Jump to content

AI in the wet - all circuits

Recommended Posts

There IS something wrong with the AI in the wet.

I've been racing these F1 games with PlayStation (nearly 20 years), and a wheel (the last 10 years), and with F1-2019, in comparison to previous F1 games, I've noticed that the AI are wrong.

How wrong? They're wrong because they're absolutely perfect in the wet.

The AI don't spin, they don't lock-up, they never go off the racing line, they're faster in a straight line; ...they are brilliant, perfect, in the wet. That's what is wrong. It's unrealistic.

I'm very competitive (with the AI) in dry conditions, but I'm no slouch in the wet either. But in F1-2019, …..I get absolutely blown-away by the AI, in both methods (see below) that I've run.

Wheel: Logitech G29

PAST                                                        CURRENT

AI: 90                                               AI: 60

View: T-Cam                                     View: Cockpit

Assists: TC (med), and ABS                 Assists: Nil

 

Just my unprofessional opinion regarding my personal experience with these F1 games over a number of years. 

  • Agree 1

Share this post


Link to post
Share on other sites
Posted (edited)

@Striker_703

It doesn't proof anything. That's the point you can't accept. It is fine that you can compare with AI in the wet. But it's not an evidence.

If i'm driving against 80 AI and i can compare in dry conditions, but i'm 3 sec slower in the wet. What does it proof? Would you say the AI is OP in the wet because i can't beat them? The only thing i would proof would be, that i'm not a good racer in wet conditions.

You can't proof anything by yourself. Maybe you are a extraordinary driver in wet conditions. To get a comparable result you need 10.000 drivers with same skills and under same conditions. Every of them should drive 100 laps in dry conditions and 100 laps in wet conditions. Then you could take the average and you would have an approach. But again that's not an evidence.

Edited by Akkan74
  • Agree 3

Share this post


Link to post
Share on other sites
Posted (edited)

@Striker_703 As a software designer you should know that bugs can be tied to very specific circumstances. Your test does not prove anything other than that the issue is not always there.

The problem might not even exist in multiplayer, or on lower AI settings like the one you use. Also, you're testing this in a practice session; online AI is absolutely not representative there. If you want to test, at least go for GP mode and short or full qualifying where AI will push and be competitive. 

I watched a fair bit of your stream and I guarantee you that with the way you drove, you would have been 3 seconds behind the AI on that same level in the session where I experienced this issue. That means it was simply not present in your sessions. And sure, I'll see if I can upload some video if I encounter the issue again in the coming days. 

Edited by Worntoathread
  • Agree 5

Share this post


Link to post
Share on other sites

@Worntoathread I'm going to be testing the tires from 2018 to 2019 to see if there's a bug in that area. As for the AI, I don't see any. I set the AI to were I was comfortable with. I done the test live, gave screens shots. The screens shots shows how close the AI and I were. So now all I'm asking is to prove a software engineer wrong that has developed AI before. All I'm asking.  

  • Disagree 1

Share this post


Link to post
Share on other sites
Posted (edited)

Have one of the top eSPort drivers to the test for you guys at 110 AI and compare the times. If you can't run the test or are unwilling. Have one of the do it. But they have to be a wheel driver. I used a G29 with no assist. If you guys really want to see if there's anything wrong with the AI have a top eSport wheel driver do the test.

Edited by Striker_703
  • Disagree 1

Share this post


Link to post
Share on other sites
Posted (edited)
7 minutes ago, Striker_703 said:

@Worntoathread I'm going to be testing the tires from 2018 to 2019 to see if there's a bug in that area. As for the AI, I don't see any. I set the AI to were I was comfortable with. I done the test live, gave screens shots. The screens shots shows how close the AI and I were. So now all I'm asking is to prove a software engineer wrong that has developed AI before. All I'm asking.  

What does 2018 have to do with it? My comment was about your testing method. Like I said, if you want to test half properly, at least do it against the offline AI in qualifying and not online practice where the AI often cruise around or barely even go out at all. You can't call a test like that reliable evidence.

Edited by Worntoathread
  • Agree 2

Share this post


Link to post
Share on other sites

Where do you have the AI set at @Worntoathread? People are complaining that the AI is faster than THEM in the wet. Here are some key words. 'The end user' meaning you or I, has set the AI to a setting of THEIR desire. Now the end user finds that the AI is faster than the end user in a different condition. That condition being a wet track. The end user as set or has not set assist. Who has set the AI people? Did Codemasters change the AI level when you got onto the wet track?  

  • Disagree 1

Share this post


Link to post
Share on other sites

If I lower the AI I'm going to be faster than the AI, if I raise the AI I'm be slower than the AI. I can do it again live so all of you can see for yourselves. How about. Or do the test the yourselves or find someone that has the skills.  

  • Disagree 1

Share this post


Link to post
Share on other sites
Posted (edited)
19 minutes ago, Striker_703 said:

Where do you have the AI set at @Worntoathread? People are complaining that the AI is faster than THEM in the wet. Here are some key words. 'The end user' meaning you or I, has set the AI to a setting of THEIR desire. Now the end user finds that the AI is faster than the end user in a different condition. That condition being a wet track. The end user as set or has not set assist. Who has set the AI people? Did Codemasters change the AI level when you got onto the wet track?  

Okay, it seems at this point you don't actually understand what the issue is. If that's the case I'm not sure if there's any point in trying to explain it further.

Myself and others made an effort to give you some good points and examples which you all decided to ignore in your responses because you only seem focused on your video tests and software engineering experience, which have little to do with the actual issue we are trying to explain here. That's where I'll draw the line as this is starting to feel like I'm talking to a brick wall unfortunately. 

Edited by Worntoathread
  • Like 1
  • Thanks 1
  • Agree 2

Share this post


Link to post
Share on other sites

I just raced in Austria here for career mode. The race started with rain during the first 5 laps. I started last due to engine component change. And it just couldn't reach anyone, not even the worst cars, being racing with the 5th best car. I restarted the session and passed traction control from full to medium and my performance improved a lot, I managed to pass a few cars, and when it dried I was 16 °. So I think the problem may be strongly linked to the complete traction control. Does anyone have this problem playing with traction control on medium or off?

Share this post


Link to post
Share on other sites
Posted (edited)

I'm going to do the test again. At Silverstone and I'll be raising the AI to 90. I'll edit this post in a moment with the feed. There won't be any feed from the camera. I want to split times visible. You'll be able to hear me though. 

I'll start the test in ten minutes. Give people enough time to join and watch if they care to. I'll be raising the AI to 90 in this test. I even sent the link out on twitter lol.

 

Edited by Striker_703
Added a link

Share this post


Link to post
Share on other sites

I am experiencing this as well. In career, 50% race at USA, heavy rain conditions, the AI were roughly 6-8 tenths faster than me per lap.

I play 105 difficulty and can keep up well in the dry, but since the patch, struggle tremendously now.

Share this post


Link to post
Share on other sites
Posted (edited)

I decided to add my 10 cents to this! For your information, I don't have the 19 game and given how broken the career mode looks I doubt  if I'll ever buy, even when all patches are out and the game is bargin basement price. I have, however, had every other game in the f1 series. It's just an interesting "debate!"

I see your stream striker. I also see you are still testing the wrong game mode, people are talking about career here, not online. Quali and race, not free practice.

I also see you've upped the difficulty and are on a different track yet still fastest in the dry so far. This at least shows a variance of AI track to track, or that your faster at Silverstone's than Catalunya.

Now we have to see what you can do in the wet...

Edited by ChasteWand
typo

Share this post


Link to post
Share on other sites
Posted (edited)

Apologies for my poor memory, both tests at Silverstone. My addled brain remembered Catalunya!

If you upped the AI by 10 then they should be in about 2 secs faster than the previous test, given 1 is about equal to 2 tenths. Or at least has been in previous games. Did you really up the AI? I didn't see any session setup in the stream as I joined when you were already lapping.

You seem not to understand ers deployment. Mode 0 no deployment, through to mode 4 highest deployment. Mode 5 is hot lap and varies in its deployment for max performance over a lap. That's how I understand it. Mode 5 will give better traction in lower gears compared to mode 4, as there is less power being produced by the PU.

Edited by ChasteWand
  • Agree 1

Share this post


Link to post
Share on other sites

I've finished the test. And I'm about to end the stream. Once I end the stream I'll be taking screen shots of the times and posting them here. You all can look over them, along with watching the replay of the stream shortly after it ends. 

Yes @ChasteWand I raised the AI to 90. You can watch the being of the stream shortly after it ends. 

Share this post


Link to post
Share on other sites

Here are the screens shot of my time sheet and all the AI's times sheets. You can look over them yourselves. Please look at the replay of the stream so you can see the beginning of it so you can see the settings that I'm using. The fastest AI on full wet tires was Marie Laursen with a 1:44.748 on lap 12. My fastest lap on full wets was a 1:44.882. -0.134 slower than the AI. Thank you to those that have watched the live stream. 

F1® 2019_20190812222335.jpg

F1® 2019_20190812222344.jpg

F1® 2019_20190812222350.jpg

F1® 2019_20190812222408.jpg

F1® 2019_20190812222429.jpg

F1® 2019_20190812222446.jpg

F1® 2019_20190812222502.jpg

F1® 2019_20190812222516.jpg

F1® 2019_20190812222543.jpg

F1® 2019_20190812222602.jpg

F1® 2019_20190812222626.jpg

F1® 2019_20190812222644.jpg

F1® 2019_20190812222700.jpg

F1® 2019_20190812222721.jpg

F1® 2019_20190812222738.jpg

F1® 2019_20190812222756.jpg

F1® 2019_20190812222811.jpg

F1® 2019_20190812222828.jpg

F1® 2019_20190812222850.jpg

F1® 2019_20190812222911.jpg

Share this post


Link to post
Share on other sites

The replay of the stream can be viewed now. You can jump to time stamp 9:00 to beginning of the settings. 

Share this post


Link to post
Share on other sites

Yes I see the AI at 90.

Congratulations on setting pretty consistent lap times compared to the AI 80 Silverstone test on the inters.

There is however a number of problems with the tests, the stream for the AI at 80 test went down and you have no screenshots. You need to go back and redo the AI at 80 test in all conditions at Silverstone, excactly as you've just done, same time of day, track temp etc. then post the screenshots. Would be interesting if you did other tracks too in exactly the same way. Monaco and Bahrain I would suggest, or other places where the AI is rumoured to be over or under powered. Then we can have a true comparison. But that would be only in the online mode of the game with AI. Multiplayer is for player v player, not player v AI, and thats not what the complaints have been about. The complaints I hear on the forum are mostly for career mode. That would need testing too!

As a man who writes code I'm sure you can appreciate that, whilst in theory, the AI is the same in all modes, code can and does interact in unpredictable ways. So to really test what some end users are complaining about you need to do this in all modes, only then can we see where the AI is at against you, and how they compare at different levels of difficulty.

Also bear in mind you are one of millions of end users and your skills and experience will differ considerably from theirs. As others have said you might be demon fast in the wet, and slow in the dry comparatively. Maybe you found a setup/exploit that gives you that speed. You have not posted your setup, another weakness in the test.

If this is to be settled then many players will have to test an post the results themselves...

Have fun doing codemasters work for them, they might listen, you might even get a free copy of the 2020 game for your effort!

 

 

  • Agree 1

Share this post


Link to post
Share on other sites
Posted (edited)

Thank you for watching the replay @ChasteWand. I understand the main complaints are about the AI in career mode. But the AI is the same in mode you're running in. The reason why I didn't do any of the test in care mode is due to car development. That plays a variable that no one looked at. The cars aren't equal. Which is why I done all the test in a un-ranked private lobby with the multiplayer car. It's all comes down to the value you have the AI set to. I'm not sure I'm be doing anymore test like this. And testing against driver/player-driver/player would mean that we both must be on the same settings. I don't use any driving assist as you seen. I would be more than happy to share my set-up for Silverstone. I'm not sure how that would help though. The cars aren't equal. So testing the AI in career mode is pointless. Car development come into play.

I would rather get a better wheel so I can see how fast I really am. I'm losing to much information coming from the G29. 

Edited by Striker_703
Added a comment

Share this post


Link to post
Share on other sites
Posted (edited)
57 minutes ago, Striker_703 said:

Thank you for watching the replay @ChasteWand. I understand the main complaints are about the AI in career mode. But the AI is the same in mode you're running in. The reason why I didn't do any of the test in care mode is due to car development. That plays a variable that no one looked at. The cars aren't equal. Which is why I done all the test in a un-ranked private lobby with the multiplayer car. It's all comes down to the value you have the AI set to. I'm not sure I'm be doing anymore test like this. And testing against driver/player-driver/player would mean that we both must be on the same settings. I don't use any driving assist as you seen. I would be more than happy to share my set-up for Silverstone. I'm not sure how that would help though. The cars aren't equal. So testing the AI in career mode is pointless. Car development come into play. 

 

 

Sorry but I disagree. As I said:

 

 

 

1 hour ago, ChasteWand said:

As a man who writes code I'm sure you can appreciate that, whilst in theory, the AI is the same in all modes, code can and does interact in unpredictable ways.

 

Testing has to be done in all modes. Car development is irrelevant to the test, you have a teammate in exactly the same car, you'd be comparing your pace to your AI teammate!

I think you misunderstand me and software testing to a certain extent. Multiplayer testing, player vs player has no bearing on this. I was simply saying that your testing the wrong mode. I understand the cars are all equal in multiplayer, but your somewhat misguided in believing that just because the AI code is the same across all modes, it'll behave in exactly the same way in each mode. There are so many variables. It might, it might not. Only testing can prove it one way or another. As you rightly say, car development is one of the variables, it could be the one that's causing the AI to be so fast in the wet. Personally I doubt it, but it's a possibility, until ruled out by testing in career mode.

Personally I'm not interested in your setup, but for a rigorous test it must be included. Then there is your hardware, the console, wheel and pedals. You mentioned that you have modded your pedals. How do we know your not gaining an unfair advantage there?

I don't wish to belittle your efforts, but all you have achieved is to show you can keep pace with the AI in varying conditions. That is either a testament to your skill as a sim racer, or that you have found a way to cheat, be that a setup exploit, a hardware mod, or software hack. To demonstrate this is really an issue and the AI are too fast in the wet for a given difficulty, is to test with a large number of players under controlled conditions. And it doesn't stop there, you have all assists off, it needs to be tested for all combinations of assists as the issue may be lurking there. Likewise it might be grip values used by the AI, or any number of other things. Only codemasters can get to the bottom of it, and they might struggle as modern games are very complex webs of code.

I

 

Edited by ChasteWand
typo

Share this post


Link to post
Share on other sites
Posted (edited)

The only way to get a good reading of the AI in career mode @ChasteWand is to have a fully developed car. On the PC side  major PC YouTube streamers had keys to do so. Pending on how the AI is coding it will always make the correct choice when it come to car development in career mode. The only way to so if that's the case is to look at the code on the PC. I don't have a PC that powerful anymore nor do I have a PlayStation 4 emulator. Now if someone from the PC side can get me a copy of the AI coding, then I could tell you what the AI is doing whilst in career mode.

Edited by Striker_703

Share this post


Link to post
Share on other sites

I agree testing would be easier with all cars being equal, however as I said before, your teamate is in an equal car in the f1 career and can be used as a benchmark at any stage of development.

Yes f2 needs testing as well. Grand Prix mode too if it still exists. I repeat myself, but I don't have the 19 game, given what I've read here on the forum I don't think it's worth the investment. I will stick to the 18 game for my f1 fix, to be honest though I prefer a more sim based game such as Project Cars or iRacing.

As I final note, I don't think this is the wisest place to be discussing hacking into codemasters games and disseminating the code to be honest!

Share this post


Link to post
Share on other sites
Posted (edited)

I currently have a career in F1 as well. Everyone will just have to follow to see how everything goes. I can live stream practice and qualification as well besides the race. There also others that are streaming a career though @ChasteWand. But I'm the only one that has proven to keep pace with the AI in this thread. 

Edited by Striker_703
Had to add another remark

Share this post


Link to post
Share on other sites
49 minutes ago, Striker_703 said:

I currently have a career in F1 as well. Everyone will just have to follow to see how everything goes. I can live stream practice and qualification as well besides the race. There also others that are streaming a career though @ChasteWand. But I'm the only one that has proven to keep pace with the AI in this thread. 

Your kind of undermining your own conclusion by saying your the only one who can keep up with the AI in the wet.

That suggests that the conclusion is the AI are in fact overpowered and you are just good in the wet given your the minority.

But that only deals with level 80 and isnt exactly a high level of AI ability.

Someone raised the point that they race at a considerably higher level. I believe they said ultimate.

You would need to test the ultimate levels on all tracks wet and dry to see any difference.

Whats interesting though is it is not disputed that the AI are not overpowered on some tracks just in the dry.

You make no mention of this discrepancy of you struggling in the dry at certain tracks, atleast not that i ever recall in your numerous responses.

You use the description of a level that is comfortable to you.

Which leads me to think your racing on a level that is not challenging enough for you as you should be struggling on the dry tracks where the AI are overpowered in the dry so your getting these results because your merely not increasing the AI level and it's become too easy for you hence why your closer to the AI in the wet.

  • Agree 2

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now

×