vtkImplicitModeller Class Reference

#include <vtkImplicitModeller.h>

Inheritance diagram for vtkImplicitModeller:

[legend]

Collaboration diagram for vtkImplicitModeller:

[legend]

List of all members.

Detailed Description

compute distance from input geometry on structured point dataset

vtkImplicitModeller is a filter that computes the distance from the input geometry to the points of an output structured point set. This distance function can then be "contoured" to generate new, offset surfaces from the original geometry. An important feature of this object is "capping". If capping is turned on, after the implicit model is created, the values on the boundary of the structured points dataset are set to the cap value. This is used to force closure of the resulting contoured surface. Note, however, that large cap values can generate weird surface normals in those cells adjacent to the boundary of the dataset. Using smaller cap value will reduce this effect.

Another important ivar is MaximumDistance. This controls how far into the volume the distance function is computed from the input geometry. Small values give significant increases in performance. However, there can strange sampling effects at the extreme range of the MaximumDistance.

In order to properly execute and sample the input data, a rectangular region in space must be defined (this is the ivar ModelBounds). If not explicitly defined, the model bounds will be computed. Note that to avoid boundary effects, it is possible to adjust the model bounds (i.e., using the AdjustBounds and AdjustDistance ivars) to strictly contain the sampled data.

This filter has one other unusual capability: it is possible to append data in a sequence of operations to generate a single output. This is useful when you have multiple datasets and want to create a conglomeration of all the data. However, the user must be careful to either specify the ModelBounds or specify the first item such that its bounds completely contain all other items. This is because the rectangular region of the output can not be changed after the 1st Append.

The ProcessMode ivar controls the method used within the Append function (where the actual work is done regardless if the Append function is explicitly called) to compute the implicit model. If set to work in voxel mode, each voxel is visited once. If set to cell mode, each cell is visited once. Tests have shown once per voxel to be faster when there are a lot of cells (at least a thousand?); relative performance improvement increases with addition cells. Primitives should not be stripped for best performance of the voxel mode. Also, if explicitly using the Append feature many times, the cell mode will probably be better because each voxel will be visited each Append. Append the data before input if possible when using the voxel mode. Do not switch between voxel and cell mode between execution of StartAppend and EndAppend.

Further performance improvement is now possible using the PerVoxel process mode on multi-processor machines (the mode is now multithreaded). Each thread processes a different "slab" of the output. Also, if the input is vtkPolyData, it is appropriately clipped for each thread; that is, each thread only considers the input which could affect its slab of the output.

This filter can now produce output of any type supported by vtkImageData. However to support this change, additional sqrts must be executed during the Append step. Previously, the output was initialized to the squared CapValue in StartAppend, the output was updated with squared distance values during the Append, and then the sqrt of the distances was computed in EndAppend. To support different scalar types in the output (largely to reduce memory requirements as an vtkImageShiftScale and/or vtkImageCast could have achieved the same result), we can't "afford" to save squared value in the output, because then we could only represent up to the sqrt of the scalar max for an integer type in the output; 1 (instead of 255) for an unsigned char; 11 for a char (instead of 127). Thus this change may result in a minor performance degradation. Non-float output types can be scaled to the CapValue by turning ScaleToMaximumDistance On.

See also:: vtkSampleFunction vtkContourFilter

Tests:: vtkImplicitModeller (Tests)

Definition at line 101 of file vtkImplicitModeller.h.


Public Types
typedef vtkImageAlgorithm	Superclass
Public Member Functions
virtual const char *	GetClassName ()
virtual int	IsA (const char *type)
void	PrintSelf (ostream &os, vtkIndent indent)
double	ComputeModelBounds (vtkDataSet *input=NULL)
void	StartAppend ()
void	Append (vtkDataSet *input)
void	EndAppend ()
int	ProcessRequest (vtkInformation , vtkInformationVector , vtkInformationVector )

virtual int *	GetSampleDimensions ()
virtual void	GetSampleDimensions (int data[3])
void	SetSampleDimensions (int i, int j, int k)
void	SetSampleDimensions (int dim[3])

virtual void	SetMaximumDistance (double)
virtual double	GetMaximumDistance ()

virtual void	SetModelBounds (double, double, double, double, double, double)
virtual void	SetModelBounds (double[6])
virtual double *	GetModelBounds ()
virtual void	GetModelBounds (double data[6])

virtual void	SetAdjustBounds (int)
virtual int	GetAdjustBounds ()
virtual void	AdjustBoundsOn ()
virtual void	AdjustBoundsOff ()

virtual void	SetAdjustDistance (double)
virtual double	GetAdjustDistance ()

virtual void	SetCapping (int)
virtual int	GetCapping ()
virtual void	CappingOn ()
virtual void	CappingOff ()

void	SetCapValue (double value)
virtual double	GetCapValue ()

virtual void	SetScaleToMaximumDistance (int)
virtual int	GetScaleToMaximumDistance ()
virtual void	ScaleToMaximumDistanceOn ()
virtual void	ScaleToMaximumDistanceOff ()

virtual void	SetProcessMode (int)
virtual int	GetProcessMode ()
void	SetProcessModeToPerVoxel ()
void	SetProcessModeToPerCell ()
const char *	GetProcessModeAsString (void)

virtual void	SetLocatorMaxLevel (int)
virtual int	GetLocatorMaxLevel ()

virtual void	SetNumberOfThreads (int)
virtual int	GetNumberOfThreads ()

void	SetOutputScalarType (int type)
virtual int	GetOutputScalarType ()
void	SetOutputScalarTypeToFloat ()
void	SetOutputScalarTypeToDouble ()
void	SetOutputScalarTypeToInt ()
void	SetOutputScalarTypeToUnsignedInt ()
void	SetOutputScalarTypeToLong ()
void	SetOutputScalarTypeToUnsignedLong ()
void	SetOutputScalarTypeToShort ()
void	SetOutputScalarTypeToUnsignedShort ()
void	SetOutputScalarTypeToUnsignedChar ()
void	SetOutputScalarTypeToChar ()
Static Public Member Functions
static int	IsTypeOf (const char *type)
static vtkImplicitModeller *	SafeDownCast (vtkObject *o)
static vtkImplicitModeller *	New ()
Protected Member Functions
	vtkImplicitModeller ()
	~vtkImplicitModeller ()
double	GetScalarTypeMax (int type)
virtual int	RequestInformation (vtkInformation , vtkInformationVector , vtkInformationVector )
virtual int	RequestData (vtkInformation , vtkInformationVector , vtkInformationVector )
void	StartAppend (int internal)
void	Cap (vtkDataArray *s)
virtual int	FillInputPortInformation (int, vtkInformation *)
Protected Attributes
vtkMultiThreader *	Threader
int	NumberOfThreads
int	SampleDimensions [3]
double	MaximumDistance
double	ModelBounds [6]
int	Capping
double	CapValue
int	DataAppended
int	AdjustBounds
double	AdjustDistance
int	ProcessMode
int	LocatorMaxLevel
int	OutputScalarType
int	ScaleToMaximumDistance
int	BoundsComputed
double	InternalMaxDistance

Member Typedef Documentation

typedef vtkImageAlgorithm vtkImplicitModeller::Superclass

Reimplemented from vtkImageAlgorithm.

Definition at line 104 of file vtkImplicitModeller.h.

Constructor & Destructor Documentation

vtkImplicitModeller::vtkImplicitModeller ( ) [protected]

vtkImplicitModeller::~vtkImplicitModeller ( ) [protected]

Member Function Documentation

virtual const char* vtkImplicitModeller::GetClassName ( ) [virtual]

Reimplemented from vtkImageAlgorithm.

static int vtkImplicitModeller::IsTypeOf ( const char * name ) [static]

Return 1 if this class type is the same type of (or a subclass of) the named class. Returns 0 otherwise. This method works in combination with vtkTypeRevisionMacro found in vtkSetGet.h.

Reimplemented from vtkImageAlgorithm.

virtual int vtkImplicitModeller::IsA ( const char * name ) [virtual]

Return 1 if this class is the same type of (or a subclass of) the named class. Returns 0 otherwise. This method works in combination with vtkTypeRevisionMacro found in vtkSetGet.h.

Reimplemented from vtkImageAlgorithm.

static vtkImplicitModeller* vtkImplicitModeller::SafeDownCast ( vtkObject * o ) [static]

Reimplemented from vtkImageAlgorithm.

void vtkImplicitModeller::PrintSelf	(	ostream &	os,
		vtkIndent	indent
	)			`[virtual]`

Methods invoked by print to print information about the object including superclasses. Typically not called by the user (use Print() instead) but used in the hierarchical print process to combine the output of several classes.

Reimplemented from vtkImageAlgorithm.

static vtkImplicitModeller* vtkImplicitModeller::New ( ) [static]

Construct with sample dimensions=(50,50,50), and so that model bounds are automatically computed from the input. Capping is turned on with CapValue equal to a large positive number.

Reimplemented from vtkAlgorithm.

double vtkImplicitModeller::ComputeModelBounds ( vtkDataSet * input = NULL )

Compute ModelBounds from input geometry. If input is not specified, the input of the filter will be used.

virtual int* vtkImplicitModeller::GetSampleDimensions ( ) [virtual]

Set/Get the i-j-k dimensions on which to sample distance function.

virtual void vtkImplicitModeller::GetSampleDimensions ( int data[3] ) [virtual]

Set/Get the i-j-k dimensions on which to sample distance function.

void vtkImplicitModeller::SetSampleDimensions	(	int	i,
		int	j,
		int	k
	)

Set/Get the i-j-k dimensions on which to sample distance function.

void vtkImplicitModeller::SetSampleDimensions ( int dim[3] )

Set/Get the i-j-k dimensions on which to sample distance function.

virtual void vtkImplicitModeller::SetMaximumDistance ( double ) [virtual]

Set / get the distance away from surface of input geometry to sample. Smaller values make large increases in performance.

virtual double vtkImplicitModeller::GetMaximumDistance ( ) [virtual]

Set / get the distance away from surface of input geometry to sample. Smaller values make large increases in performance.

virtual void vtkImplicitModeller::SetModelBounds	(	double	,
		double	,
		double	,
		double	,
		double	,
		double
	)			`[virtual]`

Set / get the region in space in which to perform the sampling. If not specified, it will be computed automatically.

virtual void vtkImplicitModeller::SetModelBounds ( double [6] ) [virtual]

Set / get the region in space in which to perform the sampling. If not specified, it will be computed automatically.

virtual double* vtkImplicitModeller::GetModelBounds ( ) [virtual]

Set / get the region in space in which to perform the sampling. If not specified, it will be computed automatically.

virtual void vtkImplicitModeller::GetModelBounds ( double data[6] ) [virtual]

Set / get the region in space in which to perform the sampling. If not specified, it will be computed automatically.

virtual void vtkImplicitModeller::SetAdjustBounds ( int ) [virtual]

Control how the model bounds are computed. If the ivar AdjustBounds is set, then the bounds specified (or computed automatically) is modified by the fraction given by AdjustDistance. This means that the model bounds is expanded in each of the x-y-z directions.

virtual int vtkImplicitModeller::GetAdjustBounds ( ) [virtual]

Control how the model bounds are computed. If the ivar AdjustBounds is set, then the bounds specified (or computed automatically) is modified by the fraction given by AdjustDistance. This means that the model bounds is expanded in each of the x-y-z directions.

virtual void vtkImplicitModeller::AdjustBoundsOn ( ) [virtual]

Control how the model bounds are computed. If the ivar AdjustBounds is set, then the bounds specified (or computed automatically) is modified by the fraction given by AdjustDistance. This means that the model bounds is expanded in each of the x-y-z directions.

virtual void vtkImplicitModeller::AdjustBoundsOff ( ) [virtual]

Control how the model bounds are computed. If the ivar AdjustBounds is set, then the bounds specified (or computed automatically) is modified by the fraction given by AdjustDistance. This means that the model bounds is expanded in each of the x-y-z directions.

virtual void vtkImplicitModeller::SetAdjustDistance ( double ) [virtual]

Specify the amount to grow the model bounds (if the ivar AdjustBounds is set). The value is a fraction of the maximum length of the sides of the box specified by the model bounds.

virtual double vtkImplicitModeller::GetAdjustDistance ( ) [virtual]

Specify the amount to grow the model bounds (if the ivar AdjustBounds is set). The value is a fraction of the maximum length of the sides of the box specified by the model bounds.

virtual void vtkImplicitModeller::SetCapping ( int ) [virtual]

The outer boundary of the structured point set can be assigned a particular value. This can be used to close or "cap" all surfaces.

virtual int vtkImplicitModeller::GetCapping ( ) [virtual]

The outer boundary of the structured point set can be assigned a particular value. This can be used to close or "cap" all surfaces.

virtual void vtkImplicitModeller::CappingOn ( ) [virtual]

The outer boundary of the structured point set can be assigned a particular value. This can be used to close or "cap" all surfaces.

virtual void vtkImplicitModeller::CappingOff ( ) [virtual]

The outer boundary of the structured point set can be assigned a particular value. This can be used to close or "cap" all surfaces.

void vtkImplicitModeller::SetCapValue ( double value )

Specify the capping value to use. The CapValue is also used as an initial distance value at each point in the dataset.

virtual double vtkImplicitModeller::GetCapValue ( ) [virtual]

Specify the capping value to use. The CapValue is also used as an initial distance value at each point in the dataset.

virtual void vtkImplicitModeller::SetScaleToMaximumDistance ( int ) [virtual]

If a non-floating output type is specified, the output distances can be scaled to use the entire positive scalar range of the output type specified (up to the CapValue which is equal to the max for the type unless modified by the user). For example, if ScaleToMaximumDistance is On and the OutputScalarType is UnsignedChar the distances saved in the output would be linearly scaled between 0 (for distances "very close" to the surface) and 255 (at the specifed maximum distance)... assuming the CapValue is not changed from 255.

virtual int vtkImplicitModeller::GetScaleToMaximumDistance ( ) [virtual]

If a non-floating output type is specified, the output distances can be scaled to use the entire positive scalar range of the output type specified (up to the CapValue which is equal to the max for the type unless modified by the user). For example, if ScaleToMaximumDistance is On and the OutputScalarType is UnsignedChar the distances saved in the output would be linearly scaled between 0 (for distances "very close" to the surface) and 255 (at the specifed maximum distance)... assuming the CapValue is not changed from 255.

virtual void vtkImplicitModeller::ScaleToMaximumDistanceOn ( ) [virtual]

If a non-floating output type is specified, the output distances can be scaled to use the entire positive scalar range of the output type specified (up to the CapValue which is equal to the max for the type unless modified by the user). For example, if ScaleToMaximumDistance is On and the OutputScalarType is UnsignedChar the distances saved in the output would be linearly scaled between 0 (for distances "very close" to the surface) and 255 (at the specifed maximum distance)... assuming the CapValue is not changed from 255.

virtual void vtkImplicitModeller::ScaleToMaximumDistanceOff ( ) [virtual]

If a non-floating output type is specified, the output distances can be scaled to use the entire positive scalar range of the output type specified (up to the CapValue which is equal to the max for the type unless modified by the user). For example, if ScaleToMaximumDistance is On and the OutputScalarType is UnsignedChar the distances saved in the output would be linearly scaled between 0 (for distances "very close" to the surface) and 255 (at the specifed maximum distance)... assuming the CapValue is not changed from 255.

virtual void vtkImplicitModeller::SetProcessMode ( int ) [virtual]

Specify whether to visit each cell once per append or each voxel once per append. Some tests have shown once per voxel to be faster when there are a lot of cells (at least a thousand?); relative performance improvement increases with addition cells. Primitives should not be stripped for best performance of the voxel mode.

virtual int vtkImplicitModeller::GetProcessMode ( ) [virtual]

Specify whether to visit each cell once per append or each voxel once per append. Some tests have shown once per voxel to be faster when there are a lot of cells (at least a thousand?); relative performance improvement increases with addition cells. Primitives should not be stripped for best performance of the voxel mode.

void vtkImplicitModeller::SetProcessModeToPerVoxel ( ) [inline]

Specify whether to visit each cell once per append or each voxel once per append. Some tests have shown once per voxel to be faster when there are a lot of cells (at least a thousand?); relative performance improvement increases with addition cells. Primitives should not be stripped for best performance of the voxel mode.

Definition at line 192 of file vtkImplicitModeller.h.

void vtkImplicitModeller::SetProcessModeToPerCell ( ) [inline]

Specify whether to visit each cell once per append or each voxel once per append. Some tests have shown once per voxel to be faster when there are a lot of cells (at least a thousand?); relative performance improvement increases with addition cells. Primitives should not be stripped for best performance of the voxel mode.

Definition at line 193 of file vtkImplicitModeller.h.

const char* vtkImplicitModeller::GetProcessModeAsString ( void )

Specify whether to visit each cell once per append or each voxel once per append. Some tests have shown once per voxel to be faster when there are a lot of cells (at least a thousand?); relative performance improvement increases with addition cells. Primitives should not be stripped for best performance of the voxel mode.

virtual void vtkImplicitModeller::SetLocatorMaxLevel ( int ) [virtual]

Specify the level of the locator to use when using the per voxel process mode.

virtual int vtkImplicitModeller::GetLocatorMaxLevel ( ) [virtual]

Specify the level of the locator to use when using the per voxel process mode.

virtual void vtkImplicitModeller::SetNumberOfThreads ( int ) [virtual]

Set / Get the number of threads used during Per-Voxel processing mode

virtual int vtkImplicitModeller::GetNumberOfThreads ( ) [virtual]

Set / Get the number of threads used during Per-Voxel processing mode

void vtkImplicitModeller::SetOutputScalarType ( int type )