I’ve carefully examined the problem, and it seems the Hit.ImpactPoint position is incorrect in the player landing check function IsValidLandingSpot(), ImpactPoint pierced the other side of the sphere…
Yes, I can confirm this happening too in my project, as I use some sphere as foot platform and my character couldn’t get on top smaller Sphere mesh anymore and always falling from it (it was fine in 5.6), it’s behaving like if I set the Sphere “Can Character Step Up On” to false even though it’s true.
The workaround for me at the moment is to change Sphere collision in Static Mesh Editor from sphere collision to capsule collision which somehow doesn’t have this problem(?)
Do you think we need to fill bug from so this problem get noticed, or usually post in this forum get noticed by Epic?
I have found some temporary solutions, the simplest of which is to set the CollisionTraceFlag in Static Mesh to UseComplexCollisionAsSimple, this will result in some performance loss.
Another approach is to inherit from the UCharacterMovementComponent class and override the IsWithinEdgeTolerance() method, making it simply return true. This should not affect normal collision detection and performance, but the collision point will still be problematic.
I think we should report this bug to Epic Games; the collision physics-related parts are probably quite serious…maybe?
Using Complex Collision as simple might be more expensive than using Capsule collision (especially if your sphere mesh is high poly), and with capsule collision you could keep use simple collision. And capsule with very short height is basically 99.99% same shape as original sphere collision.
Okay then maybe it’s best to fill the official bug form, but the last time I tried to submit one about megalights it didn’t even go through, lol. But luckily shortly after that, others reporting same bug so it got fixed in the end.
Let’s try again now hopefully this time it’s got through!