How much of a pace spread is normal and at what point would it be worth reporting as a problem? Trying to figure out which ai level i should run for a gt3 championship using basically just oldish cars (r8 lms ultra, z4, mp4, the old corvette, nissan, sls and bentley) and noticed the spread is a bit extreme on silverstone gp and bathurst (tested with ai at 110 strength). Let them run for half an hour and p1 and p19 were seperated by 5 to 6 seconds in both cases.. (On a side note seems to further confirm the observation about the nissan being very slow https://forum.sector3studios.com/index.php?threads/slow-nissan-and-mclaren-ai-in-gt3-class.14913/)
I just can confirm that it was that way from the start of GTR3 class. I made many championships especially with these older GT3 cars since the last few years and Nissan and bentley were always quite slow. How far it also depend on the "talent" files of the AI is hard to say but one bentley (the red one, don't know drivers name) was up to pace with the audis. The audis were in general quite fast (as in reality at that time). But I'm with you that a 5+ second spread even at long courses is a little bit too much for my gusto.
think it as more the ai should be able to drive fast in a straight line , so at lower levels , there braking might be earlier and there corner entry/mid corner and exit slower .. so at the end of the straight a little slower due to the exit speed as we get above 105- 110% they get better at it but generally they shouldnt be better than the player even at 120% ...alien territory if you see them getting massive/impossible advantage it be great if you could make a video and explain the track , corner ai % then we can test and fix where needed Andi
Yeah we have a similar problem in the GT1 class where the Koenig is still a second or two faster than the rest of the pack in qualifying. This tends to really spread out the field when racing What I've done is just created custom AI races leaving out the Koenig This keeps the AI much closer together in the race and I don't have these darn Koenigs racing off into the distance finish half a lap ahead of the pack. What I found strange in using this method is (and I haven't fully tested this yet) it feels like the speed of the first placed AI in qualifying is roughly the same for a set AI strength whether I include the Koenig or Not! I suspect that when you select an AI level it says; the fastest ai will do a lap of x and then adjusts the speed of the rest of the ai from that starting point. The lap time x is roughly set and the fastest car you have in your set AI of vehicles will meet that fastest lap time
But i dont see any car being particularly overpowered like your koenigsegg case. In silverstone there are 4 different cars in the top 4 covered by half a second which seems fine on a 2 minute lap. And on bathurst the audi becomes the fifth car capable of qualifying at the front. So (apart from the slow nissan) it doesnt really seem to be connected to that
I do agree the spread is too much and I feel they should fix it. But in the meantime time for some closer racing with a less spread out field you could for example just create a custom field of cars (eg those 4 different cars in the top4) and it makes the racing field much tighter and fun The second point I was making that even if you created a whole field of the slowest car I think that one would still qualify at roughly the same time as a field of the fastest cars at a given AI level. I still need to test that to be sure but at this stage it's what I'm feeling is happening tho I could be wrong LOL
Yeah i could but the whole point of this championship would be to see all those different cars around me so i dont really wanna limit my opponents to just a few different cars. (Also even then i might end up with quite a spread. I run 2 corvettes and one was 3rd and 1st in the quali sessions while the other one was 17th and 15th (out of 19 ai cars). And while one mp4 was 4th at silverstone they placed 12th, 16th, 17th and 18th around bathurst.. ) I see, might have a look at that
Just did a first test and judging by that your theory doesnt seem to be true (at least not for all cars). First had 24 gt3 ai including 5 nissans run at silverstone. The nissans ranged from 2:01.3 - mid 2:05 while the best overall was a 1:59.3. Then had 18 nissans run and they ranged from 2:01.6 - mid 2:06. Obviously have no idea whether the nissans just an exception or if its the rule tho Edit: both runs at 110 strength
Well looks like ive just stumbled across an extreme example of the ai being impossibly fast The (newest) gt3s at 120 strength around nürburgring gp fast chicane when let loose in a quali session are up to 5 seconds faster than the leaderboard record (which definitely isnt a slow one either). Think it might be a good idea to slow the ai down a little for that combo