Maybe I don’t understand how AI sensing works but it seems like a vision cone is not the way to do things.
The cone of vision in a sense makes the AI near sighted right? (more precisely, central scotoma)
Meaning, it sees further and better from peripheral vision than in front. Am I missing something? This seems bad. It feels like there should be a sphere making up the base of the cone…
Do I just not understand this thing?
I guess it works for most games and is pretty adjustable.
I’ve just attached big colliders to head joint so it follows animation, and then if target overlaps the collider, start doing a line trace to check if there is direct line of sight. Can adjust shape of the vision just by scaling the collider.
I dunno if this is better or worse, but to me it seemed simpler and more intuitive. In my case though I specifically wanted AI that you couldn’t really sneak up on, whereas in a ton of games usually you want to sneak walk and then break their neck or whatever.
Hey there @TinyTracker1! The conal vision is mostly useful for stealth games in that players are intended to be able to play inside and around peripheral vision. It’s adjustable to where you could also make it closer to realistic vision, but as you say at that point it’s somewhat similar to just dropping a vision zone. A solid example is how it’s used in this demonstration:
