DreamPartGenSemantically Grounded Part-Level 3D Generation via Collaborative Latent Denoising

Understanding and generating 3D objects as compositions of meaningful parts is fundamental to human perception and reasoning. However, most text-to-3D methods overlook the semantic and functional structure of parts. While recent part-aware approaches introduce decomposition, they remain largely geometry-focused, lacking semantic grounding and failing to model how parts align with textual descriptions or their inter-part relations. We propose DreamPartGen, a framework for semantically grounded, part-aware text-to-3D generation. DreamPartGen introduces Duplex Part Latents (DPLs) that jointly model each part's geometry and appearance, and Relational Semantic Latents (RSLs) that capture inter-part dependencies derived from language. A synchronized co-denoising process enforces mutual geometric and semantic consistency, enabling coherent, interpretable, and text-aligned 3D synthesis. Across multiple benchmarks, DreamPartGen delivers state-of-the-art performance in geometric fidelity and text-shape alignment.

• DreamPartGen unifies geometric, visual, and relational reasoning. We introduce a language-grounded collaborative diffusion framework for coherent and interpretable part-level text-to-3D synthesis. Persistent language-derived relational semantic latents remain active throughout denoising, rather than using text only as a one-shot condition.
• Duplex Part Latents and Relational Semantic Latents encode complementary information. DPLs encode modular, disentangled part geometry and appearance, while RSLs provide text-derived local refinements and global planning signals. Synchronized co-denoising through intra-part and inter-part attention maintains geometry–appearance alignment and guides consistent part assembly.
• PartRel3D provides explicit supervision for inter-part relations. We curate a large-scale relational dataset with 300K canonicalized functional and spatial triplets across 175 object categories. These annotations connect part geometry, appearance, and language through relations such as support, attach, hinge, above, touching, and aligned-with.
• Experiments demonstrate coherent and controllable part-aware generation. DreamPartGen achieves state-of-the-art performance across the reported benchmarks, with gains in geometric fidelity, text–shape alignment, and controllable part-level synthesis.