This package doesn't support loading from file, as the Fal.ai endpoint doesn't support it, for it to support that you should first upload you image to a publicly accessible host and then return the url to feed into the url and then process it. There is room for improvement and optimization, this is the Basic SDK featuring Image to 3D model inside unity.
the SDK will be updated in my free time :)
A Unity SDK for converting images to 3D models using the Fal.ai TripoSR service. Features Convert images to 3D models using URLs Asynchronous processing with status updates GLB model loading and scene integration Configurable model generation settings User-friendly UI components
Add the package to your Unity project via Package Manager
Install required dependencies:
- com.unity.cloud.gltfast (6.9.0 or higher)
- com.unity.nuget.newtonsoft-json (3.0.2 or higher)
Installation of these dependencies will be handled automatically.
Create a new GameObject in your scene Add the following UI components: text Canvas
├── Image URL Input Field
├── Process Button
├── Status Text
└── Loading Panel
Create a Model Container GameObject to hold the generated 3D models Initialize the TripoSR manager with your API key: csharp TripoSRManager.Instance.Initialize(new TripoSRConfig { ApiKey = "YOUR_FAL_AI_KEY" });
Basic Implementation
// Process an image URL
await TripoSRManager.Instance.ProcessImageUrl(
imageUrl: "https://example.com/image.jpg",
parent: modelContainer,
removeBackground: true,
foregroundRatio: 0.9f,
mcResolution: 256
);Configuration Options
removeBackground: Remove image background during processing
foregroundRatio: Scale of foreground relative to original image (0.5-1.0)
mcResolution: Marching cubes resolution for model detail (32-1024)
API Structure
TripoSRManager: Singleton manager handling initialization and processing
TripoSRApi: Handles API communication and request queuing
ImagePickerUI: UI component for image URL input and processing
GLTFast Integration: Handles 3D model loading and scene instantiation
Create a new scene with the following hierarchy:
Scene
├── Main Camera
├── Directional Light
├── Model Container
└── Canvas
├── URL Input Field
├── Process Button
├── Status Text
└── Loading Panel
Add the ImagePickerUI component to the Canvas and configure references:
[SerializeField] private InputField imageUrlInputField;
[SerializeField] private Button processButton;
[SerializeField] private Transform modelParent;
[SerializeField] private GameObject loadingIndicator;
[SerializeField] private Text statusText;The system includes comprehensive error handling through the TripoSRException class. All operations are wrapped in try-catch blocks with appropriate error messages displayed in the UI. API Response Format The system handles the following response structure:
public class TripoSRResponse
{
public ModelMesh model_mesh;
public Timings timings;
}
Import the sample scene from the Samples section in unity package manager
Recommended
- mcResolution range: 32-512 (higher values may impact performance)
- Default foregroundRatio: 0.9
API requests are queued and processed asynchronously Models are automatically positioned at the parent transform's origin
- Unity 2020.3 or higher
- .NET 4.x or higher
- Active Fal.ai API key https://fal.ai/dashboard/keys
- Internet connection for API communication