Presentation

· Contributors · Organizations · Search Program

Physics-informed Learning of Characteristic Trajectories for Smoke Reconstruction

DescriptionWe introduce Neural Characteristic Trajectory Fields, a novel representation utilizing Eulerian neural fields to implicitly model Lagrangian fluid trajectories for video-based fluid reconstruction. This topology-free, auto-differentiable representation facilitates end-to-end supervision, encompassing long-term conservation and short-term physics priors. It offers advancements in high-fidelity fluid reconstruction across synthetic and real scenes.

Authors

National Key Lab of General AI