Overlapping AMR Contour issues at AMR boundary?

Hi!

I have overlapping AMR velocity data that uses the Chombo reader. I would like to generate smooth Q-criterion contours. I’m doing the filters in the following order:

ConvertAMRdatasettoMultiblock → MergeBlocks → Gradient → CellDatatoPointData → Countor

The issue is that I am not getting smooth contours between AMR levels. There are visible seams as shown below.

Are there recommended settings, alternative filter orders, or preprocessing steps that can help eliminate these artifacts and produce smooth contours across refinement levels?

Any guidance would be greatly appreciated!

HI @ccardina

What you need to get correct gradient computation is either a Gradient filter that operates on AMR data, or a filtre to convert a AMR data to a dual unstructured grid.

I’m afraid both these feature do not exists in ParaView yet.

1 Like

Hi @mwestphal

Thank you so much for your reply.

If I compute the Q-criterion inside my simulation so that the Q field itself is already smooth across the refinement boundaries, would ParaView then be able to produce smooth contours?

In other words, am I right in thinking that the seams I’m seeing come primarily from the discontinuity in the gradient computation, rather than from the contouring step itself? Or could CellDatatoPointData introduce cracks, so even a smooth field would show seams?

For that you need a contour filter thats works on AMR data, there is one but its very specific and may not work for your usecase. You may need a generic AMR contour, which doesnt exists in ParaView yet.