Channels masks are very important feature to have, because it allows flexibility for the API user and later the artist, to combine channels from more images in different ways, to get the result. Thanks to Will for suggesting this early.
Ideas are welcomed, in general related to this. I will start with some questions: 1. How many channels should the result image have? For example, we have two input images A and B with 3 channels, but the operation is applied only to 1 channel. Should the result image R have 3 channels, or just the 1 modified channel? We could provide a bool parameter to specify this, but that would clutter the API. Imagine for each operation we will have an additional mask parameter, and this new bool parameter I just mentioned. There must be better ways to approach this. 2. Most operations can be applied per channel independently. We could force them to work only on one single channel and make them atomic. And then we could have one main function as coordinator. This function would take another function(atomic operation) as parameter, input and output images, channels mask, etc. and it will apply the operation only to the channels specified in the mask. I am just beginning a brainstorming session, so we can start off from something. Thanks
_______________________________________________ Oiio-dev mailing list [email protected] http://lists.openimageio.org/listinfo.cgi/oiio-dev-openimageio.org
