[Paraview] particle tracing and particle tails help

[Dan Lussier]

Ok - thanks for the more detailed rundown.

Cheers,

Dan


On 7-Apr-11, at 11:50 AM, Biddiscombe, John A. wrote:

> Dan, cc’ing paraview list again in case anyone else needs the  
> information
>
> 1. You suggestion of using the 'Find Data' was great, but I found  
> that I needed to use the  'ExtractSelection' before going ahead with  
> the 'ParticlePathlines' filter.  From what you described that didn't  
> seem necessary.  I suppose this came up particularly because in the  
> ParticlePathlines filter once I extracted the selection I could  
> simply set 'Mask Points' to 1 and take every particle in the  
> carefully prepared selection.  Is the extract step necessary?
>
> 2. If ParticlePathlines is the way do to track explicit particles  
> like mine, is there any use for the Particle Tracer outside of  
> tracking the motion of injected seed particles over time within a  
> vector field.
>
> 1 The find data was just a quick way of extracting particles by ID  
> over time and yes, you need the extract selection filter. The  
> particle trails filter actually has this function built in, it has a  
> secondary optional input called ‘selection’. On this input, you  
> simply provide a dataset with (say) a field called ID which must  
> match the IDs on the main dataset, if the ID exists in the  
> selection, it’ll be tracked in the main dataset. The reason I used  
> the find data instead of just mask points was really because I  
> assumed your Ids were nonsense (this is the top reason that people  
> get no trajectories), and if the Ids were no good I guessed that  
> using the find data (by ID) would for sure show me random movements.  
> I could have done it using the pathline filter and mask points (and  
> I did first time), but I was double checking to make sure things  
> were as I thought.
>
> Consider these cases for educational purposes ....
> case a: You have a single time snapshot of your particles, each has  
> an ID. You save this snapshot and select some interesting subset  
> using a selection or maskpoints or whatever analysis you desire..  
> You now use this static dataset  as the secondary selection input of  
> the trails filter and the main original time dependent dataset as  
> the primary input *** those points with the fixed Ids from the  
> selection are tracked from the dynamic dataset over T*** (note that  
> if the Ids were actually wrong anyway, you’d get random tracks).
>
> case b: You use the “find data” dialog to extract the Ids you are  
> interested in and then draw pathlines for them. The Id extraction  
> will update on each timestep and potentially give you different  
> particles (if the Ids are wrong really). What you would now see is  
> random tracks. I tried a and b to check your data and hit reply when  
> b was still visible in my screenshot.
>
> the difference between a and b, is that a selects ID’s once and then  
> builds tracks using the ID to find the particle. b extracts the  
> particles using ID at each step and constructs the trails. They’re  
> sort of done in the opposite order and should give the same result  
> for particular Ids. (you can combine a and b, but the pathline  
> filter uses map and vector structures to x-ref particles, so having  
> a large number of Ids selected will cause memory issues when long  
> trails are used).
>
>
> 2 The Particle Tracer is for advecting particles in a vector field  
> as you suggest and is of no use to you for what you are doing. Users  
> of the particletracer can add a trails (pathline) filter to view the  
> trajectories.
>
>
> JB

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.paraview.org/pipermail/paraview/attachments/20110407/34a74c43/attachment.htm>