Is it better to use "3 vertex" faces or "4 vertex" faces? And only for visible objects or also for colliders and regions?
I mostly prefer quads , it looks better.
But a quad always consists out of two triangles, so its basicly the same.
Having said this ,it counts for alll types, visible objects aswell for colliders and regions.
Additionaly an answer to an unasked question.
When to use colliders and which type?
the game engine has to check whether or not your avatar collides with an object face, otherwise you can walk through it.
if the mesh itself is used as a collider the game-emgine has to check for a collision against a long list of faces.
The reason to use a collider is to shorten this list and shorten the processor time needed for it.
therefore you can conclude that a modelled collider makes only sense, if the vertex count of that collider is lesser than the vertex count of the object you want to collide.
So if you have an object somewhere high in the air which is not in reach for any avatar, it does not need a collider at all.
if it is a small object with many faces, then place a cube or triangular tube around it , or use the bounce type "bounding box"
if the object is square then a collider makes no sense because the collider has the same number of faces and vertices as the square object.
so a rectangulair wall inside a square collider mesh is not necessary because you can use bounce type: convex hull
..and some of the other bouncing types as well.
if you look to it this way, then it might be possible for you to simplify the nodesystem for your moving door, if you only had choosen for different collider options.
So, triangulate some faces was only for PyPRP?