Track papers across labs, topics, and trends.
Search papers, labs, and topics across Lattice.
This paper addresses the failure of dynamic Gaussian representations to reconstruct small, fast-moving objects and fine textures in volumetric video by re-evaluating Gaussian allocation strategies. The authors introduce Seed Injection, which initializes new Gaussians based on temporal appearance changes and multi-view observations, and Temporal Biased Relocation, which uses time-resolved optimization signals for spatiotemporal placement. These methods, applied without increasing the Gaussian budget, significantly improve reconstruction quality in highly dynamic regions.
Small, fast-moving objects and fine textures in dynamic volumetric video are finally recoverable by intelligently allocating Gaussians based on future motion, not just current positions.
Dynamic Gaussian representations have made high-quality, real-time volumetric video increasingly practical, yet challenging motion can still break their reconstruction: small fast-moving objects may disappear, and fine textures in highly dynamic regions are often lost. In particular, simply training longer or using more Gaussians may still fail to recover small, fast-moving objects. We argue that the key bottleneck is not representational capacity itself, but where and when Gaussian primitives are allocated. Existing allocation strategies rely on current Gaussian locations and often overlook the spatiotemporal destinations of relocated Gaussians. To address these limitations, we introduce two complementary strategies. First, Seed Injection identifies under-reconstructed dynamic regions using temporal appearance changes and reconstruction errors, and initializes new Gaussians at geometrically plausible 3D locations through multi-view observations, independent of existing Gaussian locations. Second, Temporal Biased Relocation uses time-resolved optimization signals to guide the spatiotemporal placement of relocated Gaussians. Both strategies maintain a fixed Gaussian budget without modifying the underlying representation. Experiments across multiple dynamic-scene datasets demonstrate improved reconstruction quality, particularly in highly dynamic regions, recovering small, rapidly moving objects and fine-grained appearance details.