Slicer3 FlexibleLayouts
From NAMIC Wiki
Home < Slicer3 FlexibleLayouts
Key Investigators
- GE: Jim Miller
- BWH/Isomics: Steve Pieper
- BWH: Wendy Plesniak
- Kitware: Sebastien Barre (the multi-camera / multi-3D view code in the "cameraviewhell" branch has been ported to 3.4 and is being tested).
Objective
- Continue implementing infrastructure and interface for generating custom layouts in Slicer3
Approach, Plan
- Representation and Logic are currently being implemented; finish this implementation and begin testing.
- Add functionality to read/write XML specification
- Integrate with exiting vtkMRMLLayoutNode
- Begin work on building GUI to support UXP design
Progress
References
NOTES:
- Sebastien has been allowing creating multiple vtkMRMLViewerWidget or vtkMRMLCameraNode in the scenes
- previously they were singletons
- hard to patch underlying logic to take singleton status out.
- sebastien made new branch off of 3.4 with his changes
- found some bugs in trunk also (e.g. if close scene 3 or 4 times, can't exit slicer)
- He is testing now with multiple cameras and views
- Able to save, load, fiducials work ok
- will give Ron his code to test soon
- will patch 3.4 or put in 3.5 trunk after testing/fixing.
- In current implementation, new 3D views will populate tabbed 3D views...
- When populating custom layout, will have to count number of vtkMRMLViewNodes and assign to panes
- Representation for view: vtkMRMLViewNode: state for a 3D view vtkSlicerViewerWidget
- Rep for camera: vtkMRMLCameraNode: vtkMRMLViewNodes have vtkMRMLCameraNodes; ViewNodes can swap CameraNodes.
- Seb mentions problem with a render being called in a viewer that has (temporarily) no camera, due to a swap -- vtk will create a new camera that has no MRML counterpart. This is addressed as described below.
- A viewNode can use the camera from another viewNode (but in this case, a camera swap is implemented under the hood so that no viewer is without a camera.) --Can we generate an event when this swap happens in case we want to notify a user?
- Many cameras can be created and a single ViewNode can switch to use any of them.
- Single scene contains all datasets; different datasets can be exposed/hidden in each view.
- How will this work in use: how to configure a node's display parameters for each defined view on the scene?
- Node's display params can consist of kwd/val pairs where kwd is the viewerID, value is the param for that viewer.
- How to present this through the GUI in an intuitive way? Will think about this.
- How about transforming camera with xform vtkCamera API's ApplyTransform() method. Apply this in the Render Method for the viewer widget?
- Discussing difficulty of transforming camera programmatically and combining interactively applied user transforms simultaneously.
- How to encapsulate the presentation mechanism for content in any "pane" in a layout.
- IGT toplevel has had to overload too many methods -- so would be good to have a better constructed superclass for vtkSlicerViewerWidget. (Alex & Andreas) will restructure this architecture a little bit to clean it up.
- Goal: vtkKWWidget <- vtkSlicerLayoutableWidget <- vtkSlicerWidget <- vtkSlicerSliceViewerWidget
- TableViewers, plot viewers would all be subclassed from vtkSlicerLayoutableWidget.
- Layout logic/mrml underway and will be checked in soon.; Layout GUI will take a bit more time.
- Each vertex in layout tree should point to a vtkSlicerLayoutableWidgetNode