MultiViewUNet<T>

Models & Types · AiDotNet.Diffusion.ThreeD

Multi-view aware U-Net for MVDream.

Implements: ICloneable<MultiViewUNet<T>>, IParameterSource<T>

Constructors

Constructor Summary
MultiViewUNet(int inputChannels, int outputChannels, int baseChannels, int numViews, int contextDim, int? seed) Creates a new multi-view U-Net.

Properties

Property Summary
UNetNoisePredictor<T> BaseUNet { get; } Gets the base UNet for interface compatibility.
bool SupportsCFG { get; } Gets whether classifier-free guidance is supported.
long ParameterCount { get; } Gets parameter count.

Methods

Method Summary
MultiViewUNet<T> Clone() Deep-copies this multi-view U-Net, preserving trained weights.
MultiViewUNet<T> DeepCopy()
Tensor<T> PredictNoise(Tensor<T> input, int timestep, Tensor<T> conditioning) Predicts noise for a single view.
Tensor<T> PredictNoiseMultiView(Tensor<T> multiViewInput, int timestep, Tensor<T>[] viewConditions) Predicts noise for multiple views with cross-view attention.
Vector<T> GetParameters() Gets parameters.
void SetParameters(Vector<T> parameters) Sets parameters.