🎉 Launch Sale — 20% OFF all credit packages!

Get Credits
Logo of Trellis2
Trellis 2
FeaturesPricing
BlogFAQ
  1. Home
  2. Hunyuan3D
Tencent Open-Source 3D AI Generation

Hunyuan3D — Text & Image to 3D Model

Transform any image or text prompt into a production-ready 3D model with PBR textures. Powered by Tencent's open-source Hunyuan3D diffusion system — generate high-resolution 3D assets online, no GPU required.

UploadConfigureGenerate

Stage 01

Upload Image

Upload any image to convert to 3D

Stage 02

Generation Settings

Configure the quality and style of your 3D model

Generation Settings

Generation Type
PBR Materials
Polygon Count
Fine-tune500,000
Costs 3 credits per generation

What is Hunyuan3D?

Hunyuan3D is an open-source large-scale 3D synthesis system developed by Tencent. It generates high-resolution textured 3D assets from a single image or a text prompt using a two-stage diffusion architecture — a shape model followed by a texture model.

Hunyuan3D produces models with full PBR (Physically-Based Rendering) materials — base color, metallic, and roughness maps — making the output ready for games, AR/VR, and3D printing. It exports in GLB, OBJ, and PLY formats compatible with Unity, Unreal, Blender, and Maya.

Whether you're a game developer prototyping props, an e-commerce brand building3D product visualizations, or a creator exploring AI 3D, Hunyuan3D delivers production-quality 3D models in under a minute — no GPU required when using our cloud-powered platform.

Why Choose Hunyuan3D?

Hunyuan3D combines a diffusion-based shape model with a dedicated texture model to deliver high-fidelity, production-ready 3D assets.

Image to 3D Model

Turn a single photo, sketch, or AI image into a fully textured 3D model. The model excels at single-object generation with clean geometry.

Text to 3D Generation

Describe the 3D model you want with a text prompt and Hunyuan3D generates a textured mesh directly from language.

PBR Texture Generation

Automatic PBR materials with base color, metallic, and roughness maps for photorealistic relighting in any engine.

Multi-View Reconstruction

Supports multi-view image input for more accurate 3D reconstruction and consistent geometry from every angle.

Multiple Output Formats

Export as GLB for web and game engines, OBJ for traditional 3D software, or PLY for point clouds and pipelines.

Open-Source by Tencent

Free for commercial and non-commercial use. Model weights and source code available on GitHub and Hugging Face.

How Hunyuan3D Works

1

Provide Image or Text

Upload a single image — a photo, sketch, or AI-generated picture — or describe the 3D model you want with a text prompt.

2

AI 3D Generation

Hunyuan3D's diffusion-based shape model builds the geometry, then the texture model applies high-resolution PBR materials.

3

Export & Use

Download your 3D model as GLB, OBJ, or PLY with full PBR materials. Use it in games, AR/VR, 3D printing, or any 3D workflow.

Two Ways to Generate 3D Models

Hunyuan3D supports both image-to-3D and text-to-3D, giving you flexibility whether you start from a reference or an idea.

Image to 3D Model

Best for converting existing photos, product shots, or concept art into 3D. Ideal for e-commerce, archiving, and rapid prototyping.

Text to 3D Model

Best for brainstorming and creating 3D from imagination. Describe any object and let Hunyuan3D bring it to three-dimensional life.

Hunyuan3D Use Cases

From game development to e-commerce, Hunyuan3D powers 3D content creation across industries. Try it free or preview models in our free STL viewer.

Game Development

Rapidly prototype game props, characters, and environment assets. Export as GLB for Unity, Unreal, and web engines.

E-Commerce & Product Visualization

Convert product photos into interactive 3D models for online stores, AR try-on, and marketing visuals.

Film & VFX

Generate concept-art 3D assets for pre-visualization and speed up the pipeline from idea to on-screen model.

AR/VR Content Creation

Build immersive 3D content for augmented and virtual reality from simple 2D reference images.

Architecture & Design

Transform sketches and reference images into 3D models for visualization and client presentations.

Education & Research

Create anatomical, scientific, and educational 3D models for interactive learning and research.

Hunyuan3D 2.0 vs 3.1 vs 3.5

Hunyuan3D has evolved across versions. The 3.1 Pro edition powers this tool — here's how the generations compare.

Feature2.03.13.5
ArchitectureTwo-stage diffusion8-view reconstruction + watertight meshSub-60s optimized pipeline
Texture QualityStandard PBR4K PBR (metallic/roughness/normal)Up to 8K PBR textures
Mesh TopologyBasicClean quad topology + smart UV2M+ polygons, production-ready
Generation SpeedModerateRapid 2–3 min / Pro 3–6 minUnder 60 seconds
Export FormatsGLB / OBJ / PLYGLB / OBJ / PLYGLB / OBJ / PLY
LicenseOpen-sourceOpen-source (Tencent)Open-source (Tencent)

Hunyuan3D vs Microsoft Trellis 2

Both are open-source image-to-3D models. Here's how they compare on key capabilities so you can pick the right one for your workflow.

FeatureHunyuan3DMicrosoft Trellis 2
Developed byTencentMicrosoft Research
Input TypesImage, Text, or Multi-viewSingle image
Text to 3DYesNo
Polygon Control40K–1.5M adjustableAutomatic
PBR TexturesOptional (add-on)Always included
Generation SpeedFast (Pro pipeline)Fast (4B O-Voxel)
Credits per Generation2–4 (Geometry / Normal / +PBR)1–3 (512p / 1024p / 1536p)
Commercial UseYes (open-source)Yes (MIT license)

Output Formats & PBR Materials

Supported Export Formats

  • GLB — for web, game engines (Unity, Unreal), and AR/VR
  • OBJ — for Blender, Maya, and traditional 3D modeling
  • PLY — for point clouds, Gaussian splats, and research pipelines

PBR Material Maps

  • Base color (albedo) map
  • Metallic map for reflective surfaces
  • Roughness map for surface finish control

Official Resources

GitHub Repository (2.0)
Source code and installation instructions
Hugging Face Demo
Try Hunyuan3D in your browser
Hunyuan3D 2.1 Paper
From Images to High-Fidelity 3D Assets
Official Tencent Site
Tencent Hunyuan 3D homepage
ComfyUI Tutorial
Run Hunyuan3D-2 in ComfyUI
GitHub Repository (2.1)
Latest scalable 3D asset creation system

Related 3D Tools

Microsoft Trellis 2

Another open-source image-to-3D model from Microsoft Research

3D Reconstruction

Reconstruct 3D models from multiple images online

STL Viewer

Preview and inspect your generated 3D models in-browser

Frequently Asked Questions

What is Hunyuan3D?

Hunyuan3D is an open-source large-scale 3D synthesis system developed by Tencent. It generates high-resolution, textured 3D assets from a single image or text prompt using diffusion-based shape and texture models.

Is Hunyuan3D free to use?

Yes, Hunyuan3D is open-source and free. You can try it online on Hugging Face Spaces or run it locally on an NVIDIA GPU (24GB+ VRAM recommended).

What file formats does Hunyuan3D support?

Hunyuan3D exports 3D models in GLB, OBJ, and PLY formats with full PBR materials including base color, metallic, and roughness maps.

Can I use Hunyuan3D commercially?

Yes. Hunyuan3D is released under Tencent's open-source license permitting both commercial and non-commercial use. Always verify the license terms for your specific use case.

What is the difference between Hunyuan3D 2.0, 3.1 and 3.5?

Hunyuan3D 2.0 introduced the two-stage shape-texture diffusion architecture. Version 3.1 (which powers this tool) adds 8-view reconstruction, watertight meshes, clean quad topology, smart UV unwrapping, and 4K PBR textures. Version 3.5 pushes speed further with sub-60-second generation, up to 8K PBR textures, and 2M+ polygon output.

Does Hunyuan3D need a GPU?

Running Hunyuan3D locally requires an NVIDIA GPU with 24GB+ VRAM. However, you can use the free online demo or our cloud-powered platform from any device with a modern browser — no GPU required.

How long does Hunyuan3D take to generate a 3D model?

On our cloud platform, most 3D models are generated in under a minute. Local inference times depend on your GPU and the selected resolution and texture quality.

What kind of images work best for Hunyuan3D?

Clear, well-lit images of a single object with good contrast and a clean background produce the best results. The model excels at single-object generation such as props, characters, and product items.

Trellis2.app is an independent product and is not affiliated with, endorsed by, or connected to Tencent. Hunyuan3D is an open-source project developed by Tencent, released under Tencent's open-source license.

Learn more about Hunyuan3D on the official Tencent site.
Last updated: July 2026Trellis2.app Editorial TeamReviewed for technical accuracy

Ready to Generate 3D Models with Hunyuan3D?

Upload any image or describe your idea and generate a production-ready 3D model in under a minute. No GPU required — our cloud platform does the heavy lifting.

Start Creating 3D Models FreeView Pricing Plans

No credit card required. Free tier includes monthly generation credits.

Logo of Trellis2
Trellis 2

Transform images into stunning 3D models with AI-powered technology

Product

  • Features
  • Pricing
  • Blog
  • FAQ

Company

  • About
  • Contact
  • Trify3D

Legal

  • Privacy Policy
  • Terms of Service

© 2026 Trellis 2. All rights reserved.