You make yourself clear, but seem to miss the point that some of us are making that a good collision detection system would be able to deal with / minimise the effects of, lag.
You make yourself clear, but seem to miss the point that some of us are making that a good collision detection system would be able to deal with / minimise the effects of, lag.
Door-to-door racing works 99% of the time for me
Only problems with strange outcomes after collisions are when people forget the brake pedal and maybe if there is some heavy impacts but usually you just get spun around or aided into a slide.The thing I find with LFS crashes is when the car rolls and bangs around it looks full of energy like it should do but in rFactor they just sort of flop around and just eventually stop (http://www.youtube.com/watch?v=oGV2HI00juc). Anyway I'm sure Scawen will do something in the future to resolve the current problems with high speed impacts.
From what I've seen of rFactor racing you seem to be able to drive through parts of car (mainly wheels) which would obviously make collision detection easier
... http://www.youtube.com/watch?v=3ghL_VIMZzk - http://www.youtube.com/watch?v=fSBJ10UOWf8 (3.40 onwards)Keiran
It is clear, I just don't agree with you. I have seen flyng cars in single player too, it isn't just limited to curbs and walls.
I think that Blowtus summed it up well.
Maybe the net code is "advanced" but the multiplayer experience and door to door racing is definetelly NOT.
The so called SIM doesn't simulate body suspension and tyre deformation graphically, so how can i enjoy driving door to door with someone when his car is gliding on the track? It's cartoonish as cartoons can be, ridicolous.
How am I missing the point? I already questioned that there might not be much else to be done other than to "dumb up" the netcode to make it smoother.
Because as said earlier, the cars are updated online packet by packet, and when one person lags and cannot send/receive packets to the server, there are chances that a car will "blink/appear" inside another car. How does that connect with the collision detection? Think about it, you have two cars side-by-side down a straight, and the one on the right lags and 'morphs' into the other car for a very quick moment. The car is beyond the collision detection's control, and the update of a car is briefly 'blinked' inside the car on its left. Like as if it quickly teleported partway inside the car on its left. Only then would the collision detection have to go haywire because two cars are morhped together! The packets and updates are not smooth enough at times so that a collision is done frame by frame. If you look carefully, a car will suddenly appear inside another car... and that goes beyond just fixing the 'collision model' :rolleyes:. This is a lag issue, and if just any detection system were improved for collisions, you certaintly couldn't account for a car lagging inside another... how would that be resolved? Should cars be allowed to morph for a certain set time of milliseconds? So that if there is lag, it will not affect the other person's car? Only thing I can think of to be honest. It isn't a step-by-step collision, these "lag collisions" are like 1 frame if you ever watch them in a replay. A car will appear, and the car being affected is just using that update as having the car inside it, making it fly, spin, tumble, whatever you want to call it.
EDIT: And if this 'max millisecond' rule was done, it would make the netcode very strange, because you would see some cars lagging, but also morphing with other cars. It may improve some things, but how can you tell someone is lagging online and make that recognizable through code? This is why the only logical solution I can think of would be to make the netcode smoother by not sending so many updates per car. And this may have to happen when we get a larger max grid allowed on the track/server. A bit unfortunate, because I like how many updates of a client's car are given to you when playing online in LFS.... as opposed to other simulations which look a bit 'dumbed' up for a good reason. And with rFactor in particular, the amount of updates given on that are not as high as LFS (like Boris just said above). This doesn't mean it is bad, it could be for reasons like we are discussing here (and to allow a larger grid for the host). IIRC, in rFactor you don't see numerous things like you do in LFS (steering input, active suspension play (this is nicest in LFS for viewing clients), complete tire heat and wear, etc).
People didn't vote "Best Multiplayer" on BHM for no reason....
EDIT2: And this doesn't look all that different from racing on LFS.... aside from the fact that every car seems to take every turn like it is Turn 1 frenzy
http://youtube.com/watch?v=T9PpBYFQoYQ. But still, my point is that both games are fun online, so I don't see why LFS's "flying cars" are the reason to say one is better than the other.Yeah, you're getting it, and yes I sort of munged (s'(t=1) * dt) into that ds1.
s'() is just the equations of motion. F=ma stuff. (x' = v, v' = a)
BTW, my simplified notation will break down if we push on it too hard...call it
quasi-notation.
And your line of thought is leading the right way. For backwards euler we solve
the system of linear equations we derive (using a Taylor expansion) from the
system of non-linear equations we get from asking
'what is s(1) such that s(1) - s(0) - s'(1) * dt = 0'
How do we find that? Well, we can guess, plug the numbers in and see if we are
close. If we miss, we can take that miss information (the residual), and somehow
use that to guess again. And again.
In effect we simply iterate educated guesses until we get the answer within
tolerance. These guesses are often based on following characteristic
eigenvectors (just think orthogonal deltas of the state vector to try) of the system,
which is to say a conjugate gradient solver. This iterating is why semi-implicit is
more expensive than explicit. (We could not iterate, using gaussian elimination,
since we linearized the problem, which is in principle more deterministic, but in
practice CG is faster for these sparse systems).
Since I've gone this far, the linear system is an approximation of the non-linear
system. We could do an outer iteration to solve that system, using linear system
solves as non-linear guesses. Way more expensive, but that is what a full
implicit solver (like Newton's method) would do.
That has to be the absolute shortest description of implicit integration I've ever
seen, so I don't know if it is understandable at all.
Peptis, I agree implicit integration will not solve the lag induced collision problems,
but it would make run of the mill stiff (in the mathmatical sense) situtations more
stable. There are lots of factors in deciding on an integrator. My main point is that
IF a moddable simulator chooses to do so, it would be useful to know as a modder
with respect to directions of exploration with mod parameters or extensions.
I now think that we are actually talking about different things.
You seem to be talking about cars lagging online and reappearing inside each other for a fraction of a second. You then go on to discuss what should happen in such a situation: Should we consider this a collision? What should be done about it?
I am talking about an actual collision. One where two cars have genuinely collided (or a car has collided with a wall), in this case lag is not the problem. The issue here is what do we do in the case of such a 'genuine' collision?
These issues aren't mutually exclusive, but they are both important and the final solution needs to accomodate both of them.
I completely agree. However, from my experience I do think that the current integrator is probably doing a good job, I just think that it is being fed the wrong information in some rare situations.
Anyway, I agree with an earlier poster, you should have posted here a lot sooner. Welcome
Thx, I think I got a good idea of what it's about and I don't intend to go any further unless I need it for my own purposes some day. If anything it will sure make me sound smart.
Perhaps we are
The only reason why I brought up the lag + morphing cars is because that is the most logical explanation for a flying car during door-to-door racing. Collisions that you speak of with other cars can happen at a normal rate and turn out to be a typical collision you'd expect to see online in LFS. It is just that, to me, the flying cars are more connected to the lag + insta-morphing issue. Can be a number of things that create such problems however... all of which have been discussed by everyone here.
I never said anything remotely like that. I think LFS is better online for a lot of reasons.
I know what you are talking about. I quoted you, didn't it? Barriers = reproducible at will. In a race = uncommon.
How is it not?! If a car is inside another car, a collision has occurred. That is all collision detection code does: "is any part of volume X inside volume Y?"
The discussion at hand is about what LFS does with those sorts of collisions. For instance, client X gets an update from the network telling him that car Y is inside our car. How far is he into our car? What was his velocity vector when he struck us? We can use that to create an approximation of a realistic result, right? Right now, LFS does some wacky things under some circumstances, but I'm sure that will be worked out by S3.
I think you mean not as much data per update (rather than fewer updates, which would make teleporting issues worse).
That wasn't even directed towards you. :zombie:
It is so hard to talk on this forum. Comments are not directed at particular people, and they think I am talking to them :doh: Even harder to explain something that is dependent on personal experience & knowledge too it seems
That *is* collision detection. Every collision involves objects inside other objects temporarily - lag just increases the depth. So if you have a system that can deal with interpenetration (collision) effectively, lag shouldn't cause extra dramas, except for a bit of warping around as corrections are applied, which we already have anyway.
The url contained a malformed video id.
That's from a poorly made mod (in that particular regard).. regular cars and well made mods don't have that problem.
That is complete BS.. I have heard so many LFS fanboys say that- of course it shows suspension movement graphically
illepall tire deformation it doesn't.. but that's not exactly something other drivers look at in real life. LFS over does it in the deformation department, and all tires react the same, no matter if they are F1 tires or street tires. not exactly realistic either. Door to door racing on-line is better in rF no matter what- because of the far superior net-code. Do you even own rF?
I think what Boris was trying to say is that rFactor makes online opponents look rather 'simplified' and smooth, and they don't show a lot of things like LFS does. Tyre deformation shouldn't matter, but things like steering input, tire heat, wear/flatspots, etc... all of those updates are shown in LFS and very close to what you would experience seeing on your own system live. Whereas from what I've seen on rFactor, it is rather simplified and toned down. I've watched my friend play online and spectate other drivers, as well as watched lots of videos of rFactor's online play, and I have to agree that it is pretty smooth looking.
I don't see how that is not obvious when playing online, it is clearly visible that these things happen. But it isn't as bad as you think we are trying to make it sound. And it isn't an "LFS Fanboy" saying :rolleyes: Get that out of your head dammit!!!! (and especially you!!!) It is an OBSERVATION.
I don't see how that is not obvious when playing online, it is clearly visible that these things happen. But it isn't as bad as you think we are trying to make it sound. And it isn't an "LFS Fanboy" saying :rolleyes: Get that out of your head dammit!!!! (and especially you!!!) It is an OBSERVATION.
Fanboyism ^
Imho the barrier-flip and the lag-induced fly-to-the-moon-collisions are actually the same thing?
When you hit a barrier it seems that parts of the car are _inside_ the barrier (e.g parts of a wheel, just look at the massive tyre flex going on).
Which is exactly the same thing that happens when you lag into someone elses car.
I also have the feeling that the cars do not absorb enough energy through damage in those (or in all?) cases. Well, and barriers of course don't absorb anything..
So what's actually wrong?
1. When it's caused by lag you can't do anything about it, but in the other cases (barriers) it just shouldn't happen, somewhere _is_ a flaw in the collision detection.
2. When a car is found to overlap too much (as in: the collision should've been detected earlier), for whatever reasons, the force generated should be reduced/have a sane maximum.
3. When the force is applied more of it should go into deformation (be it visible or not) and less into motion. Actually simulating how much a body deforms on impact (and how much crash energy it thus absorbs) seems to be very hard in real-time (and I doubt it's done at the moment?). And then it's not that important either (it's a racing simulation after all, not a crash simulation). From the results I guess there's a "simple" formula which depending on masses, speeds and impact angles decides how much energy goes into deformation - maybe that needs a tweak..
When you hit a barrier it seems that parts of the car are _inside_ the barrier (e.g parts of a wheel, just look at the massive tyre flex going on).
Which is exactly the same thing that happens when you lag into someone elses car.
I also have the feeling that the cars do not absorb enough energy through damage in those (or in all?) cases. Well, and barriers of course don't absorb anything..
So what's actually wrong?
1. When it's caused by lag you can't do anything about it, but in the other cases (barriers) it just shouldn't happen, somewhere _is_ a flaw in the collision detection.
2. When a car is found to overlap too much (as in: the collision should've been detected earlier), for whatever reasons, the force generated should be reduced/have a sane maximum.
3. When the force is applied more of it should go into deformation (be it visible or not) and less into motion. Actually simulating how much a body deforms on impact (and how much crash energy it thus absorbs) seems to be very hard in real-time (and I doubt it's done at the moment?). And then it's not that important either (it's a racing simulation after all, not a crash simulation). From the results I guess there's a "simple" formula which depending on masses, speeds and impact angles decides how much energy goes into deformation - maybe that needs a tweak..
For some reason the URL decided to included the end braket, http://www.youtube.com/watch?v=oGV2HI00juc
So both mods are poorly made then, the so such highly acclaimed CTDP F1 mod?
http://www.youtube.com/watch?v=WBVxoRLZS2c - what put the BMW on its side :S Contact from the following car couldn't have been enough to scratch a rear bumper.
http://www.youtube.com/watch?v=IyOC5ewZPe0 - :S Strange collision there. Car teleports into it.
http://www.youtube.com/watch?v=uXiA2HlXhDw - strange reaction for a kerb and a car that goes into the ground.
http://www.youtube.com/watch?v=iAvGgfuEW2w - again strange collision behaviour
http://www.youtube.com/watch?v=teXRv-tHfKY - again strange collision behaviour and reactions.
The reactions in rF are hardly much more realistic than LFS but in a different way. I can race normally in LFS with 99% of the time nothing strange happens. Last time I can remember getting strange crash behaviour was hitting the wall hard at AS Club on the exit of the fast S bit in the FOX. But when the car rolled and banged about it actually looked like what it probably would. Rather than roll effortless till it eventually stops which is what happens 99% of the time rFactor. I very rarely have a crash in rFactor that I think is convincing, I'm always left thinking "was that it".
The suspension movment and body roll in rFactor can hardly be called realistic or anywhere near LFS, call people fanboys if you want for saying what is fact. It doesn't look very convincing and it is often not noticeable that the car is actually moving around. Racing in LFS you can actually see when someone in fronts car is moving around and looking loose at the rear and it all looks natural, something I haven't witnessed in rF yet.
Well, i am a fanboy, but i am telling you the facts here. You can't honestly say that rFactor shows suspension movement in Multiplayer? I played Lupo Mod pretty much online, and there is NO suspension movement. Only movement is the tire, the car itself is static. And you are saying that is realistic??
I was LMAO when i witnessed that, it was so arcadey. Also that "perfect" net code was making the cars move so unpredictably, and everyone there had stable connection. I rather have the car disappear for the milisecond, then to have it bouncing on the track unrealistically.
I also watched a lot of PCC, Megane's videos and it's all the same. Really, really funny looking... and if someone can get immersed in that kind of multiplayer enviromennt then Kudos to him..
watchin bwx calling anything fanboyism proves in forums u can say whatever u want to say, no matter what your knowledge of things is.
leave the propaganda for bhms bwx.
leave the propaganda for bhms bwx.
Agreed. I haven't seen this great netcode people speak of in rF, nor realistic dynamics (in single player, multiplayer, replays or anything).
Lol, so true!
:munching_ This is entertainment at it's finest.
LOL- he said himself he was a fan-boy.. So I guess I was right.
The difference between me and a "fan-boy" is that I have an open mind about all race sims and don't pigeonhole any of them into certain categories or have some delusion that one is a sim and one isn't because of some subjective "feeling" someone has.
rF does some things better, LFS does some things better. DUH! The fan-boys of either of the games are the ones who lose out.
It's also funny that people that don't even own both games are posting about the differences between them.
Suspension movement should be the least of your worries when comparing the on-line capabilities and net-code of a race sim anyway. I never noticed a difference in suspension movement when playing on or off line in rF.. but there definitely is suspension movement simulated in rF. I've heard people say there isn't so many times and they are just plain wrong.
propaganda:
information, ideas, or rumors deliberately spread widely to help or harm a person, group, movement, institution, nation, etc.
Looks like the fanboys who say rF doesn't have suspension movement are the ones spreading propaganda.
Show me, as I've never seen any, in any rFactor-movie I've seen... And no, those camera shaking effects which should simulate a bump don't count...
Yes, please show us BWX.. I don't want to be blind fanboy... i wanna see it.
Rfactor vs LFS
(1872 posts, started )
FGED GREDG RDFGDR GSFDG