[cairo] Rendering Intent names

Chris Murphy lists at colorremedies.com
Tue Mar 16 22:16:17 PDT 2010


Well, it's probably not simple to do exactly right, but it doesn't need to be exactly right. It also depends on the versions of the source and destination profiles. If they're both ICC v4 profiles, then yeah you can get away with perceptual rendering for most anything that's "photographic" on a basic level. If one of them is v2, then the default needs to be relcol + bpc, until otherwise proven that perceptual would be better and that would be important areas of fine detail which are highly saturated. So you'd need something that could guess at what in an image would be an important area of fine detail - the highly saturated part is cake to figure out in comparison. Detail is slightly harder. Important detail is a lot harder. But it doesn't have to be perfect.

The saturation intent is next to useless in practice because there's nothing that says how it should behave. And therefore we can't know in advance how it behaves and therefore if it might do a better job on a certain category of images, even photographic ones. It's also possible that some profile's saturation intent will cause large hue shifts, in order to preserve or enhance saturation. And still for other profiles to simply reference the perceptual intent table instead of having a unique saturation intent table.


Chris Murphy



On Mar 16, 2010, at 5:45 PM, James Cloos wrote:

>>>>>> "CM" == Chris Murphy <lists at colorremedies.com> writes:
> 
> CM> An algorithmic approach to determining the [rendering] intent
> CM> automatically I think is a better default than either one size
> CM> fits all, or altering existing terminology.
> 
> Now that is a tremendous idea.
> 
> can you provide any insights as to what the code should look at to
> determine which intent is better?  Useful heuristics?  Et cetera?
> 
> I suspect that contone images (such as jpegs) should lean towards
> perceptual.  And perhaps if all of the colours come from a simple
> palette (ie, all fully saturated, or all from the 8bit psuedo-
> colour web palette) that may suggest saturated or relative.
> 
> Or not?
> 
> Perhaps we should allocate a wiki page to collect ideas on automating
> intent defaults?
> 
> -JimC
> -- 
> James Cloos <cloos at jhcloos.com>         OpenPGP: 1024D/ED7DAEA6

Chris Murphy
Color Remedies (TM)
New York, NY
----------------------------------------------------------------------
Co-author "Real World Color Management, 2nd Ed"




More information about the cairo mailing list