multiple particle tracers from single (programmable) pointsource

I have time dependent data and use a pointSource as a source for a particle tracer. The pointsource has the option “Number of Points”, and “Radius” for a random point cloud. However, If I use more than 1 point or a radius that is not zero, my point source does not work with the particle tracer. Should this work?

I could create a point source with multiple points with a programmable source/filter. If I want to change the coordinates every timestep, will that only work with the programmable filter? It seems that the source is accessed once, and the filter every timestep, is that correct?

Hi @bigfooted ,

Yes it should.

I’m a bit confused about what is not working.

Hi @mwestphal,
If I change radius to 0.001 for instance , I only see 1 static point that is not moving with the flow anymore. If I change to 0, I see particles moving with the flow, and new particles added every time step.
This is with paraview 5.10.1-799-gb5dd20f80e
I also checked with the 5.10.1 binary, same result.

What is the number of points ?

If I use 1 point, I see 1 static point, if I choose 10 points, with a nonzero radius, I see a point cloud of 10 points, but they are not moving. It also does not matter if I put Force reinjection every NSteps to 0 or 1.

please share a statefile and data to reproduce the issue.

The case is on github:

You can get the folder without the rest of the testcases by:

svn export

It contains the statefile and 6 timesteps, 80Mb

It show (in my case) that the programmable filter works correctly and injects a couple of points every time step that are transported downstream with the flow. The pointsource results in static points only

Here is what I’m able to generate with your data:

Please note that:

  1. Your data is 2D, you need to project your point source on it (using a calculator)
  2. Your time resolution is very low, resulting streamlines are not great
  3. Your velocity is quite high, I had to divide it by 10 to keep the particle in domain long enough to trace something
1 Like

Thanks @mwestphal, I did not realize that I had to map the 3D velocities and coordinates to 2D. The point source now also works. Here is a result where I use the programmable filter on a larger dataset: