Class NeRFNetwork<T>

Namespace

AiDotNet.Diffusion.Models

Assembly

AiDotNet.dll

Neural Radiance Field network for 3D representation.

public class NeRFNetwork<T>

Type Parameters

T

The numeric type.

Inheritance

object

NeRFNetwork<T>

Inherited Members

object.Equals(object)

object.Equals(object, object)

object.GetHashCode()

object.GetType()

object.MemberwiseClone()

object.ReferenceEquals(object, object)

object.ToString()

Constructors

NeRFNetwork(int, int, int?)

Initializes a new NeRF network.

public NeRFNetwork(int hiddenDim = 64, int numLayers = 4, int? seed = null)

Parameters

hiddenDim int

Hidden layer dimension.

numLayers int

Number of hidden layers.

seed int?

Optional random seed.

Properties

ParameterCount

Gets the total parameter count.

public int ParameterCount { get; }

Property Value

int

Methods

GetParameters()

Gets all network parameters.

public Vector<T> GetParameters()

Returns

Vector<T>

Initialize()

Initializes network weights.

public void Initialize()

QueryDensity(double, double, double)

Queries the density at a 3D point.

public double QueryDensity(double x, double y, double z)

Parameters

x double

y double

z double

Returns

double

Render(CameraPose, int)

Renders an image from the NeRF at a given camera pose.

public Tensor<T> Render(CameraPose cameraPose, int resolution)

Parameters

cameraPose CameraPose

The camera pose.

resolution int

Output image resolution.

Returns

Tensor<T>

Rendered image as tensor [H, W, 3].

SetParameters(Vector<T>)

Sets network parameters.

public void SetParameters(Vector<T> parameters)

Parameters

parameters Vector<T>

UpdateParameters(Tensor<T>, CameraPose, double)

Updates network parameters using gradient.

public void UpdateParameters(Tensor<T> gradient, CameraPose cameraPose, double learningRate)

Parameters

gradient Tensor<T>

cameraPose CameraPose

learningRate double