Number of crashes???

Have one question for a looong time.
There are a lot of freaky "fly out cars" on some of circuits at the same turns.
Is there any way to become the number of crashes like them progressively smaller?

==========================================================
NATAZONE OFFICIAL WEBSITE - ||| now.by.ru |||
NATAZONE'S GP3-2000 PAGE:
||| gp3.now.by.ru ||| or ||| natazone.narod.ru/gp3.htm |||
NOW AZLK 2004 with Gordeev and Panin

I don't know.
I just HATE how often the cc cars go off track at that 2 fast corners at Sepang, before uphill hairpin. Maybe it's a cc line problem, I don't know.

See ya

Hmm... Not only in Sepang! Hungaroring, San Marino... A LOT of tracks!!!
Does anyone know the reason & how to solve this problem? Thanx!

==========================================================
NATAZONE OFFICIAL WEBSITE - ||| now.by.ru |||
NATAZONE'S GP3-2000 PAGE:
||| gp3.now.by.ru ||| or ||| natazone.narod.ru/gp3.htm |||
NOW AZLK 2004 with Gordeev and Panin

This sort of thing happens at the Indianapolis, Buenos Aires, Fundidora Park, Jacarepagua, Paul Ricard circuits, and a number of other GP3 tracks, but certainly not at most of them, and specifically not at the tracks that either of you have mentioned. Is it that a couple of CC offs during a race are considered to be "a lot" by you but ones that occur on almost every single lap don't bother you?

Anyway, it is a CC-line problem.

Hi
Do you mean when the cc-cars crash on the barriers by themselves?

In many tracks I noticed this strange effect and I sadly use to call it "the kamikaze syndrome". By this effect every race 12 cars or more, while aproaching a turn instead of slow down and brake they push the accelerator hiting the barrier at the max speed. This happen massively in my game in Buenos Aires, Suzuka and in some other tracks... where everytime only 6 cars finish the race.
The last one is my Welkom track where the first turn and the other hard brakes use to be a car-parking after 40 laps.

This happen also in straight sections. In this case cc-car can control again the situation. So it shouldn't be related to cc-line or coaching cc-car command because these are related mostly to collisions without a meaning between two cars entering a turn.

Making some tests on the magicdata in these days I noticed an interesting value marked with the arrow that seems to be the origin of the shame, but I need someone expert with magicdata to confirm my theory:

512 ; cc random perf range min
2048 ; cc random perf range max
80 ; some chance. same chance multiplied with race perc. <------------
64 ; segment count, range 1 to 256
7 ; sectors to pit in 1

I can assume that this value is a sort of frequency when a cc-car loose control for a while to simulate the error of the driver at breaking (?). More high is the value more times happen that a car loose control.
In many tracks this value is, let's say high i.e. 80. Well, reducing it to 10, maybe it is a coincidence but I noticed that the number of retires by the previous effect is massively decreased!
I tried to make many races at Welkom, Buenos Aires and other tracks changing that value to 10 and 80 many times and everytime it was set to 10 there were less number of retires by accidents.


I also noticed that increasing the cc shuffler, the fastest lap is not done at the second lap only or into the first 5, but is improved all along a stint, til the cars enter the pits for refueling.
In almost all magicdata this value is very low 1000-4000, but increasing it to 15000 or more the result seems to be similar to the reality.

15892 ; fuel consumption 1
15865 ; fuel consumption 2
18621 ; cc shuffler??? <------------

Major Tom once told me that it is the range between the fstest and the slowest car on track. Anyway it seems to simulate something similar to the lost of weight in race. by this now is possible to apply the pit strategy like in the real races because delayng the stop of 2-3 laps, a car can overtake another already refueled.


I don't know if these particular are already known or anyway correct so I hope someone can check them and tell...

bye



Edited 2 time(s). Last edit at 04/26/2004 07:46AM by R_Scandura.

Real big thanx to Ricardo for a good & complete answer!


==========================================================
NATAZONE OFFICIAL WEBSITE - ||| now.by.ru |||
NATAZONE'S GP3-2000 PAGE:
||| gp3.now.by.ru ||| or ||| natazone.narod.ru/gp3.htm |||
NOW AZLK 2004 with Gordeev and Panin




Edited 1 time(s). Last edit at 04/26/2004 06:22PM by Natazone.

Unfortunately this does not solve the problem.
It does alleviate it a bit on some tracks and considerably more on othere, but the problem remains in effect on all tracks that I tested (eight of them) to quite a large extent, even when the "AI error factor" is reduced to zero. At Indianapolis, for example, half the field nearly always crashes out of each race, while the rest nearly all have offs at the same corner leaving the entire surviving field strung out over many laps by the end of a full length race. It really does seem to me to be that the CC-line has to be altered in most of these cases, most likely to make the AI brake earlier for the problem corners.

desc65=some chance. same chance multiplied with race perc.

this is AI error, I had always thought that it maybe connected to the AI crashing out
in gp3, this was confirmed to me when I was so pissed off at all the crashes at suzuka
in gp4 I tried changing it to zero and it solved the problem, no crashes.

I've never had any problems setting desc65 to zero, so it's strange the problems
you're getting with the cars crashing out JM, do you let the AI qualify before running
the race, I've found it improves the AI mistakes especially in the opening laps.

____________________________________________________________________

desc66=segment count, range 1 to 256
in gp2 this worked really well, it's the distance the AI waits to enable a safe spin
recovery, rather than just ploughing back on the track, I've never had any luck with
this one in gp3.

__________________________________________________________________________

cc shuffler is an old description I used in gp2 days, it works slightly differently in
gp3/gp4 and slows down the AI pace as the race progresses, it's only really noticeable
with 100% races, values above 8192 and below 1000 should be avoided as you will
start to get weird results, it effects qualifying too. a value of 1000 should make the AI
competitive throughout the whole race.




hockenheim 1999 - 2003



[www.belini.fsnet.co.uk]

"desc65=some chance. same chance multiplied with race perc.

this is AI error, I had always thought that it maybe connected to the AI crashing out
in gp3, this was confirmed to me when I was so pissed off at all the crashes at suzuka
in gp4 I tried changing it to zero and it solved the problem, no crashes."

I'm discussing GP3 here, just so there is no misunderstanding.
Set this number to zero and run some full-length races at Fundidora Park, Indianapolis, Jacarepagua, or (the Pen's) Paul Ricard, and you will see the AI repeatedly running off the circuit at the very same corners again and again, less than they do when the number is higher, but still much too much. It would be really terrific if the problem could be solved this easily, however, wishing that it would will not make it so. Even some original Microprose tracks are not above reproach here and do appear to need some CC-line improvement.

there's no misunderstanding, most of the magic in gp4 is the same as gp3 anyway.

of those tracks mentioned I only have the gp32k indy original, desc65 @ zero
as you can see, one DNF at indy, and the suzuka I'm building at the moment had
no DNF's.

I'm probably getting better results as I did the tests running the belini grip
patch which does make it harder for the AI to screw up, but since I discovered
the use of desc65 in gp4 I've always applied the same to gp3 tracks with no
problems and much lower crash rates even without the patch.

poor line can cause big problems, straying too far from the values used in
the original magic can also screw things up.

But I think the main culprit is the slipstreaming effect which is too
advantageous and creates an unreal amount of overtaking opportunities.












work in progress
sakhir 2004



[www.belini.fsnet.co.uk]

The crashes that I am discussing are almost always confined to one particular corner on a track and generally occur in the midst of a turn and regardless of whether the AI are alone or in the close company of others, so slipstreaming is not a factor here at all.

I do generally have much lower track grip values in the magic data than the original GP3 values, so as to achieve realistic laptimes and cornering speeds and this may have something to do with the different results that you and I are getting, but as this has no adverse effect at all on most tracks, but only on generally one particular corner in the tracks that I mention above and a few others, I really think that therefore it's more of CC-line problem.

that does sound like a line problem if they're going off on there own but if
you've made the grip too low that will also contribute.


desc42=track grip
values below 16000 and above 18500 cause problems

desc51=added to car.field_108. related to cc grip factor desc52
basically AI grip/cornering ability values shouldn't be
lower than 64512 maximum=0 (65536)

desc52=grip???
min=128 maximum=768 any other values can give strange results

desc54=?? cc grip factor ace
try and keep this @ 500+

desc74=timing factor? (dw)
as you probably know is pole time

desc75=timing factor finetune? (dw)
this is what you should use to correct the lap times, keeping the
grip higher and adjusting desc75 should stop the AI falling off, if
this fails then it's the track.




work in progress
suzuka 1999 - 2003


[www.belini.fsnet.co.uk]

Excuse me Belini, but how do you mean "pole time" (the desc74)? How does that value affect pole time?

And please don't decrease the "desc65" too much in your phenomenal tracks, cause drivers are supposed to make some errors during the race

Ah yes, and can you please tell us how can we reduce the effects of slipstreaming in GP32k? Namely I don't use the tracks you made right now, for my PC is still too lame for that (I plan to upgrade it just so I can try your fabulous works! ) and since the belini grip patch doesn't work well with magic data inside original tracks, which values should we alter then? THX

desc74=timing factor? (dw)

the only thing I know for certain is that it's used for calculating multiplay
turns and fuel, but it does have other effects on the way the game runs.

1m 18.201 = 78201

_______________________________________________

you can program the magic to work with the belini patch and physics, but
some things are different

desc51=added to car.field_108. related to cc grip factor desc52
is the ONLY value that changes the AI grip

contrary to what I've written earlier this needs to be lower when using the
patch, anything from 63000 to 65536. the exe is patched to make up the
deficit along with an internal setting for player grip.

desc54=?? cc grip factor ace
now alters both player and AI grip




hockenheim 1999 - 2003


[www.belini.fsnet.co.uk]

1m 18.201 = 78201
aha I get it, it's the pole time. Moreover, I have found out that if you play a little with the "timing factor?" and the "timing factor finetune?" you can get a realistic pole time at such tracks as original Monza 1998, or original Silverstone 2000 for example. Those pole times are unrealistic because it rained in qualifying and they are way off of what they should be in dry conditions. Seems that GP3 creators didn't bother to notice that at all, they just "copied" the times that were achieved in reality so basically those laptimes turned out to be so realistic that they are actually unrealistic

(Hope you can understand what I'm saying, I know my English is a little raw...)

Edited 1 time(s). Last edit at 04/28/2004 04:49PM by Speed_Addict.

I have a question for all you GP3 masters. In a few tracks (Schneeburg 1999 and Bern Bermangarten for example) my car is fast for the first lap of the race. Then it loses 2 to 3 seconds, and the car slows down a lot, while the CC cars keep their pace, or better it. What could be the problem?

---------------------
Quote: Joey

Quote

I think Bush will win. I also think that when he does, I will start looking at properies in other parts of this world. Kerry's campaign has been so inept. Despite the fact that people are now held in police custody without ever being charged, 2 million people have lost jobs, the rest of the world hates America, and the line between church and state has never been blurrier, this whole ****ing election is about semantics and John Kerry being a flip-flopper. Americans are ****ing morons, and the Democrats running campaign are among this group.

hello!

i change the values in failures for magic data.

I input the value "0" to all failures but four cars failed in the race.

As I control those abandons?
It has as I do that? or only moving in the options of the game?




thanks