[Openicc] Linux CM ideology, was: meta data in test chart

[Chris Murphy]

On Jan 28, 2011, at 7:03 PM, Hal V. Engel wrote:
>> 
> At the very end of this he was asked "How does this affect color management?"   
> He said some things about "true color" mode, I think meaning 8bpc, and not 
> talking at all about color management at least as we would understand that 
> term.  In affect saying if we are using 8bpc color then there is nothing left 
> to do for color.  Either he just does not want to deal with the issue or he 
> does not understand the issue.  I am not sure which. 

Well, my opinion after watching the video is that he does not understand the issue, in that he does not equate "color management" to a context of:

1. RGB numbers are almost completely ambiguous. The color you get for a set of RGB values is meaningfully, visually different depending on the device they are displayed on.

2. Color management provides a mechanism for bringing unambiguous color meaning to those RGB values.

3. Color management provides a mechanism for transformation of RGB values in order to preserve color "appearance"*.


*a very limited color appearance preservation capacity but without question superior to doing nothing to those RGB values as they float from  display to display.

I agree that "true color" equates to 24bpc and he seemed to be referring to precision and number of channels, nothing about transformations or association to CIE color at all. So either he doesn't understand that world, or simply didn't make the connection with the question. Either way that is reassuring because if he does totally understand ICC/CIE color transformations and does not find them of value at all then that would be an actual barrier to progress. I don't get that impression.


> 
> The other interesting thing is he says that with the new architecture they do 
> not know what to do with plug ins for the composting system and he in affect 
> said that using composting plug ins was "crazy".   In the past he said that CM 
> really belonged in the composting layer.  But the only way to do that 
> presently is to create a plug in.  So where does that leave us?  If plug ins 
> are going away that means we need to somehow get CM into the core composting 
> software.  How do we make that happen and how long will that take?

Yeah my thoughts as well. I'm unclear on the time frame for this. Or where it gets implemented. It seems to me this could be done opt-in very efficiently and perhaps much easier than other opt-in mechanisms if EGL could be enabled for color management. That would be one way to back this off to opt-in, and simply let everything else remain not-color managed. That is not exactly the end of the world, even though for mobile phones and tablets and home televisions, I see as much benefit to users and manufacturers to having fast efficient color management (even just matrix-matrix transforms without full lookups) as on the desktop for professionals. EGL may be a place for this.

Otherwise for an opt-out system, seems like the color management would occur in the external compositing manager.


Chris Murphy