TilePreprocessor<T>
Models & Types · AiDotNet.Diffusion.Preprocessing
Tile preprocessor for ControlNet conditioning.
Inheritance: DiffusionPreprocessorBase<T> → TilePreprocessor<T>
For Beginners
This creates a blurry version of your image that ControlNet uses to keep the same colors and general layout, while letting the AI add sharp details. It’s commonly used for upscaling and detail enhancement.
How It Works
Produces a blurred/downsampled version of the input image for tile-based ControlNet conditioning. This guides the model to preserve overall color and composition while allowing fine detail regeneration.
Constructors
| Constructor | Summary |
|---|---|
TilePreprocessor(int blurRadius) |
Initializes a new tile preprocessor. |
Properties
| Property | Summary |
|---|---|
ControlType OutputControlType { get; } |
|
int OutputChannels { get; } |
Methods
| Method | Summary |
|---|---|
Tensor<T> Transform(Tensor<T> data) |