On 04/25/2013 12:22 AM, Graeme Gill wrote:
In other words, when using ICC-compliant software (e.g. Photoshop, Gimp), should the vcgt be loaded, or will it result in double correction (and thus bad results)? Is there a consistent behavior between vendors in this regard?Yes, it should be loaded. What other purpose would it have if it is included in the profile ?
Well, it could have been designed the following way: if your workflow fully supports ICC, then ignore vcgt (you don't need it); if it doesn't, then just load the vcgt LUTs. One could make the argument that since vcgt is non-standard, you cannot rely on users loading it, and so the rest of the ICC profile should not be made dependent of a non-standard extension.
Right now if a system supports ICC but not vcgt (which is a completely valid situation since vcgt is not part of the ICC standard), then it will not get the correct result because the rest of the profile describes the display's behaviour assuming the vcgt is applied.
On a related note, I guess ICC-aware PC software could have some special logic that basically looks at the currently loaded hardware LUTs, and then makes processing decisions based on whether the vcgt from the profile is loaded or not, so that the result is always correct. I am not aware of any software that does that, though.
As far as I know, applying the vcgt usually results in the following: - No change in gamut, because it's not possible to alter gamut using RGB 1d-LUTs, you need a 3d-LUT or a matrix for that;Not technically true - it can reduce/change the absolute 3D gamut volume by moving the white point. You can't shift the primaries chromaticities with 1D curves though.
Yes, that's what I meant.
Am I correct? If I am, is there any rationale behind the seemingly arbitrary choice of D65 besides the fact that it is the standard sRGB white point?Many people use something other than D65, so that their display is a better visual match to print under a D50 WP lamp.
I'm new to ArgyLL, so pardon me if I'm going to ask a dumb question - how can I specify a white point other than D65 when generating an ICC profile with vcgt tables? colprof doesn't seem to have an option for that.
And a final question: is calibrating the white point without adjusting the primaries (e.g. when using non-ICC-aware software on a system with the vcgt loaded) such a good idea? Could it result in a worse perceived result because the white point is moved away from the center of the gamut? I've read something about chromatic adaptation but I'll admit I'm way over my head here…I'm not sure what you mean by "without adjusting the primaries" - changing the white point using calibration always adjusts the primary intensities. The gamut width is basically
Yes, actually I meant the primary chromaticities. I realize that I'm a bit sloppy with the terminology - sorry about that, I'm quite new to color concepts.
determined by the primary chromaticities, and useful systems have primaries pretty close to the spectrum locus boundary. If you look at a chromaticity plot of the daylight locus (preferably a u'v' one, not an xy one!), the difference between D50 and D65 is not very significant relative to the gamut boundary, and the change in relative colorimetric gamut shape is small.
Understood. Thank you for your very useful answer. -- Etienne Dechamps