Maniphest T96941

Geometry ndoes Raycast "light leaking"
Closed, Resolved
Actions

Assigned To
None
Authored By
baoyu (baoyu)
Apr 1 2022, 10:28 AM
Tags
  • BF Blender
  • Geometry Nodes
Subscribers
baoyu (baoyu)
Philipp Oeser (lichtwerk)

Description

System Information
Operating system: Windows-10-10.0.22000-SP0 64 Bits
Graphics card: NVIDIA GeForce RTX 2080/PCIe/SSE2 NVIDIA Corporation 4.5.0 NVIDIA 471.68

Blender Version
Broken: version: 3.2.0 Alpha, branch: master, commit date: 2022-02-21 23:28, hash: rBfcda858e3200

Description
Last edit: IT IS NOT A BUG

//
A is box which is big enough to wrap the icosphere B. Geomtry nodes on top of A, when casting rays from A in the inversed normal direction to B, the "hit position" sometimes return "0,0,0"s.
When A's subdivision been set to 2 or above, the bug can be seen.

Blend file for test:

raycast bug.blend154 KBDownload

Event Timeline

baoyu (baoyu) created this task.Apr 1 2022, 10:28 AM
baoyu (baoyu) added a project: Geometry Nodes.
Philipp Oeser (lichtwerk) changed the task status from Needs Triage to Needs Information from User.Apr 1 2022, 1:00 PM
Philipp Oeser (lichtwerk) added a subscriber: Philipp Oeser (lichtwerk).

That normal is the normal on the evaluated geo [which is the Cube], no?


So you are basically shooting in the (negative directions of the sides of the cube, thus missing the icosphere sometimes is expected.

If you are after something like the Target Normal Project mode of the actual shrinkwrap modifier youl'll probably have to get normals on closest elems on the target in a different way.

Does not seem like a bug to me, makes sense?

baoyu (baoyu) added a comment.Apr 1 2022, 2:30 PM

No, the normal is the box's normal. I made some hleping arrow to help visualizing the inversed normal of the box. Just play with the "Shrinkwrap factor" to see the problem.


raycast bug.blend158 KBDownload

Philipp Oeser (lichtwerk) added a comment.Apr 1 2022, 2:54 PM
In T96941#1333572, @baoyu (baoyu) wrote:

No, the normal is the box's normal.

Box is the Cube, no?

I made some hleping arrow to help visualizing the inversed normal of the box. Just play with the "Shrinkwrap factor" to see the problem.

Exactly, Now many of these (I circled some in red) dont hit the icosphere. So missing is expected, no?

baoyu (baoyu) added a comment.Apr 1 2022, 3:03 PM

Indeed! Thanks for pointing out my blind spot.
Well, if I switch the "subdivide mesh" to "subdivide surface", the points will probably hit the icosphere.

baoyu (baoyu) closed this task as Resolved.Apr 1 2022, 3:05 PM
baoyu (baoyu) updated the task description.