends in Lab space, etc.).

Some of the models are only valid in XYZ (matrix),
because XYZ is modeling the physics of additive light.

Some of the newer tags introduced in ICCV4 allow for
extra stages that could more transparently present one
color spaces as the PCS while using another for the actual model.


And last but not least, does Argyll have a tool to edit profile metadata, I
have several profiles with identical description, which is rather confusion.


No, sorry, not at the moment. Not to hard to write (see icc/iccrw.c for a hard
coded version that modifies certain tags), but not something I've got around to.

Graeme Gill.

Other related posts: