Research map / Graph learning / Knowledge and temporal graphs
MultiMat: Multimodal Program Synthesis for Procedural Materials using Large Multimodal Models
Jonas Belouadi, Tamy Boubekeur, Adrien Kaiser
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add incremental tree search with backtracking at inference (ablation) | VLMaterial[SBS] · Unconditional generation | Node Error Ratio (x100): 33.953 -> 14.846 | Improved |
| condition on whole-graph visualization (Graph) instead of text-only program | MultiMat · Unconditional generation | KID (x100): 14.155 -> 2.365 | Improved |
| use mixed conditioning (text plus node preview images) instead of text-only | MultiMat · Unconditional generation | Node Error Ratio (x100): 14.846 -> 8.923 | Improved |
| use graph conditioning instead of mixed conditioning (ablation) | MultiMat · Unconditional generation | Node Error Ratio (x100): 8.923 -> 15.024 | Hurt |
| use graph conditioning instead of mixed conditioning (ablation) | MultiMat · Inverse generation, no parameter optimization | DreamSim (x100): 34.922 -> 36.609 | Improved |
| add gradient-based parameter optimization with differentiable rendering (ablation) | MultiMat · Inverse generation | DreamSim (x100): 36.609 -> 40.367 | Improved |
| add gradient-based parameter optimization with differentiable rendering (ablation) | VLMaterial[SBS] · Inverse generation | DreamSim (x100): 31.344 -> 31.348 | No clear effect |
| add parameter optimization (ablation) | MultiMat[Graph] · Inverse generation | KID (x100): 2.801 -> 14.886 | Hurt |
| use multimodal conditioning instead of text-only | MultiMat · Repair statistics | Nodes repaired by conversion (%): 12.26 -> 3.51 | Improved |
Closest papers by meaning
- VLMaterial: Procedural Material Generation with Large Vision-Language Models (ICLR 2025 · similarity 0.92)
- Make-it-Real: Unleashing Large Multimodal Model for Painting 3D Objects with Realistic Materials (NeurIPS 2024 · similarity 0.87)
- ShapeCraft: LLM Agents for Structured, Textured and Interactive 3D Modeling (NeurIPS 2025 · similarity 0.87)
- MatFuse: Controllable Material Generation with Diffusion Models (CVPR 2024 · similarity 0.87)
- 3DCodeBench: Benchmarking Agentic Procedural 3D Modeling Via Code (NeurIPS 2026 · similarity 0.86)
- cadrille: Multi-modal CAD Reconstruction with Reinforcement Learning (ICLR 2026 · similarity 0.86)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
