[Paraview] Visualizing complex vibration modes

Wed Jan 6 16:48:09 EST 2016

I can't think of a reasonable way to do it with just paraview.simple. paraview.simple really just manages the pipeline by creating readers, filters, representations, and writers. It would have to construct a filter or some chain of filters that computed displacements based on time, and I can't think of any set of built in filters to do that. (The Salome distribution seems to have some added filters, so there may be something there I don't know about.)

I guess technically you could use paraview.simple to pull data out of the pipeline and manipulate it directly. That's not a very good way of doing it, and you will be working directly with a VTK object just like in the programable filter, so it's not really any better anyway.

-Ken

From: Nima Maftoon <n_maftoon at yahoo.com<mailto:n_maftoon at yahoo.com>>
Reply-To: Nima Maftoon <n_maftoon at yahoo.com<mailto:n_maftoon at yahoo.com>>
Date: Wednesday, January 6, 2016 at 2:25 PM
To: Kenneth Moreland <kmorel at sandia.gov<mailto:kmorel at sandia.gov>>, "paraview at paraview.org<mailto:paraview at paraview.org>" <paraview at paraview.org<mailto:paraview at paraview.org>>
Subject: [EXTERNAL] Re: [Paraview] Visualizing complex vibration modes

Thanks Ken!

Is there any Paraview solution rather than a VTK solution? Like using paraview.simple?
Nima

________________________________
From: "Moreland, Kenneth" <kmorel at sandia.gov<mailto:kmorel at sandia.gov>>
To: Nima Maftoon <n_maftoon at yahoo.com<mailto:n_maftoon at yahoo.com>>; "paraview at paraview.org<mailto:paraview at paraview.org>" <paraview at paraview.org<mailto:paraview at paraview.org>>
Sent: Tuesday, January 5, 2016 11:05 AM
Subject: Re: [Paraview] Visualizing complex vibration modes

You have to jump through a couple of hoops to get it, but this expression will return it:

self.GetInputInformation().Get(vtk.vtkStreamingDemandDrivenPipeline.UPDATE_TIME_STEP())

-Ken

From: Nima Maftoon <n_maftoon at yahoo.com<mailto:n_maftoon at yahoo.com>>
Reply-To: Nima Maftoon <n_maftoon at yahoo.com<mailto:n_maftoon at yahoo.com>>
Date: Tuesday, January 5, 2016 at 6:31 AM
To: Kenneth Moreland <kmorel at sandia.gov<mailto:kmorel at sandia.gov>>, "paraview at paraview.org<mailto:paraview at paraview.org>" <paraview at paraview.org<mailto:paraview at paraview.org>>
Subject: [EXTERNAL] Re: [Paraview] Visualizing complex vibration modes

Hi Ken,

Thanks for your detailed explanation. Any hints about how to get access to the animation time?

Nima

________________________________
From: "Moreland, Kenneth" <kmorel at sandia.gov<mailto:kmorel at sandia.gov>>
To: Nima Maftoon <n_maftoon at yahoo.com<mailto:n_maftoon at yahoo.com>>; "paraview at paraview.org<mailto:paraview at paraview.org>" <paraview at paraview.org<mailto:paraview at paraview.org>>
Sent: Monday, January 4, 2016 1:49 PM
Subject: Re: [Paraview] Visualizing complex vibration modes

I suggest that your filter sets the TIME_RANGE anyway. It is good practice for a filter that responds to time updates to advertise as such by providing either TIME_STEPS or TIME_RANGE. I can't remember all the nitty-gritty details of how the streaming demand driven pipeline decides when filters need to be updated, but if you don't set a TIME_RANGE in RequestInformation then I fear there is a chance that ParaView/VTK will not update your filter when the time step changes.

All the TIME_RANGE does is specify a continuous range of valid time values. It does not actually specify and time steps. That is appropriate for your filter because the displacement is computed on the fly. I would set the TIME_RANGE to be one period of the vibration.

Setting a TIME_RANGE is unlikely to hurt anything. If no reader or filter has provided time steps, then ParaView might adjust the animation start and stop time to include the time range provided. If the reader has already provided time steps, then ParaView by default sets itself to "Snap to TimeSteps" mode, and it will just always use the time steps. Regardless, the animation engine controls the time range and time steps within ParaView. The TIME_RANGE you provide is just a suggestion. If the animation engine defines a time range outside of the TIME_RANGE you given, your filter will still be given those "out of bounds" time steps, which is no problem for your filter since the motion is periodic.

-Ken

From: Nima Maftoon <n_maftoon at yahoo.com<mailto:n_maftoon at yahoo.com>>
Reply-To: Nima Maftoon <n_maftoon at yahoo.com<mailto:n_maftoon at yahoo.com>>
Date: Monday, January 4, 2016 at 11:14 AM
To: Kenneth Moreland <kmorel at sandia.gov<mailto:kmorel at sandia.gov>>, "paraview at paraview.org<mailto:paraview at paraview.org>" <paraview at paraview.org<mailto:paraview at paraview.org>>
Subject: [EXTERNAL] Re: [Paraview] Visualizing complex vibration modes

Hello Ken,

Thanks a lot for your response. There are some good ideas in your message for me to work on. I didn't know about SLAC at all. I see what you mean about the reader but I prefer to follow your suggestions the python way for now because I think it is less of a headache.

Let me transfer your suggestions for the reader to ones to be implemented in a python programmable filter. I guess instead of advertising the TIME_RANGE I would use whatever time range and time steps that the animation engine sets. Those can be controlled with number of frames, etc. I think time sweep should be done by the animation engine as well. Now I am again stuck with my original problem of accessing the time information to drive the sin function.
Thanks again

Nima

________________________________
From: "Moreland, Kenneth" <kmorel at sandia.gov<mailto:kmorel at sandia.gov>>
To: Nima Maftoon <n_maftoon at yahoo.com<mailto:n_maftoon at yahoo.com>>; "paraview at paraview.org<mailto:paraview at paraview.org>" <paraview at paraview.org<mailto:paraview at paraview.org>>
Sent: Monday, January 4, 2016 11:41 AM
Subject: Re: [Paraview] Visualizing complex vibration modes

Nima,

I think your question is very open ended and there are multiple ways to implement what you want.

If it were me and I had enough control over the development of the ParaView components, I would add the capability of animating mode shapes in the reader. I don't see a MED reader in my binary distribution of ParaView, so I am assuming that the version of ParaView that comes with Salome as a custom plugin that provides this MED reader (as well as other facilities). Perhaps you can enhance the MED reader to understand mode shapes with different phases and animate them.

The way this would work is that the MED reader would advertise that its data is available over a certain time range by sending a TIME_RANGE key down the pipeline during the RequestInformation phase. This is enough to tell ParaView to set up an animation for this time range. The next time RequestData is called on your reader, an UPDATE_TIME_STEP key will be set with the current time value for the animation. The reader can then compute the displacement vector with the "cos(wt) + b cos(wt + pi/2)" using the given time value for t. I would have the reader also internally apply the displacement vector to the point coordinates. This implementation is a bit of a headache for the developer, but it creates a nice seamless interface for the user. Load the file and it just works. No scripts or extra processing necessary.

I know of two readers in ParaView that support mode shapes: Exodus II and SLAC. Of these two, the SLAC reader is closer to your use case because it also supports mode shapes defined with complex values and different phases. The source code for the SLAC reader is in ParaView/VTK/IO/NetCDF/vtkSLACReader.cxx, so you can consult the solution there.

If you are not in a position to make modifications to the MED reader, the next best solution would be to create a filter that animates the mode shape for you. The filter would work basically the same as I described above for the reader except that the data comes from the filter input rather than a file. Such a filter can be implemented with the programmable filter and then encapsulated in a plugin without having to compile any custom C++ code (http://www.kitware.com/blog/home/post/534).

Hope that helps.

-Ken

From: ParaView <paraview-bounces at paraview.org<mailto:paraview-bounces at paraview.org>> on behalf of "paraview at paraview.org<mailto:paraview at paraview.org>" <paraview at paraview.org<mailto:paraview at paraview.org>>
Reply-To: Nima Maftoon <n_maftoon at yahoo.com<mailto:n_maftoon at yahoo.com>>
Date: Monday, January 4, 2016 at 6:41 AM
To: "paraview at paraview.org<mailto:paraview at paraview.org>" <paraview at paraview.org<mailto:paraview at paraview.org>>
Subject: [EXTERNAL] Re: [Paraview] Visualizing complex vibration modes

Hello,

I saw that the two links about the history has been attached together making a broken link. I corrected them below:
http://www.salome-platform.org/forum/forum_10/587894859

http://www.paraview.org/pipermail/paraview/2011-December/023538.html

I hope this time they are transferred correctly.
Nima

________________________________
From: Nima Maftoon <n_maftoon at yahoo.com<mailto:n_maftoon at yahoo.com>>
To: "paraview at paraview.org<mailto:paraview at paraview.org>" <paraview at paraview.org<mailto:paraview at paraview.org>>
Sent: Wednesday, December 30, 2015 4:57 PM
Subject: Visualizing complex vibration modes

Hello all,

I am trying to visualize complex vibration modes with Paraview. Unlike in a real mode, points in a complex mode have different phases. Please see previous discussions on this subject:
http://www.salome-platform.org/forum/forum_10/587894859
http://www.paraview.org/pipermail/paraview/2011-December/023538.html

The Paraview version included in Salome has a macro for modal visualization. I've put that macro and a MED file containing both real and imaginary parts of the mode shapes and another two other MED files for separate real and imaginary parts in the following shared folder.

<http://1drv.ms/1TqggEK>
http://1drv.ms/1TqggEK

The real mode animation in the mode macro is based on applying ExtractSurface(), ScaleVector(), WarpByVector() in cascade and animating the ScaleVector using a sinusoidal interpolation in a CompositeKeyFrame. However for a complex mode two components should drive the WrapByVector filter.

I describe my thoughts, which didn't solve the problem, below:

If the complex mode shape is a+jb, the animated mode shape should have a form like a cos(wt)+b cos(wt+pi/2). w is only for visualization and can be 2pi, 1, etc. t can come from the animation time. I learned about GetAnimationTime () that can be accessed inside a PythonAnimationCue (). Another way of getting the time is through CompositeKeyFrame for real and imaginary parts where "Phase" for one of them should be set to pi/2 (it is not clear to me whether the phase should be defined as deg or rad for CompositeKeyFrame). I guess at least one of my problem is that I don't know how to access data of the imaginary and real fields in the complex file or data of the two separate reader sources and do math operation on them to make a new field out of them and feed it to a wrap filter.

Any thoughts would be appreciated.
Nima

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/paraview/attachments/20160106/4ac56969/attachment.html>