Hello chaps.
I have been doing a computer vision class where I proposed a natural vision to agents into games.
The idea is to use some stuff I learned from class and apply to 2d vision on a top down map, generate an histogram and AI from there.
Here a video of a prototype, all glb stuff ;)
Sorry, you lost me at - "a natural vision to agents into games." What?
I guess you are trying to create a sort of heightmap, based on the 2D map and using that to create some sort of AI based on the terrain?
is it like Fog of War?
I'm not sure (Erico correct me if I'm wrong), but this it's something like a map on you can localize your enemies?¿....
The interface for now it's very simple, but sometimes it's cool leave very simple the things...
Comment us, we don't have too much clear your idea ;)
hehe sorry I used all wrong words here, I must be going nuts.
It is just a system which gives the enemy a 45 field of view.
Enemy can only see what is inside that and work from there.
So in the video, he tries to go to the target, in case it is blocked, he goes on analyzing what he sees and choose newer directions based on that.
I tried to make it work like if it was a human trying to do the same, so there is a lot of ´looking around´ to be sure.
At the very end, imagine a top down game where npcs can only work around by what they see.
Some games use it or have used such before, so it nothing new.
...but fun to implement and look at ;)
This is some very great work :nw:
Hopefully someday I can turn that into a realtime game with plenty of inteligent NPCs.
I was doing an article about it, but it feels empty without a game to go along. ;)
Would be nice if this could be pushed into a little robot.