Indycar AI feasible?

wowbagger · May 20, 2024

Is anyone doing/has done recently an Indycar season offline?

I'm wondering if it's worth the effort to hunt down all the tracks and get things set up, or whether there are some gotchas (like AI not able to handle full course yellows on ovals, or lack of passable AIWs, etc) that mean it's better to not bother getting stuck into it.

Cheers

wowbagger · Jun 5, 2024

Just to answer my question ...

Yes, a 2024 Indycar season is feasible, with a few caveats.

- Toronto doesn't have a passable AIW for the tight twisty bits, but Vancouver seems to work well enough.
(I did a lot of AIW editing for rF1 way back when, but have never got back into it for rF2 and so just run with what I can find.)

- And Belle Isle is standing in for Streets of Detroit (which is a horrible track anyway).

- Cleveland Burke Lakefront Airport which is one of my favourite tracks going back to ICR2, instead of Nashville.

The ovals are doable if:
- you make sure damage is low (I used around 50%)
- have immobile damaged cars disappear after a couple of seconds

-The AI strength varies wildly from road course to oval. You might have to run 95% on some road courses, but 120% on 1 or 2 of the ovals.

- The AI Dallaras have ridiculous braking distances. You can adjust the braking strength in the player json - but if you change it too much they have difficulty making some of the corners (turn 12 IMS, for example). The alternative is to turn on the lowest ABS setting which brings you closer to what the AI can pull off every single corner.

If I can remember any other player.json settings I changed, I'll append them here.

Happy for any tips anyone might have, specific to racing the Dallara IR18s.

Durge Driven · Jun 5, 2024

Never forget first day in servers at Indy.
I clocked the fastest lap time all day, was in another tenth to everyone else.
Caught perfect double tow with immersion off the scale.
Moving from walls like real seemed to work as well.

Best tip I can give which would be obvious to you ( maybe not some others) is to cross the white lines, don't straddle them.

Result is identical to the real thing.

P.S. Portland is a awesome race , pits okay, behaviour okay.

Of course that all depends on the learning they get.
Restart quick races over and over you just in for pain.

I will usually just let practice and qualifying run with AI and start from the back, set AI strength and aggression where winning should be nigh impossible but perfect conditions maybe possible which is "what I get" with ISIMotor2.5. In other engines these tiny variation are just "noise" to me.
I mean I like doing full sessions but just not enough hours in the day and simply more fun racing.
New combos I will drive full sessions or at least qualify but reduce speed to start at back anyways.
Simply don't get half the enjoyment starting at front unless you set impossible strengths which will just have cars everywhere getting ploughed into each other.
Of course this has improved through S397.

AI is just one of the things that LMU does way better so full sessions I gravitate to LMU because they can be as good as racing.