Performance and Magic Data

nice thread, you guys have outstanding ideas. A question for me would be, is it possible to use the failure editor in conjuction with the enhaced performance files, so that you have the chance to create more realistic failures for the teams in the seasons? i know that the failure editor works only with non championship races and championships, but it would be interesting to know, if it's possible to connect the performance files with the failure editor for a little bit more realistic

@ quickslick - I've never used the GP4 Failure Editor tool, but looking at the help file I wouldn't have thought so. It's more for editing saved games isn't it? The performance files have already finished their work by then.

yes, you're right, it's more or less for saved games. You can start a non championship race and then before the start of the race is, you have to save the race and then you can edit the failures for the drivers. So my question is answered

Noog Wrote:
-------------------------------------------------------

> I like it too.
>
> When you say the failure doesn't fit, do you mean
> that the total number of failures for a driver is
> very different to what it was in real-world, or do
> you mean the reasons for breaking down differ to
> those in the real world?

I met in some mods the car retires always. Ok some times it happens in the real world too, but this a game "what ever that means"

> Zandvoort 67 its just great. I put some ads for
> 79, I thing. And I little modify the pit
> entrance.
>
>
> Do you find the CC cars too slow at the last
> corner at Zandvoort? If you do, try changing
> desc51 (cc aggressiveness) in the MD file from
> 60024 (the default in mine I think) to 0.

I cant remember any more, but I thing was because the entrance was to short for breaking, or it was because is had much crashes if someone was overtaking and other was going to the pit(like patrese and Berger in Estoril 1992 or93).
Yes zou right the cc are too slow in the last curve. I will to try the desc51, maybe I dont need it to change the pit.

Thanks, huskyman49 and Noog for Bathurst I will test it.

The Kleine Nordschleife its one of my fav too, shame it can be 22KM the all Nordschleife. I like Pau, and Kylami 1982 too. There a lot of great tracks here,



Edited 2 time(s). Last edit at 12/23/2018 11:16AM by landex.

quickslick Wrote:
-------------------------------------------------------
> yes, you're right, it's more or less for saved
> games. You can start a non championship race and
> then before the start of the race is, you have to
> save the race and then you can edit the failures
> for the drivers. So my question is answered


Yes. It should create the failures during the race not before it starts. And the funny thing is: It determines the lap and the track sector when the failure appears. If the failure appears mor at the ending of the lap, its probability of getting to pits higher. In failure editor only stays the lap and the failure.
But, and thats why I love this games, if you can go to pit, you can repair it all, and the failure doesnt appears anymore along the race.(It could appear again, dont you thing?)

Failure rates and retirements.

If a car always retires and you want to change that, open the appropriate PF, find the team you want to change and adjust the failure probability. Look at the values the other teams are using and base an estimate on that if you like (staying within the range 0 to 24578 for a race length of 100%, according to PRBlanco's v1.2 spreadsheet).

That's the easy way and it should resolve your problem.

There are more accurate ways of going about things but it's trickier than it seems; basically because in the real world a car fails to finish for a number of reasons. They may have had a mechanical failure, crashed out or they may never have started the race at all! Unfortunately however, there's no way currently to tell GP4 to make driver X crash out more than driver Y, so at the moment I use the failure probability variable to account for all reasons, ie in a ten race season, if driver A broke down 5 times and crashed out 5 times the driver would initially be assigned 100% a failure rate, not 50% as it 'should' be.

Rene is working on something which might allow us to simulate a driver's crash likelihood in the future, but for now, I use the failure rate value to represent both mechanical failure AND driver failure - bearing in mind that some drivers might ACTUALLY crash out in the GP4 race already anyway, so you have to build that into your analysis as well. It gets quite complex when you start to delve into it.

One tip for now, I'd say, is that you can't assess the appropriateness of your failure values over one race, or even one season. You need plenty of data to work with because it takes a while for the trends to show through:

Consider Driver A with a low failure value and Driver B with a high one. Both could fail to finish the same number of times in any given season, but if you played the same season 10 times you'd definitely see the data converging around the probability determined by the failure rate, ie Driver A finishes far more often than Driver B, over time.

If you just want to correct that one race I'd recommend the easy way, covered above. You'll be close enough.

Zandvoort - Yes, I changed that pit lane entry behaviour too. I think I increased the desc67 (sectors to pit in) value from 0 to 8 in the MD file. The CC cars being too slow around the last corner is separate issue, which is what I was referring to earlier. Being a beta release there's a lot to fix at Zandvoort, but it was the circuit that first got me messing with MDs and I enjoy being able to drive it in the virtual world without fear of serious injury! Some of these old tracks were just wild and it's a wonder anyone lived to tell the tale.

I enjoy Kyalami '82 too, but I don't know Pau well enough to enjoy it yet, although it certainly looks good. I agree about the number of quality tracks in GP4. The community has been blessed with much talent.

Now I checked my Zandvoort to change the desc67 and realize that it has no MD in .dat. There are some tracks who has no MD. I use to load any MD-file, but this is no good. Is it a way to generate it considering the .dat?

Thanks about the PF retiring. I assume that(range 0 to 24578) 0-never retires, and 24578 always. Would you say 12289 its 50% of retiring?

There is something I forget to say and its very anoing, its the Black flags: They exists for Human car only. P.e.: if some sectors are black flagged and CC goes to thru the Bank at any velocity, it doesnt get any flag! This is really annoying.

A failure at the start could simulated by random program who puts fuel to 0

Yes, the community is bless with talents, iam still learn and far from it.

Merry Xmas and good healthy 2019 for every one!

@Noog could you pass me a link to Paulo's spreadsheet?

---

Some mods
F1 1996 | F1 2002 | F1 2007 | F1 2011 | F1 2013 | F1 2015 | F1 2018

Always retiring should be value 16384. It scales with the race percentage, so if you do a 50% race, use value 32768.

René Smit, Independent Software Developer.
Download my GPx tools here.

If you set value "0"? In fact I tried, but AI is so clever that withdrawals happen the same based on causal race accidents.



Edited 1 time(s). Last edit at 12/24/2018 01:16PM by mralpha.

SDI Wrote:
-------------------------------------------------------
> Always retiring should be value 16384. It scales
> with the race percentage, so if you do a 50% race,
> use value 32768.


Haaaa, it scales with % of the race. I always drive at 100%. I met if I want that car has a probility of retiring of 50% in a 100% Race, what should be the formula? Or what shoulde be a formula for 1x100%Race for Xpercentage?

SDI this explains what said before about the mods. If the PF of the mods were made for 50% race, and I always drive at 100%...

If the mods were made for 50%, you should cut the value in half for 100%.

chance% = max(20, race_distance%) * value / 16384

René Smit, Independent Software Developer.
Download my GPx tools here.

Thanks SDI, I will try it.

Nice to read that 'my' Zandvoort '67 beta version is still liked by some members. It's almost 11 (!) years ago that I created the track for GP4. At that time I was a n00b regarding cc-lines and magic data. I can create a better cc-line and a custom magic data set for Zandvoort '67 if there is a demand for it.

---

My Grand Prix 4 Files
I'm a total dick. How many people can say that?

@ Turbo Lover

Well that gets my vote. That would be a real treat!

Any chance you might talk us through how you'd go about that too? I'd love to learn track editing but I just don't get it; especially CC lines.

@ René

Ah-ha! Yes of course. 16384. I remember now. It WAS that originally in the spreadsheet. I must have changed it a long time ago and forgot I did it. My thinking at the time, IIRC, was that I was trying to account for the fact that some cars 'fail' by crashing, so I increased the value by 50% to compensate, which I then marked as 'default' in my spreadsheet because that's what it represented - but only to me!

@ Landex.

Firstly, René is absolutely right, of course. The correct figure is 16384 for a 100% race.

I use a 100% failure rate value of 50,000 for my 15% races. (Much longer than that and I start making mistakes!). The mathematically accurate figure would be 109227 (unless René says otherwise!), but mine is lower because I don't WANT a driver to retire in every race, even if they did in real life.

The 50,000 figure I use means that the least reliable cars will tend to fail in about about 8 of 17 races, but remember that they will sometimes fail with repairable ailments, like throttle etc, which WON'T be recorded as a failure to finish. Conversely however, they will crash out - let's say maybe twice a year on average - and those crashes out WILL be recorded as a failure to finish. The 50k figure @ 15% seems to give rise to an average of about 7 actual failures to finish per worst car per season, along with 5 or so races in which they had to pit to fix a repairable problem or a broken wing - all of which leaves them just five races to get some points on the board. It works for me, but as always I'm taking certain liberties with the historical accuracy again which might not suit some people!

3, Black flags

Don't know. I've noticed the issue with CC cars not getting one but generally speaking I'm the only one who tries to cheat!

4. Start line failures

An interesting idea. Cars stalling on the grid would be a nice touch; also if the CC cars jump started once in a while too. Perhaps if we asked René nicely...?


@ Kedy89.

I'm afraid not. I tried to find a link to it the other day but drew a blank. I thought I had an original somewhere but apparently not. The spreadsheet I use is an evolution of PRB's. I'd like to offer you a copy, but it's still very much in development (ie it's hard to use and some bits don't work as well as they might) and I don't have the time to support it out in the field. I definitely intend to produce a user-friendly version of it at some point but it'll be a while yet. I will try to get on with it though, because I'd like to push at least a little something back into the community some day.



Edited 2 time(s). Last edit at 12/24/2018 07:48PM by Noog.

@Noog: 109227 would be correct, but the maximum value is 65535 (the values are 16 bits).

René Smit, Independent Software Developer.
Download my GPx tools here.

Quote
SDI
@Noog: 109227 would be correct, but the maximum value is 65535 (the values are 16 bits).

So under 25% race length a 100% failure value can't work properly? How interesting, as always.

Indeed.

René Smit, Independent Software Developer.
Download my GPx tools here.

@Turbo Lover

Your Zandvoort67 its great! Like I said before its one of my favorites. I remember, I put some yellow kerbs and ads to make it like 1979(I think).

And I feel like Noog, I would like to learn building tracks like you.
My main issues are the pits and some graphics who appear and disappear(like Trees, Buildings) ingame.

Just one question: The pit end has to lower or higher than the track height?

@Noog

"I don't WANT a driver to retire in every race, even if they did in real life."
Me too, but that exactly whats happen in some mods. Not 100 but 80% of the races.

@landex: I also want to learn to build tracks, especially Sebring would be my first project. I would like to partner with others like you and share the work... I have already converted parts of Sebring from GTR2 and I'm still waiting for Paulos part two of his highly recommendable online tutorial to proceed: [www.youtube.com]