Children CM have colliders to avoid collision, it’s ok if there is occlusion. How do I set that up?
I also added a CM collider to the blend list to avoid collision during transition and it’s going through stuff when the child camera is active.
How do these extensions interact?
How do you avoid penetrating objects at all cost (occlusion ok)
these are the children cameras
If you put a collider extension on a BlendList (or any manager-style vcam with children) the collider will apply individually to each child vcam. That is to say, when each vcam evaluates its position it will invoke the common shared collider and will be pulled out of obstacles. Additionally putting collider extensions on the children is redundant and wasteful.
During a blend, the already-resolved vcam positions are interpolated, and it is possible in that case for the camera to pass through obstacles. Blends are not subject to collision resolution.
One way to be totally sure that the final camera is never inside an obstacle is to add a custom behaviour to the camera itself (the one with the brain) that pulls the camera out of any intersecting obstacle. You would have to do the resolution in the script’s LateUpdate, and set the script’s execution order to be after CinemachineBrain.
Collision should work during blend: by showing as a homonymous UI gizmo in various components you set the expectation that the extension will work the same way in all components. So applying a different logic is breaking this expectation.
