DASH: 4D Hash Encoding with Self-Supervised Decomposition for Real-Time Dynamic Scene Rendering

2025-07-25

AI Summary

Dynamic scene reconstruction has long suffered from feature ambiguity and rendering artifacts due to the restrictive planar assumption. This paper proposes a real-time reconstruction framework leveraging multi-resolution 4D hash encoding and self-supervised static-dynamic decomposition. Departing from low-rank constraints, it employs 4D hash encoding for flexible spatiotemporal modeling; introduces a lightweight self-supervised decomposition network to explicitly separate static and dynamic components; and incorporates spatiotemporal smoothing regularization to suppress deformation artifacts. Evaluated on real-world dynamic scenes, the method achieves real-time rendering at 264 FPS while significantly improving geometric fidelity and visual quality. It establishes new state-of-the-art performance in terms of both quantitative accuracy and qualitative realism.

Technology Category

• Computer Vision: Motion & Tracking
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning
• Planning, Routing, and Scheduling: Replanning and Plan Repair

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Dynamic scene reconstruction is a long-term challenge in 3D vision. Existing plane-based methods in dynamic Gaussian splatting suffer from an unsuitable low-rank assumption, causing feature overlap and poor rendering quality. Although 4D hash encoding provides an explicit representation without low-rank constraints, directly applying it to the entire dynamic scene leads to substantial hash collisions and redundancy. To address these challenges, we present DASH, a real-time dynamic scene rendering framework that employs 4D hash encoding coupled with self-supervised decomposition. Our approach begins with a self-supervised decomposition mechanism that separates dynamic and static components without manual annotations or precomputed masks. Next, we introduce a multiresolution 4D hash encoder for dynamic elements, providing an explicit representation that avoids the low-rank assumption. Finally, we present a spatio-temporal smoothness regularization strategy to mitigate unstable deformation artifacts. Experiments on real-world datasets demonstrate that DASH achieves state-of-the-art dynamic rendering performance, exhibiting enhanced visual quality at real-time speeds of 264 FPS on a single 4090 GPU. Code: https://github.com/chenj02/DASH.

Problem

• Dynamic scene reconstruction challenges in 3D vision
• Low-rank assumption causes poor rendering quality
• 4D hash encoding leads to hash collisions

Innovation

• Self-supervised decomposition for dynamic scenes
• Multiresolution 4D hash encoder
• Spatio-temporal smoothness regularization