NeRFOptions

Options & Configuration ยท AiDotNet.Models.Options

Configuration options for NeRF models.

Inheritance: ModelOptions โ†’ NeRFOptions

Constructors

Constructor Summary
NeRFOptions()

Properties

Property Summary
bool UseHierarchicalSampling { get; set; } Whether to draw fine samples where the coarse pass found density (hierarchical volume sampling).
double LearningRate { get; set; } Learning rate of the optimizer.
double RenderFarBound { get; set; } Far bound of each ray, in scene units.
double RenderNearBound { get; set; } Near bound of each ray, in scene units.
int ColorHiddenDim { get; set; } Hidden width of the view-dependent color head.
int ColorNumLayers { get; set; } Number of hidden layers of the view-dependent color head.
int DirectionEncodingLevels { get; set; } Number of frequency bands of the positional encoding of the viewing direction.
int HiddenDim { get; set; } Hidden width of the density MLP.
int HierarchicalSamples { get; set; } Number of additional fine samples per ray drawn from the coarse density (N_f in Mildenhall et al.
int NumLayers { get; set; } Number of fully connected layers of the density MLP.
int PositionEncodingLevels { get; set; } Number of frequency bands of the positional encoding of the 3D position.
int RenderSamples { get; set; } Number of stratified coarse samples per ray (N_c in Mildenhall et al.