3D Generation Model

Hunyuan3D V2 Multi-View

Hunyuan3D V2 Multi-View is a 3D generation model from WaveSpeed that converts multi-view images into textured 3D meshes.

Start Building with Hunyuan3D V2 Multi-View View All Models

Publisher

WaveSpeed

Type3D

Context Window50,000 tokens

Released2026

Price$0.01-$0.03/model

Provider

WaveSpeed

3dtext-to-3dimage-to-3dwavespeed

Try Hunyuan3D V2 Multi-View →

About Hunyuan3D V2 Multi-View

Multi-view image-to-3D mesh generation

Hunyuan3D V2 Multi-View is a 3D generation model published by WaveSpeed that takes up to four directional images — front, back, left, and right — and produces a 3D mesh from them. It is based on Tencent's Hunyuan3D V2 architecture and is available on MindStudio at a cost of $0.01–$0.03 per generated model. The multi-view variant is specifically designed to accept images from multiple angles simultaneously, which gives the reconstruction pipeline more geometric information to work with compared to single-image approaches.

The model supports configurable inference steps, guidance scale, and octree resolution, giving developers control over output quality and detail level. An octree-based representation is used internally to encode 3D geometry, and the model can optionally output a textured mesh. It is well suited for workflows in game asset creation, product visualization, and any pipeline that requires converting real-world or rendered objects into 3D geometry from multi-angle photography.

Capabilities

What Hunyuan3D V2 Multi-View supports

Multi-View Input

Accepts up to four directional image inputs — front, back, left, and right — to reconstruct 3D geometry from multiple angles simultaneously.

3D Mesh Output

Generates a 3D mesh from the provided images, with an optional textured mesh output selectable via a toggle input.

Octree Resolution Control

Exposes an octree resolution parameter that lets developers tune the geometric detail and memory footprint of the output mesh.

Inference Tuning

Supports configurable inference steps and guidance scale inputs, allowing trade-offs between generation speed and output fidelity.

Reproducible Generation

Accepts a seed value so that outputs can be reproduced exactly or varied systematically across runs.

Ready to build with Hunyuan3D V2 Multi-View?

Get Started Free

FAQ

Common questions about Hunyuan3D V2 Multi-View

What inputs does Hunyuan3D V2 Multi-View require?

The model accepts up to four image URLs corresponding to the front, back, left, and right views of an object. You can also provide a seed, inference step count, guidance scale, octree resolution, and a textured mesh toggle.

How much does it cost to use this model?

Pricing is $0.01–$0.03 per generated 3D model, depending on the configuration used.

Does the model support text-to-3D generation?

The multi-view variant is specifically designed for image-to-3D workflows using directional image inputs. Text-to-3D is listed as a tag associated with the broader Hunyuan3D V2 family, but this variant's inputs are image URLs rather than text prompts.

What is the context window for this model?

The model has a context window of 50,000 tokens and a maximum response size of 10,000 tokens.

Can I control the level of detail in the output mesh?

Yes. The octree resolution parameter lets you adjust the geometric detail of the output. Higher resolutions produce finer detail but may increase processing time.

Resources