I remember a while back someone posting a paper talking about the use of three photos from different angles being used to produce a normal map. It was created for archeologists as a replacement to the expensive depth cameras they would otherwise need to get detailed height map info. Anyone remember this?

I'm not sure what paper you are talking about but there's this tutorial by Ryan Clark which seems pretty close to what you want to do.

http://www.zarria.net/nrmphoto/nrmphoto.html

Not sure if this is what you are looking for, but quite cool nonetheless.

http://www.newscientist.com/article/...2?DCMP=youtube

You Tube

Yeah that's the vid! Thanks!!

They say albedo map, but they seem to mean diffuse map? They also seem to be using 'reflectivity' in the physics sense, instead of the cg sense we usually mean. They say 'a measure of the reflectance', when I think they're trying to say 'a measure of the diffusely reflected light' aka, diffuse map.

Cool technique, but they're essentially generating a heightmap, no? Wouldn't normal map data still provide a more accurate version of the surface?

Generating bump/normal map from a 2D image is nothing new. Generating geometry from 2D images is nothing new either.

What interests me is using two differently lit photographs of the same motif. Would be nice to try it out for yourself.

Originally Posted by glib

They say albedo map, but they seem to mean diffuse map? They also seem to be using 'reflectivity' in the physics sense, instead of the cg sense we usually mean. They say 'a measure of the reflectance', when I think they're trying to say 'a measure of the diffusely reflected light' aka, diffuse map.

Cool technique, but they're essentially generating a heightmap, no? Wouldn't normal map data still provide a more accurate version of the surface?

So I was also wondering about this as in the TF2 tech papers they refer to the diffuse as the albedo map and I had never heard of that before.