Latest resources

2021 Fictional/Concept Paint Schemes Pack #1 DAT Designs
A collection of my fictional/concept paint schemes designed for the 2021 NASCAR Cup Series season
5.00 star(s) 1 ratings
Downloads
38
Updated
2018 Formula 1 Carset BryceO84
2018 Formula Carset for the FORMULA 1 Mod by VSR.
0.00 star(s) 0 ratings
Downloads
17
Updated
66 - CrashClaimR.US Camry - CAL daniele
5.00 star(s) 1 ratings
Downloads
73
Updated
66 - Unsponsored Camry - VEG1 daniele
5.00 star(s) 1 ratings
Downloads
67
Updated
61 - Pit Boss Camry - PHO1 daniele
5.00 star(s) 1 ratings
Downloads
81
Updated

Problems with ROVAL 2018 even with the patch.

D0rant

Active Member
SRD Member
Messages
188
Reaction score
177
It must be poorly done .LP files. There's only so much track.ini can fix. LP is what sets the norms of behavior. If those are corrupted from the get-go, then the AI never stood a chance to behave and race properly.
 

TChapman500

Quit SRD
SRD Member
Messages
296
Reaction score
465
Definitely an LP problem. At least we know where the trouble spots are. If I had to guess, I would say that the min and max race lines are too close to each other in the chicane and that other trouble spot. The crashes at the chicane probably caused the crash in that other location. Also, the chicane is really only for single-file traffic and the AI don't really have any concept of backing-out. So a tweak to the AI lines may only slightly mitigate the problem. You will still have the issue of what if two cars go through it double-file followed immediately by a car going through it single-file?
 

D0rant

Active Member
SRD Member
Messages
188
Reaction score
177
They generally follow the race.lp when they are running single file (unless squeeze percentage is tweaked in track.ini to promote more random AI lines) but when they run side by side, to understand the racing room they have they rely on the min and max race .lp files. If those were made too close, every time when the AI runs side by side they think they run out of room and slam into each other.
 

RJMik

Active Member
SRD Member
Messages
83
Reaction score
33
Lately I've found that on road courses the ai_drafting_distance is often the culprit for multiple crashes. My version of the early access roval has 1.25 as that value. Try going up to 1.35 and see if that helps, then 1.45 if it doesn't. It may not fix this particular problem but it's worth a try.
 

canadienhits

The Dominator, Cup champ.
Staff member
Paint Crew
Messages
13,383
Reaction score
9,081
Definitely an LP problem. At least we know where the trouble spots are. If I had to guess, I would say that the min and max race lines are too close to each other in the chicane and that other trouble spot. The crashes at the chicane probably caused the crash in that other location. Also, the chicane is really only for single-file traffic and the AI don't really have any concept of backing-out. So a tweak to the AI lines may only slightly mitigate the problem. You will still have the issue of what if two cars go through it double-file followed immediately by a car going through it single-file?
Pretty much, as you see, one car usually slows for the chicane, and the other one is barreling in and either sideswipes it or hits it square on the back, either resulting in both cars spinning.
 

RacerXero84

Obnoxious old fart
SRD Member
Messages
3,921
Reaction score
5,104
Pretty much, as you see, one car usually slows for the chicane, and the other one is barreling in and either sideswipes it or hits it square on the back, either resulting in both cars spinning.

I'm thinking the LP race file was created with someone blasting through that area, so when 1 car has the preferred line they think they can zip through, but get's caught up with the 2nd car that doesn't and needs to slow in order to make the series of turns. I'm thinking a new race LP with a more moderate attack on that section (and maybe more conservative in other hazard zones as well, possibly throughout the entire course) would alleviate the issue, and from there just play with the AI grip and drag in the INI to augment the difference to obtain a proper level of speed through those zones and make it more uniform.
 

TChapman500

Quit SRD
SRD Member
Messages
296
Reaction score
465
The speed at which the player recorded the LP is not the speed at which the AI navigate the track. That speed is ONLY for a visual reference from raceline.lp. As far as I can tell, the AI will follow the LP as fast as they can with the setup that they're given. The rest is from the AI difficulty as well as the INI values. But we do need to redo at least parts of the AI lines (the in-game editor has a splice feature to replace part of an LP line with new position data).
 

D0rant

Active Member
SRD Member
Messages
188
Reaction score
177
The speed at which the player recorded the LP is not the speed at which the AI navigate the track. That speed is ONLY for a visual reference from raceline.lp. As far as I can tell, the AI will follow the LP as fast as they can with the setup that they're given. The rest is from the AI difficulty as well as the INI values. But we do need to redo at least parts of the AI lines (the in-game editor has a splice feature to replace part of an LP line with new position data).

Not exactly true. The .LP files do not store information on the speed/throttle/brakes used while recording the lines EXCEPT the race.lp which does require "Player's/Recorder's" performance to be correct in terms of where to brake, where to let go of the throttle etc.
 

MWick4

SRD SuperTruck Series Director
SRD Member
Messages
2,636
Reaction score
2,387
Honestly John, I can’t remember if I used the updated LP files. Maybe reinstall the track and lower AI aggression to like 95 (I think that’s what I have it at) and don’t use the updated files. It gave me some trouble during testing, I may have just gotten lucky with the truck series race tbh.
 

fortine_oo

Senior Member
SRD Member
Messages
727
Reaction score
711
As usual with ai, the troubles are not the result of one or two maladies, but many. Further complications arise when their interaction compounds the trouble.
I haven't seen/experienced the trouble to anywhere near the degree of canadienhits. For me, if the ai get through the first lap cleanly (and they usually do), the ai spread out single file and there are no incidents thereafter. I never went more than 10 laps because the ai were just too spread out to create contact.
I suspect the ai wrecking for canadienhits is mostly caused by the ai ratings for the cars in the roster.
[ In close racing (and especially at plate tracks), any ai with less than a 50 minimum rating (for DRIVER/TRACK TYPE, DRIVER/CONSISTENCY, VEHICLE/CHASSIS) is likely to be involved in an accident when the exe selects/applies a rating below 50.] AI Driver Ratings - General Information

That's not to say there aren't deficiencies in the ai development.
There is a problem with the lp's, but T17 is the real problem. Due to the track construction/segments, the ai are being pulled to cut the corner and they are prone to losing control due to the curb and the grass. Plus, they shouldn't really be running there anyway if you're looking for any kind of realism. That racing line also allows them to undercut the Player, thereby giving an unfair advantage to the ai. ]

There are track.ini values that need some attention, also.

The download contains a lightly edited track.ini (edited lines noted by: ++++++++++++++++) to discourage wrecks and increase passing, and a little tweaking to even up the ai with the Player. Plus, lp's.
 

Attachments

  • Charlotte Roval_Early_14 mini edit.7z
    92.3 KB · Views: 53

canadienhits

The Dominator, Cup champ.
Staff member
Paint Crew
Messages
13,383
Reaction score
9,081
Thanks fortine. Glad you are weighing in on this.
 

canadienhits

The Dominator, Cup champ.
Staff member
Paint Crew
Messages
13,383
Reaction score
9,081
Ok, after a lap 1 crash, everything ran smoothly. Thanks all.
 

fortine_oo

Senior Member
SRD Member
Messages
727
Reaction score
711
Ok, after a lap 1 crash, everything ran smoothly. Thanks all.

Where was the crash?
Strictly ai or were you involved?

I ran 2 races in three separate game starts (42 ai, Strength Manual at 100%), i.e. I started a race session and ran 3-5 laps, then started a new race and ran 3-5 laps, then exited the game, started a new race session, etc.
Never had any collisions.

Did you check your ai ratings?
 

canadienhits

The Dominator, Cup champ.
Staff member
Paint Crew
Messages
13,383
Reaction score
9,081
Exiting the chicane, strictly the AI. But after the caution it ran green the rest of the way. My ai ratings are real life cup drivers MAxgrafx Cup ratings.
 

D0rant

Active Member
SRD Member
Messages
188
Reaction score
177
I like the Dooper and MasGrafx Cup formulas cause it helps to spread the field better but at the same time for the road courses I would suggest using some other formula because IMO they spread out too much. Not to mention that the mentioned formulas have low ratings for poorly performing cars and drivers (which means they are more likely to screw up and do something stupid) that might be the reason you have those crashes.
 

fortine_oo

Senior Member
SRD Member
Messages
727
Reaction score
711
Exiting the chicane, strictly the AI. But after the caution it ran green the rest of the way. My ai ratings are real life cup drivers MAxgrafx Cup ratings.

I looked a those ratings while trying to remedy problems at some track.
I don't understand why the "real world racing" Roster Rates require the Driver/Track-type ratings to be so low. Most if not all of the Track-type minimum values are below 50, and even some of the maximum values are below 50. This low-balling has little or no impact on the final results and leads to unnecessary accidents.

I use ai ratings that give me repeatable results in levels of performance in order have some predictability in developing ai with Player-like tendencies.

If you're looking for a wide variety of winners and finishing orders, I'd suggest using a ratings set developed by GurtonBuster (used in the 2015 HTC Series v2.0 carset - http://www.simracingdesign.com/threads/2015-htc-series-v2-0.57401/).
For more information: ai Ratings Roster ALTERNATIVE
 

D0rant

Active Member
SRD Member
Messages
188
Reaction score
177
I looked a those ratings while trying to remedy problems at some track.
I don't understand why the "real world racing" Roster Rates require the Driver/Track-type ratings to be so low. Most if not all of the Track-type minimum values are below 50, and even some of the maximum values are below 50. This low-balling has little or no impact on the final results and leads to unnecessary accidents.

I use ai ratings that give me repeatable results in levels of performance in order have some predictability in developing ai with Player-like tendencies.

If you're looking for a wide variety of winners and finishing orders, I'd suggest using a ratings set developed by GurtonBuster (used in the 2015 HTC Series v2.0 carset - http://www.simracingdesign.com/threads/2015-htc-series-v2-0.57401/).
For more information: ai Ratings Roster ALTERNATIVE

Generally the reason people use those formulas is that they help to spread out the AI field quicker, thing is that can also be achieved via papy.ai but a lot of people do not know that, so in order to simulate more realistic racing where the AI don't generally race in packs 12-15 laps into a race (at least with the older NASCAR packages that never happened for sure in real life) they use the first method.
 

canadienhits

The Dominator, Cup champ.
Staff member
Paint Crew
Messages
13,383
Reaction score
9,081
Where was the crash?
Strictly ai or were you involved?

I ran 2 races in three separate game starts (42 ai, Strength Manual at 100%), i.e. I started a race session and ran 3-5 laps, then started a new race and ran 3-5 laps, then exited the game, started a new race session, etc.
Never had any collisions.

Did you check your ai ratings?

Generally the reason people use those formulas is that they help to spread out the AI field quicker, thing is that can also be achieved via papy.ai but a lot of people do not know that, so in order to simulate more realistic racing where the AI don't generally race in packs 12-15 laps into a race (at least with the older NASCAR packages that never happened for sure in real life) they use the first method.

Actually the reason I use the Maxgrafx Cup formula is that it's in NRatings, and it seems to be the best at simulating real life drivers for a full season.
 

fortine_oo

Senior Member
SRD Member
Messages
727
Reaction score
711
Generally the reason people use those formulas is that they help to spread out the AI field quicker, thing is that can also be achieved via papy.ai but a lot of people do not know that, so in order to simulate more realistic racing where the AI don't generally race in packs 12-15 laps into a race (at least with the older NASCAR packages that never happened for sure in real life) they use the first method.

Two things:
1) "... the reason people use those formulas ..."
I think people use those ratings because they believe those ratings are "realistic", and they have no desire to create their own ratings.

2) "... via papy.ai ..."
Are you referring to the papy_ai.ini ?
I investigated editing many of those lines. I found little improvement there, and even if there was some improvement, it didn't apply to all tracks, in fact, it was sometimes detrimental to the point of being a problem. Anything that needs to be adjusted for any track can be accomplished through the ptf, track.ini, and lp's, barring the NR2003 idiosyncrasies locked in the code/physics engine.
 
Top