Atticus' Workthread - [RE-RELEASED] Updated Road Atlanta (1997) Beta - All Links on p. 1

Turbo Lover Wrote:
-------------------------------------------------------
> Thanks mate, it's a big shame that the modern
> track doesn't have challenging corners anymore.
> Maybe the most soulless Tilke track.


Yeah, it turned out to be that - it looked better on paper. But yes, apart from the Esses, it's literally ALL low-speed corners - even the second part of the Esses was tightened up a bit compared to how the initial rebuild design looked like... I think the classic layout is much more diverse, more challenging and faster. (That is not to say the work of FlashFormula on the new layout is bad, I think it's quite good, it's just that the layout itself is woeful.)

---

My workthread - [www.grandprixgames.org]
Full of classic F1/non-F1 track layouts

My blog about F1 performance analysis - [thef1formbook.wordpress.com]

I very much hope that this will actually blow some minds...





It's been sitting on my hard drive since May, but I did some more work on it before the motorsport world ramps up with the annual Macau GP and I plan to release it the days before the event itself (which is this weekend).

Now, I do have some problem with it, and I'd very much welcome some feedback.

The smaller issue is that the Melco hairpin creates too big a gap between cars - it does so in real life too, so that overtaking is indeed almost impossible into Lisboa (despite the huge straights), but it's a lot tighter in the real world than it is in here. In reality, you do have to be perfect to maintain position at the end of the main straight(s), here the AI isn't even close. Things I've tried: adjusting the CC-line, the alleged 'slipstream' value in the magic data (it's not really a slipstream, it's that field_c2 value deducted from something) and playing around with strategies (to make some cars lighter on certain stages of the race).

The bigger issue is that the circuit replicates those infamous Lisboa mass crashes all too often - the AI crashes there 99% of the time, no matter what. They accelerate up to 360-365kph at the start with the pack creating that big bubble of air (they don't do it anymore after negotiating Melco for the first time) and then they always overshoot an outbreaking attempt somehow. It's either the overtaker who bumps the guy on the outside into the Lisboa barriers, or it's the guy on the outside who simply runs wide and crashes. It seems the AI just can't take a narrowing track and a deceleration from 370kph to 100kph (it might be programmed to borderline navigate the Monza chicane safely in terms of aggressiveness which is a lot shorter distance from pole to T1). Things I've tried: turn Careful AI on/off and decrease AI errors.

---

My workthread - [www.grandprixgames.org]
Full of classic F1/non-F1 track layouts

My blog about F1 performance analysis - [thef1formbook.wordpress.com]



Edited 15 time(s). Last edit at 11/19/2015 03:16PM by Atticus..

You might want to discuss with klausfeldmann about Macau actually or take a look at what he did for it, he has a beta version of the track that I've played around with. A link for it is in my database/here (hopefully he is ok with me putting a link!): [drive.google.com]

It was never officially released as such as its DAT only, but from the running I did on it the cars stayed together at the Melco hairpin and even attempted overtakes at turn 22 after it. I also don't remember their being massive crashes at Lisboa but I didn't do enough testing to get a good read on it. It's probably worth a look to compare your solutions!

_________________________________________________

For a list of EVERY download for GP4, look here: [docs.google.com]

Thank you.

We've already been talking about each others work (via PM) - even back in May. I took a good look at his files, but he has not built a CC-line yet, so I can't really compare.

---

My workthread - [www.grandprixgames.org]
Full of classic F1/non-F1 track layouts

My blog about F1 performance analysis - [thef1formbook.wordpress.com]

Yeh apologies I realised this right after sending. I actually built a cc-line based off of his track, which was what I ended up testing. The cc-line I made generated the results I mentioned above. Its by no means perfect as I'm no expert, but it did a job of getting the cars moving round the circuit. I imagine it's nowhere near what you have so far, but you are welcome to take a look in case it does help: [www.mediafire.com]

EDIT: Oops wrong account... sent by truecrysis



Edited 4 time(s). Last edit at 11/17/2015 01:09AM by F1_Manager.

F1_Manager Wrote:
-------------------------------------------------------
> Yeh apologies I realised this right after sending.
> I actually built a cc-line based off of his track,
> which was what I ended up testing. The cc-line I
> made generated the results I mentioned above. Its
> by no means perfect as I'm no expert, but it did a
> job of getting the cars moving round the circuit.
> I imagine it's way off what you have so far, but
> you are welcome to take a look:
> [www.mediafire.com]
> s1ct08.dat
>
> EDIT: Oops wrong account... sent by truecrysis


Haha, OK, thank you, I'll check it out.

---

My workthread - [www.grandprixgames.org]
Full of classic F1/non-F1 track layouts

My blog about F1 performance analysis - [thef1formbook.wordpress.com]

I thought up removing the customised magic data altogether and things instantly changed dramatically - lap 1 crashes are less frequent (although still annoyingly usual), and cars can follow better which results in more overtakes.

There's a lot of magic data changes, starting with the track grip, air pressure, fuel, CC shuffler, to wing levels, setups and bump factors and strategies (the original is Montreal) - I've researched every value as best as I could so I will now revisit my sources to see if there are some leeways in certain values.

---

My workthread - [www.grandprixgames.org]
Full of classic F1/non-F1 track layouts

My blog about F1 performance analysis - [thef1formbook.wordpress.com]



Edited 1 time(s). Last edit at 11/17/2015 01:36AM by Atticus..

Embedding : Remove the S in HTTPS:// that's all

Briere Wrote:
-------------------------------------------------------
> Embedding : Remove the S in HTTPS:// that's all


Thank you, it worked.

---

My workthread - [www.grandprixgames.org]
Full of classic F1/non-F1 track layouts

My blog about F1 performance analysis - [thef1formbook.wordpress.com]

OK, I switched the AI from low downforce to medium downforce in the magic data and it helped the main problem, the crashes at the starts, immensely. There's still more than usual and you will definitely be surprised by some scenarios, but hey, it's quite frequent there in real life as well.

Medium downforce is not necessarily unrealistic either. In fact, it gives better laptimes in az F1 car than low downforce - presumably cars just use low in reality to protect position into Lisboa and it doesn't matter if you are slower elsewhere as there's zero opportunity to pass. It's especially not a disadvantage if everyone opt for shallow wings, which is the case usually. That said, this means that you WILL have the opportunity to go lower with YOUR wings in-game and pass the AI at will - but I cannot make it playable otherwise. :/ (And you will still have to be able to keep up with cc cars in the twiddly bits.)

I also decreased the AI aggressiveness factor a bit, that helped a lot as well.

Passing (among AI cars) is still a problem though, but that may stay like that if all else fails - it's not uncommon in Macau either.

@truecrysis: I inspected your .dat closely and figured that my Melco can actually bear with a tighter Radius 1 value in the CC-line - although due to my approach to the corner, it cannot take less than 2250 (as opposed to your 2070). It means that, with a wider Radius 2 value, the CC-line now promotes better acceleration out of the turn, it emphasises exit speed, which is good and important as well - but it makes only a minimal difference, if any at all, sadly. (Fisherman's and R bends are better in my version I think - they are also increasing radius curves on the exit which is how it should be due to the long straights that follow them. I noticed AI cars closing up nicely in them, just not enough to make a move at Lisboa, so I think they're alright. I may still make them increasing radius the whole way though.)

---

My workthread - [www.grandprixgames.org]
Full of classic F1/non-F1 track layouts

My blog about F1 performance analysis - [thef1formbook.wordpress.com]



Edited 1 time(s). Last edit at 11/18/2015 11:37AM by Atticus..

Practice and qualifying sessions are now underway at Macau as we speak, so if you want to take a lap or two yourself, you now have the opportunity to do so in Grand Prix 4 as well - and on a layout that reflects its real world counterpart to the smallest of details.

Please note the bugs that I wrote about in the gp4form file and above in this topic and be careful if you adjust settings such as slipsteam (in GP4 Tweaker) or physics as anything other than vanilla is untested in this respect. I won't say the game will crash - but the AI may just as well do so...

DOWNLOAD LINK is on page 1 as usual.

Enjoy and have fun.

---

My workthread - [www.grandprixgames.org]
Full of classic F1/non-F1 track layouts

My blog about F1 performance analysis - [thef1formbook.wordpress.com]

As some of you may have already noticed, Google Chrome finds the .rar file malicious and blocks it for some reason.

I feel I have to emphasise that I by no means intend to cause harm to your computer, I have already pulled the file down, ran multiple scans through it (avast! and COMODO) and I can confirm that it is 100% that it is a false positive.

I am working on a solution as we speak and I hope I will be able to find a fix very soon.

Stay tuned.

EDIT: It's not a Macau-specific problem, now I can't even download my earlier works either. It must be that the Chrome auto-detectors find my whole Mediafire account unreliable - or, temporarily, the whole Mediafire hosting service unreliable. (Currently testing the latter with other files.)

EDIT 2: Nope, it's just my account.

EDIT 3 (25th November 2015): The problem just miraculously resolved itself. Has anybody else actually had this, or it was just me messing up something with my Chrome? :P

---

My workthread - [www.grandprixgames.org]
Full of classic F1/non-F1 track layouts

My blog about F1 performance analysis - [thef1formbook.wordpress.com]



Edited 3 time(s). Last edit at 11/25/2015 08:15PM by Atticus..

OK, here is the situation:

- Google Chrome currently blocks all my files on my Mediafire account for some reason. It claims they are malicious - contain malwares/unwanted softwares - but I specifically checked each and every one of them with the latest version of avast! and I can safely say they are clean. No harm has been done to your computer by downloading my earlier works.

- Google has a manual review request function to which one can apply with his or her site for them to check it looking for malware. Problem is, I cannot find my individual Mediafire account ID, it's just not listed anywhere (the title bar only reads [www.mediafire.com]).

- I will now contact Mediafire to tell me if I can somehow see my unique account ID and if they provide me that then I can contact Google to maybe review my account. I believe this is the only way to get a guarantee that my files are safe and clean.

- If all else fails, I will port all my uploads to another hosting service, although that will not guarantee my 'innocence' in your eyes, so that's definitely a fallback option at this stage.

I am sorry for the inconveniences. If you still believe me, please try using another web browser to download my tracks in the meantime - I'll restore the Macau download for you as well (now knowing that it's not the culprit).

EDIT: The problem somehow went away by itself. Chrome allows downloads from my account once again. Yay.

---

My workthread - [www.grandprixgames.org]
Full of classic F1/non-F1 track layouts

My blog about F1 performance analysis - [thef1formbook.wordpress.com]



Edited 2 time(s). Last edit at 11/25/2015 08:16PM by Atticus..

Thanks very much! Just downloaded and will give it a spin over the weekend, much appreciated!

Edit: God damn it, I have got to log out of this account more often...



Edited 1 time(s). Last edit at 11/20/2015 12:17AM by F1_Manager.

F1_Manager Wrote:
-------------------------------------------------------
> Thanks very much! Just downloaded and will give it
> a spin over the weekend, much appreciated!
>
> Edit: God damn it, I have got to log out of this
> account more often...


Haha, thank you.

As for my security issue, this is for the F1 Virtual guys up here: I know you also use Mediafire as a hosting service, did you maybe have any similar issues with them in the past?

EDIT: Nevermind, it works now. (Somehow.)

---

My workthread - [www.grandprixgames.org]
Full of classic F1/non-F1 track layouts

My blog about F1 performance analysis - [thef1formbook.wordpress.com]



Edited 1 time(s). Last edit at 11/25/2015 08:17PM by Atticus..

Can I ask what value you changed to 'make the ai go from low downforce to medium'? Does this value effect the player?

Atticus. Wrote:
-------------------------------------------------------
> OK, I switched the AI from low downforce to medium
> downforce in the magic data and it helped the main
> problem, the crashes at the starts, immensely.
> There's still more than usual and you will
> definitely be surprised by some scenarios, but
> hey, it's quite frequent there in real life as
> well.
>
> Medium downforce is not necessarily unrealistic
> either. In fact, it gives better laptimes in az F1
> car than low downforce - presumably cars just use
> low in reality to protect position into Lisboa and
> it doesn't matter if you are slower elsewhere as
> there's zero opportunity to pass. It's especially
> not a disadvantage if everyone opt for shallow
> wings, which is the case usually. That said, this
> means that you WILL have the opportunity to go
> lower with YOUR wings in-game and pass the AI at
> will - but I cannot make it playable otherwise. :/
> (And you will still have to be able to keep up
> with cc cars in the twiddly bits.)
>
> I also decreased the AI aggressiveness factor a
> bit, that helped a lot as well.
>
> Passing (among AI cars) is still a problem though,
> but that may stay like that if all else fails -
> it's not uncommon in Macau either.
>
> @truecrysis: I inspected your .dat closely and
> figured that my Melco can actually bear with a
> tighter Radius 1 value in the CC-line - although
> due to my approach to the corner, it cannot take
> less than 2250 (as opposed to your 2070). It means
> that, with a wider Radius 2 value, the CC-line now
> promotes better acceleration out of the turn, it
> emphasises exit speed, which is good and important
> as well - but it makes only a minimal difference,
> if any at all, sadly. (Fisherman's and R bends are
> better in my version I think - they are also
> increasing radius curves on the exit which is how
> it should be due to the long straights that follow
> them. I noticed AI cars closing up nicely in them,
> just not enough to make a move at Lisboa, so I
> think they're alright. I may still make them
> increasing radius the whole way though.)

RHill Wrote:
-------------------------------------------------------
> Can I ask what value you changed to 'make the ai
> go from low downforce to medium'? Does this value
> effect the player?

Strictly speaking, that itself does not, but I did tweak the player car setup as well to make it similar to the AI. So now, for example, you can't reach 355kph before Lisboa without touching your setup in-game, that's not the default for this track. (But as I've said before, it's not a problem, the ideal aero setup is actually medium downforce here.)

---

My workthread - [www.grandprixgames.org]
Full of classic F1/non-F1 track layouts

My blog about F1 performance analysis - [thef1formbook.wordpress.com]

Atticus. Wrote:
-------------------------------------------------------
> RHill Wrote:
> --------------------------------------------------
> -----
> > Can I ask what value you changed to 'make the
> ai
> > go from low downforce to medium'? Does this
> value
> > effect the player?
>
> Strictly speaking, that itself does not, but I did
> tweak the player car setup as well to make it
> similar to the AI. So now, for example, you can't
> reach 355kph before Lisboa without touching your
> setup in-game, that's not the default for this
> track. (But as I've said before, it's not a
> problem, the ideal aero setup is actually medium
> downforce here.)

When you say you have set medium downforce in the magic data for the ai, what value are you referring to?

RHill Wrote:
-------------------------------------------------------
> When you say you have set medium downforce in the
> magic data for the ai, what value are you
> referring to?


Um, wing levels? :P

(And the gear ratios accordingly, of course.)

EDIT: It's not that hard, really. They are the first few values in the .md3 files as CMagic and GP4 Builder both indicate. (If you were trying to ask that, specifically.)

---

My workthread - [www.grandprixgames.org]
Full of classic F1/non-F1 track layouts

My blog about F1 performance analysis - [thef1formbook.wordpress.com]



Edited 1 time(s). Last edit at 11/25/2015 12:20PM by Atticus..

Hi guys,

I've fixed a rather major bug in Monaco Historic which caused often 60-70% of the field to crash out of a full-distance race at the Chicane.

Unfortunately, this meant that I must had to remove the old wall on the inside of the complex to make room for the AI to navigate it when passing lapped cars. (So the cars can now use all the space that is available nowadays.) This and a reduction of AI errors in the magic data greatly decreased the number of DNFs, but sadly made the old chicane considerably easier to navigate. If you still prefer the tighter version and you don't bother the AI killing themselves off during longer races, I've included the old configuration in a separate folder. (It's still scary.)

Also, in Imola Classic, I've fixed the CC-line at Piratella so that the AI does not decelerate as awkwardly as before and the line flows better on the run down from Acque Minerali to Rivazza (creating an overtaking spot for the AI in the process).

If I were to update the files, I took care to improve on a number of little details, such as the driving lines, strategies, jip mappings, setup and some track cams in Monaco and track cams, strategy and some verge inconsistencies near the pit entry in Imola.

I think that's basically it for 2015 from me - but watch this space as I have some nice surpises (of the Macau release magnitude) in line for next year.

Starting a few days before the 2016 F1 testing season...



P.S.: Thank you for the jip mappings again, Turbo Lover, they are precisely for those works next in line.

---

My workthread - [www.grandprixgames.org]
Full of classic F1/non-F1 track layouts

My blog about F1 performance analysis - [thef1formbook.wordpress.com]



Edited 1 time(s). Last edit at 12/13/2015 03:12PM by Atticus..