DensiCrafter: Physically-Constrained Generation and Fabrication of Self-Supporting Hollow Structures

TL;DR

DensiCrafter addresses the challenge of generating lightweight, self-supporting hollow 3D structures from multimodal inputs by integrating a continuous density-field representation into a pretrained Trellis generator and optimizing it with three differentiable, simulation-free physical constraints plus a mass-regularization term. The method preserves the outer surface while carving interior cavities, yielding hollow structures whose density field is optimized over a restricted domain and subsequently converted back into a high-fidelity outer mesh with an inner cavity. Empirical results show up to 43% material mass reduction and improved upright stability, with high geometric fidelity and successful real-world FDM printing, demonstrating fabrication-ready, physics-aware generation without changing the underlying generative architecture. Overall, the work bridges 3D generative modeling and manufacturability by embedding density-based topology considerations directly into the generation pipeline, enabling practical, self-supporting designs.

Abstract

The rise of 3D generative models has enabled automatic 3D geometry and texture synthesis from multimodal inputs (e.g., text or images). However, these methods often ignore physical constraints and manufacturability considerations. In this work, we address the challenge of producing 3D designs that are both lightweight and self-supporting. We present DensiCrafter, a framework for generating lightweight, self-supporting 3D hollow structures by optimizing the density field. Starting from coarse voxel grids produced by Trellis, we interpret these as continuous density fields to optimize and introduce three differentiable, physically constrained, and simulation-free loss terms. Additionally, a mass regularization penalizes unnecessary material, while a restricted optimization domain preserves the outer surface. Our method seamlessly integrates with pretrained Trellis-based models (e.g., Trellis, DSO) without any architectural changes. In extensive evaluations, we achieve up to 43% reduction in material mass on the text-to-3D task. Compared to state-of-the-art baselines, our method could improve the stability and maintain high geometric fidelity. Real-world 3D-printing experiments confirm that our hollow designs can be reliably fabricated and could be self-supporting.

Figures (10)

• Figure 1: Pipeline of our method. We optimize this density field using differentiable losses that embed simulation-free physical constraints, yielding a material-efficient, self-supporting result.
• Figure 2: Self‐supporting structures generated from diverse inputs by optimizing the internal mass distribution; all examples remain upright under simulation.
• Figure 3: Our method produces self‐supporting, material‐efficient hollow structures.
• Figure 4: Text-to-3D generation results (each pair: left: without gravity; right: under gravity).
• Figure 5: Image-to-3D generation results (each pair: left: without gravity; right: under gravity).