[CREATE] CREATE Digest, Vol 164, Issue 6

[Johann ELSASS]

Hi Paul, 

Indeed, for pixel formats that would not be supported by PNG like CYMK, it makes sense to use another format. For backward compatibility we could still have the current PNG, and on top of that the EXR/other source. So a program that doesn't handle CMYK could still read and play with the layers. 

Or if we consider that it takes too much space to keep the PNG in this case, we could, in the case of non-RGB format only, drop the PNG completely.

About effects, I agree about providing two images: the source image and the image with effects. Note that if we make the effects using SVG, then this is the same as providing the rendered PNG and the SVG source. Indeed, we can put an image inside an SVG and apply effects to it.

Thanks for clarifying what you meant about linear transforms. That's what I though but I was confused by the word "mask" as for me it is a grayscale image to apply as alpha channel.

Same remark as before, linear transform can be done using SVG.

Implementing all of that with SVG might be complicated. Obviously there is choice to make: either use SVG to its full potential or provide other mechanism like linear transforms and effects in the layer structure.

In all cases, it seems to me essential for backward compatibility to have both the rendered PNG and the source (SVG, EXR, image before effects or transform).

Regards,

Johann