Automobilista 2 September 2024 Development Update - The Road to V1.6 Pt2

I've read multiple posts in here, this place has become overly toxic. It's like a massive pile on. When the update comes out and if you don't like it, try and put a team together and create your own game. There is nothing stopping any of you but goodness me, settle down. I used to enjoy reading the posts but seriously I've never seen such a bunch of ingrates.

 

The main complaint isn't about reiza or the update but how overtake_gg advertise it

 

i could only watch on 240p on my phone from the airport wifi last night, but had another look today and whilst I won't go on about all the stuff that has already been said regarding stream quality, the AI looks wicked fun to race against. This was echoed by many in the comments.

I think it is a mistake to focus on some specific problem you think the AI has and see if you can see it in THIS VIDEO and go OMG IT'S TERRIBLE. The AI behaviour in response to slower cars is very convincing. In IMSA races it's not like the LMDHs just power past the GTs every time, because the speed differential with the smaller slower cars makes them extremely dangerous for one's race ( see last week's INDY race as a case in point - or the end of Sebring where the BMW GT3 ruined I think Bourdais' run at first place). I don't see glitchy AI, I see realistic caution (however explicitly programmed), and the immersion really is there. Especially the bit racing the Lambo and the Porsche.

 

I can only speak of the current build. I have had 100s of races against the AI, with many of the classes in game. I think most of the problems people have with the AI can be avoided in two steps.

Firstly run qualifying sessions of at least 20 minutes and don't skip the session before the end.

Secondly is to not do the usual 8 overtakes on lap 1, show some restraint and give the AI a lap to sort themselves out.

I would say with the current build I have no big problems with the AI 8 out of 10 races.

 

Maybe next week. This weekend is the last big ACC 24h-race.

 

Aint nobody got time for that! I joke, I do that, but my guess is that not many do.

 

To have a good experience with AI (especially in v1.6) and the most engaging races, it is crucial to have the AI level set at your level of performance or very slightly above. Ideally, the AI level should be calibrated by the player on Barcelona NC in a quali session of at least 20 mins, pushing hardest as possible with the SC 2024 and F-Ultimate Gen2/F-3 to match the top AI best lap time. For example, I have determined that (with v1.6) my AI level is 110 for the former and 105 for the later. Thus depending on car, I am using anything between 105-110.
When the AI is properly matched to the player level for a track/car, it makes races much more interesting and realistic as each overtake may be difficult and risky (especially on narrower tracks) and the AI is more likely to pass you before entering a corner if you are a bit too slow. With such tweaks, I had very good engaging 30 mins AI races with Group C cars in v1.6, especially on the slower tracks where it is hard to overtake (Cascais, Jerez) and it was almost never possible to troll AI to overtake it braking late before a curb.
It is also true that the AI behavior is best evaluated as a whole on a race at least 30 mins long: you might see it making mistakes, going off-track and rejoining the track, have a mechanical failure (if enabled in settings), a FCY happening redistributing the dynamics of the race. Just a handful of laps is not enough even if their behavior especially on first lap is much better.

 

So many racing games where this fixes a bunch of issues... AC, Raceroom, GTR2, rF2, AMS2!


I did have a concern from the stream though at watching several AI cars at Road Atlanta stay fixed to the one in front with little apparent effort to get past - that would be my biggest negative from the stream (along with IMSA and Reiza liveries being mixed together, but that's very much a personal thing).

Positives definitely outweighing though. And that's always going to be the case when 4-5 months' worth of hype gives way to reality - the key is to look at what's improved and not what isn't yet, but that's why communication such as blogs is important IMO, keeps a focus on what's realistic and what isn't.

 

I don‘t feel like the comments in this thread are toxic, I feel more like people are a little bit disappointed by what they saw in that stream compared to the ai video Reiza released. Reizas video was a single class race at monza, a track which allows overtakes more naturally than Road Atlanta. Besides that, the Streamer ran the AI level maybe a little bit too high for him and it was a multiclass race.
Just this week I was racing LMU on the new released COTA Track with Hypercars and GTE. I was in third position with a 6 second gap to P2 and when the race leaders closed in on the GTE cars, I was behind them in just a couple of corners (3 or 4) and after we got to the long straight, I overtook the race leaders and the GTE cars in one go. What I want to say: It seems like this is more like a common problem in racing games.

But I doubt that this is much better in multiplayer/ online races. Drivers in slower class cars don‘t respect blue flags to much and if they are fighting for positions you better stay away from them or there is a high change you get bumped into.

 

Very helpful, thanks.
Would be lovely if there was an in-game recommendation of AI setting based on your lap times in some sort of an evaluation mode. But probably asking for a lot. I'll try your method for now

 

The problem with multiplayer is that it doesn’t last for very long. 2-3 month of players and after the excitement, they play other things. Just take a look at le man ultimate: best game ever but 200-300 players stat in average. Hope reiza will release 1.6 before petit le man.

 

No doubt, multi class is surely one of the hardest things to get in any racing game AI, there's never been one to completely nail it (I haven't tried iR's) and why I almost exclusively drive the slower class in multiclass AC or AMS2 races.

GTR2 and AMS1 both had their moments but only if you took a load of time to fiddle with all sorts of settings and physics workarounds (in GTR2, the best way was to effectively make Ai-to-AI collisions impossible by reducing checks so the cars would be brave enough to overtake) - with the complexity of newer titles I get why this is tricky.

 

Hope it wouldn’t delay 1.6.

 

I agree and have been doing it for quite a while after having a brief discussion about this with Crimson on these forums. My race events always skip practise, do a 20 min quali and then race and I've had mostly excellent racing with the AI over the last couple of years.

 

People were disappointed for good reason and after waiting 8 months since the last update we wanted to see something other than F1 at Hockenheim, Group C at Kyalami, F1 at Nurburgring ... we hoped the focus would be on new stuff, and people were criticizing the quality of the stream overal, 720p at 30 fps, the camera blocking the view of the new UI, and especially that driving of F1 on at midninght on a track that was not made for it. People were expecting a more professional approach and the stream was obviously bad because Overtake took both parts down and not scheduled a new one.

 

One of the things with AI that doesn’t get mentioned very often is what are peoples expectations of where they will finish when running against them.

If I start mid pack and finish top 5 that’s probably been a decent race. If I start top 5 and I’m leading after a few laps I’ll quit the race and run it at a higher difficulty as its just hot-lapping at that point.

Starting at the back with AI set too low and trying to overtake the entire field also seems a bit pointless.

I’d rather occasionally scrape a win or top 3 and be challenged than consistently win the races.

 

I am also not sure what are many people's expectations when it comes to AI. Like the bunching up people are describing not realizing that those situations happen constantly in real life too.


What i hope is that the AI can eventually figure out a way through the race to make the pass. I can also say that from racing online there are many races where i followed someone for 10 minutes or even 30 and none of us made a pass on anybody, we just drove like a train because we were on equal skill and pace at all times with nobody making any mistakes, so why is it considered unrealistic when AI do that too ?

 

Also, I don't think people are criticising Reiza. The stream was by Overtake. They are an independent media entity. The failure of the stream was all on Overtake's shambolic organisation. Not the guy doing the stream but the lack of planning and focus that obviously hadn't happened beforehand.

 

I fully agree with doing a proper qualifying session to see what the pace of everyone compared to the player actually is.

But i do not agree with showing restrain, because why should i ? If i am able to exploit the AI and find an easy way through then heck yeah i will do it, at least in the current 1.5.X.X. version it is way too easy to do so and i am curious how different it will be in 1.6.

Though i have my suggestion what helps show restrain - set damage type to Full and scaling to Maximum, then you will actually have to show much more respect because now if you misjudge a corner and bash someone's rear you get proper aero or even suspension damage for careless driving.

 

Damn you are right, they removed all of it