Sirocco Racing : New Sim Demo

Hi Folks, As it seems to be a slow news day in RAS-land I thought i'd spam my new demo :) **Note - The collision detection code isnt finished, so only the tyres collide with the surface. This causes problems with walls / trees as the tyre thinks it has been compressed by ~1000mm and so rebounds madly ;). Shift-R quickly i'd suggest :).** http://www.siroccoracing.com Here are the improvements since the last version: - a.. Multiplayer - the start of a Client-Server based multiplayer system a.. Multiplayer Chat a.. Force Feedback - based around lateral tyre force a.. New Car - the JTCC NSX modelled superbly by CD_QuikSilver a.. Differential - basic adjustable viscous diff added a.. Engine Braking - adjusted to be dependent on engine speed a.. Problem fixed - error when calculating the speed of the engine a.. Problem fixed - tyre forces being applied in slightly the wrong direction If you call into VROC i'll turn on the multiplayer server. It currently automatically tries to connect to the server, if you want to stop this set "startasclient" in ./data/options.ini to 0. You can also create your own server by changing startasserver to 1 in options.ini. I will upload the dedicated server to the downloads section in a couple of hours. Have fun :) All the best, Ash http://www.siroccoracing.com

On Mon, 24 Mar 2003 14:33:41 -0000, "Ashley McConnell" <[email hidden]> wrote: >Hi Folks, > >As it seems to be a slow news day in RAS-land I thought i'd spam my new demo >:) Hehe, I think the LFS 0.2 demo would have done that, but it's still kinda quiet about it. ... >http://www.siroccoracing.com >Here are the improvements since the last version: - > >a.. Multiplayer - the start of a Client-Server based multiplayer system Nice! Do you already handle a bit of timing problems? (server time vs. client time); prediction, extrapolation and that sort of stuff? >a.. Force Feedback - based around lateral tyre force From a quick run, it felt as if the FF was a bit strong; so I quickly got hard-left or hard-right forces. Although that is ofcourse a car-specific thing, you still get more quickly into these wobbling left-right vibration things when cruising on a straight piece of road... >a.. Problem fixed - tyre forces being applied in slightly the wrong >direction Heh, don't I know about those problems. ;-) Got something like that I think in my latest version (v050 beta 11). Severe car drag at low-speed (perhaps a direction error when feeding the relaxation formulae). One thing I noticed while glancing over a car.ini file is that the relaxLengthLon > relaxLengthLat. Which seems odd, as mostly it's the other way round (say L_lat=0.91, L=lon=0.091). Might be a bug, but might be a butterfly after all. :) Ruud van Gaal Free car sim: http://www.racer.nl/ Pencil art : http://www.marketgraph.nl/gallery/

"Ruud van Gaal" <[email hidden]> wrote in message news:[email hidden]... > On Mon, 24 Mar 2003 14:33:41 -0000, "Ashley McConnell" > <[email hidden]> wrote: > > >Hi Folks, > > > >As it seems to be a slow news day in RAS-land I thought i'd spam my newdemo > >:) > > Hehe, I think the LFS 0.2 demo would have done that, but it's still > kinda quiet about it. > Hopefully i'll get a chance to give it a run tonight :))

Quoted text (10 lines)

> ... > >http://www.siroccoracing.com > > >Here are the improvements since the last version: - > > > >a.. Multiplayer - the start of a Client-Server based multiplayer system > > Nice! Do you already handle a bit of timing problems? (server time vs. > client time); prediction, extrapolation and that sort of stuff? >

I have done a little of all of that, some of it myself and some from the library i am using (http://www.rakkarsoft.com). It is by no means any good yet, but I think i have the basis for a good amount of players. I will of course have be drastically reduce the number of bits i use to send everything, just want to get it working for the moment before looking at compression. Any progress on Racer becoming multiplayer goodness? That would be fun :). Good thing about starting multi is that replays and ghost-cars should drop out fairly easily (simply a pretend network car!). It'll be a start anyway :)

Quoted text (7 lines)

> >a.. Force Feedback - based around lateral tyre force > > From a quick run, it felt as if the FF was a bit strong; so I quickly > got hard-left or hard-right forces. Although that is ofcourse a > car-specific thing, you still get more quickly into these wobbling > left-right vibration things when cruising on a straight piece of > road...

Those settings are fine for my MOMO, I get a fairly similar feeling to when I am driving GPL. Its not too strong really, although with your MSFF(?) the maximum force may be different. As a test you could simply change the strength (0.3 currently IIRC) to perhaps 0.1 or 0.2.

Quoted text (8 lines)

> > >a.. Problem fixed - tyre forces being applied in slightly the wrong > >direction > > Heh, don't I know about those problems. ;-) Got something like that I > think in my latest version (v050 beta 11). Severe car drag at > low-speed (perhaps a direction error when feeding the relaxation > formulae).

Axes, you cant live with em.....you cant live without em ;) > > One thing I noticed while glancing over a car.ini file is that the > relaxLengthLon > relaxLengthLat. Which seems odd, as mostly it's the > other way round (say L_lat=0.91, L=lon=0.091). > Might be a bug, but might be a butterfly after all. :) > From what I read from the paper (a while ago now!) was that the relaxation length was the amount of the circumference that it took the tyre to produce its "normal" (i.e. the force if we werent using the relaxation model). I thought that the longitudnal would take longer to get up to this speed, but thinking about it now - this effect is probably due to the flex in the tyre which would be greater laterally? I'll fiddle with the file and see what comes out :) All the best, Ash http://www.siroccoracing.com

I'll whip you online soon, just too damn busy right now. Will download, will practise and will leave you in the dust when we meet in Sirocco Cyberspace :) Mikkel Ashley McConnell <[email hidden]> wrote in message news:b5n5b6$3al$[email hidden]... > Hi Folks, > > As it seems to be a slow news day in RAS-land I thought i'd spam my newdemo

Quoted text (34 lines)

> :) > > **Note - The collision detection code isnt finished, so only the tyres > collide with the surface. This causes problems with walls / trees as the > tyre thinks it has been compressed by ~1000mm and so rebounds madly ;). > Shift-R quickly i'd suggest :).** > > http://www.siroccoracing.com > > Here are the improvements since the last version: - > > a.. Multiplayer - the start of a Client-Server based multiplayer system > a.. Multiplayer Chat > a.. Force Feedback - based around lateral tyre force > a.. New Car - the JTCC NSX modelled superbly by CD_QuikSilver > a.. Differential - basic adjustable viscous diff added > a.. Engine Braking - adjusted to be dependent on engine speed > a.. Problem fixed - error when calculating the speed of the engine > a.. Problem fixed - tyre forces being applied in slightly the wrong > direction > > If you call into VROC i'll turn on the multiplayer server. It currently > automatically tries to connect to the server, if you want to stop this set > "startasclient" in ./data/options.ini to 0. You can also create your own > server by changing startasserver to 1 in options.ini. I will upload the > dedicated server to the downloads section in a couple of hours. > > Have fun :) > > All the best, > Ash > http://www.siroccoracing.com > >

On Mon, 24 Mar 2003 16:24:15 -0000, "Ashley McConnell" <[email hidden]> wrote:

Quoted text (9 lines)

> >"Ruud van Gaal" <[email hidden]> wrote in message >news:[email hidden]... >> On Mon, 24 Mar 2003 14:33:41 -0000, "Ashley McConnell" >> <[email hidden]> wrote: >> Hehe, I think the LFS 0.2 demo would have done that, but it's still >> kinda quiet about it. > >Hopefully i'll get a chance to give it a run tonight :))

Hopefully I'll get a new computer before I try. ;) ... >> Nice! Do you already handle a bit of timing problems? (server time vs. >> client time); prediction, extrapolation and that sort of stuff? >> > >I have done a little of all of that, some of it myself and some from the >library i am using (http://www.rakkarsoft.com). Interesting, never heard of that. I have mostly based my (beginning) code on Quake source and Phil's HawkNL semantics. > It is by no means any good >yet, but I think i have the basis for a good amount of players. I will of >course have be drastically reduce the number of bits i use to send >everything, just want to get it working for the moment before looking at >compression. Any progress on Racer becoming multiplayer goodness? There is some working, but it doesn't interpolate yet (so an update frequency of 20Hz still gives shocky cars), and there are is no time infrastruture yet (I'd better think that through and understand it before I start, though I have some idea through a book about DX7 programming, not sure about the title anymore). > That >would be fun :). Good thing about starting multi is that replays and >ghost-cars should drop out fairly easily (simply a pretend network car!). Indeed. :) Although I do have replays, I am a bit hesitant on sharing the code for network play. In network play, bandwidth limits and compression start to matter, whereas for replays it's not that important. But still, it may be worthwhile to get the two mixed a bit; at least for event generation. A ghost car I would do with 2 pointers in the replay: the start and endframe. Still, there are enough syncing issues and drawing issues. First have to get a new gfx engine in, to avoid (gfx)code that gets thrown away later on. ... >Those settings are fine for my MOMO, I get a fairly similar feeling to when >I am driving GPL. Its not too strong really, although with your MSFF(?) the >maximum force may be different. They make a lot of difference; for the Mac I have a Logitech Wingman Formula Force GP, and that gives hardly tough forces. I like the MSFF in that respect. > As a test you could simply change the >strength (0.3 currently IIRC) to perhaps 0.1 or 0.2. Ok, thanks. I now & then think about a hardcore test; use a measuring spring and attach that to the wheel. Then apply a force to the wheel, and input the force that is measured into the computer, so the computer can really generate the 'exact' forces. Really simple, but a bit hardcore and for high-end only. But I discovered a bug in my FF code (still there), where I just add Mz of the front wheels together and use that for the FF force. I instead should divide by the steering ratio. I discovered that when I actually felt a high-end wheel that generated the exact force you feed it, and it turned out 5Nm is pretty much! :) (and my numbers are roughly around 40Nm). >> Heh, don't I know about those problems. ;-) Got something like that I >> think in my latest version (v050 beta 11). Severe car drag at >> low-speed (perhaps a direction error when feeding the relaxation >> formulae). > >Axes, you cant live with em.....you cant live without em ;) They s... eh, rotate all the time! ;-) ... >From what I read from the paper (a while ago now!) was that the relaxation >length was the amount of the circumference that it took the tyre to produce >its "normal" (i.e. the force if we werent using the relaxation model). Right. > I >thought that the longitudnal would take longer to get up to this speed, but >thinking about it now - this effect is probably due to the flex in the tyre >which would be greater laterally? Yup, compare longitudinal stiffness to lateral stiffness, or imagine pushing a tire sideways and longitudinally. The SAE950311 document also shows these values. The bigger lateral relax len would make response a bit softer. Ruud van Gaal Free car sim: http://www.racer.nl/ Pencil art : http://www.marketgraph.nl/gallery/

"Mikkel Gram-Hansen" <[email hidden]> wrote in message news:3e80ff37$0$52122$[email hidden]... > I'll whip you online soon, just too damn busy right now. > Will download, will practise and will leave you in the dust when we meetin > Sirocco Cyberspace :) > > Mikkel > > You probably will, I have been soundly thrashed in the past - 48secs!!! Thats Impossible! :) All the best, Ash http://www.siroccoracing.com

On Mon, 24 Mar 2003 16:24:15 -0000, "Ashley McConnell" <[email hidden]> wrote: ... >I have done a little of all of that, some of it myself and some from the >library i am using (http://www.rakkarsoft.com). It is by no means any good >yet, but I think i have the basis for a good amount of players. Damn, now you've got me all excited about network play again. ;-) RakNet doesn't have a Unix edition yet, but I'm helping them out with that. Seems like a nice library, but I've got a nice basis of my own, but still, interesting. I did some dead reckoning yesterday and this much improved the choppy multiplayer updates in Racer. Have a method on paper now which should give supersmooth warpfree holy-moly interpolation. :) Ah, you never know until you see it running ofcourse. It's based on the generic DIS concept (just to send pos,vel,acc and interpolate from there as pos'=pos+t*vel+0.5*acc*t^2), and then adds splines to warp smoothly upon packet updates (from a paper somewhere, I think it was Gamasutra). Interpolating the car's quaternions and restarting the spline halfway, possibly throttling the spline advancement (in case the car is lagging too much) should do the trick to deal with it all. Can't wait to implement it... Not really difficult either fortunately. > I will of >course have be drastically reduce the number of bits i use to send >everything, just want to get it working for the moment before looking at >compression. Any progress on Racer becoming multiplayer goodness? It was already working by now, just looked awful. Now updates are already looking reasonably playable, and wheel heights (no sinking into the track), wheel rotation (just derived from car velocity right now) and smoke/skidmarks are all done now (smoke derived from slipratios that I send). Ruud van Gaal Free car sim: http://www.racer.nl/ Pencil art : http://www.marketgraph.nl/gallery/