Photogrammetry objects

Started by sjefen, March 29, 2022, 06:02:58 PM

Previous topic - Next topic

WAS

Quote from: Dune on March 31, 2022, 02:07:35 AMI think glossiness and specular (even bump to some degree) can't be derived from one photo (so I actually don't believe in any software that does that), because any caught pixel is just a tone (can be just a whitish color, or a reflection on a dark green area). I think it can only be derived from several photo's from several angles, which can see the differences of that one particular pixel/area in changing angles.

Almost always is, though. Even in PBR unless you're scanning in multiple imaging types (besides just a camera) they delit and doctor out as much shadow and lighting, and just bring back in what they think is appropriate and matched real world.

You're only getting truly dynamic specular/reflection based mapping from procedural software like Substance, which ofc is still just fake, but accurately modeled after your topology. Which is probably what a lot of people use for PBR too.

Dune

Yeah. But even if it's not completely accurate, the overall specularity/gloss gives a realistic feel. Nobody actually checks if a pixel is really accurately colored.