33^4 should be enough!

I have a slightly unusual situation - I have a proprietry monitor 
calibration scheme that I'm trying to emulate with ICC. The scheme I 
have gives me the primaries of the monitor, the white point and very 
precise brightness values for every step of r, g and b separately (3 
* 256). My first thought was to pretend to have an IT8 chart with 768 
patches on it which seems like it would work from your comments below.



>Hi,
>Well, first you have to take in account that profiles does not contain
>sample points but a model for the colorspace translation.
>One particular implementation of this model, could be a 3D grid containing
>samples at fixed slices. If these are what you are referring, there is
>AFAIK no limit in ICC spec, but a cube side of 33 - 48 points is the
>practical limit. Note that, in RGB, a side of 33 points does represent
>33 * 33 * 33 * 3 = 107811 samples!! On CMYK, the numbers
>grow to enormous: 33 * 33* 33* 33 * 4 = 4743684!!!!!
>On 6-inks profiles, anything higher that 6 points is a little monster.
>
>Photoshop does work fine with this type of profiles, in fact, all
>profiles pre-revision 4 can be only of two different flavors: LUT-based
>(this ones) and matrix-shaper. Rev 4 will introduce a lot more variants.
>
>If you are referring to IT8 files, well, I've checked up to 1000 samples,
>I assume this would be enough for most applications  ;-)
>
>Regards,
>Marti.


-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
Lcms-user mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/lcms-user

Reply via email to