A Diffusion Transformer that tries to learn the underlying physics of reconstruction and
rendering. It
autonomously learns
shadow casting, specular
highlights, and inter-reflections, though not as precise
as ray-tracing, but all neurally learned.
Bottom row showing environment maps under which we relit the object. Results above are generated with 4-8
sparse views of the objects.
(This webpage contains a lot of videos and interactive viewers. We suggest using Chrome or Edge for the
best experience.)
