GraphUpdateObject vs Scan performance at runtime

Hi, thanks for this great plugin!

We are using the Pro version and currently run a pretty big (5000 node) PointGraph in v4.2.17

We are adding PointNodes to the graph procedurally at runtime, which works great, but as soon as we try to update a part of the graph with:

GraphicsUpdateObject from script, using “UsePhysics = true”, it takes multiple seconds for a small box.

running a full Scan takes 150ms or so, which seems really strange. I was wondering if you could point us in some direction to where we could optimize this, seems like updating 1-2 nodes should be near instant?

here’s a screenshot of a box affecting just 1 node with usephysics enabled (update walkable, etc is all false)


Updating a point graph is sadly kinda inefficient due to a lack of constraints. When updatePhysics=true It has to check all nodes that could potentially have connections going through the given bounding box, which is all nodes in the graph. I’m not sure why it takes longer to update it than scan it from scratch though, I would expect it to be around the same time.

Your graph looks very much like a grid, though. Is there a reason you are not using a grid graph? It would be a lot more efficient.