Smol-GS: Compact Representations for Abstract 3D Gaussian Splatting
By: Haishan Wang, Mohammad Hassan Vali, Arno Solin
Potential Business Impact:
Shrinks 3D scenes to tiny sizes, keeping detail.
We present Smol-GS, a novel method for learning compact representations for 3D Gaussian Splatting (3DGS). Our approach learns highly efficient encodings in 3D space that integrate both spatial and semantic information. The model captures the coordinates of the splats through a recursive voxel hierarchy, while splat-wise features store abstracted cues, including color, opacity, transformation, and material properties. This design allows the model to compress 3D scenes by orders of magnitude without loss of flexibility. Smol-GS achieves state-of-the-art compression on standard benchmarks while maintaining high rendering quality. Beyond visual fidelity, the discrete representations could potentially serve as a foundation for downstream tasks such as navigation, planning, and broader 3D scene understanding.
Similar Papers
SUCCESS-GS: Survey of Compactness and Compression for Efficient Static and Dynamic Gaussian Splatting
CV and Pattern Recognition
Makes 3D pictures smaller and faster to use.
TinySplat: Feedforward Approach for Generating Compact 3D Scene Representation
CV and Pattern Recognition
Makes 3D scene models much smaller.
MemGS: Memory-Efficient Gaussian Splatting for Real-Time SLAM
CV and Pattern Recognition
Makes drones build better 3D maps with less memory.