[Networking] Interpolating replicated movement data in Pawn, without modifying source

[HateDread]

@Acren;99002

Yeah, I don’t use Replicate Movement. It just doesn’t work for me. Causes desync right off the bat.
So what I do is disable that and just replicate the properties I need or call RPCs.

You said clients send their input events to the server? Just do whatever the server does with them but locally when you send them.

In the client’s RPC or OnRep, if you do nothing, should just cause the client and server simulations to run completely independent of each other, which will eventually desync.

Right, but I don’t see how to intercept and utilise the physics calculations.

Imagine we’re only doing the local physics interpolation to server physics (and not keeping track of past stats and so on… that’ll come later). I definitely do send the input event to the server, but also execute it locally, so my client is definitely doing physics simulations with the same inputs. Should I then tick the Pawn on the client, to get the physics updated, and store this locally, then tick on the server to get the replicated/changed version, then next client Pawn tick, interpolate? I don’t yet know how to control who/what/when ticks the Pawn to this detail, yet. Any ideas?