I'm not sure. I'm tempted to advise just ignoring the channel mask idea for
now. As you said, I'm not really sure how channel masks should operate -- do
they mean to just leave the destination unmodified for the unspecified
channels? Or have the destination lack the channels entirely? I don't know
what's useful, if either. So let's ignore it for right now.
We can debate channel masks separately and come up with a strategy we like that
we could then apply to all the IBA functions, or none if nobody wants to speak
up as a champion for the idea. (How much will it really be needed, in the
context of how oiiotool or similar IBA-based utilities are likely to operate?)
-- lg
On Jul 16, 2012, at 4:25 PM, Stefan Stavrev wrote:
> Sorry, yes I will combine blending with over. I was thinking
> of these other options we have and got confused.
>
> What about channels masks, like Will suggested for contrast?
> Apply the operation only to those channels specified in the
> channels mask. If alpha channel is in there then apply
> blending formula with over, otherwise just apply blending
> formula without taking alpha into consideration.
>
> Then a new question comes up, how many channels does
> the result image have? Same as before, or just the modified
> channels? So if it had 3 channels but we only did blending
> on 1 channel, does the result image have 3 or 1 channel...
> We could even have a bool parameter to specify this,
> whether to keep same number of channels or if the result
> image should only have the modified channels.
>
> I will install Nuke and play around with it to get some more ideas.
> You can see here that you can pick which channels the operation
> is applied to: Nuke
>
> Will, would be nice to hear your opinion on this.
>
> Thanks
>
--
Larry Gritz
[email protected]
_______________________________________________
Oiio-dev mailing list
[email protected]
http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org
