vtkProgrammableAttributeDataFilter - manipulate attribute (cell and point) data via a user-specified function

Super Class: vtkDataSetToDataSetFilter

Description:

vtkProgrammableAttributeDataFilter is a filter that allows you to write a custom procedure to manipulate attribute data - either point or cell data. For example, you could generate scalars based on a complex formula; convert vectors to normals; compute scalar values as a function of vectors, texture coords, and/or any other point data attribute; and so on. The filter takes multiple inputs (input plus an auxiliary input list), so you can write procedures that combine several datset point attributes. Note that the output of the filter is the same type (topology/geometry) as the input. The filter works as follows. It operates like any other filter (i.e., checking and managing modified and execution times, processing Update() and Execute() methods, managing release of data, etc.), but the difference is that the Execute() method simply invokes a user-specified function with an optional (void *) argument (typically the "this" pointer in C++). It is also possible to specify a function to delete the argument via ExecuteMethodArgDelete(). To use the filter, you write a procedure to process the input datasets, process the data, and generate output data. Typically, this means grabbing the input point or cell data (using GetInput() and maybe GetInputList()), operating on it (creating new point and cell attributes such as scalars, vectors, etc.), and then setting the point and/or cell attributes in the output dataset (you'll need to use GetOutput() to access the output). (Note: besides C++, it is possible to do the same thing in Tcl, Java, or other languages that wrap the C++ core.) Remember, proper filter protocol requires that you don't modify the input data - you create new output data from the input.

Caveats:

This filter operates on any combination of the filter input plus a list of additional inputs (at a minimum you must set the filter input via SetInput()). It is up to you check whether the input is valid, and to insure that the output is valid. Also, you have to write the control structure for the traversal and operation on the point and cell attribute data. By default the output point and cell data will be copied through from the input point data (using reference counting). You can control this using the output's CopyAllOff() flag, or by using individual flags for each point data field (i.e., scalars, vectors, etc.) The output of this filter is the abstract type vtkDataSet, even if your input is a concrete type like vtkPolyData. Thus you may need to use vtkCastToConcrete to obtain the output as a particular concrete type, or one of the special methods of the superclass (e.g., vtkDataSetToDataSetFilter::GetPolyDataOutput) to retrieve output of the correct type. The filter correctly manages modified time and network execution in most cases. However, if you change the definition of the filter function, you'll want to send a manual Modified() method to the filter to force it to reexecute.

Methods:

Detailed Method Descriptions: