Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Decoder Generates Manufacturable Structures: A Framework for 3D-Printable Object Synthesis

Abhishek Kumar

TL;DR

The paper tackles the challenge of generating printable 3D geometries by introducing a constraint-aware, decoder-based framework. It employs a variational decoder that maps latent codes to voxel grids at a fixed resolution $64^3$, with explicit manufacturability constraints such as a $45^\circ$ overhang limit, minimum wall thickness $t_{\min}=2$ mm, and connectivity requirements embedded in the decoding process. A four-block upsampling architecture with constraint-aware normalization and a composite loss $\,\mathcal{L}=\mathcal{L}_{\text{recon}}+\lambda_1\mathcal{L}_{\text{manuf}}+\lambda_2\mathcal{L}_{\text{kl}}$ guides learning, producing binary, printable occupancies after a threshold $\tau=0.5$. Empirical validation on 500 test objects across 5 categories and 50 printed samples demonstrates improved printability, reduced overhang violations, and practical inference speed, indicating potential for real-time, manufacturable design synthesis in additive manufacturing.

Abstract

This paper presents a novel decoder-based approach for generating manufacturable 3D structures optimized for additive manufacturing. We introduce a deep learning framework that decodes latent representations into geometrically valid, printable objects while respecting manufacturing constraints such as overhang angles, wall thickness, and structural integrity. The methodology demonstrates that neural decoders can learn complex mapping functions from abstract representations to valid 3D geometries, producing parts with significantly improved manufacturability compared to naive generation approaches. We validate the approach on diverse object categories and demonstrate practical 3D printing of decoder-generated structures.

Decoder Generates Manufacturable Structures: A Framework for 3D-Printable Object Synthesis

TL;DR

The paper tackles the challenge of generating printable 3D geometries by introducing a constraint-aware, decoder-based framework. It employs a variational decoder that maps latent codes to voxel grids at a fixed resolution , with explicit manufacturability constraints such as a overhang limit, minimum wall thickness mm, and connectivity requirements embedded in the decoding process. A four-block upsampling architecture with constraint-aware normalization and a composite loss guides learning, producing binary, printable occupancies after a threshold . Empirical validation on 500 test objects across 5 categories and 50 printed samples demonstrates improved printability, reduced overhang violations, and practical inference speed, indicating potential for real-time, manufacturable design synthesis in additive manufacturing.

Abstract

This paper presents a novel decoder-based approach for generating manufacturable 3D structures optimized for additive manufacturing. We introduce a deep learning framework that decodes latent representations into geometrically valid, printable objects while respecting manufacturing constraints such as overhang angles, wall thickness, and structural integrity. The methodology demonstrates that neural decoders can learn complex mapping functions from abstract representations to valid 3D geometries, producing parts with significantly improved manufacturability compared to naive generation approaches. We validate the approach on diverse object categories and demonstrate practical 3D printing of decoder-generated structures.
Paper Structure (12 sections, 3 equations, 3 figures, 1 table)

This paper contains 12 sections, 3 equations, 3 figures, 1 table.

Table of Contents

  1. Introduction
  2. Methodology
  3. Decoder Architecture
  4. Manufacturability Constraints
  5. Network Design
  6. Training Procedure
  7. Visualizations and Decoder Process
  8. Results
  9. Quantitative Evaluation
  10. Manufacturability Validation
  11. Discussion
  12. Conclusion

Figures (3)

  • Figure 1: Decoder architecture pipeline showing progressive upsampling from latent code to final 3D structure with integrated manufacturability constraints applied at each stage.
  • Figure 2: Comparison of unconstrained generation (left) showing problematic overhangs versus decoder-constrained output (right) with valid $45°$ support angles.
  • Figure 3: Examples of decoder-generated 3D-printable structures: (1) precision bracket with support geometry, (2) lattice structure optimized for weight reduction, (3) organic form with smooth manufacturability.