Dusting the Phantom: Gaussian Allocation by Spatiotemporal Destination Recovers Fast-Moving Detail
Minhyeok Roh · hf · 2026-10-19
Researchers from Yonsei University, ETRI, POSTECH, and Seoul National University argue that in dynamic Gaussian representations, vanishing small fast-moving objects and lost fine textures stem not from representational capacity but from where and when Gaussian primitives are allocated — existing strategies rely on current positions and overlook spatiotemporal destinations of relocated Gaussians.
They propose two complementary strategies: Seed Injection identifies under-reconstructed dynamic regions via temporal appearance changes and reconstruction errors, initializing new Gaussians at geometrically plausible 3D locations through multi-view observations; Temporal Biased Relocation relocates primitives toward their spatiotemporal destinations. Together they recover small fast-moving objects and fine textures without simply training longer or adding more Gaussians.
More from Research
- Dream2Discern: Self-Generated Hard Examples Predict Multi-Agent Failures With 90%+ Accuracy — RutgersU · 2026-10-19
- BridgeWAM: latent bridge queries decouple video and action experts, 2.47x faster robot inference — Fudan-University · 2026-10-19
- NVIDIA's FC-SWE: Failure-Conditioned RL Boosts Multi-Attempt Resolution for SWE Agents — nvidia · 2026-10-19
- AVC-Reward: an event-centric framework for evaluating audio-video generation consistency — ZhejiangUniversity · 2026-10-19
- PremiseCheck Fact-Checks Peer Reviews at GPT-5.5-Level Accuracy for Half the Cost — Dingjie Song · 2026-10-19
- SciΠvot benchmark: frontier models prefer flawed research plans and lag human scientific judgment — Youran Sun · 2026-10-19