GPL Patch Readme (I'm disappointed)

Can't believe they didn't improve the collision detection! At least I didn't see anything in the readme about it. I guess it's safe to assume that any tiny collision may still result in one or both cars turning into a fireball leaping 500 feet into the air. Glad to see they improved the networking, but really, the poor collision detection was my main concern. Naturally improved networking may improve this indirectly, but it seems to me there are some fundamental problems with collisions regardless of how good your connection is. -A **** Posted from RemarQ - http://www.remarq.com - Discussions Start Here (tm) ****

> it >seems to me there are some fundamental problems with >collisions regardless of how good your connection is. I noticed the same thing and was disappointed too. Open wheel racing is no place for "trading paint" but the colossal online crashes that result from a tiny touch make that part of GPL one of the worse simulated in history, imho

I think the one part that made CPR actually quite good online.. was the collision detection that made it indusive for WARPING ONLINE RACING... THAT is was GPL needs.. the collision detection is GREAT for racing against the AI.. but for online racing.. it is WAY too touching... it needs to be relaxed... and then the online racing would be fantastic!!!! Oh well.. patch 1.2 ??? :-) James Pickard Melbourne, AUSTRALIA PS.. I still can't wait.. should be great...

Quoted text (15 lines)

> Can't believe they didn't improve the collision detection! > At least I didn't see anything in the readme about it. I > guess it's safe to assume that any tiny collision may still > result in one or both cars turning into a fireball leaping > 500 feet into the air. > Glad to see they improved the networking, but really, the > poor collision detection was my main concern. Naturally > improved networking may improve this indirectly, but it > seems to me there are some fundamental problems with > collisions regardless of how good your connection is. > -A > **** Posted from RemarQ - http://www.remarq.com - Discussions Start Here (tm) ****

> indusive ???? ~daxe -----------== Posted via Newsfeeds.Com, Uncensored Usenet News ==---------- http://www.newsfeeds.com The Largest Usenet Servers in the World! ------== Over 73,000 Newsgroups - Including Dedicated Binaries Servers ==-----

On 08 Jun 1999 15:15:32 PDT, "Larry Linthicum" >I noticed the same thing and was disappointed too. Open wheel racing is no >place for "trading paint" but the colossal online crashes that result from a >tiny touch make that part of GPL one of the worse simulated in history, imho Actually IMO that makes it one of the best simulated. The problem is that "Car A" and "Car B" are almost touching - if they touch a gentle collision will (and can) occur. However, a slight warp happens and "Car A" lands in the same position as "Car B". Obviously the collision energy to put it in that position must be enormous, and the engine responds in kind - that is wholly accurate as far as the elasticity of the collision and conservation of energy, and wholly inaccurate insofar as the way the cars moved (warped). What we need to do for simulating real life is introduce whatever "kludges" are necessary to get around the problems caused by latency and warping problems, and that is something that products like MTM, Powerslide etc do superficially very well. That _reduces_ the pure simulation aspect, although with the benefit of improved gameplay. I'd say the gameplay is more important, but I'm very wary in the manner in which many of the kludges are implemented. Everyone's opinion on this will vary, making it a particularly difficult area to address. Cheers! John

Quoted text (16 lines)

>On 08 Jun 1999 15:15:32 PDT, "Larry Linthicum" >>I noticed the same thing and was disappointed too. Open wheel racing is no >>place for "trading paint" but the colossal online crashes that result from a >>tiny touch make that part of GPL one of the worse simulated in history, imho >Actually IMO that makes it one of the best simulated. >The problem is that "Car A" and "Car B" are almost touching - if they >touch a gentle collision will (and can) occur. However, a slight warp >happens and "Car A" lands in the same position as "Car B". Obviously >the collision energy to put it in that position must be enormous, and >the engine responds in kind - that is wholly accurate as far as the >elasticity of the collision and conservation of energy, and wholly >inaccurate insofar as the way the cars moved (warped).

I have been thinking about this a little, and what appears to cause warps is that when there is an error between the way GPL has predicted the position of the car and the way having received a packet saying where the car actually was a short while ago (and predicting its position now from that?) it immediately rushes the car to the new position. I actually think it may be slightly clever than that (not sure on this one - Randy?) but it drives the car very rapidly but not instantenously to the new position (the reason I say this because I saw an on-line warp the other day where the car seemed to come past me at about 1000mph with a sound like wind). What would be nice would be that the sim engine was much gentler with those little warps (so when you see the skid marks they are not so jagged) by driving the car more carefully between the last predicted and new predicted/actual position. Alan http://www.chandler.u-net.com

I strongly disagree. I'm well aware of WHY these collasal crashes happen, but I still believe its possible to prevent them without "ruining" the physics model. At least that's what I believe until someone from Papy explains why it is not possible. -A **** Posted from RemarQ - http://www.remarq.com - Discussions Start Here (tm) ****

I think this brings us back to the need for the OnOffline type discussions we had months ago. It seems the Papy prediction is a little moronic in that it only predicts in straight lines i.e. if 2 cars are side by side in a corner and the inside car drops a few packets then it just goes straight into the outside car. I hope that in future that sims have some "benefit of the doubt" built into them. By this I mean that in the event of packet loss/major lag, that the cars would follow inside or outside lines much the same way as the AI does rather than just shooting off in a straight line as I am assuming they do in GPL. I think collision should only occur when the packets from both cars indicate that they have indeed occupied each others space. Are my assumptions right Randy? i.e. is the prediction simply linear tangents from the last know direction? This seems to be the case to me, especially thinking of how fishtailing cars off the grids seem to veer far further than they really do due to their orientation not necessarily being indicative of their direction. And the many times the outside car is taken out by the inside car in corners with the inside car not even realizing there was contact.

Quoted text (30 lines)

> On 08 Jun 1999 15:15:32 PDT, "Larry Linthicum" > >I noticed the same thing and was disappointed too. Open wheel racing is no > >place for "trading paint" but the colossal online crashes that result from a > >tiny touch make that part of GPL one of the worse simulated in history, imho > Actually IMO that makes it one of the best simulated. > The problem is that "Car A" and "Car B" are almost touching - if they > touch a gentle collision will (and can) occur. However, a slight warp > happens and "Car A" lands in the same position as "Car B". Obviously > the collision energy to put it in that position must be enormous, and > the engine responds in kind - that is wholly accurate as far as the > elasticity of the collision and conservation of energy, and wholly > inaccurate insofar as the way the cars moved (warped). > What we need to do for simulating real life is introduce whatever > "kludges" are necessary to get around the problems caused by latency > and warping problems, and that is something that products like MTM, > Powerslide etc do superficially very well. That _reduces_ the pure > simulation aspect, although with the benefit of improved gameplay. I'd > say the gameplay is more important, but I'm very wary in the manner in > which many of the kludges are implemented. > Everyone's opinion on this will vary, making it a particularly > difficult area to address. > Cheers! > John

> > indusive >???? >~daxe probably CONducive... :-) __ Put your message in a modem, and throw it in the ***-sea... remove SPAM-OFF to reply.

well, altho' marked "rare," inducive is in my _OED_ (but note the "c"): "tending to induce"

Quoted text (12 lines)

> > > indusive > >???? > >~daxe > probably CONducive... :-) > __ > Put your message in a modem, and throw it in the ***-sea... > remove SPAM-OFF to reply.