Latest resources

2020 ARCA Menards Series East Carset BryceO84
2020 ARCA Menards Series East Carset for the Gen6BR_cts mod.
5.00 star(s) 3 ratings
Downloads
31
Updated
18 - Monster Energy Supra - BRIS1 Hendrick9Fan
5.00 star(s) 2 ratings
Downloads
40
Updated
18 - Monster Energy Supra - CLT1 Hendrick9Fan
5.00 star(s) 2 ratings
Downloads
36
Updated
18 - Monster Energy Supra - DAR1 Hendrick9Fan
5.00 star(s) 2 ratings
Downloads
33
Updated
18 - Monster Energy Supra - PHO1 Hendrick9Fan
5.00 star(s) 2 ratings
Downloads
34
Updated

Has anyone ever played with these lines in the papy ai.ini file?

DaleTona75

Unhealthily Obsessed With Backmarkers
SRD Member
Messages
382
Reaction score
492
I've recently been poking around in the papy_ai.ini file and have found some interesting things.

Now I know that messing with the papy_ai.ini file can have serious adverse affects to the driving experience, and that ratings/track.ini should be the main focus for tweaking AI's, but there are some things I want to know if anyone has a clue what they do.

First off is under the behavior section where you find these lines.
crashing_recovery_alt = 0.6 ; altitude threshold below which will attempt recovery from crashing mode
crashing_recovery_pitch_roll = 0.96 ; safe pitch/roll velocity v. surface normal cosine for recovery from crashing mode
crashing_recovery_slip_angle = 0.35 ; slip angle threshold (~ 20 degrees)below which will attempt recovery from crashing mode
crashing_recovery_yaw_velocity = 0.05 ; yaw vel. threshold below which will attempt recovery from crashing mode

I have a hunch that this may have an effect on how easily drivers give up/how hard they try to save the car/stay in the throttle during a spin/wreck, but I would like to know from a more experienced source.

Then in the driver tab, I found these, which REALLY intrigue me.
ai_strength_to_traction_circle = 1.0
ai_strength_to_aero_drag = 0.3
ai_strength_to_engine_power = 0.2
ai_strength_to_traction_circle_spread = 8.0
ai_strength_to_engine_power_spread = 4.0
ai_strength_to_aero_drag_spread = 4.0
driver_qualifying_to_traction_circle = 0.02
driver_qualifying_to_aero_drag = 0.03
driver_qualifying_to_engine_power = 0.015
driver_roadcourse_to_traction_circle = 0.03
driver_shorttrack_to_traction_circle = 0.03
driver_speedway_to_traction_circle = 0.03
driver_superspeedway_to_traction_circle = 0.03
vehicle_aero_to_aero_drag = 0.13
vehicle_engine_to_engine_power = 0.07
vehicle_chassis_to_traction_circle = 0.05
ai_strength_to_driver_consistency_spread = 3.0
driver_consistency = 0.95
driver_consistency_to_driver_commitment_dev = 0.03
driver_consistency_to_dlong_lookahead_dev = 0.4
driver_consistency_to_dlat_lookahead_dev = 0.75
driver_consistency_to_left_offset_dev = 0.08
pitcrew_speed_to_pitcrew_speed = 0.1
pitcrew_consistency_to_pitcrew_consistency = 0.9
driver_finishing_to_driver_consistency = 0.4

Ratings tests by people such as fortune_oo have found that certain driver ratings catagories have more influence than others, and I've always wondered why and how? If these numbers were to be changed, could you make track type ratings as strong/important to a drivers performance as engine ratings?

Finally under the mechanical section I found these.
mechanical_failure_chance = 12.0 ; chance in 10000 (!!) per ACTIVE AI car for induced problem/failure each interval
mechanical_failure_interval = 540.0 ; average interval in ticks to check for mechanical problem/failure (gets randomized)

Again going back to fortune_oo's ratings tests, he found the reliability has a negligible effect on a drivers retirement rate. I have played with this along time ago, and recall having more mechanical related DNF's by editing these lines. Could this be the key to emulating the high dropout rates that were common in the 70's/80's?

Do any of you have more concrete knowledge of/experience with what these do? I'm interested in trying to learn as much as I can about how this game ticks and what I can change to manipulate racing, so any knowledge is good knowledge.
 

casssp

Active Member
SRD Member
Messages
186
Reaction score
94
Well, in terms of mechanical failures I got to mess the .ini and have some experience. How I play 10 to 20% of laps, I made my own configuration:

mechanical_failure_chance = 5000.0 ; chance in 10000 (!!) per ACTIVE AI car for induced problem/failure each interval
mechanical_failure_interval = 250.0 ; average interval in ticks to check for mechanical problem/failure (gets randomized)

In the first line, it means 5.000 chances in 10.000, that is, 50% of chances of failure.

For the second line, I understood it as an internal ticking clock. After x ticks, there is an incidence to take the failure, so I suppose that would go to the first line which could be a 50% chance. As my races are 10 to 20% laps, it works very well with 200 and 250 to simulate many failures as in the old races. I mean, it would be possible to run a few laps and failures as if it were a lot of laps. If you put 150 then, almost all cars will broken down in a few laps.


Also, all values below must be divided so that they reach 100:

; mechanical problem chances, should total 100
suspension_problem_chance = 4.0 ; % chance any problem involves suspension problem
tire_problem_chance = 4.0 ; % chance any problem involves tire problem
fuel_system_problem_chance = 4.0 ; % chance any problem involves fuel system problem
engine_problem_chance = 4.0 ; % chance any problem involves engine problem
brake_problem_chance = 4.0 ; % chance any problem involves brake problem
suspension_failure_chance = 10.0 ; % chance any problem/failure involves suspension failure
tire_failure_chance = 10.0 ; % chance any problem/failure involves tire failure
fuel_system_failure_chance = 10.0 ; % chance any problem/failure involves fuel system failure
engine_failure_chance = 10.0 ; % chance any problem/failure involves engine failure
brake_failure_chance = 10.0 ; % chance any problem/failure involves brake failure
fuel_leak_chance = 10.0 ; % chance any problem/failure involves fuel leak
coolant_leak_chance = 10.0 ; % chance any problem/failure involves coolant leak
oil_leak_chance = 10.0 ; % chance any problem/failure involves oil leak
 

jxeboy

Active Member
SRD Member
Messages
75
Reaction score
118
I have done quite a bit with these:

ai_strength_to_traction_circle = 1.0
ai_strength_to_aero_drag = 0.3
ai_strength_to_engine_power = 0.2
ai_strength_to_traction_circle_spread = 8.0
ai_strength_to_engine_power_spread = 4.0
ai_strength_to_aero_drag_spread = 4.0

The top 3 effect how much the driver ratings matter. For example if you raise the "aero_drag" from .3 to .5, drivers with a higher aero rating will be faster. I haven't done extensive testing on this. However, since the formulas available today for rating in softwares like NRatings were built to use the standard numbers, editing these produces maybe unrealistic results because those ratings work differently.

The second set of settings with 'spread' in them control how much better or worse a driver is based on the ratings. So for example if you raise the aero_drag_spread from 4.0 to 10, you will notice a much more spread out field at aero-dependant tracks. I have used these settings to make the field have many more lapped cars if I am simulating older nascar races, or even made these numbers smaller to make the pack tighter like today's racing.
 

Rufio2031

BBMC Tracks
SRD Member
Messages
225
Reaction score
674
The top 3 effect how much the driver ratings matter. For example if you raise the "aero_drag" from .3 to .5, drivers with a higher aero rating will be faster. I haven't done extensive testing on this. However, since the formulas available today for rating in softwares like NRatings were built to use the standard numbers, editing these produces maybe unrealistic results because those ratings work differently.
TL;DR - ai_strength determines what specifically makes all AI tougher or easier. All the others are to what effect a specific driver rating has on the behavior of a specific driver/vehicle.

My interpretation is slightly different. The lines you mention aren't really correlating to the driver ratings. The "ai_strength_to" lines are a crazy variable calculation that defines when you raise the AI strength, what about them gets better?

For example, if you raised the ai_strength_to_traction_circle the AI would gain more grip as you raised the ai strength; that's what makes them faster in the end result.
If you raised ai_strength_to_aero_drag the AI would have less drag as you raised the ai strength; that's what makes them faster in the end result.

The way I see it, all the lines kind of follow this logic:
What effect does ai_strength have on traction_circle?
What effect does the driver_shorttrack rating have on the traction_circle?
What effect does the vehicle_aero rating have on the final aero_drag?
 

casssp

Active Member
SRD Member
Messages
186
Reaction score
94
ai_drag_modifier = 1.25 ; > 1.0 = more drag, which is slower

Taking advantage of the topic and still on the topic, if I want to simulate the aerodynamics of the squared/boxed cars (until 1988 for example), which value would be ideal for ai_drag_modifier ? At the moment I have used 1.25 for default...
 

Rufio2031

BBMC Tracks
SRD Member
Messages
225
Reaction score
674
Taking advantage of the topic and still on the topic, if I want to simulate the aerodynamics of the squared/boxed cars (until 1988 for example), which value would be ideal for ai_drag_modifier ? At the moment I have used 1.25 for default...
I think we'd need more about what you're trying to accomplish.

If you're just trying to slow the vehicles down:
In my opinion, the closest equivalent to the track.ini's ai_drag_modifier would be ai_strength_to_aero_drag. But it's a moving variable. The effect that it has depends on the ai_strength you use. So if you're moving the ai_strength around you're going to end up with slightly different results. On top of that the track.ini ai_drag_modifier can still alter that value in the end.
It sounds like what you're really looking for is a new physics model, which is a completely separate topic.
I think your best option is to leave these alone and continue do what you're doing in the track.ini.
Also know none of this has any effect on your car. It's basically saying I'm driving a cup 2003 car, but all the AI drives a snowplow.
 

casssp

Active Member
SRD Member
Messages
186
Reaction score
94
Understand. Now it is explained why I was getting the results you talked about, that moving on one side mess the other. One depends on the other after all.
In fact I was wanting the cars to break a lot as if they were long runs but playing 10-20% laps (I managed to do that) and I was also thinking about making the cars a little heavier, without accelerating or braking so quickly, messing in track.ini. So messing here and there I already felt it and you ended up confirming my hypothesis.

I created a default base below to adjust the tracks I'm going to play. It has worked well...

ai_accel_modifier = 0.35 ; acceleration grip efficiency
ai_decel_modifier = 0.75 ; braking grip efficiency
ai_fuel_use = 1.25 ; > 1.0 = more fuel consumed
ai_grip_modifier = 1.00 ; > 1.0 = more grip
ai_drag_modifier = 1.25 ; > 1.0 = more drag, which is slower
ai_line_modifier = 1.00 ; dangerous hack, don't use unless you are a trained professional

Also I forgot to mention that I also messed track_tire and track_xx_grip values.

But anyway, I think I got something close to what I was looking for. More than that it only works if you edit the physics as you said.

Thank you!
 
Last edited:

Matt Baker

Active Member
SRD Member
Messages
120
Reaction score
407
As a poor example, when I want to get certain speeds achieved to closer match real life, I reduce the drag on the car's setup then adjust the AI to the track. Example: my Michigan track. The track behaves the best when your spoiler angle is between 45-50 degrees instead of the 70 degrees that's typically used at speedway tracks, because the AI have been tuned to the higher speeds that the player car can achieve. You can do the reverse to slow the car down by increasing the drag of the setup (nose as low as it goes, rear as high as it goes, max spoiler) then slow the AI down.
 

casssp

Active Member
SRD Member
Messages
186
Reaction score
94
As a poor example, when I want to get certain speeds achieved to closer match real life, I reduce the drag on the car's setup then adjust the AI to the track. Example: my Michigan track. The track behaves the best when your spoiler angle is between 45-50 degrees instead of the 70 degrees that's typically used at speedway tracks, because the AI have been tuned to the higher speeds that the player car can achieve. You can do the reverse to slow the car down by increasing the drag of the setup (nose as low as it goes, rear as high as it goes, max spoiler) then slow the AI down.
Your idea seems to be very interesting, especially if I combine your idea with the idea of increasing the track temperature to 110 F which makes the car lose some power. Well... I'll try. Thank you!
 
Top