Hunyuan3D-Buffalo 1.0

A Unified Multimodal Model for Scalable 3D Generation,
Understanding, and Editing

Junliang Ye^*, Kenkun Liu^*, Guocun Wang^*, Yang Li^*†, Yansong Qu^*, Chunshi Wang^*,
Jingwei Xu, Yunhan Yang, Zibo Zhao, Jiachen Xu, Jiaao Yu, Lifu Wang,
Zhihao Liang, Zhuo Chen^†, Chunchao Guo^†

Tencent Hunyuan3D

^† Project Leaders

Yang Li, Zhuo Chen, Chunchao Guo

^* Core Contributors

3D Editing

Junliang Ye, Guocun Wang, Yansong Qu, Yang Li, Chunshi Wang, Kenkun Liu

Text-to-3D

Kenkun Liu, Junliang Ye, Yang Li

3D Understanding

Guocun Wang, Junliang Ye, Kenkun Liu, Yang Li

Hunyuan3D-Buffalo 1.0 Pipeline

Hunyuan3D-Buffalo 1.0 is a unified 3D multimodal framework for 3D understanding, text-to-3D generation, instruction-guided 3D editing, and part-level 3D generation. It supports diverse tasks within a single pipeline by connecting language, 3D representations, and generative 3D modules.

The framework uses a shared Hunyuan3D-VLM backbone to bridge 3D QA, grounding, generation, editing, and part-level decomposition. Based on this unified representation, Hunyuan3D DiT modules further enable scalable multimodal generation, high-quality 3D editing, and part generation.

3D Understanding

3D QA and grounding with language-guided reasoning.

Text-to-3D

Generate 3D assets from text prompts.

3D Editing

Edit 3D objects with natural-language instructions.

Part Generation

Extact semantic parts via language.

Instruction-Guided 3D Editing

Compare the source mesh and the edited mesh under user-specified editing instructions.

Editing Instruction

Source Mesh

Before Editing

Edited Mesh

After Editing

Case 1 of 1 · Hover a thumbnail to preview its edited result.

Part-Extraction with Language

Visualize the whole mesh together with all part components, shown either combined or exploded.

Part Instruction

Whole Mesh

Original

Exploded Parts

Separated Components

Case 1 of 1