Text, image, or sketch to 3D with Rapid and Pro modes; optimize meshes and segment parts for clean, usable 3D assets.
3D prompts work best when they describe silhouette, materials, and output intent. The stronger the build brief, the more usable the resulting asset becomes.
Best results start with a clear subject, materials, and output goal.
Hunyuan 3D V3 / V3.1 is Tencent's AI 3D generator on Pixio. It turns text, a single image, multiple views, or a sketch into production-ready 3D meshes with PBR materials, configurable polygon counts, and optional part segmentation. Rapid mode is faster and lighter; Pro (V3.1) delivers higher fidelity and supports more input types—ideal when you need clean, engine-ready assets for games, viz, or animation.
Hunyuan 3D V3 / V3.1 is Tencent's AI 3D generator on Pixio. It turns text, a single image, multiple views, or a sketch into production-ready 3D meshes with PBR materials, configurable polygon counts, and optional part segmentation. Rapid mode is faster and lighter; Pro (V3.1) delivers higher fidelity and supports more input types—ideal when you need clean, engine-ready assets for games, viz, or animation.
| Mode | Input | Best for |
|---|---|---|
| Text to 3D | Prompt only | Concepts from scratch—describe shape, material, and style |
| Image to 3D | Single image or multi-view images | When you have a reference; multi-view improves accuracy |
| Sketch to 3D | Sketch drawing (Pro/V3.1) | Fast concept blockouts, silhouette-driven generation |
| Option | Values | Notes |
|---|---|---|
| Mode | Rapid, Pro (V3.1) | Rapid for speed and cost; Pro for highest fidelity, sketch input, and part segmentation |
| Face count | 40K – 1.5M | Higher for hero assets; lower for LODs or real-time engines |
| Output type | Textured (PBR), low-poly, geometry-only | Depends on backend; textured is full 4K PBR where supported |
| Topology | Triangle, Quad-dominant | Quads better for rigging, animation, and retopology |
| Part segmentation | On / Off (when available) | Splits mesh into editable parts for cleanup or reuse |
Credits and exact ranges depend on Rapid vs Pro and resolution; check the model card in Pixio for current values.
Rapid is tuned for fast iteration and lower cost—use it for concept validation and drafts. Pro (V3.1) adds higher resolution (e.g. up to 1.5M faces), sketch-to-3D input, multi-view refinement, and part segmentation for clean, component-based assets. Switch to Pro when you need the best single-asset quality, quad-dominant topology for animation, or export-ready PBR for engines.
Text-to-3D: Describe subject + shape + material + style in one clear sentence. Example: "A vintage brass desk lamp with a green glass shade, art deco style, matte metal and glossy glass."
Image-to-3D: Use a clear reference with good silhouette and even lighting. Multi-view (several angles of the same object) reduces artifacts and missing geometry.
Sketch-to-3D: Simple, readable silhouettes and main contours work best; the model infers depth and detail from the sketch.
| Scenario | Best choice |
|---|---|
| Highest-quality image/text/sketch to 3D with PBR and face control | Hunyuan 3D V3 / V3.1 |
| Multi-view input and part segmentation | Hunyuan 3D V3 / V3.1 or Tripo |
| Quick single-image draft or simple asset | Image to 3D (generic) or Meshy |
| Full pipeline: text/image to 3D + rigging, segmentation, plugins | Tripo or Meshy |
| Style-only retexture or remesh of existing geometry | Meshy (remesh/retexture) |
Lead with silhouette before detail.
Materials help the model resolve form more clearly.
Say what the asset is for: product, game, animation, visualization.
Refinement should serve the pipeline, not just aesthetics.
A strong 3D prompt defines silhouette, materials, and final use so the result feels buildable instead of vague.
Describe the subject, silhouette, scale, and material language so the asset has a clear physical identity.
Use images or multi-view inputs when the object shape needs to survive more accurately.
Improve the asset once the core shape works so it fits better into game, product, or visualization workflows.
Hunyuan 3D V3 / V3.1 is strongest when the prompt reads like a build spec instead of a loose concept caption.
Use it for product forms, environment props, stylized assets, or 3D pipelines that need a strong starting mesh.
When refining, optimize toward the final destination instead of trying to solve everything in the first prompt.
Say what the object is and how it should read at a glance before chasing detail.
Describe the surface language and what the asset is meant for so the model has a stronger target.
Once the form is correct, improve readiness for texturing, animation, or export instead of starting over.
Use image generation first when you need a clearer concept frame before turning it into an asset.
Once the form works, stylization tools can push the asset into a more distinct final language.
