Yeah, I thought I hadn’t expressed myself clearly…
Mmph: Yes, I’m quite aware of saving layered xcf (or psd) files. That’s what I do now. I have my textures saved as a multi layered xcf and save out copies of individual layers to be used in blender. That’s standard. I’ve been doing that since I started, back in the days of photoshop 3 and 3ds4.
Fligh: I’m also aware of OpenEXR (and RPF/RLA) , and use them a lot. They’re very useful for compositing and I was very happy when blender started using OpenEXR as standard.
What I was getting at was an image format geared towards textures, instead of compositing. Unless I’m missing something, at the moment the extra channels in an OpenEXR file (the Z buffer, etc.) are mainly used for compositing.
I’m talking about an image format that has a channel for normal, specularity, displacement etc. as well as the standard RGBA channels.
Perhaps OpenEXR can already do this, because of it’s “flexible and expandable structure”?
I guess what I’m talking about is a workflow thing. All the trial and error involved in, for example, making a normal map that matches and compliments your colour map, and the fact that you have to constantly switch back and forth between your paint program and your 3d program to see how your texture is working, and saving many different copies of what should be the same image file could be dramatically reduced and replaced with a more intuitive, interactive way of working.
The gimp plugin idea was just an extension of this, but as blender’s 3d paint tools and gl previews get better they would probably serve the same same purpose.