Spherecast has some unpredictable imprecision. In circumstances where there should be a steady and single intersection point, there can be a noticeable discrepancy between that point and the returned RaycastResult’s position.
In the following scenario, there is logically a single point of intersection.
However, in the following video, you can see the point (the small black part) jitter.
This discrepancy can create a lot of problems on something like a character controller.
Hello, Im replying here because I cant post bug reports, the original message is in the shapecast announcement post, I’ll just copy paste it:
It seems to be a bug with spherecast, or it’s something that is not documented.
If there’s a wall close to the spherecast it would literally not detect any hit at all, it doesnt depend on the distance to the wall it just seems to be really specific “dead spots” where it wont detect basically anything, the way I got these spots was by raycasting and reflecting a sphere multiple times
Here’s a video
Here’s a place with one spot I found, the script runs and prints the result (nil) SpherecastBug.rbxl (86.8 KB)
if it’s not a bug then Im not really sure why this happens
This is happening I’m quite sure with blockcast as well. I am finding many “dead spots” in my map, where block casts (which do not start by intersecting geometry) cannot find any geometry block-casting straight down. Very similar feeling to the video you sent above.
