{"slug": "stackgen3d-stable-3d-block-stack-generation-from-multiview-silhouettes", "title": "StackGen3D: Stable 3D Block Stack Generation from Multiview Silhouettes", "summary": "A research team (RIPL) posted arXiv paper 2610.31803v1 introducing StackGen3D, a two-stage framework that generates physically stable 3D block assemblies from a given block inventory and three orthographic silhouettes. StackGen3D's physics-informed graph neural network (PI-GNN) predicts block support relations and curates a dataset of stable structures, on which a conditional flow-matching model is trained; at inference, lightweight geometric selection and refinement reduce inter-block penetration without physics simulation. Best-of-10 selection raised physical stability from 45.3% to 72.4% at a silhouette IoU of 0.970, and the PI-GNN generalized from simple stacks to complex structures substantially better than a parameter-matched multilayer perceptron.", "body_md": "arXiv:2610.31803v1 Announce Type: new \nAbstract: Generating physically stable 3D structures from visual specifications requires resolving geometric ambiguity while accounting for physical interactions among objects. We present StackGen3D, a framework for generating stable block assemblies conditioned on a given block inventory and three orthographic silhouettes. Our approach consists of two stages. First, we develop a physics-informed graph neural network (PI-GNN) that models support relations between blocks and serves as a fast stability predictor for curating a diverse dataset of stable structures. Second, we train a conditional flow-matching model on this dataset to capture the multimodal distribution of block configurations consistent with the target silhouettes. At inference time, lightweight geometric selection and refinement reduce inter-block penetration without physics simulation. Preliminary experiments show that the PI-GNN generalizes from simple stacks to complex structures substantially better than a parameter-matched multilayer perceptron. For structure generation, Best-of-10 selection improves physical stability from 45.3% to 72.4% while achieving a silhouette IoU of 0.970. These results highlight the potential of combining relational physical reasoning, scalable data curation, and conditional generative modeling for creative 3D assembly.\n  Project website: [https://ripl.github.io/StackGen3D-web/](https://ripl.github.io/StackGen3D-web/)", "url": "https://wpnews.pro/news/stackgen3d-stable-3d-block-stack-generation-from-multiview-silhouettes", "canonical_source": "https://arxiv.org/abs/2610.31803", "published_at": "2026-10-26 04:00:00+00:00", "updated_at": "2026-10-26 04:12:02.151910+00:00", "lang": "en", "topics": ["artificial-intelligence", "machine-learning", "ai-research", "neural-networks", "generative-ai"], "entities": ["StackGen3D", "RIPL", "arXiv", "PI-GNN"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/stackgen3d-stable-3d-block-stack-generation-from-multiview-silhouettes", "markdown": "https://wpnews.pro/news/stackgen3d-stable-3d-block-stack-generation-from-multiview-silhouettes.md", "text": "https://wpnews.pro/news/stackgen3d-stable-3d-block-stack-generation-from-multiview-silhouettes.txt", "jsonld": "https://wpnews.pro/news/stackgen3d-stable-3d-block-stack-generation-from-multiview-silhouettes.jsonld"}}