Bounds issue in vtkSplatterPlotFactory

[Michael Reuter]

The vtkSplatterPlotFactory seems to be having issues when generating the bounds for the resulting VTK dataset. The issue seems to center around that the number of actual points used is always less than the number of points requested. However, the resulting data always says it has the number of points requested. I think this is making the calculation of the bounds unstable. We should try to fill the number of requested points completely.

[Michael Reuter]

Picture showing bounds of data from SplatterPlot view generation.

[Michael Reuter]

If filling out the number of requested points requires too many trip through the data, since the number of actual points can be calculated ahead of time we can just use that for the number of requested points. If that winds up being larger than the number of requested points, then just use the original request.

[Dennis Mikkelson]

It should work to pass through the boxes one time, recording the information about the points we will actually use in vectors, then make correctly sized vtk data structures and fill them out with the data we saved.

[Dennis Mikkelson]

RE 7278_Bounds_issue_in_vtkSplatterPlotFactory

The points, signals and number of points per vtk cell is now recorded in vectors while going through the list of boxes above the specified threshold. With that information, vtk data structures are initialized to hold exactly the number of points and cells that will be used. The points, signals and number of point per cell info is then copied into the right-sized vtk data structures. This fixes the problem with calculating proper bounds for the splatter plot data and consequently fixes the problem with data disappearing from the splatter plot view when pick mode was created. NOTE: The original splatter plot code also created vtk data structures for the signals and vtk cells that were too large, but since picking was not done, the problem was not discovered. That original problem was carried forward in the code when the strategy for including points was revised to produce a useful view.

refs #7278

Changeset: 5ce0dd13e51b310e8805d49e1945e51688aa23b4

[Dennis Mikkelson]

RE 7278_Bounds_issue_in_vtkSplatterPlotFactory

Fix doxygen error

refs #7278

Changeset: e2c67ae2d5b5d62358cf7498ce7984ec960034f3

[Dennis Mikkelson]

NOTES TO TESTER:

The branch name is: feature/7278_Bounds_issue_in_vtkSplatterPlotFactory

1. Start MantidEV (Intefaces->SCD Event Data Reduction)
2. On the Select Data tab, choose an event file, such as systemtests/Data/TOPAZ_3132_event.nxs. Set Max |Q| to Map to MD to 25 and select Apply Lorentz Correction. Press Apply.
3. On the Find Peaks tab, choose "Find Peaks", set Estimated Max of a,b,c to 15, set Number of Peaks to Find to 500, set Min Intensity to 100. Press Apply.
4. On the Find UB tab, choose "Find UB Using FFT", set Lower Bound to 3, Upper Bound to 15, Tolerance to 0.12. Press Apply.
5. Select the Point Info tab.
6. In MantidPlot right click on the MD workspace and bring up the Vates Simple Interface.
7. Change to the splatter plot mode.
8. Choose Pick Mode. The data should remain in view!
9. Point at a peak using the mouse and press "p". The data about the peak will appear on the Point Info tab.

[Michael Reuter]

This is now working great! Thanks Dennis.

[Stuart Campbell]

This ticket has been transferred to github issue ​8124