Hunyuan 3D V3.1 — Image or Text to 3D, Done Right
Production-Grade 3D Models from an Image or a Prompt
Hunyuan 3D V3.1 turns either a single image or a plain-text description into a complete, textured 3D model. Drop in a product shot or character concept and the image-to-3D pipeline reconstructs the geometry and wraps it in coherent textures — or skip the image entirely, describe what you want ('a stylized low-poly red fox, game-ready'), and the text-to-3D pipeline builds the model from scratch. Either way you get the same Pro-grade output: faithful silhouettes, sensible topology, and optional PBR materials (base color, roughness, metalness) so the model reacts correctly to lighting in any modern engine. Export as GLB for the web and game engines or OBJ for traditional DCC tools, and preview the result interactively right on this page before you download.
The two input modes cover different starting points. Image-to-3D is the right call when you already have a reference — a photo, a render, or concept art — and want the model to match it closely; a single, well-lit image with the subject clearly separated from the background gives the cleanest reconstruction, and you can optionally add extra views (back, sides, top) to sharpen the result. Text-to-3D is for when you're starting from an idea rather than a picture: write a clear, concrete prompt — name the object, its style, and the details that matter — and the model composes geometry and textures from language alone, no asset hunting required.
PBR materials are the headline upgrade for realism in both modes. With PBR enabled, the model produces the full map set a physically based renderer expects, so metal reads as metal and matte surfaces stay matte under changing light — drop-in compatible with Unreal, Unity, Blender, three.js, and any glTF viewer, with no manual material pass. You also control the polygon budget, from a lightweight 40k-triangle mesh for real-time use up to 1.5 million triangles when you need every contour for close-up renders or 3D printing, plus a geometry-only mode when you just want clean topology without textures.
The workflow stays simple. Pick a mode, provide your image or prompt, choose your detail level and whether you want PBR, then generate and inspect the model in the built-in interactive viewer — orbit, zoom, and check the silhouette from every angle. When it looks right, download the GLB (or grab the OBJ) and move straight into your engine, render, or slicer. Every generation is also saved to your dashboard gallery so you can return to it later.
How to Generate a 3D Model with Hunyuan 3D V3.1
Pick a Mode & Provide Input
Choose Image to 3D and upload a clear reference image, or switch to Text to 3D and describe the model you want — name the object, its style, and the key details.
Choose Detail & PBR
Pick your polygon budget, and toggle PBR materials on if you want realistic, physically based textures for engine lighting — or choose geometry-only for clean, untextured topology.
Generate & Download
Click Generate, then orbit and zoom the result in the interactive viewer. Download the GLB or OBJ — every model is also saved to your dashboard gallery.
Hunyuan 3D V3.1 Technical Specifications
| Provider | Tencent |
| Platform | Cloud API (hosted) |
| Input Modes | Image-to-3D and text-to-3D |
| Input | Single image, or a text prompt (up to 1024 characters) |
| Output Formats | GLB and OBJ |
| PBR Materials | Optional — base color, roughness, metalness |
| Polygon Count | 40,000 to 1,500,000 triangles |
| Preview | Interactive in-browser 3D viewer |
| Commercial Use | Supported |
| Processing | Asynchronous, typically 1–2 minutes |
Why Choose Hunyuan 3D V3.1
Image or Text to 3D
Start from a reference photo or concept art with image-to-3D, or from a plain-language prompt with text-to-3D — one model, two ways in, the same production-grade result with faithful silhouettes and sensible topology.
True PBR Materials
Optional base color, roughness, and metalness maps mean the model reacts correctly to lighting in Unreal, Unity, Blender, and any glTF viewer — no manual material pass.
Engine-Ready Export
Download GLB for the web and game engines or OBJ for traditional DCC tools, and preview interactively before you commit — what you see is what you ship.
Hunyuan 3D V3.1 vs Other Image-to-3D Methods
| Feature | Hunyuan 3D V3.1 | Trellis 2 | Photogrammetry |
|---|---|---|---|
| Input | Image or text prompt | Single image | Dozens of photos |
| PBR Materials | Yes — optional | Texture only | Manual / baked |
| Output | GLB + OBJ | GLB | Mesh + textures |
| Detail Control | 40k–1.5M triangles | Resolution presets | Capture-dependent |
| Setup Effort | One image or one prompt | One upload | Rig + capture session |
| Best For | Game & product assets | Clean-topology models | Real-world scans |
What Can You Create with Hunyuan 3D V3.1
Game Assets
Turn concept art, reference photos, or a written brief into real-time game props, characters, and environment pieces with a polygon budget tuned for your engine.
Product Visualization
Convert a single product photo into an interactive 3D model for e-commerce, configurators, and marketing — with PBR materials for accurate surfaces.
3D Printing
Generate high-polygon meshes from an image or a prompt and export OBJ for slicing, prototyping, and physical models of objects, miniatures, and collectibles.
AR & VR
Produce lightweight GLB assets that drop straight into AR experiences and VR scenes, viewable in any glTF-compatible runtime.
Idea or Concept to 3D
Have a 2D concept? Use image-to-3D. Have only an idea? Describe it with text-to-3D. Either way, get a rotatable 3D version in minutes — ideal for pitching, previs, and design iteration.
Animation & Film
Build set-dressing props and background assets fast, then refine in your DCC tool of choice with clean OBJ exports and usable UVs.
Related AI Models
Frequently Asked Questions About Hunyuan 3D V3.1
What input does Hunyuan 3D V3.1 accept — images or text?
Both. Use the Image to 3D / Text to 3D switch at the top of the generator. Image-to-3D reconstructs a model from a reference picture (and optional extra views), while text-to-3D builds one from a written description of up to 1024 characters. Both modes produce GLB and OBJ output with optional PBR materials.
What are PBR materials and should I enable them?
PBR (physically based rendering) materials describe how a surface reacts to light using base color, roughness, and metalness maps. Enable PBR when you want realistic results in engines like Unreal, Unity, Blender, or three.js. If you only need quick geometry or plan to texture the model yourself, leave it off — or choose geometry-only mode for untextured topology.
How much detail can I get?
You choose the polygon budget in both modes. A lower count produces a balanced mesh suitable for real-time use, while a higher count pushes the triangle count up for close-up renders and 3D printing — the model supports up to 1.5 million triangles. Higher detail takes slightly longer and costs a few more credits.
How long does generation take?
Usually one to two minutes for either mode. Generation runs asynchronously — progress is shown on the page and the finished model is saved to your dashboard gallery, so you can navigate away and come back to it.
Can I use the models commercially?
Yes. Models generated with Hunyuan 3D V3.1 support commercial use, including games, products, client deliverables, and monetized content. Download your file within 7 days — gallery media is cleaned up after that window.
How is this different from Trellis 2?
Hunyuan 3D V3.1 accepts both images and text prompts, emphasizes high-fidelity geometry with optional PBR materials, and exports both GLB and OBJ — a strong choice for engine-ready and printable assets. Trellis 2 is image-only and focuses on clean topology with explicit resolution and texture-size control. Try both on the same subject and keep the result that fits your pipeline.