[argyllcms] Re: Patching profiles with non-D50 illuminants for Absolute Colorimetric

Alastair,
 
> In a recent post in the Fluorescence thread, Gerhard said:
> "Btw, keep in mind that Argyll records the mediaWhitePointTag in the
> "traditional way", i.e. *not* adapted from the viewing illuminant to
> D50. 

Just to make sure I follow the arguments here, is the adaptation required
strictly when the colorimetry of the measurements is calculated for
something *other* than D50? Let's say I measure the color patches using
D50/2 then, there is no adaptation required, right? But if I measure the
color patches using A/2 or D65/2 then, what I understand the v4 spec says is
that the media white still needs to be encoded in D50/2. Therefore, the only
way to go from A/2 or D65/2, for instance, to D50/2 is through a chromatic
adaptation like Bradford.

Is this the correct interpretation of the specs?

> Thus, if you use a printer profile generated with "profile -iA ..."
> for proofing with ICC-absolute intent, you may not get, what you expect,

Because, in that case, the colorimetry encoded in *this* Media White Point
does not correspond to the original spectral reflectance integrated with D50
illuminant and the 2 degree Observer but rather with the original spectral
reflectance integrated with A illuminant and the 2 degree Observer
chromatically adapted from A to D50 using Bradford.

Is that what's going on?
 
> if your expectation is an *illuminant relative* reproduction, as the
> term ICC-absolute is interpreted by the ICC.

This almost makes me feel like the next version of the spec should require
only the spectral reflectance of the media and let the CMM at run time do
the adaptation when required. At least, there would be no confusion of what
the media white point tag would contain.

> In order to get the ICC
> desired behaviour, you need to patch the mediaWhitePointTag and
> mediaBlackPointTag accordingly in the Argyll generated profile."

Does this "patching" need to be more difficult than opening the profile in,
say, ProfileEditor and changing the Media White Point in XYZ or Lab?

> How would one go about doing this?  I've generated printer profiles
> using measurements made with D50 illuminant and D65 illuminant.  The D65
> ones give noticeably more pleasing perceptual and rel. col. results, but
> aren't much use for proofing as they stand. :)

ProfileEditor?

> How does one go about patching the white and black point tags?

Why would you need to change the black point? Is that actually used in v4
profiles? As far as I can remember it was not of much use in v2 profiles.

> Are 
> there any appropriate tools or do they still need to be written? :)

Could an hex editor be used?

> Mathematically, how would the white and black points be calculated?

Some CAT, I presume.
 
> Would scaling a reference D50 whitepoint to match the Y component of the
> measured point be sufficient or is it more complicated?

I think all three primaries need to change.

> All the best,
> --
> Alastair M. Robinson
> 

Regards,

Roger Breton  |  Laval, Canada  |  graxx@xxxxxxxxxxxx
http://pages.infinit.net/graxx

Other related posts: