Information for Customizing AI drivers in AMS2

Things took longer than planned, but then you guys know all about that!

Attached are some screen shots of the practice sessions I ran. I didn't drive, just left them to it. AI at 107, medium aggression, 1.09x wet scale, 1 hr practice session, default date, Spa 2022 (a long track for 9 AI not to trip over each other). Each driver had the same quali/race skill and a range of wet skills, each driver named according to skill, and all in the same model car. I have also attached the last test custom driver xml file, please note I am using custom skins, the excellent Immersion Modding Group's 1995 IndyCar World Series v2.0 which can be found on overtake.gg.

So the wet skill parameter does indeed work, but not as I would have hoped. Moreover, I can see why I perceived it as not working. The difference between 100 and 0 wet skill is small, too small. About 10 race/quali difference for 100 wet difference. It is a tiny difference in comparison to quali or race skill given the same range. A driver with a middle quali/race skill and high wet skill will never match, let alone beat, a high quali/race skill/low wet skill driver. It can and does happen in real life racing. This seems to be because wet skill is a percentage of quali/race skill (as described in the first post on this thread) rather than an absolute. I hope this can be looked at and improved. We have the ability to scale wet/dry performance in the race options now, so I would hope this would be possible and would allow a much more realistic implementation of the differences between drivers in wet and dry conditions.

 

Thanks for the reply. I didn't realise the range was specific to series, perhaps this also skewed my perception of the difference. I don't really understand how a 1s range can be a 2s range. The experiment illustrates just over 1s in the wet, which is what we are discussing, not the dry! Mistakes I get, which is why I gave them a long session to try and even things out. If I understand you correctly, the 50/100 driver should make less mistakes relative to the 100/0, for a slight advantage in the wet.

I did omit some pertinent information by mistake. Weather was 1 slot rain, livetrack wet preset, real time. No acceleration on time or weather, no mistakes multiplier. Rain intensity was 0.4, by shared memory. The track evolved from wet to soaked quickly and remained as such for the whole session. All AI drivers were set for every available skill to be identical except wet skill, the only variable.

I think a 2s range in the wet (for this pro series) would be more appropriate, but I would hope for more available as a custom option (like the mistake multiplier), maybe as far as 50% of the spread available with the dry range from 0 to 100 quali/race. That is to say, at best, a 50/100 driver could just about match 100/0.

No demands, just an observation and (passionate) feedback.

 

EDIT: Delete.