All Models
Hunyuan3D V3
Hunyuan3D V3 generates textured 3D models from text or images, with optional PBR materials.
Call the API endpoint to get the latest, detailed pricing and pricing estimates.
Call the API endpoint to get the latest, detailed pricing and pricing estimates.
Call the API endpoint to get the latest, detailed pricing and pricing estimates.
Hunyuan3D V3
Hunyuan3D V3 is Tencent’s 3D model from the Hunyuan lab. It builds a textured mesh from a text description or from images. Physically based materials are available and off by default.
Face count runs from 40,000 to 1.5 million, with 500,000 as the default. You can choose Normal, LowPoly, or Geometry generation, and triangles or quads.
From an image, you can add left, right, and back views of the same object. Extra views help the model reconstruct sides it would otherwise have to invent.
What it's good for
Product meshes and props that will go into a real pipeline, including real-time work when you pick LowPoly. It is especially useful when you have more than one photo of the object — a bottle, a shoe, a device — and you want the back and sides to match the references instead of being guessed.
Model ID
tencent/hunyuan3d-v3
Output Type
3D
Pricing
$0.25
Provider
Tencent
Hosting
Protoface
Hunyuan3D V3
Hunyuan3D V3 is Tencent’s 3D model from the Hunyuan lab. It builds a textured mesh from a text description or from images. Physically based materials are available and off by default.
Face count runs from 40,000 to 1.5 million, with 500,000 as the default. You can choose Normal, LowPoly, or Geometry generation, and triangles or quads.
From an image, you can add left, right, and back views of the same object. Extra views help the model reconstruct sides it would otherwise have to invent.
What it's good for
Product meshes and props that will go into a real pipeline, including real-time work when you pick LowPoly. It is especially useful when you have more than one photo of the object — a bottle, a shoe, a device — and you want the back and sides to match the references instead of being guessed.