Instead of the cars snapping into position when a new position packet arrives over the net (as seen in LFS), the sim smooths out the position change when new packets arrive so you don't get that jerky movement. This neccessarily introduces some more latency on top of what you already get over the net and may also make the cars look somewhat detached from the laws of physics, but it looked rather well done in that clip.
LFS could absolutely benefit from something like that, yes. For one it'd make door to door racing against a lagging player a little safer.
What exactly has changed compared to 1.02 with the Scazzato sound? Less distortion at least, but I still get the feeling I'm hearing audio track from some onboard video where sound is very low quality and distorted by wind noise. Interesting coincidence how similar it sounds (in 1.02 at least) but I don't think that is the intention though.
At least good that the rumours (some months ago at rsc) weren't true, that Scazzato would have been excluded completely. Not like that there are too many games trying to do it (sounds) the other way (synthesized).
Apparently too long since last time played / watched video - not really after all but there is something similar. I guess biggest difference is simply different sound samples (and thats why it sounds so differencet compared to 1.02 crossfade, at first glimpse).
A shame to loose Scazzato, just because of the potential. Then again without patch V/W in LFS I would have gladly seen synth sounds go in LFS too.
Despite the claims of the fanboys, i see stupid stuff every time i watch iRacing, and this isn't on the rookie servers. It's far from the haven of clean racing that some claim it to be. Quite frankly if it had been me loosing a shedload of SR because of some of the glory boys idiot driving i've seen in the last two days i'd have told them to stick their subscription where the sun dont shine.
At least with CTRA you know you wont be on the end of a punishement for someone elses mistakes, and the idiot drivers get a small ban to go away and think about their driving standards
As for LFS I think the latency does not cause much trouble. The cars snapping into new positions isn't really a problem, the much bigger problem is the car collisions in general and the way small tap sometimes gets translated into huge tap. Even you have cars colliding with very small speed differentials you still have the risk that the cars may explode way from each others (extreme case: moonflights). Especially collisions between multiple cars can be extremely problematic in LFS (just like coding wise as well). I don't think smoothing would do much to that problem simply because there is always latency caused by something (physics engine refresh rates for example).
As I see it, that "smoothing" would be more of an eyecandy thing as the much bigger problems are elsewhere.
As for other sims, I think iracing has this smoothing thing in their sim and imho it doesn't work really well. It's a lot harder to tell what the car in front of you is doing than it is in LFS. I don't really know about nk enough to say if smoothing could improve some issues there...
I think the nature of the racing over the internet will always mean that racing centimetres apart from someone else will always be unpredictable at best. Not until everyone is on ultra low latency fibre optic connections anyway.
I have to say i much prefer iRacing and nKPros ultra smooth approach over LFS rather jittery movements.
Maybe it's a bit different things to different people. I really dislike the way the cars move in iracing, like no mass or inertia, the cars almost like transform from one place to another, like what you see the car doing in front of you is never what the car is actually doing. In LFS you can kinda see what's going on by looking at the other car. Maybe I'm so used to the LFS netcode that I just can't "see" the jerkyness but most of the time it doesn't bother me at all.
It's a combination of both I think. General collisions certainly need improvement as even against AI it's a bit erratic, but I think part of the reason you get these explosions is that when a new position packet arrives you effectively get a car's worth of mass moving to the correct position in one 100th of a second. If another car happens to be in the way when that happens the pure momentum involved results in a serious shunt.
Smoothing would reduce the distance a car could move each physics step, and that would probably reduce the risk of getting physics explosions as well.
the packets would still tell the engine that the cars are inside each other
the smoothing is just done for the graphics and doesnt change the mess of physics youll have to deal with in any way
as for the jerkiness of lfs i fully agree with hyper... the way cars both locally and remote behave in lfs is probably one of the main reasons why lfs has become such a movie generator its simply the only sim that looks convincing from the outside
Yes, but not that far inside each other after a single physics step. You'd avoid cases where the car suddenly moves, for instance, 1 meter sideways across a single physics step when a packet arrives and all that energy get transferred in any collisions that may occur as a result. Instead you'd get a less extreme collision as the engine slowly moved the car 1 meter sideways over a number for physics steps.
This assuming the smoothing was also applied to active physics bodies in the simulation and not just a visual thing of course.
Would it be that much worse than the current "syncing" though? Seems every client is running physics for itself and then it just handles any bad predicitons the best it can once updated car positions arrive. Doesn't seem like collisions are synced in any meaningful way. Or am I missing something?
There are really two different things that are indirectly tied to each others. First you have the collision model which seems to be having problems when two car bodies intersect. The other is the prediction code which tries to guess the future with the info it gets and has got. The smoothing is just a way to make the error on prediction code less visible. Surely smoothing could lessen the effect of two car bodies intersecting but it can do only so much.
I'm not sure. Shotglass brought up collision syncing and I don't think I quite understand what he meant. As far as I know there is no syncing going on other than what you get through position packets. Nothing specifically relating to collisions at least.
Indeed, which is what I've been saying all along. It would help, but the general collision response code would need work as well before it's properly stable. Smoothing would only bring multiplayer collisions up to the level of the single player ones.
Not really. What smoothing does it just smoothing, it has almost nothing to do with the cars intersecting due to latency or because of mis-guessed prediction, it just makes the transitions of the cars slower due to eliminating the sudden snaps of other cars into new positions when new packet is recieved from the host. Which just means that the cars won't snap inside other cars as quickly. Quickly is quite relative term though because there is bound to be some overlap when two cars are close to each others. The issue is not how to avoid overlap, the issue is how to deal with it