(03-12-2021, 07:37 PM)checkpoint10 Wrote: @samsepi0l, I would be interested in your take on this one, since you've done some work on the track.txt values.
I think I would be able to help more if I knew what was Mauricio's position on the track at the very moment when the yellow flag was thrown.
Personally, when I experience a problem I like to load up the replay and observe every car that misbehaves position at the split second when the yellow was thrown. The DLONG values of where the cars are at this moment of time will give you the clues you need to understand.
I feel like I have to check my notes every time I work on the track.txt's.
Here is the post from our work at Caesars that I reference the most:
https://www.icr2.net/forum/showthread.ph...88#pid9288
Now- we finished Rio in Summer 2020 right? I made some AI newer than that last year, and I've since learned more but nothing earth shaking. I did Vancouver (boblogsan's newer version) and Road Atlanta and I continue to improve. I swear every time I do a new track I feel I get better and it makes me want to Re-do everything old I've ever done.
I can tell you that the LPs and the track.txt are very interlocked. Strange behavior can be attributed to both.
I feel the wiki is only partially correct in that one of the values (9, 10, 11) has to do with crossing (the PANIC LP) so that the car can drive off of the racing surface to enter the pits. An easy way to test this is to change the pit window line on the track.txt to something stupid low like a 5 lap pin window, so the cars have to pit really frequently.
I just started a season with my 1985 carset, and I plan on doing a race at Rio so I'm sure I will make a new version of the AI for myself and all of you to share.