Neural Trajectory Fields for Dynamic Novel View Synthesis

TL;DR

This paper introduces DCT-NeRF, a neural representation that models dynamic scenes by learning smooth trajectories, enabling high-quality, consistent novel view synthesis of time-varying scenes from limited photographs.

Contribution

The paper presents DCT-NeRF, a novel coordinate-based neural method that captures dynamic scene trajectories for improved view synthesis of moving scenes.

Findings

Enables high-quality rendering of dynamic scenes.

Ensures temporal consistency between frames.

Outperforms previous methods in dynamic view synthesis.

Abstract

Recent approaches to render photorealistic views from a limited set of photographs have pushed the boundaries of our interactions with pictures of static scenes. The ability to recreate moments, that is, time-varying sequences, is perhaps an even more interesting scenario, but it remains largely unsolved. We introduce DCT-NeRF, a coordinatebased neural representation for dynamic scenes. DCTNeRF learns smooth and stable trajectories over the input sequence for each point in space. This allows us to enforce consistency between any two frames in the sequence, which results in high quality reconstruction, particularly in dynamic regions.