Too many wrecks at Bristol

[RaCeRKiD5]

Well its that time in my offline league where we race at Bristol. My problem isnt really the amount of wrecks as much as the amount of cars taken out. I would start the race off really well and then when one wreck comes it takes out half the field. Ive tried playing with literally everything in the track.ini and ai ratings, read all the tutorials and I am still dumbfounded. Any help is greatly appreciated.

 

[D0rant]

Are you using the papy ai by Revamped Reloaded? Because a few days ago I was running Dover 2013 on my offline race and the cars would literally wreck every 2-3 laps with the original papy ai in place, as soon as I replaced it with RR papy ai the race would be rather normal with green flag runs extending to 40-50 laps. You could also raise these lines in the papy ai file. driver_consistency_to_dlong_lookahead_dev = 5.6
driver_consistency_to_dlat_lookahead_dev = 5.75 (these are my values I don't remember the original ones but it seems like the cars manage to avoid big wrecks this way).

P.S. Those lines affect how far the AI sees a car that is stopped on the track. A value too high will cause even dumber wrecks though but I think 3 to 6 could be the right path. The usual values IMO are way too low as the AI seems to be blind and continiously wreck into a wrecked car because their field of view is very limited. By raising those values they will see further.

 

[InsideLine]

Someone please sticky all these bristol AI threads.......last time I checked the search bar is working fine..........................

 

[D0rant]

Ok, I've just had a race at Bristol 2012 Day FSE track and learned something new about the AI wrecking too much. The cars often would slide of to the wall by themselves and that would cause pile up crashes. The solution to this is the line in the track.ini:

ai_inverse_slipcurve_k = 0.75 ; how much slip angle AI need to corner. smaller = more, more = increased drag

the original value in the track was 0.18 which would cause the AI to randomly lose grip in the corner which is kind of realistic but since it's happening every 2 to 5 laps it becomes impossible to get a decent green flag run. I would suggest raising the value in your track.ini.

Note that the higher the value the harder it will be to spin the cars in front of you. Lower value will make the ai wiggle in the turns which is also pretty realistic yet isn't practical for racing. The value of 1 will glue the cars on the track so I would suggest to try something from 0.4 to 0.7. I've just ran a 150-lap Sprint Cup race under the value of 0.85 and I still managed to get into a few cars and cause two cautions, two more cautions came out in the second half of the race because of AI incidents which caused 7 cars not to finish the race (IMO pretty realistic). 4 cautions overall in 150 laps is what I like to see (it's better than having 15 cautions like it does without tweaking things)


p.s. Also make sure the ai_panic_decel value isn't over 10 as it can cause huge pile ups. My panic decel is 6.9.

 

[eX29]

I just tried a race online with a friend with ai_inverse_slipcurve_k = 0.5 which is the middle of what you suggested and the racing was a lot more fun. It resulted in three total cautions, one caused by my friend and the last caused by me for over driving a turn, along with the other due to a car blowing up and stacking up the field.

I did make other changes before BUT ended up with 16 cautions in a 50% race due to slower lap down cars starting on the inside and a few cars driving down pit road, back onto the track and straight into pit road over and over.

With the adjustment to ai_inverse_slipcurve_k line it made it more enjoyable overall with long green flag runs.

Thanks! (y)

 

[D0rant]

No problem, I'm glad it helped

 

[J.R.Franklin]

I don't know what versions of Bristol 2012 FSE that you guys are referring to. The values in the track.ini for Bristol 2012 FSE and Bristol Night 2012 FSE for "ai_inverse_slipcurve_k =" were set at 0.23. They were never set as low as 0.18 or as high as 0.75, so that would lead me to believe that you have either tried altering those values yourself or got an altered version of the tracks from some other source than Sean and Fred's site.

The following is a quote from the "IMPORTANT-Bristol A.I. README.txt" which I wrote and included with the track:

The A.I. for this track were constructed using the following:

1. a default papy_ai.ini,
2. the default fast setups for each physics (CUP, GNS and CTS)
3. the opponent strength set at 97%.
4. Ratings for the car sets used in testing and in developing the A.I. for this track are from the NRatings program using the MasGrafx Full Season V3.1 formulas.


The track was extensively tested by numerous people prior to release and no one reported the problems you are experiencing. Therefore I would guess that the excessive crashing you're experiencing is a result of not adhering to items #1 and #4 above.

If people who have downloaded the Bristol 2012 FSE tracks are not experiencing the excessive crashing issues that are being reported by those in this thread then I'd highly recommend that you not make the suggested changes that are being recommended by them.

If you are experiencing issues with the track, then I would suggest that you thoroughly read the README about the A.I. that came with the track first before fiddling around with the track.ini settings.

 

[D0rant]

Perhaps you're right about the value not being at 0.18, I guess a few weeks ago I lowered it myself to 0.18 and then wrongly claimed it was the original value. Please note I am using 2012 FSE Bristol tracks, I don't know if there's the 2013 ones. Anyway, I was using both default papy ai and revamped papy ai and the result was the same, the cars would slide into the wall by themselves, at first I thought it was the the squeeze pcnt problem, I lowered it to a 0 but the problem repeated itself again and again until I put in a higher value in the inverse slipcurve and it was fine ever since then

P.S. And I was using the CUP Masgrafx formula just like I always do (I'll use Ethone's formula for SS and short tracks from now on). My opponent strength was set at 90 to 93 as I'm not really good at Bristol and I always use the fast setups as I find those comfortable to drive.

 

[J.R.Franklin]

Perhaps you're right about the value not being at 0.18, I guess a few weeks ago I lowered it myself to 0.18 and then wrongly claimed it was the original value. Please note I am using 2012 FSE Bristol tracks, I don't know if there's the 2013 ones. Anyway, I was using both default papy ai and revamped papy ai and the result was the same, the cars would slide into the wall by themselves, at first I thought it was the the squeeze pcnt problem, I lowered it to a 0 but the problem repeated itself again and again until I put in a higher value in the inverse slipcurve and it was fine ever since then

P.S. And I was using the CUP Masgrafx formula just like I always do (I'll use Ethone's formula for SS and short tracks from now on). My opponent strength was set at 90 to 93 as I'm not really good at Bristol and I always use the fast setups as I find those comfortable to drive.

Oops, I typed 2013 out of force of habit. The tracks really are 2012. I've corrected that above.

I'm not really sure why you're experiencing something that most users are not. It could also be a result of you using the altered papy_ai from Revamped Reloaded rather than the original/retail papy_ai.ini. We never really tested the tracks with the altered papy_ai.ini from Revamped Reloaded, so I don't know what the a.i. would behave like using that one instead. But for the most part, when people use that Revamped Reloaded papy_ai.ini it introduces problems with other tracks that are not RR tracks.

The behavior of the a.i. sliding up into the walls going into the corners sounds like you may have raised the "ai_grip_modifier" from the 1.02 setting we had to some higher value.

The other thing that sometimes causes the a.i. not to be able to hold themselves down on the Lp lines is increasing the "ai_squeeze_pcnt" to a higher value than the 0.70 setting we had.

Outside of those 2 things, I'd have to believe that it is more a ratings issue and a track issue. Maybe the aggression ratings are set higher in your carsets. I just don't know, but most end users are not seeing the problems you are having. Sure hope you can figure it out because the tracks are about the best and most accurate versions of Bristol ever made for NR2003 and the racing experience against the a.i. are a flat-out blast.

 

[D0rant]

I'm definititely not whining I had these issues and if others do not it's perfectly fine. Having spent over 7 years with this wonderful game I've learned some tricks to make the ai act exactly the way I want to and I have already fixed the issues with the ai I had. And yes the 2012 Bristol tracks are wonderful, it's like I'm really there, great detail!

P.S. And I'm aware of your work when it comes to updating this game and I'm very thankful for it

P.S.S. Most of the modern tracks (FSE, Smiffsden, RR) work great with revamped reloaded papy_ai. I tweak some things here and there but overall it's fine.

 

[starsxxx]

p.s. Also make sure the ai_panic_decel value isn't over 10 as it can cause huge pile ups. My panic decel is 6.9.

For the panic_decel, It kind of depends on your ai_dlongpad_scale and the ai_decel_grip. I don't know a lot about the dlongpad scale. I know it gives a different result of the panic decel.

 

[Hendrick4life]

Dude....

 

[MattyO]