[vtkusers] How to visualize thousands of actors with good performance

[David Doria]

On Mon, Sep 27, 2010 at 9:28 AM, Affolter Martin <martin.affolter at ntb.ch>wrote:

> Jim,
>
> thanks for your reply. You're right, we're using one pipeline for each
> element, which obviously results in a large number of pipelines. With this
> approach we are able to select and highlight every actor (mesh, vertex,
> edge, ...) individually. With VTK 4.4 we never really had any performance
> issues so it was not necessary to change the way we visualize things.
>
> Used actor-interaction:
> - Selection requests, that for each shown graphic-element the corresponding
> geometry-element can be identified.
> - Highlighting requests, that for each geometry-element the corresponding
> graphic-element can be identified and altered (color, transparency, etc).
> - Identification must be fast
>
> And here is some additional information:
> - The initial polydata is never manipulated. If the geometry-data changes,
> the old polydata is erased and new pipelines are created.
> - If the geometry-data is being transformed (identity-transformation only),
> the corresponding actors are also transformed (usermatrix)
>
> We could consider combining polydata to use fewer actors as you suggested.
> However, this change would generate a lot of work in our system. I was
> hoping that we could restrict the pipeline execution (according to our
> needs) and thus make it run faster. Do you know how to do that? What classes
> do you use for the combination of polydata?
>
> Thanks
>
> Martin Affolter
>
>
To combine PolyData objects you can use vtkAppendPolyData:
http://www.vtk.org/Wiki/VTK/Examples/Cxx/Filters/CombinePolydata
<http://www.vtk.org/Wiki/VTK/Examples/Cxx/Filters/CombinePolydata>
David
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.vtk.org/pipermail/vtkusers/attachments/20100927/d28ac106/attachment.htm>