Solution fix

video input $\rightarrow$ depth/motion

[논문] HyperNeRF https://hypernerf.github.io/

Main point: HyperNeRF proposes to align frames towards a hyperspace for topologically varying scenes.

Video of the HyperNeRF project:

[논문] ECCV 2022(oral) PREF: Predictability Regularized Neural Motion Fields

PREF Overview:

Real-time Solution?

history video + flame input $\rightarrow$ depth/motion

향후 계획: demo of our solution