GoDe: Gaussians on Demand for Progressive Level of Detail and Scalable Compression
This addresses scalability and adaptability issues in 3D scene representation for real-time applications, offering an incremental improvement over existing methods.
The paper tackles the high storage and VRAM demands of 3D Gaussian Splatting for novel view synthesis by proposing a model-agnostic technique that organizes Gaussians into hierarchical layers for progressive Level of Detail, enabling a single model to scale across compression ratios without re-training and with minimal quality loss.
3D Gaussian Splatting enhances real-time performance in novel view synthesis by representing scenes with mixtures of Gaussians and utilizing differentiable rasterization. However, it typically requires large storage capacity and high VRAM, demanding the design of effective pruning and compression techniques. Existing methods, while effective in some scenarios, struggle with scalability and fail to adapt models based on critical factors such as computing capabilities or bandwidth, requiring to re-train the model under different configurations. In this work, we propose a novel, model-agnostic technique that organizes Gaussians into several hierarchical layers, enabling progressive Level of Detail (LoD) strategy. This method, combined with recent approach of compression of 3DGS, allows a single model to instantly scale across several compression ratios, with minimal to none impact to quality compared to a single non-scalable model and without requiring re-training. We validate our approach on typical datasets and benchmarks, showcasing low distortion and substantial gains in terms of scalability and adaptability.