I have a 4 player networked game using one host-client and other connecting clients (2D, although that doesn’t really matter). My characters can move, sprint, and dodge. What I’ve noticed is that general walking and sprinting sync pretty well, but dodging creates an obvious problem.
Is there a better way to sync a fast movement than by lerping via Network Transform? For instance, when a character dodges, should I produce the dodge for the local client and then send the dodge command to server to then be applied to each remaining client? They would ultimately lerp to the resulting server-state, but in the interim would it help lessen the lerp, or by the time the dodge command got to the end-clients would the lerping have already done the majority of that anyway? Does anyone know of any good reference materials or guides for getting better basic transform syncing?
I would greatly suggest you to drop the NetworkTransform. It doesn’t really do interpolation. It’s more of a prototype tool.
As for syncing such precise movement. If you have a server authoritive movement model, it shouldn’t be an issue. Otherwise, you have a few options.
Option 1 is to send a dodge instruction. This could however create wierd behavour if the dodge start positions aren’t thr same. So I guess include that.
Option 2 is to just run the sim on the client and trust yur movement sync code to take care of it.