3 ms·
So, basically, Microsoft Research's Photosynth tech, but where the input is a bunch of videos instead of a bunch of photos, and the result is an "animated" 3D m
by derefr 13y ago
So, basically, Microsoft Research's Photosynth tech, but where the input is a bunch of videos instead of a bunch of photos, and the result is an "animated" 3D model of the environment instead of a static one?
- GrantS 13y agoIn some ways, yes, though while the approach you describe has been explored ( http://www.inf.ethz.ch/personal/ataneja/ModelingDynamicScenes.pdf http://www.inf.ethz.ch/personal/ataneja/ModelingDynamicScene... ), that's not quite what concentric mosaics and other light field approaches are doing. One strength of light field methods is that there is no need to model the scene. (Photosynth, on the other hand, spends much of its computational budget determining the 3D position of sparse feature points and the cameras observing them.)