topic Is this Raycast normal behavior? in Bifrost Forum

Is this Raycast normal behavior?

brandaosa — Fri, 14 Jun 2024 08:41:57 GMT

I have an irregular surface (mountains) where I will scatter points.
The goal was just to cull points display behind each other along camera line. So I thought about the raycast node. But it was a bit costly to make the setup to work so I am here to double check.

I have to first remove points where normals were facing out the camera.
But the problem is that for the mountain surface, some faces will still face the camera even if they were culled a surface nearer the camera.
I tried to raycast from the camera to the remaining points. It works by clearing the points culled by nearer surfaces but instead of deleting these points, the node reproject it on the nearer surface (on the video, you can see the "dancing/sliding" blue circles) So to get the effect I needed, I had to compare the points original positions with the points after raycast (dancing) and remove those which distance were > 0.

GRAY POINTS displays the intended result

(view in My Videos)

Notice points will not overlap on camera. Below, the holes on the mesh can be see more clearly.

(view in My Videos)

Why raycast is reprojecting points back?

Re: Is this Raycast normal behavior?

sepu6 — Sat, 15 Jun 2024 04:46:37 GMT

Without looking at a file, difficult to help. It could be also the method that you are projecting or a bug. Also, worth having a look at these threads.

https://discord.com/channels/872260298508222534/1065636496637435924

https://discord.com/channels/872260298508222534/1067839241331150889

Re: Is this Raycast normal behavior?

brandaosa — Tue, 18 Jun 2024 09:31:02 GMT

by the way the file, except for the montain = a plane with two wave modifiers - can be replicated by looking at the graph. I will check if this is the behaviour in a different version. THanks

Re: Is this Raycast normal behavior?

jason-t-brown — Thu, 20 Jun 2024 13:41:17 GMT

The "dancing circles" you are seeing are the results of the location scope - it's showing you where the rays are hitting on the surface closest to the camera. If you turn off "circles at sample" on the location scope, they will disappear.

The method I use for doing this still uses a raycast, but is simpler - what I do is set everything up so that the rays are being cast at the camera, then look for any rays that register a hit. You have to be careful that you have `min_distance` on the raycast compound set to a number above zero, or the hits will happen at the very start of the raycast.

Hope this helps