I3DDiffusionModel<T>

Interfaces · AiDotNet.Interfaces

Interface for 3D diffusion models that generate 3D content like point clouds, meshes, and scenes.

Implements: IDiffusionModel<T>

Known implementations: ThreeDDiffusionModelBase<T>

For Beginners

3D diffusion models create 3D objects instead of flat images.

Types of 3D generation:

  • Point Clouds: Sets of 3D points that form a shape
  • Meshes: Surfaces made of triangles (like in games)
  • Textured Models: Meshes with colors and materials
  • Novel Views: New angles of an object from one image

How it works:

  1. Text-to-3D: Describe what you want → 3D model
  2. Image-to-3D: Single image → full 3D model
  3. Score Distillation: Use 2D diffusion to guide 3D optimization

Applications:

  • Game asset creation
  • Product design visualization
  • VR/AR content generation
  • Architectural modeling

How It Works

3D diffusion models extend diffusion to generate three-dimensional content, including point clouds, meshes, textured models, and full 3D scenes. They are used in computer graphics, game development, and virtual reality.

This interface extends IDiffusionModel with 3D-specific operations.

Properties

Property Summary
bool SupportsMesh { get; } Gets whether this model supports mesh generation.
bool SupportsNovelView { get; } Gets whether this model supports novel view synthesis.
bool SupportsPointCloud { get; } Gets whether this model supports point cloud generation.
bool SupportsScoreDistillation { get; } Gets whether this model supports score distillation sampling (SDS).
bool SupportsTexture { get; } Gets whether this model supports texture generation.
int DefaultPointCount { get; } Gets the default number of points in generated point clouds.

Methods

Method Summary
Mesh3D<T> GenerateMesh(string prompt, string negativePrompt, int resolution, int numInferenceSteps, double guidanceScale, int? seed) Generates a mesh from a text description.
Mesh3D<T> ImageTo3D(Tensor<T> inputImage, int numViews, int numInferenceSteps, double guidanceScale, int? seed) Generates a 3D model from a single input image.
Mesh3D<T> PointCloudToMesh(Tensor<T> pointCloud, SurfaceReconstructionMethod method) Converts a point cloud to a mesh.
Tensor<T> ColorizePointCloud(Tensor<T> pointCloud, string prompt, int numInferenceSteps, int? seed) Adds colors/normals to a point cloud.
Tensor<T> ComputeScoreDistillationGradients(Tensor<T> renderedViews, string prompt, int timestep, double guidanceScale) Computes score distillation gradients for 3D optimization.
Tensor<T> GeneratePointCloud(string prompt, string negativePrompt, int? numPoints, int numInferenceSteps, double guidanceScale, int? seed) Generates a point cloud from a text description.
Tensor<T>[] SynthesizeNovelViews(Tensor<T> inputImage, ValueTuple<double, double>[] targetAngles, int numInferenceSteps, double guidanceScale, int? seed) Synthesizes novel views of an object from a reference image.