[Paraview] Bug in volume rendering of unstructured grids on server
cquammen at cs.unc.edu
Fri Jan 13 17:09:55 EST 2012
My problems with volume rendering of unstructured grids appear to have
been resolved by running
git checkout v3.12.0
git submodule init
to ensure all the submodules are brought to the right version when
trying to compile ParaView at the v3.12.0 tag. I assume there was some
mismatch between the version of VTK ParaView needed and what was
checked out when I didn't update the submodules.
On Thu, Jan 12, 2012 at 10:10 AM, Cory Quammen <cquammen at cs.unc.edu> wrote:
> Forgive me, I should have specified how I was running the server. Here
> is the command I used that resulted in the strange behavior:
> mpirun -np 4 pvserver -display :0 --server-port=11111 --use-offscreen-rendering
> It seems to work fine for all rendering aside from volume rendering of
> unstructured grids.
> On Thu, Jan 12, 2012 at 2:04 AM, Utkarsh Ayachit
> <utkarsh.ayachit at kitware.com> wrote:
>> Can you try running pvserver as follows:
>> mpiexec <....> pvserver --use-offscreen-rendering
>> Does that help?
>> On Tue, Jan 10, 2012 at 5:52 PM, Cory Quammen <cquammen at cs.unc.edu> wrote:
>>> Dear all,
>>> I am attempted to do client/server rendering of an unstructured grid
>>> using ParaView 3.12 that I have compiled myself on a RedHat system
>>> with two NVIDIA Quadro FX 5000 graphics cards. I have built ParaView
>>> from the v3.12.0 tag. To summarize, the problem is that only a subset
>>> of the unstructured grid data is rendered.
>>> Here is what I have done so far to attempt to identify the problem.
>>> 1). Start the server on 4 cores. Start the client and connect it to the server.
>>> 2). Add a Wavelet source with the default parameters. I have the Edit
>>> -> Settings -> Render View -> Server -> Remote Render Threshold set to
>>> 0 MBytes to force server side rendering, so 4 windows titled
>>> "Visualization Tookit - OpenGL" appear right away. They are stacked
>>> neatly on top of each other and the contents of the windows are all
>>> 3). Change the Representation to Volume. Everything renders in
>>> ParaView as expected. The content of the rendering windows from the
>>> server stay black.
>>> 4). Add a Tetrahedralize filter connected to the Wavelet source.
>>> 5). Change the Representation to Volume. Now the problems begin.
>>> First, the content of the rendering windows is no longer black. It is
>>> the same as the content in the ParaView display except that the
>>> background color in the server window is black instead of the gray
>>> background I have in the ParaView display. Also, only a subset of the
>>> data is volume rendered. If I move the top window in the stack of
>>> windows from the server, I see more data in ParaView, and in the
>>> window that has been revealed is a different subset of the data set.
>>> If I rearrange the windows so that they are all on top of all other
>>> windows, I can see the whole data set volume rendered. However, if any
>>> of the windows are covered up or go offscreen, then the portion of the
>>> data set that is covered or off the screen does not appear in
>>> 6). If I change the Representation back to Surface, the view in
>>> ParaView now suffers the same way as when the Representation was
>>> Volume. The same is true for Outline, Points, Surface with Edges, and
>>> Wireframe representations.
>>> Clearly, the volume representation for unstructured grid data sets is
>>> causing a change in the way rendering is carried out on the server. If
>>> I can guarantee that the windows from the server are uncovered, then
>>> volume rendering of unstructured grids seems to work. However, this
>>> seems unnecessary considering how the other representations don't
>>> appear to have this requirement.
>>> Can anyone shed some light on what might be going on here?
>>> By the way, I have confirmed that I have the same problem on a
>>> different system with Fedora 14 installed.
>>> Cory Quammen
>>> Research Associate
>>> Department of Computer Science
>>> The University of North Carolina at Chapel Hill
>>> Powered by www.kitware.com
>>> Visit other Kitware open-source projects at http://www.kitware.com/opensource/opensource.html
>>> Please keep messages on-topic and check the ParaView Wiki at: http://paraview.org/Wiki/ParaView
>>> Follow this link to subscribe/unsubscribe:
> Cory Quammen
> Research Associate
> Department of Computer Science
> The University of North Carolina at Chapel Hill
Department of Computer Science
The University of North Carolina at Chapel Hill
More information about the ParaView