When testing cl_autotimenudge, I found that my field of vision would become very jerky at times, usually during battles. This doesn't always affect the FPS counter, but the game/input/display becomes very stuttery, almost like the engine is having difficulty calculating the correct time nudge.
Core 2 duo 3Ghz, Nvidia 120GT, 8gb memory, windows 7, firefox
same man)
Negative timenudge works by extrapolating (predicting) player's position. There will always be some missed predictions of position of a player.
Even strafe jumping player will be a bit jerky, because he accelerates, but your client can only predict him moving by a constant speed. So each prediction of his position will be wrong, and he will be warped a tiny bit forward.
Dodging is even worse, because your client predicts enemy going in certain direction, but then gets "corrected" by the server than the player has moved in the opposite direction in the mean time. This is something that can't be fixed, ever.
On low values of timenudge it's not noticeable, but cl_autotimenudge enabled TN of negative ping divided by half. So if you play on ping 80, it enables TN -40 (or is it -52,5? I never know).
isn't the lowest setting for timenudge -20 though?
This is a side effect of the high timenudge values that can be set when autotimenudge is 1. Since cl_timenudge -20 is already a bit stuttery, you can imagine that autoTN going lower would be even worse.Originally Posted by Storm
I find though that on laggy servers I aim better in general with autoTN 1 despite the graphical unpleasantness.
It would be great if we could have the effects of timenudge without the stuttering, but that may be a pipe dream.
For cl_timenudge, yes.
However cl_autoTimeNudge is capable of internally setting a value of lower than -20.
Note: cl_autoTimeNudge 1 will override any current value of cl_timeNudge.
I don't think you guys fully get what I'm saying, I don't mean the enemy models are jerky/warpy at all. What I mean is that I experience a feeling of FPS lag or input lag randomly, when it is enabled. The fps counter stays at 125 though.
Quakelive servers are running at 40Hz (40fps), but clients are running at a higher speed (probably 125fps for most of the players), so frame interpolation occurs with normal gameplay.
(Imo), when you are using autonudge, your client can't keep this interpolation synced with the server (since it's autoadjusting itself realtime on the fly), and that's what you percieve as a fps drops.
We do get what you're saying. It happens to everyone.
This may be behind the reasoning for limiting cl_timenudge to -20. Lower than this may have caused these side effects.
so use autotn 1 and timenudge 0?
But they are, they have to be, unless you're playing on sub-30 ping where it wouldn't even be noticeable (but then why would you use autotimenudge in the first place?)
FPS counter doesn't lie. What you observe must be prediction errors, which fall in two categories:
- other players being mispredicted, the effect is jerky movement/warping of other players,
- your own position being mispredicted, the effect is "rubber-banding" of your own model.
If you move in a direction but get hit (by anything, even machinegun has a bit of knockback, not to mention getting hit with RG/RL), your position will be corrected by a server too. So your own movement becomes jerky when taking damage. The higher your ping, the lower the timenudge, the higher the symptoms.
These effects will never go away completely. It's not a bug. That's simply how TN works... It's your decision whether to use it or not. If not the problems, it would be the default long time ago.
Posting Permissions
Reply With Quote
