Local or Relative Vector Sensitivity?

[Bullit]

I have made a setup - see the image -where i can have particles move only if the geometry deformation they are influenced by pass a certain threshold.
In short it ignores the small(in Y) values bumps. So a filter pass based on Y value.

Now this works for simple situation where the surface is an horizontal plane and i can use the Y value. But if this is a mountain or other irregular object i can't use Y as a source.
In short i need to be able to find small spikes even if they are inferior in Y to top of the mountain. Any idea how to achieve this?

sensitivity.jpg

[Mathaeus]

I have made a setup - see the image -where i can have particles move only if the geometry deformation they are influenced by pass a certain threshold.
In short it ignores the small(in Y) values bumps. So a filter pass based on Y value.

Now this works for simple situation where the surface is an horizontal plane and i can use the Y value. But if this is a mountain or other irregular object i can't use Y as a source.
In short i need to be able to find small spikes even if they are inferior in Y to top of the mountain. Any idea how to achieve this?

If surface is, let's say continuous, without sharp extrusions like in pic, that could be mountain - perhaps something based on angle, like curvature map compounds.
Otherwise, checking the Y of point neighbors, or spreading a small array in circle around particle. Second is a bit expensive. But, last two could returns something unexpected, spikes become filtered by some arbitrary distance.
So, I'd say, curvature map, or even the baked ambient occlusion.

[Bullit]

Thanks for your ideas. Unfortunately it will have sharp cliffs.