Profile Connection Space

[LoneGoose]

Should the PCS of the input profile ideally match that of the destination profile?

If Adobe RGB has XYZ as it's PCS and a printer profile has L*a*b* does that degrade quality? Probably the CMM that has to handle this problem, I suppose (?).

If my profiling package only spits out output profiles with Lab as a connection space there doesn't seem much I can do about it ... I'm just interested.

[meddington]

>If Adobe RGB has XYZ as it's PCS and a printer profile has L*a*b* does that degrade quality?

I would think this unlikely. CMM's handle the conversion (XYZ <-> LAB) and the modes are quite interchangable. I would think any inaccuracies due to the math are surely invisible to the naked eye.

[rich apollo]

I don't think it's gonna' hurt you. You might have better luck posting this question to the Color Forums at Chromix.com

You got me wondering about some stuff, so I started rooting through some profiles. All the working space profiles I looked at (with the exception of PhotoGamutRGB_avg6c) use XYZ as the PCS. The difference between this profile and the other working space profiles is that it is LUT based and actually carries the tag of a printer profile. The others are matrix-based and carry no rendering intent tags.

Everything that I looked at that had A2B and B2A tags was LUT-based and listed Lab as the PCS (this included scanner, monitor, and output profiles). Everything that was matrix-based and had no rendering intent tags, listed XYZ.

[meddington]

As a test, round-tripping XYZ to LAB and back to XYZ, the results are accurate to 5th decimal place. So long as the CMM is competent in the conversion, I can't see this manifesting itself into a visual issue.

[SNYCI_44]

Long overdue reply. The ICC specifications state that the PCS can be only and either CIELab or CIEXYZ as the values are directly convertable, so all CMMs will calculate correctly.

EHA