I have been thinking that it would be possible to render a view of a face with an ortographic camera and then apply it as a uv mapp.

The camera would need to be aligned to the face.

All other objects would need to be invisible, yet still cast shadows.

You would need to do quite a lot of faces to 'do' a complex object.

This could be usefull for landscapes, uv maps could be painted onto after and you would not end up with a unuseable mesh as with radiosity.

I think it would be cool it this process could be automated with a plugin that renders each face and applys it as a uv map.

I have heard of other programs doing this, has anyone got any ideas, or point to add?

I have done this once with a simple building demo, rendering each face manually in Radiance and then manually UV mapping them (although fairly simple as it was all straight flat projections). I did at the time think it could be scripted (I'm not sure how you script internal rendering in Blender) but I haven't had a chance to try. I can post my demo when I get home but that won't be for another couple of weeks.