There's lots of talk of server-side prediction, but I haven't seen a client-side alternative implemented anywhere.

My game uses mouse-based movement, so it's easy to predict the player's location 150ms from when the event is sent. Wouldn't this be more accurate than using old data on a receiving client's side?

Is there something fundamentally wrong with this approach that I'm missing? I'm thinking of implementing it, but the lack of existing info gives me cold feet.

  • \$\begingroup\$ It's probably more that it's pretty straight-forward to determine where everyone will be in a short period of time, when you know everyone's destinations. Not sure how much discussion one can make out of it. :) You'll see a lot more work being done on the fine points of dead reckoning, where the clients don't actually know where exactly other networked players are going, because that's a much trickier problem to solve. \$\endgroup\$ Commented Nov 1, 2012 at 7:40
  • \$\begingroup\$ There was actually some talk about this not so long ago here: gamedev.stackexchange.com/questions/31288/… \$\endgroup\$
    – Marton
    Commented Nov 2, 2012 at 9:29
  • \$\begingroup\$ Here's a more general question on the same topic -- just to demonstrate this has been done / thought about. \$\endgroup\$
    – Anko
    Commented Mar 1, 2014 at 11:46


You must log in to answer this question.

Browse other questions tagged .