GenSplatCodec: Feed-Forward Gaussian Splatting Compression via One-Step Diffusion
2026-07-27 • Computer Vision and Pattern Recognition
Computer Vision and Pattern Recognition
AI summaryⓘ
The authors focus on improving a method called 3D Gaussian Splatting, which helps recreate 3D scenes efficiently but produces a lot of data. Current compression methods lose important details at low file sizes, while generative models alone don’t keep different viewpoints consistent. To fix this, the authors propose GenSplatCodec, which combines a simplified scene structure with appearance information and uses a new generative decoding approach guided by the geometry. They also develop a training process that helps the system learn better, resulting in improved quality versus file size performance.
3D Gaussian Splattingscene reconstructionGaussian compressiongenerative modelsrate-distortion performanceview consistencygeometry-guided decodingfeed-forward codechierarchical geometry control
Authors
Qiang Hu, Zhenlong Wu, Lei Huang, Zihan Zheng, Xiaoyun Zhang, Wenjun Zhang
Abstract
Feed-forward 3D Gaussian Splatting (3DGS) enables scalable scene reconstruction without per-scene optimization, yet produces dense Gaussians that are costly to store and transmit. Existing feed-forward Gaussian compression methods formulate decoding as deterministic representation recovery, which becomes inadequate at low bitrates when high-frequency textures and view-dependent appearance are discarded. Although generative models offer a promising alternative, using them as standalone post-processing decouples generation from the transmitted scene structure, thereby compromising cross-view consistency. To address these limitations, we propose GenSplatCodec, a unified feed-forward Gaussian codec that reformulates low-bitrate Gaussian compression as geometry-guided generative decoding. We present a detail-aware feed-forward Gaussian coding scheme within a dual-stream formulation, where the resulting compact Gaussian structural stream is complemented by a lightweight reference appearance stream. We further introduce a geometry-guided one-step generative decoding approach that jointly exploits decoded structural and appearance cues through hierarchical geometry control to reconstruct high-fidelity and view-consistent novel views. Finally, we develop a three-stage optimization strategy that stabilizes the learning of the unified codec and adapts the generative decoder to codec-derived structural and appearance cues. Extensive experiments across multiple datasets demonstrate that GenSplatCodec consistently achieves superior rate-distortion (RD) performance over existing methods.