Handling larger amounts of AI over a network?

Options
Hi,
Does anybody know the most effective way to handle a large amount of AI over a photon network?
Essentially, upon testing and trialling an enemy system. The enemies are largely similar to actual players in the game, in that they both send, position, rotation, animation and attributes (health etc).

Considering, even with optimization and speaking to people about smooth network gameplay. Even at 6 players in my game, I can still see a lack of smoothness in areas of networked players. Photon itself says that a game of 32 players would be difficult.

I planned on having around 50-100 AI at a time.
I did think, perhaps I could spawn the enemy on the master client, then send the AI's target or path if you like, then send that across to all the players. Essentially, I wouldn't be sending position information at all, just it's path and then all the animation,position etc would be done locally. The problem with that is, if I used a navmesh to move AI (which is likely), Unitys pathing system actually calculates slightly different every time.

I have actually tested this offline and it does infact calculate a different route everytime, even with the AI is in the exact same position.

The best answer I guess would be: Photon can in fact handle 100 Photonviews sending transform,rotation and attributes just fine. I will test this myself today but I wondered if anybody had a solution if this isn't doable?

Thanks

Comments