Brands Hatch GP

Discussion in 'General Discussion' started by Paul Knapp, Sep 25, 2014.

  1. peterchen

    peterchen Registered

    Joined:
    Oct 25, 2011
    Messages:
    2,099
    Likes Received:
    287
    Don´t know why it is at "0" by default. It should always be "2", yes.

    Mistakes "0" is a little bit boring and less lifelike in my opinion.
    Has not much influence on times, as long as you don´t set it to values higher than "0.7" or so.

    Higher values create more variation, not less.
    I have it on about 0.4 but I´m not the fastest driver.
    0.45 leads to a variation of 1,5 sec+ for Civic (depends on track certainly!)
    This value influences the absolut times!
    I´ve noticed that higher values can lead to faster times! Must not always be the case though...
    Mess around with it and learn. Note the times with different values and compare.
    I´ve not yet messed around with it in the new build. With my older installation this function was broken
    but with my new installation it seems to work again.

    But isn´t all that explained properly in the "defenitive AI settings"-thread that I posted?
     
  2. Spinelli

    Spinelli Banned

    Joined:
    Jan 28, 2012
    Messages:
    5,290
    Likes Received:
    32
    Peter, I know that Mistakes at "0" means hardly any mistakes, it says so right in the setting's description :)

    About AI limiter, I think you're right, Closer to 1 is more variable as the following explain:

    source -> http://www.nogripracing.com/forum/showpost.php?p=1242353&postcount=57

    source --> http://www.nogripracing.com/forum/showpost.php?p=1242355&postcount=58


    But them, look what this person has to say NOTE: he is talking about oval racing, maybe that's why he seemed to have different results
    source --> http://www.thepits.us/board/viewtopic.php?t=9221
     
    Last edited by a moderator: Oct 22, 2014
  3. hexagramme

    hexagramme Registered

    Joined:
    May 25, 2013
    Messages:
    4,245
    Likes Received:
    194
    I'll try and change the AI mistakes to 0.25 and see what happens.
    Thanks for the tip! :)

    As for AI limiting I was getting sick and tired of seeing the AI grinding on the brakes everywhere to artificially keep their times down.
    Plus racing in a tight pack of 25 cars only spread out by a few seconds on a road course is ridiculous imo.

    The less limiter for me the better. Keeps it more realistic and natural.
    I've seen very few races irl where and entire field stays together on a road course all through a race... ;)
     
  4. Spinelli

    Spinelli Banned

    Joined:
    Jan 28, 2012
    Messages:
    5,290
    Likes Received:
    32
    Ok so the lower the limiter the more variation ahhhh I keep reading conflicting info. Peter said you need it higher for more variation between drivers to have them spread out more but hexa, you're saying it's better to go lower (make sure you don't use 0.0 though, that will be too perfect).

    Anyways, whether it's higher or lower, I advise everyone to just try both themselves. As long as we know what setting needs to be adjusted (AI limiter, (0.1 - 1.0) that's the most important thing :)
     
  5. hexagramme

    hexagramme Registered

    Joined:
    May 25, 2013
    Messages:
    4,245
    Likes Received:
    194
    Yeah man I see what you say. I guess it depends on what people mean when they say 'variation'. Whether it's variation in one single AI, where its speed varies throughout the race in waves, or variation as in a spread out field.

    But yeah, limiter at 0 might be too perfect. I'll try a bit more. I really need to start experimenting more with values. I tend to forget all that is at our disposal. :)
     
  6. endonesia

    endonesia Registered

    Joined:
    Oct 5, 2010
    Messages:
    79
    Likes Received:
    0
    Yes, I got the AIW update from Pete, but for me the AI is still very slow. This is with the Camaro GT3 @ 105%

    [​IMG]

    Any thoughts Peter?
     
    Last edited by a moderator: Oct 24, 2014
  7. peterchen

    peterchen Registered

    Joined:
    Oct 25, 2011
    Messages:
    2,099
    Likes Received:
    287
    EDIT: Eeerhm, seems like I´ve only altered Indy and not GP! Line on GP track is still catastrophic. That wasn´t me! My memory is really bad! *shame*
    I forgot that I wanted to do GP and haven´t done it yet!!
    Measure Indy please. How is the difference there?
    (between you and AI and between old and new)
    Sorry!!! :eek::eek::confused:

    Will make GP on monday!
    *facepalm*
     
    Last edited by a moderator: Oct 24, 2014
  8. endonesia

    endonesia Registered

    Joined:
    Oct 5, 2010
    Messages:
    79
    Likes Received:
    0
    No worries.

    Indy layout is spot on @ 100% strength :)
     
  9. peterchen

    peterchen Registered

    Joined:
    Oct 25, 2011
    Messages:
    2,099
    Likes Received:
    287
    Puuuh! good.
    Wait ´til monday then.
     
  10. Marc Collins

    Marc Collins Registered

    Joined:
    Jan 11, 2012
    Messages:
    3,159
    Likes Received:
    162
    Where is this AIW update?

    EDIT: I see now.

    Am I the only one who thinks it's absurd to worry about fixing AIW on a third party track that has no authorization in the first place? Just distribute your updated version (thanks for doing it!) and as long as it is clearly labelled, no one is for the worse. The "author" of the track can include it in a future version (that rarely come), or not, and no harm is done either way.

    A more complicated for users, but easier for permissions method is to just release it as an update. Then only those who want the update can apply it. Again, no harm done.

    There are way too many sub-standard tracks out there that need help and way too many good tracks that need AIW help. If you have the talent to fix these, Peter, we should be encouraging your work on as many tracks as possible! They need it.
     
    Last edited by a moderator: Oct 24, 2014
  11. DurgeDriven

    DurgeDriven Banned

    Joined:
    Mar 20, 2012
    Messages:
    6,320
    Likes Received:
    43
    Why don't they take all their no authorization stuff and shove it where the sun can't shine.

    At least GPL always kept its dignity not trolloped around the gutters.



    lol gessus
     
  12. Marc Collins

    Marc Collins Registered

    Joined:
    Jan 11, 2012
    Messages:
    3,159
    Likes Received:
    162
    Is the minimum 24 laps required to build the wisdom files based on changing the sample size setting?

    I just ran a 2 hour practice session with an average of 38 laps for the AI and no wisdom files. But my sample stints is set to the default 5:

    "AI Calibrate Sample Size":5,

    Of course I had previously set this to 2:

    "Autocalibrate AI Mode":2,
     
  13. peterchen

    peterchen Registered

    Joined:
    Oct 25, 2011
    Messages:
    2,099
    Likes Received:
    287
    In which folder you looked for the .wis files?
     
  14. Marc Collins

    Marc Collins Registered

    Joined:
    Jan 11, 2012
    Messages:
    3,159
    Likes Received:
    162
    OK, the mystery continues...

    I was looking in the wrong place--under the track folder in User instead of the car folder. So I go to look at the car, NSX in this case, and I see five sub-folders, based on the colour of the car (used instead of numbers I guess). Why would a .WIS need to be distinguished by the colour or number of the car is beyond my comprehension, but let's let that pass. Inside the sub-folders I see not only the .WIS file for the track where I ran the 2 hour test practice, but for every track I have driven the car at. Now I have no way to know which .wis files are "real" versus template ones that seem to be created as soon as you drive the car somewhere.

    I ran the practice with a full field of 21 AI (22 including me) that included all of the "colours" of NSX. Yet only two of the five have anything but a blank template when I look inside the contents of the files. So why/how would only 2 cars gain wisdom during this session instead of all of them?

    And lastly, is it 24 laps that must be experienced by the AI before a real wisdom file gets created/updated? What does this sample size line mean/do?
     
  15. peterchen

    peterchen Registered

    Joined:
    Oct 25, 2011
    Messages:
    2,099
    Likes Received:
    287
    +1 on documentation.
    But Marc, I really wonder what´s your problem is!???
    You don´t have to do anything with this files! (other than eventually copy a complete .wis file from car #1 to other cars (of the same kind))
    Just be happy with the automated learning they do in this mode at every session! Thats a big plus!

    Sample size is the amount of times the AI will test certain corners/parts of the track.
    The higher it is, the higher will be the accuracy the AI learn. Will lead to bigger .WIS-files.
     
  16. GauchoRS

    GauchoRS Registered

    Joined:
    Nov 23, 2011
    Messages:
    471
    Likes Received:
    9
    yes, I agree.
    there is a big hole where feature documentation should be and requesting this makes them angry about. :(
     
  17. Marc Collins

    Marc Collins Registered

    Joined:
    Jan 11, 2012
    Messages:
    3,159
    Likes Received:
    162
    Sorry Peter, but this isn't all so obvious.

    Why do only some of the cars in a 2 hour practice session get a populated wisdom file and others get only a blank template?

    Is there a limit to how many cars can gain wisdom within a particular session (in other words it's a waste of time running more than x AI for this purpose)?

    Is there a FPS impact for "Autocalibrate AI Mode":2,

    If sample size determines the amount of times AI check a corner/something, then presumably it should affect how many laps must occur before they can wise-up? A smaller sample size should allow a wisdom file after fewer laps, but it should be expected to be less wise than one with a higher sample rate. However, if many cars can't get a populated sample file even after 2 hours and 40 plus laps, lowering the sample may be a better alternative so that at least some learning occurs.

    Or, are we expected to leave it cranked-up, accept that only a few will learn because that's the way it's supposed to work, and then copy and paste .wis files?

    Unless each of us wants to do literally hours and hours of experimentation to answer these questions, it would be nice if we could share results. It isn't as easy and just turn it on and forget it.

    Also, even assuming it was that easy and we actually got a full range of .wis files consistently, though they took hours and hours to produce, it would still make sense to share these publicly so that people could benefit from community experience without every single person running hours-long practice sessions on every track for every car class, no????
     
  18. magicaldog

    magicaldog Registered

    Joined:
    Oct 17, 2015
    Messages:
    57
    Likes Received:
    0
    Where can I get Brands Hatch GP for rF2 ?

    Previous links in this thread does not seem to work anymore

    I need it badly

    Any help is greatly appreciated
     
  19. PRC Steve

    PRC Steve Registered

    Joined:
    Sep 27, 2012
    Messages:
    1,301
    Likes Received:
    11
  20. magicaldog

    magicaldog Registered

    Joined:
    Oct 17, 2015
    Messages:
    57
    Likes Received:
    0
    Is that the smooth version with rF2 mesh ?

    If it is I love ya ;)
     

Share This Page