D-Prism: Differentiable Primitives for Structured Dynamic Modeling

2026-04-18

AI Summary

Existing dynamic modeling approaches struggle to simultaneously capture the geometric structure and rigid motion of structured dynamic objects, such as multi-part assemblies or articulated mechanisms, with high fidelity. This work proposes a novel method that extends differentiable primitives to dynamic scenes by integrating 3D Gaussian splatting with a deformation network. Specifically, Gaussians are anchored onto the surfaces of differentiable primitives, and an adaptive strategy dynamically adjusts the number of primitives during optimization. This enables joint refinement of both geometry and motion states. The proposed approach substantially outperforms existing unstructured representations on structured dynamic scenes, achieving notable improvements in geometric reconstruction accuracy and motion tracking consistency.

Technology Category

• Computer Vision: Motion & Tracking
• Search and Optimization: Sampling/Simulation-based Search
• Intelligent Robots: Motion and Path Planning

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Capturing both geometry and rigid motion for structured dynamic objects, like multi-part assemblies or jointed mechanisms, remains a key challenge. Existing dynamic methods, such as deformable meshes or 3DGS, rely on unstructured representations and fail to jointly model suitable geometry and articulated motion. Primitive-based methods excel at structured static scenes, but their dynamic potential is still unexplored. We propose D-Prism, the first framework to achieve high-fidelity structured dynamic modeling by extending differentiable primitives to the dynamic domain. Specifically, we bind 3DGS to primitive surfaces, leveraging their respective strengths in appearance and geometry. We introduce a deformation network to control primitive motion, ensuring it accurately matches the object's movement. Furthermore, we design a novel adaptive control strategy to dynamically adjust primitive counts, better matching objects' true spatial footprint. Experiments confirm that our method excels at structured dynamic modeling, providing both structured geometry and precise motion tracking.

Problem

• structured dynamic modeling
• geometry and motion
• articulated objects
• primitive-based representation
• dynamic 3D reconstruction

Innovation

• differentiable primitives
• structured dynamic modeling
• 3D Gaussian Splatting
• articulated motion
• adaptive primitive control