Overview
Rodin Hyper3D provides multiple 3D generation modes optimized for different use cases—from fast prototyping to higher-quality model output. Modes will vary in generation time, mesh density, texture resolution or cost. The model supports both image-based and text-based inputs, with options for multi-image workflows (multi-view).
The tables and descriptions below outline the key features of each model and generation mode for Rodin:
Model | Description | Avg. Time | Features | Extras |
|---|---|---|---|---|
Rodin Sketch | Fast prototyping with low-res output | ~20 sec | 1K textures, basic geometry, simple UV mapping, low-poly | — |
Rodin Regular | Balanced, customizable 3D model generation | ~70 sec | Adjustable polygons, 2K textures (Base Pack) | High Pack available |
Rodin Detail (soon) | Enhanced version of Regular with greater detail | >70 sec | 2K textures, adjustable polygons | High Pack available |
Rodin Smooth (soon) | Clearer, sharper output with slightly longer generation | >70 sec | 2K textures, adjustable polygons | High Pack available |
Rodin Generation Modes
(i) Rodin Sketch
Designed for rapid 3D asset prototyping, Rodin Sketch delivers basic geometry with quick turnaround. Output is limited to GLB format with triangular meshes. Textures are available in Shaded or PBR formats
(ii) Rodin Regular
Offers detailed 3D model generation with control over polygon count and texture quality. Outputs may support multiple formats (GLB, OBJ, FBX, etc.) and include both triangular and quad meshes depending on the format. Textures can be requested as Shaded or PBR.
(iii) “High Pack“ Variant
Available for Rodin Regular, this enhances models with 4K textures and high-polygon geometry—approximately 16 times the poly count of standard outputs.
Input Modes
a. Image-to-3D
Automatically triggered when image files are submitted. Rodin will generate a 3D model based on the uploaded images. You can add a prompt in addition to the image, or drag and drop the input image to the propt box, to generate a relevant prompt in seconds.
b. Text-to-3D
This mode will activate when no input image is provided. Instead, a descriptive text prompt is required to drive the generation process.
Multi-Image Support
When multiple images are submitted, Rodin switches to multi-image generation. Two configuration modes are available:
Concat Mode (Default)
Use when images depict different views of the same object. This mode helps Rodin reconstruct a consistent 3D representation from various angles. One or more images are required (you can upload multi-view images in any order, regardless of the order of view.)
Fuse Mode
If you are uploading images of multiple objects, fuse mode will try to combine features from the different objects in images into one single generated model.
Quality (mesh density)
This parameter defines the generation quality. Possible values are high (50k faces), medium(18k faces), low(8k faces), and extra-low(4k faces). The default is medium. For Rodin Sketch, the value will fixed to medium.
Below are four variants of the same 3D character, generated with “extra-low” to “high” quality (higher face count)
Was this helpful?