FeCoSplat: Feedback-Guided Compression for Feed-Forward 3D Gaussian Splatting

2026-09-27

AI Summary

This study addresses the high storage and transmission costs of feed-forward 3D Gaussian Splatting, as well as the difficulty existing compression schemes face in balancing coding efficiency with decoding complexity. To this end, we propose a rendering-feedback-guided intermediate feature compression paradigm. Specifically, our method employs a two-stage feature compression strategy that constructs a rendering feedback loop to further refine intermediate features, alongside a lightweight implicit state predictor designed to reduce decoding overhead. Experimental results demonstrate that the proposed framework achieves superior rate-distortion performance at low bitrates while requiring only 3.45M parameters for reconstruction at the receiver side, effectively reconciling the competing demands of efficient encoding and lightweight decoding.

Abstract

Feed-forward 3D Gaussian Splatting (3DGS) enables efficient novel-view synthesis from sparse multi-view images, yet its representations remain costly to store and transmit. Existing approaches compress either the input images, incurring heavy receiver-side reconstruction, or the reconstructed Gaussian primitives, which are difficult to compress due to their heterogeneous and irregular attributes. We instead compress compact intermediate features, providing a better balance between compression efficiency and receiver-side complexity. Based on this paradigm, we propose FeCoSplat, a feedback-guided compression framework for feed-forward 3DGS. FeCoSplat first compresses multi-view features to obtain an intermediate 3DGS, whose rendered views are used as feedback to guide a second-stage compression for further refinement. The resulting bitstreams are decoded into a compact implicit state, from which the final Gaussian primitives are reconstructed with a lightweight predictor. Experiments demonstrate that FeCoSplat achieves favorable rate--distortion performance, particularly at low bitrates, while requiring only 3.45M parameters for receiver-side Gaussian reconstruction. Code will be released soon.

Problem

• 3D Gaussian Splatting
• compression
• feed-forward
• novel-view synthesis
• rate-distortion

Innovation

• 3D Gaussian Splatting
• Feedback-Guided Compression
• Feed-Forward
• Intermediate Feature Compression
• Novel View Synthesis