Skip to main content

NetField

C++: phynexis::ml::shape::NetField
Python: phynexis.ml.shape.NetField
Header: src/ml/shape/net_field.hpp

Description

Base class for neural-network field representations. Holds the neural network model, field type, and surface extraction configuration.

Constructors

NetField()

Default constructor with FieldType.SDF.

NetField(type)

Creates a field with the specified type.

Parameters:

ParameterTypeDescription
typeFieldTypeField type (SDF or Laplace)

Methods

init_from_model_files(model_file)

Initializes the field from a saved model file.

Parameters:

ParameterTypeDescription
model_filestrPath to the model file

initialize()

Initializes the field from memory.

Properties

PropertyTypeAccessDescription
net_modelNeuralNetread/writeNeural network model
field_typeFieldTyperead/writeField type
surface_configSurfaceConfigread/writeSurface extraction configuration

Nested Types

SurfaceConfig

Configuration for surface extraction.

PropertyTypeDefaultDescription
extraction_methodExtractionMethodSphericalProjectionMethod for extracting the surface
projection_tolerancefloat1e-5Projection convergence tolerance
projection_max_iterationsint100Maximum projection iterations
alphafloat0.0Alpha shape parameter
use_alpha_shapeboolFalseUse alpha shape for extraction
grid_resolutionint64Grid resolution for marching cubes
iso_valuefloat0.0Isosurface value

SurfaceConfig.ExtractionMethod

ValueDescription
SphericalProjectionSpherical projection method
MarchingCubesMarching cubes method