1. This site uses cookies. By continuing to use this site, you are agreeing to our use of cookies. Learn More.

General AI Performance Discrepancy Reports (No Discussion)

Discussion in 'Automobilista 2 - General Discussion' started by Renato Simioni, Sep 14, 2023.

Thread Status:
Not open for further replies.
  1. Marnus Zwarts

    Marnus Zwarts Boost Creep AMS2 Club Member

    Joined:
    Jan 6, 2021
    Messages:
    244
    Likes Received:
    93
    F classic G2 race transition from wet to dry:
    There is a large discrepancy between player and AI to AI.
    See screenshots for settings.
    Only myself and 2 cars started on wets (after thunderstorm in practice and rain for the start) The rest on slicks
    The slick dudes dropped like flies at 20+sec per lap
    6 laps in four AI drivers went in for wets.
    Halfway through the race the rain stopped and the track dried up.
    By 75% race distance there was a solid dry line with little to no water on track
    I stopped for slicks
    No matter how gently I accelerated, the (still on slicks mostly) whistled past me like i was driving an Uno. Any amount of gentle acceleration cause the rear to completely loos grip and spin. I started losing 40s per lap and the dudes still on slicks from the start annihilated the rest of us.
    upload_2023-9-23_13-12-45.jpeg upload_2023-9-23_13-12-45.jpeg upload_2023-9-23_13-12-45.jpeg
     
    • Agree Agree x 1
    • Informative Informative x 1
  2. Marnus Zwarts

    Marnus Zwarts Boost Creep AMS2 Club Member

    Joined:
    Jan 6, 2021
    Messages:
    244
    Likes Received:
    93
    upload_2023-9-23_13-22-13.jpeg upload_2023-9-23_13-22-14.jpeg upload_2023-9-23_13-22-14.jpeg upload_2023-9-23_13-22-14.jpeg
     
    • Informative Informative x 1
  3. Marnus Zwarts

    Marnus Zwarts Boost Creep AMS2 Club Member

    Joined:
    Jan 6, 2021
    Messages:
    244
    Likes Received:
    93
    upload_2023-9-23_13-27-50.jpeg

    Could custom AI drivers and skills have an influence on this?
    Apologies for the multiple posts....
     
    Last edited: Sep 23, 2023
    • Informative Informative x 1
    • Friendly Friendly x 1
  4. CrimsonEminence

    CrimsonEminence Custom Title Staff Member AMS2 Club Member

    Joined:
    Nov 4, 2018
    Messages:
    9,896
    Likes Received:
    10,208
    Please use default AI and no custom AI for tests.
     
    • Agree Agree x 3
  5. Dylan Hale

    Dylan Hale Well-Known Member AMS2 Club Member

    Joined:
    Jan 10, 2018
    Messages:
    1,851
    Likes Received:
    1,128
    You might want to take a look at 1988 Imola, (and probably the other Imola Variants.) It seemed like the AI could both brake later and accelerate faster out of the hairpins on either end of the track and out of the final chicane on the start finish straight, even while riding on the curves or grass. Racing the same cars, on the same settings at Barcelona shows me I'm roughly on pace.

    Was using the Formula Classic Gens 2 and 4
    Clear conditions, with high track usage.
    Ai at 100% and mid agression.aggression.
    Also did about 30 minutes of practice before hand trying setup changes to get up to pace... (if that matters?)
     
    Last edited: Sep 23, 2023
  6. Mazdaspeed

    Mazdaspeed Well-Known Member AMS2 Club Member

    Joined:
    Aug 25, 2022
    Messages:
    1,526
    Likes Received:
    781
    Baseline Strength: 108/Max Aggression
    Car Used:F-USAG1&2
    Track: Barcelona
    Session: Qualifying/Race

    AI loses only 1 second from qualy to race, while I lose 2-3 seconds, however it seems the AI dont really fuel the cars for qualy in these classes, you see some with 29l others with 55l and so on. But anyway that doesn't make a difference because come race time, the AI cars with 100l do the same times as some that have 50l, so the fuel weight is definitely not doing much to affect the pace of the AI. Photos are from the G1 test, but the results are pretty much the same, I didn't test G3, but you have to assume they also have the same issue as the cars are pretty similar.

    On a positive note I did the test with the F-Usa 2023 and it was perfect! Automobilista 2 Beta 9_25_2023 1_03_07 AM.png

    Automobilista 2 Beta 9_25_2023 12_52_50 AM.png
     
    • Agree Agree x 1
  7. Mazdaspeed

    Mazdaspeed Well-Known Member AMS2 Club Member

    Joined:
    Aug 25, 2022
    Messages:
    1,526
    Likes Received:
    781
    Baseline Strength: 108/Max Aggression
    Car Used:BMW Procar
    Track: Barcelona
    Session: Qualifying/Race

    AI is a bit too slow on the dry and way slower on the wet, they lost 1.8 seconds compared to the qualy time, while I lost about 7 tenths (it might have been a case of a bad qualy for me and the AI is simply too slow overall) while in the wet I lost 11 seconds while the AI lost 13. The race began with light rain and a moist track, which lasted about 10 laps then it got wetter and we all change tires on lap 13. Dunno if that affected the AI dry pace at the beginning, it didn't to me.
    Automobilista 2 Beta 9_25_2023 10_10_44 PM.png Automobilista 2 Beta 9_25_2023 10_46_55 PM.png Automobilista 2 Beta 9_25_2023 10_48_18 PM.png
     
    • Like Like x 1
    • Informative Informative x 1
  8. kkdrummer

    kkdrummer Well-Known Member AMS2 Club Member

    Joined:
    Jun 25, 2016
    Messages:
    1,602
    Likes Received:
    771
    Not sure about gen 1 and 2, but there is a problem with the fusa 23.

    First off, my ai baseline strength with stcokcar23 at Barcelona is 109. However, I'm a second faster than the fusa 23 ai at 109. I need to bump it up to 113 to be evenly matched at lows 1.25.
    Sec 1 is where I beat them, the other 2 sectors they are faster.

    Then in the race, 30 laps, clear weather and 40 degree track temp. The fist 10 ai were faster then me. I got dropped from 1st to 5th, I just couldn't keep up with their pace. I was doing 1.27 highs at that point. It was only after lap 8 or 9 that I was at the same pace as the fast ai. I pitted at lap 16, again I was slower than the AI after pitting, only towards the end of the race pace was more balanced.
     

    Attached Files:

  9. Mazdaspeed

    Mazdaspeed Well-Known Member AMS2 Club Member

    Joined:
    Aug 25, 2022
    Messages:
    1,526
    Likes Received:
    781
    Strange, I had no issues with them, had a rather nice time and was battling for the lead with pretty equal laptimes with a full tank. Did you start the race on the softer tires?
     
  10. kkdrummer

    kkdrummer Well-Known Member AMS2 Club Member

    Joined:
    Jun 25, 2016
    Messages:
    1,602
    Likes Received:
    771
    Yep, softs. The AI only uses softs so it makes sense to go for softs as well.
     
  11. Mazdaspeed

    Mazdaspeed Well-Known Member AMS2 Club Member

    Joined:
    Aug 25, 2022
    Messages:
    1,526
    Likes Received:
    781
    Gave them a go again and you are right in that they are slower compared to the base strength, however I was able to have a great time hanging to a 1 second lead for some 10 laps, never being under threat of getting overtaken and having the best lap of the race at 113 strength. Aero wash is strong in this class, so if you get passed, you lose a good 5+ tenths of your ultimate pace!

    Automobilista 2 Beta 9_26_2023 11_30_16 PM-min.png Automobilista 2 Beta 9_26_2023 11_05_48 PM-min.png
     
    Last edited: Sep 28, 2023
  12. kkdrummer

    kkdrummer Well-Known Member AMS2 Club Member

    Joined:
    Jun 25, 2016
    Messages:
    1,602
    Likes Received:
    771
    I was trying to establish the AI' raw pace in my previous test. Yes, if I manage to stay within draft distance I can keep up with them better, but without their draft their are father than me on a fuel tank. Also, I was on default setup.

    Fusa23 at Fontana. 113. The AI is too slow, mainly coming out of the bankers. I'm half second faster than the AI in clean air.
     

    Attached Files:

    • Informative Informative x 1
  13. Dicra

    Dicra Local Gamepad Ambassador AMS2 Club Member

    Joined:
    May 1, 2020
    Messages:
    1,579
    Likes Received:
    1,111
    20230928121014_1.jpg 20230928123351_1.jpg F-USAG2/Reynard-Ford performance qualifying vs race @110 strengh/high aggression. I'm not sure whether this is actually the pace they are able to go because I don't know if they were fuelled to the brim, which I was. Imo it is NOT sensible to go for full fuel with these cars because the tire wear is horrible; I'd estimate they're done after 10/11 laps at Bathurst, because there was already very notable tire wear by lap 8.

    EDIT: Very similar at Brasilia (which is much better AI-wise than Bathurst), about 1.5 seconds loss relative to the AI from quali to race plus big tire wear.
     
    Last edited: Sep 28, 2023
    • Agree Agree x 1
    • Informative Informative x 1
  14. Dicra

    Dicra Local Gamepad Ambassador AMS2 Club Member

    Joined:
    May 1, 2020
    Messages:
    1,579
    Likes Received:
    1,111
    20230928175624_1.jpg 20230928174239_1.jpg F-Reiza 110/High Aggression AI in "light rain" conditions on dry tires vs F-Reiza 110/High Aggression AI in dry conditions. I gain around 2.5 seconds (admittedly my laps in the rain weren't the best and there was probably around a second in them, but it's almost impossible to drive perfectly in these very difficult conditions, which the AI easily manage, and then also are very quick compared to normally).

    EDIT: Apparently the screenshot I made of the light rain times is broken; I'll retry and post one as soon as I have something representative.

    EDIT2: Apparently, light rain and light rain are not at all the same. The AI stay a bit quicker in relation to the player in a similar way though. The lap I did was one of the best I've ever done in light rain conditions.
     
    Last edited: Sep 28, 2023
    • Informative Informative x 1
  15. Dosda Pierrick

    Dosda Pierrick Active Member AMS2 Club Member

    Joined:
    Jun 9, 2023
    Messages:
    201
    Likes Received:
    41
    Baseline Strength: Tested with 105, medium and 120 max agression
    Car Used: Any Copa Classic B
    Track: Londrina Long
    Session: Qualifying/Race
    The ai are very slow in any situation here
     
    • Informative Informative x 1
  16. kkdrummer

    kkdrummer Well-Known Member AMS2 Club Member

    Joined:
    Jun 25, 2016
    Messages:
    1,602
    Likes Received:
    771
    Montreal, Sprint at 107% high Aggression which is my baseline. I was 1.2 father than the AI in Quali. I am pretty sure I could have shaved off another 0.5 of my best lap.
    Sector 2 is especially poor from the AI. They slow down a lot at the exit of the S after the bridge.
     

    Attached Files:

  17. Dicra

    Dicra Local Gamepad Ambassador AMS2 Club Member

    Joined:
    May 1, 2020
    Messages:
    1,579
    Likes Received:
    1,111
    20231008152439_1.jpg Baseline Strength: ? Normally 110 at least.
    Aggression: High
    Strength used: 105
    Class: F-USAG3
    Fuel in the car: 76 liters at race start.
    My vehicle: Lola-Toyota, which in the hands of the AI was more than 2 seconds quicker per lap. In quali it was 1.4 seconds gap; I don't have a screenshot of that.

    AI in this class is just far too quick compared to the player, much quicker out of every slow corner; they're gaining around half a second out of every exit of every slow corner. Braking performance seems alright, maybe even slightly too weak, but their pace out of corners destroys racing with this class.
     
    • Informative Informative x 1
  18. F_B

    F_B Well-Known Member AMS2 Club Member

    Joined:
    Apr 25, 2016
    Messages:
    862
    Likes Received:
    798
    GT1 @ VIR Full, 103% strength, 50% aggression, medium rubber.
    I'm 21st in qualifying, driving the McLaren

    setup:
    more front arb, less rear arb
    2 clicks more rearwing
    front + rear -1mm height
    +3 springs front
    preload -2 clicks
    clutches 2

    - AI has often much better acceleration out of T1
    - AI can also take the rollercoaster a bit better than me

    [​IMG]

    GT1_AI_Report_VIR.jpg
     
    • Informative Informative x 1
  19. Mazdaspeed

    Mazdaspeed Well-Known Member AMS2 Club Member

    Joined:
    Aug 25, 2022
    Messages:
    1,526
    Likes Received:
    781
    Baseline Strength: 108/Max Aggression
    Car Used:F1 ClassicG1
    Track: Imola 88
    Session: Qualifying/Race

    Tested the AI fuel pace and in Brands Hatch it was pretty close to being equal to the player bar the Lotus. Imola however is not close, the Lotus are again much faster, I think they definitely didn't get updated so will ignore them for now (they are also OP in general). The issue here was that the generic cars lost some 9 seconds, while I lost a pretty realistic 6 seconds. I forgot to take a photo of the qualy times, but I did a 1:27.4(could have done a 27 flat), while the closest generic was Keke...ehmm Keijo with a 1:27.7. Now the fastest generics like Pereira and Marshall weren't able to do a representative lap before retiring in qualy, but the best Mansell...ahem, Marshall could manage in the race was a 1:36.2 in clear air, while I manage a 1:33.1 using normal fuel mixture.

    I dunno if the AI conserve their fuel right now, but that might be one reason why they where so slow, if not, well it seems the fuel issue might be track dependent as well.

    Automobilista 2 Beta 10_14_2023 8_07_52 PM-min.png
     
    • Informative Informative x 1
  20. Mazdaspeed

    Mazdaspeed Well-Known Member AMS2 Club Member

    Joined:
    Aug 25, 2022
    Messages:
    1,526
    Likes Received:
    781
    The GT1 AI is much faster in general than most classes no matter the track, I have to reduce them to 100% to be able to challenge for podiums.
     
    • Informative Informative x 1
Thread Status:
Not open for further replies.

Share This Page