[vtkusers] HDF5 with VTK

Shakes Shekhar.Chandra at sci.monash.edu.au
Thu Jan 24 20:24:28 EST 2008


Hi guys,

Don't know if this helps but perhaps good for the record. In my limited 
experience with HDF5, the third party software I found useful were the 
Intel Array Viewer for viewing 3D HDF5 files and HDFView for 2D images. 
I have used the later more and found it good for images and the former 
not terrific.

Cheers
Shakes

Ogedengbe, Emmanuel wrote:
> All these contributions are very helpful for me. I just need to take
> time to identify which format will be compatible with the formulation of
> my CFD code. Much appreciation is rendered.
>
> Emmanuel
>
> -----Original Message-----
> From: vtkusers-bounces+emogeden=nrcan.gc.ca at vtk.org
> [mailto:vtkusers-bounces+emogeden=nrcan.gc.ca at vtk.org] On Behalf Of
> Dominik Szczerba
> Sent: Wednesday, January 23, 2008 12:44
> To: vtkusers at vtk.org
> Subject: Re: [vtkusers] HDF5 with VTK
>
> On Wednesday 23 January 2008 18.39:21 Eric Monson wrote:
>   
>> Since I don't know anything about CFD data, I may be throwing out a  
>> naive answer here, but in the past few months I've been dealing with  
>> (not too large) data sets in HDF5 format and viewing them in ParaView.
>>
>> There are two routes that have worked for me. First, XDMF is a format
>>     
>
>   
>> which can be read by ParaView (www.xdmf.org -- although the site seems
>>     
>
>   
>> to be a bit down right now, the format is in active development). To  
>> use Xdmf with existing HDF5 data you just have to write an XML file  
>> which describes the topology and geometry of the data set, and then  
>> within that XML you can refer to the spot in the HDF5 file where the  
>> data resides. So, it's a pretty easy way to quickly get up and going  
>> with viewing your data.
>>     
>
> There is a major limitation in this format: it does not allow to specify
>
> ordering for the data. So, if you are unlucky to have your data
> generated 
> using matlab or fortran you are left behind.
>
> -- Dominik
>
>   
>> One caveat is that current versions of ParaView do not automatically  
>> support time series of Xdmf files (you would have to build it in  
>> yourself using the directions in the ParaView Wiki -- 
>>     
> http://www.vtk.org/Wiki/Animating_legacy_VTK_file_series) 
>
>   
>> .
>>
>> Also, native time support within the Xdmf format is currently under  
>> development, but at an early stage, so I don't know when you could  
>> expect this to be ready.
>>
>> The second route I've used, just as an inefficient quick fix, is to  
>> write a Python script which reads in the data from the HDF5 file and  
>> writes out individual XML VTK files for each time step (I've done  
>> image data *.vti and unstructured grid *.vtu), plus a PVD file which  
>> describes what the "time step" is for each file. The latter is what  
>> ParaView loads, and can be used then for animations. Since this was my
>>     
>
>   
>> first Python script, I'm sure it's not very pretty, and it's also not
>>     
>
>   
>> very generic since I was in a rush to get some data converted for a  
>> presentation, but I'm willing to send it to you as an example if you'd
>>     
>
>   
>> like.
>>
>> -Eric
>>
>> -----------------------------------------------------
>> Eric E. Monson
>> Duke Visualization Technology Group
>>
>>
>> On Jan 23, 2008, at 10:34 AM, burlen wrote:
>>
>>     
>>> You are on target to point out the issues that the one will face  
>>> when a high resolution 3D results are required, lots of large files
>>>       
>
>   
>>> are indeed a challenge which should be anticipated for in the design
>>>       
>
>   
>>> of the CFD code. However, judging from Emmanuel's initial mail, he  
>>> is not asking about design issues he will face when implementing a  
>>> CFD code, he is rather asking how he can get a run up on the screen.
>>>       
>
>   
>>> My impression is that he just needs something simple to get him up  
>>> and going.
>>>
>>> CFD code design consideration aside, assuming that his results are  
>>> in HDF5 format, if he can get a HDF5 to VTK converter up and  
>>> running, then down the road he can extend that into a PARAVIEW  
>>> reader(thereby bypassing the extra file io) if needed(like you have
>>>       
>
>   
>>> done), so that is a win win.
>>>
>>> It sounded like he hadn't even installed HDF5 and H5Utils onto his  
>>> workstation, so that is probably the first order of bus. At least  
>>> then he can use the H5dump to see the contents of his files. Then  
>>> install VTK and/or Paraview. In any case it sounds as though he has
>>>       
>
>   
>>> some work in front of him to get up and going...
>>>
>>>       
>>>>> Michael
>>>>>
>>>>> On Jan 22, 2008, at 10:18 AM, Ogedengbe, Emmanuel wrote:
>>>>>
>>>>>
>>>>>           
>>>>>> After installation of VTK for use with purely C++ on windows, I  
>>>>>> have
>>>>>> been struggling with how to get started with reading my CFD  
>>>>>> datasets
>>>>>> for visualization. I have done some research with HDF5 but having
>>>>>> problem with having to go into increasing number of third party
>>>>>> software and libraries. Recently, I learnt that H5utils can
>>>>>>             
> convert
>   
>>>>>> HDF5 datasets into the formats required by other free
>>>>>>             
> visualization
>   
>>>>>> software like VTK. But it's only for UNIX systems, though it  
>>>>>> appears
>>>>>> help can come from CygWin .
>>>>>>
>>>>>> Can anyone help? I just want to visualize simple 2D/3D grid-based
>>>>>> scalar and vector flow fields simulated with my native codes.
>>>>>>
>>>>>>
>>>>>> Thanks in anticipation
>>>>>>
>>>>>> Emmanuel
>>>>>>
>>>>>> _______________________________________________
>>>>>> This is the private VTK discussion list.
>>>>>> Please keep messages on-topic. Check the FAQ at:
>>>>>>
>>>>>>             
>> _______________________________________________
>> This is the private VTK discussion list. 
>> Please keep messages on-topic. Check the FAQ at: 
>>     
> http://www.vtk.org/Wiki/VTK_FAQ
>   
>> Follow this link to subscribe/unsubscribe:
>> http://www.vtk.org/mailman/listinfo/vtkusers
>>
>>     
>
>
>
>   




More information about the vtkusers mailing list