GitLab

[hbm] Fix crashes when importing an FE file with unexpected empty line in element/node definition -> now reports an error

[kriging] Enabled the intermediate targets in Kriging module again, options for it can be shown/hidden by a checkbox

It was working after all, only a small bug, see previous commit.

[kriging] Kriging with intermediate targets can now be done when only one of the targets (skin or bones) is defined; automatic nugget off by default

[kriging] Decimation based on relative displacement now off by default, moved it below the homogenous decimation

[element quality] Number of elements in each range now reported correctly only for element types for which the metric is defined

[kriging] Added a property to show/hide the UI for "Intermediate targets" -> hid it in Kriging module (does not work there anyway)

also small updates for the documentation of surface registration

[vtkDisplay] Fixed Z and Y clipping planes being switched; X clipping plane now the default and correctly shown in the GUI when clipping is activated for the first time

[kriging] Fixed minor bug in homogenous decimation - in some cases there would be one last voxel in Y direction missing

# Conflicts:
#	CHANGELOG.md

[meshoptimizer] Method for Mesquite optimization of only selected elements - not currently used anywhere, only testing; custom timeout limits for the optimizer

The time limit was set to 600 for the untangler, kept that, but can be changed in the future

[hbm] Several fixes for exporting double values to fixed formats - in some cases large numbers were rounded or output as more digits than fixed limit

Updated the test data to be more robust and test this

- added tests for vtkPIPERFilters, currently has only one simple test for the registration_test
- added vtkOBJWriter - copy-pasted code from VTK 8.0, once we move to new VTK version it should be replaced
- documentation for the registration + minor fixes to the existing kriging documentation

[selection] Renamed "picking" to "selection", added "export selected elements as .vtu" and "deselect bone elements", updated GUI documentation

Also added missing documentation on clipping plane and removed from documentation display options that no longer exist

Lacks documentation, test cases, not sure if all features work correctly.

Note: the largest feature of this release are the regular expressions in parser, but unit tests are missing for it. It should be added once the piper child model in radioss is released (should be coming soon), which should include plenty of examples for the regexp keywords

Merge remote-tracking branch 'origin/69-error-while-running-piper-script-for-position-by-fe-simulation' into regexp_formatrules

[element quality] Fixed a bug causing elements with quality values exactly equal to boundaries of ranges to not be included in either of the ranges.

[hbm] Some small optimization of FEModelVTK (loops over VTK objects, unnecessary vector copies in setMeshAsGenericMetadatas etc.)

[hbm] Optimized the mechanism for switching between NID and Index-based node mapping -> faster loading of models

Specifically:

- Avoided unnecessary re-creation of the ID maps - only do it if a node was added or removed from the model since last call
- Calling the ID conversion function (bottleneck) is now parallelized per entity type
- Replaced map by unordered_map - the sorting of std::map is not used in any way and unordered_map was slightly faster
- Various micro optimizations

In total, loading of the first visualization (calls the setMeshDefINDEX several times) of GHBM went from ~17s to ~8s on my PC

Mostly just extracting calls to various size() functions outside the loops and using a constant number as the termination condition instead.
Measured a significant speed-up (on MSVC) in most cases (mainly the vtk ones, should be done this way everywhere).

Moved vtkLaplacianPolyData and vtkSurfaceDistance into a shared library "vtkPIPERFilters" for our vtk + eigen filters

More similar "one-file" filters are to come, no need for a separate library for each of them

[kriging] Enfore the rule of "can't combine interpolate displacement + use drift" in the GUI as well, also set interpolate displacement as default (usually it is better)

[kriging] Fix box-box neighbourhood being set wrong in VtkSelectionTools + accordingly modified Kriging code to process the neighbourhood correctly

Boxes are now correctly set as neighbours only if they share at least a part of a face. For the purpose of neighbourhood extension, it takes boxes that are direct neighbours and also their neighbours (except for the ones that are directly opposite to the shared face) in order to get also boxes that share at least an edge (this will catch also some boxes that don't share anything at all in some cases, but will not take many points from those anyway as they are far away)
Also removed some commented / unused code that counted with number of all points as a splitting criterion for boxes
Added visual debugging code that can be enabled via a macro (a proper unit test would be better though...)

[kriging] Deformed CPs are now stored only if the deformed model is stored in history (i.e. not cancelled) and rather than name of target CPs the name of the model in history is used (as all the targets contribute to deformation of each source CP set)

[kriging] The kriging module now suggest names for control point sets equal to the name of the file from which they are loaded

# Conflicts:
#	CHANGELOG.md
#	src/application/module/kriging/KrigingModule.h
#	src/application/qml/ModuleParameter.qml
#	src/lib/kriging/IntermediateTargetsInterface.h

[kriging] Only control points are being taken into account when computing the size of overlap region instead of all points

For know it's only commented out in case the allPoints locators are useful in potential future improvements of the method