Performance of FlyingEdges3D versus Contour Filter?
Hello everyone, @cory.quammen @seanm
I have a situation where I would like to replace the use of Contour Filters in my application with that of FlyingEdges3D. However, when I measured the performance I realised that the former was outperforming the latter which to me seems rather odd. Is this true in some cases that Contour Filter can be faster or is there any particular setup required for Flying Edges to be faster (maybe I'm missing something). I have attached a zip file of a test application where I process a poly data with both and measure the time it takes to execute each time. To run the app:
- change path to the result folder in main.cpp (this is optional, only required if you care to see the final result of the processing)
- change path to the 'chipped_chamber.vtk' file in main.cpp (required)
- build (make) & run ./FlyingEdge_versus-ContourFilter
the first rendering will be done with Flying Edges (see the terminal for the output) close the view for the second view to execute using Contour Filter (check time measurements in terminal)
- minimum required CMake version 3.15
Repro_App_FlyingEdge_vs_ContourFilter.zip
- The following app tends to produce the result of Contour Filter being faster.
- please, note this uses a different poly data than the previous app. I'm not sure what significance that has on the overall result.
- It's been streamlined, with no rendering, just quick execution of both algorithms sequentially. (See output in terminal)
- I have also included a screenshot of what the poly data looks like.
to execute:
- change path to the result folder in main.cpp (this is optional, only required if you care to see the final result of the processing)
- change path to the 'brain.vtk' file in main.cpp (required)
- rendering has been disabled to make execution faster, to enable this define the undefined macro 'render_output' in main.cpp
- build (make) & run ./FlyingEdge_versus-ContourFilter_brain