[vtkusers] 2D visualization question(s)...

[Terry J. Ligocki]

I've been using VTK for some time for a variety of 3D visualization
tasks and at the moment I'm using the "official" 3.1.2 release.  I am
currently trying to do some fairly complicated 2D visualizations of some
2D and 3D data sets.  For this work I need to be able to do several
things:

   * Overlay many images which have various positions and sizes with
     respect to each other (representing data in different places at
     different resolutions).
   * "Draw" on top of them (to show bounding boxes, etc.) and have the
     images show through everywhere something isn't drawn.
   * Zoom and pan everything together.

I've been looking into several ways to do this using strictly 2D
portions of VTK, mixing 2D and 3D portions of VTK, and using strictly
3D portions of VTK.  This has lead me to the following questions:

   * Is there a fairly complete and detailed description of the VTK
     image classes, the semantics/meanings of their methods, and their
     interactions?  I have the VTK book and users guide but these don't
     provide nearly enough detail for what I'm trying to do.  The
     "doxygen" documentation does allow me to find out what is available
     to me in a fairly comprehensive way but it isn't intended to give
     me the semantics information I need.  For example, a "vtkActor2D"
     has a "SetLayerNumber" function which "Sets the layer number in the
     overlay planes into which to render".  Unfortunately, this doesn't
     tell me how this effects this 2D actor's interaction with other 2D
     (or 3D) actors (it didn't seem to but I'm not sure how it was
     supposed to).  I have, at least, a dozen of these at the moment.

     My goal here is NOT to complain or point out problems.  It IS to
     find something I can look at instead of peppering this mailing list
     with lots of specific misunderstandings, code fragments, and
     questions.  If I can't find anything further that I can look at
     (short of the source code, :-) I will start asking (this is not a
     threat, :-).

   * Has anyone been down any of these paths, what did they learn, and
     what did they end up doing?  I am concerned about performance
     because some of these data sets may lead to images which are large
     (i.e. > 1K x 1K).  I have verified that the 2D VTK image
     "stuff" can handle this fairly well but I'm not sure what will
     happen if I try to use the 3D VTK functionality and "map" the data
     onto a mesh of quads or triangles (generating millions of them),
     render these, and try to zoom and pan.  Also, the data can't be
     interpolated (user's requirement) across the quads.

If possible, I don't want to use much (anything?) outside the standard
VTK distribution but I won't mind hearing about other possibilities.
I'm currently using the Tcl/Tk interface to VTK and need to use this for
now (again, information about other things would be fine).  I'm using
VTK for the 3D version of this work so I want to stick with it for the
2D version so I can present the users with a similar interface and
underlying framework.

Thanks for your time and help with this.  I will send a follow up on
this when I know more, have made a decision, and seen what the pros and
cons are.  That's all for now...

        Terry J. (Ligocki, tjligocki at lbl.gov, LBNL/NERSC Vis. Group)