Yes I do believe you are right but depends on the right circumstances. If you only have a single poly set to double then it's fine, if you have more than 2 polys very close together then it can be a problem. I think some of you have a problem in some cars with flicker on the guardrail posts. Going on my above theory that could be because the material and texture used on the guardrail and posts is the same. I need double sided for the guardrail but not the posts. So I'll make the posts a different material and single sided and see if that helps. EDIT: No one has mentioned the AI so I guess no news is good news.
Tested with Spark F2 against AI in a 10 minute race with naturally progressing real road + real road time scale = 15x. I drove car #48 and, since there are only 3 cars with defined .RCD, I selected two AI so they drove remaining two cars #22 and #23. AI strength/aggression/limiter = 100/25/0. I had a pretty terrible race, achieving my fastest lap time of the race (1:15.671) on lap three and making many mistakes on the remaining laps. AI fastest laps were achieved on lap 8 and were 1:15.659 and 1:15.617. So they're running a little slower than I but that is not unexpected since I normally run around 105%-110% strength. Based on this short race, the cars appeared to use a good line and drove well with no unexpected slowing which is good. Cars did exceed track limits on occasion, for example, running wide to the outside on T1, but not every lap and sometimes to the inside so felt realistic. I'll test again when my concentration is better and get some more laps in and crank up the AI strength to give them a good workout and report back although might be a few days. So far so good, though! Many thanks for your excellent work on this track
AI Report: 10 AI Clio Cup, 100% Difficulty, 50% Aggression 15 minute practice, 3 lap qualify, 30 minute race, standing start, progressive real road at normal time My best lap 1:17.0 AI best lap 1:19.0 I started at the back of the pack for the race. Watched them do an orderly start and moved up the typical 5+ places before the first turn. Was too optimistic and while getting myself gathered up for T2, got punted from behind. Catch back up to Tim Wheatley, who also had a minor off, tail him for a lap before passing and then I promptly overdrive T2 again! Repeat chasing that Wheatley AI, passing, and then begin working my way through the field instead of going off <ahem>. Witnessed two AI incidents... first was at the diving left turn a few turns after the bridge (T9) where one AI tried passing another AI, clipping the inside curb, and getting unstable. Second was side-by-side on the bridge and the righthand one clipped the bridge at the exit; I thought 'that was stupid' and then nearly did the same thing on the next lap without the benefit of a car beside me, LOL, so probably not as dumb as it seemed. AI overbrake at T1 and make it easy to slip under them between T1 and the tight T2. No evidence of a "block" line, either, so there's always room. AI also brake at the crest of the hill (T10) before diving down into the last turn, again making it easy to slip by them (better timing required than T1 pass). They also seem to park briefly in the final turn. Not certain, but the AI clipped barriers when they entered the pits. Reviewing the replays, the bridge exit is a problem and was clipped 3x in the race. T1 is a center of action when there's a crowd, with cars aggressively using the inside curb, but apart from making the AI not overbrake, there's probably no reason to consider it a problem for the Clios.
Thanks Emery very detailed. So over all they aren't that bad. Just need to look mainly at the bridge exit.
Did another race using same settings as my last post but with '69 G4 with 20 AI @ 100% and starting from the back to move through the field. My fastest lap was in the high 12s which was 1.6 seconds faster than the next fastest AI car and was easily able to move through the field to finish 9th even with several offs and mistakes. However, my goal was to mainly observe AI behavior more closely as I moved through the field and my experience echoed Emery's. Here are incidents from Log Analyzer. Race incidents Pos Driver Vehicle Contacts Other Contacts Overall Contacts 1 Kevin Knorpp 16 18 34 2 Sal Fuller 6 21 27 3 Layne Fuller 12 9 21 4 Filip Babineaux 6 3 9 5 Gaetan Legastelois 6 6 6 Sam Adams 6 6 7 Sloan Bread 6 6 8 Jonas Edwards 4 4 9 Chuck Van Horne 3 3 10 Stacy Lee 2 2 11 Frank Fauvel 2 2 12 Chuck Schultz 2 2 ...and graphs...
No there isn't!! To be honest I haven't mastered AIW it's a tricky beast. I thought I may have stumbled apon a descent AIW with this one I made, but maybe not. I'll add a block path!! I need binks AIW tool
@PSR Im interested to know your thoughts on the graphs. I always run AI at 120 strength. My take on it is to me they look a little plain in places. BUt if the racing is close and not many places to pass that would be what you could see. But over all a few things need smoothing out.
There are several people on here who I am sure would be happy to help with AIW for such a good quality track (meaning it has a modder committed to supporting it and fixing/tweaking things). It's a whole black art that is difficult to master if you are trying to pay attention to the myriad of graphics and modelling issues. Just post a public appeal and I bet they will appear (maybe with a little help from me). Or, you can use this as your chance to learn it yourself. Some of us can post good links to threads where the gory details have been discussed.
I usually look for extreme position or gap changes can indicate a serious problem AI have with navigating the track. In this case, I don't see any obvious issues other than possibly a lack of passing but this is not the easiest track to pass on and also higher aggression could change things. So looks like block path and tweaks to address emery's other observations are the right approach and should also improve lap times.
Seems like we see things in a similar way. Nice to feel your on the same wave length as others . I'll make those adjustments and see how it goes
