Rodin: the Most Controllable Text/Image-to-3D AI

<aside> 💡

What’s new in Gen-2.5:

Controllable Thinking Effort lets you choose between speed（4s~80s） and detail across five generation modes.
- Extreme High can deliver skin-texture-level detail with detailed inputs, reaching up to industry’s first 10M+ polygons.
- Extreme Low can generate up to 1M-polygon or lower-poly models in around 4 seconds, making it ideal for fast iteration and API workflows.
A new 3D-native texturing architecture brings full PBR materials, without blind spots, and more realistic surface results, reaching up to industry’s first 12K texture.
All modes support batch generation, and Low + X10 is a great setup for exploring 10 outputs at once.
Smart Low-poly now supports more artist-style topology for workflows that need cleaner, more usable meshes.
Manual/auto Bang-to-parts and Refine: manually split models into parts, parts can be further splitted and refined, support 3rd party models import.

View API Doc here

DCC Bridge (Blender, Unity, UE, Maya, Godot, Issac, Omniverse, C4D, 3dMax) here

</aside>

INPUT

Upload the image(s) you want to use for 3D. You can upload up to 5 images. Use “Remix Gen” if you want style change, object replacement or others.
Choose from different “thinking effort”, “presets” and “batch generation amount”.

GIF5 - Hyper3D Rodin Different Thinking Effort.gif

Click on “Generate”
Geometry Generation: Adjust the image weight (we allow up to 5 images to be uploaded at one time), and choose the geometry option such as “smooth edges”. “Redo” for free until you find a satisfied outcome; “history” to choose among different redo versions; only charge when you confirm the geometry.
Material Generation: “Redo” for free until you find a satisfied outcome; “history” to choose among different redo versions; only charge when you confirm the material.
Pack: You can choose to pack only the geometry or both the geometry and the material.

Write a text prompt to generate an image preview, redo until meeting the satisfying generated image, then follow the same steps as the image to 3D.