[Bf-committers] Map Camera Coordinates to UV layer patch

[Roland Hess]

It's the hoops and the workflow that bug me. When doing camera mapping, 
I find it much easier to just build my initial forms, drop into camera 
view with a background image, then align everything right there. Done. 
If I could bake that camera view directly into UV coords, then my job is 
done, and with a familiar, easily tweakable result.

Brecht made the comment in the patch tracker that this is really UV 
wrapping functionality, and should be included as an unwrap option. 
After going through the code for the various unwrap methods, I'm 
thinking that the best approach would be this:

Have "Unwrap from View" simply do this correctly when the view is a 
camera view. It's what you would expect anyway, but the unwrap code 
doesn't really deliver. There's even a note in the source about how 
"scaling gets weird" when you do it from camera view. I'd propose that I 
just fix that "weirdness" so that when you "Unwrap from View" in a 
camera view, you get what you should have gotten all along: a direct 
translation of the 2D representation of the mesh in the camera view in 
the UV Editor.

With that fixed, there is no need for sticky coordinates and they can 
just go away.
> The main issue I have with UV Projection modifier is that in order to
> have it bake to a uv layout that you can paint over you have to jump
> through some crazy hoops. If you apply the modifier it projects to view
> which is useless if you would like to paint the backside of the object.
>
> Briggs walked me through a very complex setup about 6 months ago to bake
> to a custom defined uv layout that I had made. That was good but after
> leaving it for a few months there is no way I could duplicate it. I did
> not understand the reasoning behind why it was so hard to do but Briggs
> described it as two sepearte systems not communicating well.
>
> Ideally you could bake what is projected to a UV layout.
-- 

Roland Hess
harkyman