Forum Discussion

DRudd13's avatar
DRudd13
Seasoned Ace
4 years ago

Re: AI Difficulty Calculator for F1 22 (F1Laps)

@FILaps I've just done COTA my next track. In time trail, my average over 5 valid laps was 1.35.401 which is level 95. In Grand Prix mode with Lando on level 75 I should be around 1.7-2.0 seconds a lap faster than the fastest car, is that right? But my OSQ time was 1.37115 which is level 79 might have made a couple of mistakes but still should be good enough to challenge for pole. I was 7th. Max on pole done a 1.36.443 which works out at level 84 still lower than my time trail but 9 above the 75 level I'm playing on. Teammate done a 1.38.569 level 69. to the race 25% I finished 6th but because Pérez had a problem and Max DNF. I set the best time of 1.39.411 fastest lap on lap 2 of softs, but I did an opposite strategy start med to soft. While most went soft med. Charles who won done a 1.39.595 on old meds. Pictures to follow. But now running more tests. I can't make any sense of it, though.

4 Replies

  • FILaps's avatar
    FILaps
    New Traveler
    4 years ago

    @DRudd13 wrote:
    @FILapsI've just done COTA my next track. In time trail, my average over 5 valid laps was 1.35.401 which is level 95. In Grand Prix mode with Lando on level 75 I should be around 1.7-2.0 seconds a lap faster than the fastest car, is that right? But my OSQ time was 1.37115 which is level 79 might have made a couple of mistakes but still should be good enough to challenge for pole. I was 7th. Max on pole done a 1.36.443 which works out at level 84 still lower than my time trail but 9 above the 75 level I'm playing on. Teammate done a 1.38.569 level 69. to the race 25% I finished 6th but because Pérez had a problem and Max DNF. I set the best time of 1.39.411 fastest lap on lap 2 of softs, but I did an opposite strategy start med to soft. While most went soft med. Charles who won done a 1.39.595 on old meds. Pictures to follow. But now running more tests. I can't make any sense of it, though.

    Thanks for the detailed feedback, I appreciate it! I think that there are two things going on:

    1. For cota and Japan, or more generally for tracks later in the season, we don't have as much data as for the earlier tracks yet. I will look into your examples and see if we can fix any discrepancies as soon as possible.

    2. We've noticed some inconsistencies in the AI's behavior. We're still looking into it - not quite sure how to make sense of it yet. 

    Keep sharing your observations and we're going ot investigate them!

  • DRudd13's avatar
    DRudd13
    Seasoned Ace
    4 years ago


    Thanks for the detailed feedback, I appreciate it! I think that there are two things going on:

    1. For cota and Japan, or more generally for tracks later in the season, we don't have as much data as for the earlier tracks yet. I will look into your examples and see if we can fix any discrepancies as soon as possible.

    2. We've noticed some inconsistencies in the AI's behavior. We're still looking into it - not quite sure how to make sense of it yet. 

    Keep sharing your observations and we're going ot investigate them!


    It's in my best interest to get the AI sorted out, as I'm not starting my team till it makes sense in Grand Prix mode. Going to run 3 tests at Bahrain, Spain and Belgium, I'll let you know my findings. For now here are my COTA pictures

  • DRudd13's avatar
    DRudd13
    Seasoned Ace
    4 years ago

    Another 2.

  • ScarDuck14's avatar
    ScarDuck14
    Legend
    4 years ago

    @DRudd13  Thanks for your message..  unfortunately I suffer severe insomnia and I’m now at the point where I’m completely exhausted.  So tried to read your posts.  But just can’t focus.  When I’ve had a few days sleep and refreshed I’ll have another go.

    But did note you said you thought every 10 increase in ai equals 1 second.  Last year this in my team would always seem the be the case although ai in last 5 laps off races would find almost a second a lap. Whereas I couldn’t be sure I blame rubber banding.  But could also be the case as the race went on I’d lose focus and not increasing my pace 

    Anyway this year the ai increase by 10.  Isn’t. One second.  From what I can tell and until the torque update is out I can’t test cause with out tc I’m  slow as no confidence: accelerating.  and with TC I’m slow because off amonnt off power the TC limits when accelerating. Either way I can’t find a happy medium.  And I only test ai in relation to my pace and not it’s overall pace. As ai is different on every track I’d change it ever track 


    But from what I get back so far is 10 is not 1 second anymore.     Rough example it’s More every 10 increases by 1 second up to 50-60. Then gradually 10 - 1 second next 10 - 1.1 next  10 -,1.2.  But when you get to 100 - 105 is 1 second and 105  - 110 another second 

     Although I can’t be sure as I’m struggle to beat 80 although did pole Jeddah. At 80 in just my second midfield my team race and went on to win.  But ai does seem beatable at Jeddah compared to other tracks


    last Year was easy to calculate. Once I had a teammate with 90+ Pace normally lando.  Whatever times he posted on mediums in practice I’d adjust using 10 = 1 second: and unless either off is had a issue.  Qualifying always qualified within half second with lando majority off the time beating me.  In 50% race would always finish within a few seconds off eachother with me beating him more than half the time.

    So. For me personally my method worked awesome for me:   And reflected what I knew last year.  Ai were faster in qualifying compared to race

    And for the conspiracy theorists…, with this game have adaptive ai for lower difficulty levels when on casual:  there could be an argument that higher levels have some level off adaptive ai implementation.  Where as whatever your teams ranking plus your what lap times your posting.. the ai pace matches to reflect it.  For instance teams ranked higher than you. Depending on how much.  Will adjust and  lap just that bit faster than.  Chase any higher ranked  team car that is front off me.  And almost impossible for me to catch.  Despite how strong I am on that track.   And they always stay just ahead:  always going just fast enough to keep me behind.  And also answer why with ers and drs there are many reports off not being able to catch and pass the ai.

    It would be a way off making it appear the ai is consistent track to track.  until Jarno rain and Monaco master got a wet race at Monaco in his my team   He struggled against ai. I’ve seen you tube gurus actually admit defeat and power ai. Although some seem to be doing better than others on 110 but they are lying 

    what’s I noticed last year from forum posts was people starting my team careers and setting the ai not taking or wanting  into account that my team car is a bottom ranking car. 

    And just a random final thought:  god Miami is a hateful rubbish nasty  track to drive with the current torque and acceleration model as it is.   Games  were fun once  over 40 years off gaming and sure games have frustrated me over the year  but the other night was the first time I got so angry with a game I  rage quit.

    To tired to focus on reading anything and to tired to stop my brain to stop thinking.  I wish it would shut up 

About F1® Franchise Discussion

Join other players talking about classic F1® games here.4,995 PostsLatest Activity: 4 hours ago