[MVRC] Mantium Virtual Racecar Challenge 2016

[Ft5fTL]

I shaved off some frontal downforce because the car had massive oversteer at the entry of the corners. And i messed up the lap. Ai is way faster than me

Edit: Seems like i edited the wrong part of the video for the exterior cam. I should stop doing things when i need to sleep.

[CAEdevice]

https://www.youtube.com/watch?v=_Brl36K-odU

I shaved off some frontal downforce because the car had massive oversteer at the entry of the corners. And i messed up the lap. Ai is way faster than me

Edit: Seems like i edited the wrong part of the video for the exterior cam. I should stop doing things when i need to sleep.

Very good job! I wonder what would happen if we could see our races in such a 3D environement

[Ft5fTL]

Drafting would change things drasticly on front. If anyone interested i can host a 1 hour race and upload the replay file for someone who has a better connection to upload on youtube.

[CAEdevice]

I shaved off some frontal downforce because the car had massive oversteer at the entry of the corners.

Have you imported the aero properties only or also the mechanical features (es. CoG?).

[Ft5fTL]

I shaved off some frontal downforce because the car had massive oversteer at the entry of the corners.

Have you imported the aero properties only or also the mechanical features (es. CoG?).

The only thing that i didnt change was the suspension geometry and the tires. Since both are from a Lmp mod it should be fine. The engine has lower power output due to lack of hybrid system. I tried to compensate that with giving the engine extra 90 bhp. So instead of producing 540+144 bhp, it produces around 630 bhp. For the other features, Virtual stopwatch saved me some guessing work.

[CAEdevice]

Very interesting and well done.

In my experience (2015 KVRC car + AssettoCorsa), the car was easier to drive with the CoP slightly behind the CoG (about 50mm). I alsobhad problems with the tye model: KVRC/MVRC tires are a bit too narrow compared to power and weight.

[Ft5fTL]

After driving the car on a simulator, the corrected CoP value makes sense. I couldnt drive my stock car due to the 1.610 CoP, it was oversteering like a drift car. Had to lower the front wing by 2 clicks (around 300N) for a better driving experience.

[Alonso Fan]

Must have taken some serious work getting that into rfactor. Kudos

[CAEdevice]

I wish you a happy new year, low residuals , high downforce and, above all, an astonishing df/dr ratio

[LVDH]

https://www.youtube.com/watch?v=_Brl36K-odU

I shaved off some frontal downforce because the car had massive oversteer at the entry of the corners. And i messed up the lap. Ai is way faster than me

Edit: Seems like i edited the wrong part of the video for the exterior cam. I should stop doing things when i need to sleep.

This is super cool. I really hope that some day we can easily import the cars and properties into a realistic car simulator and let the AI run the races. This would encourage some interesting fine tuning. It would be possible to sacrifice some down force and efficiency to improve drag. While it would increase the lap time, overtaking with the car would be easier due to the higher top speed. In the end that then might win. Apart from that we get nice videos instead of just numbers.
So who will start working on a process that we can automate to achieve this small goal?

[Ft5fTL]

I can teach the basics about rF2 modding and supply with modding resources but i dont think i can spare enough time for add each competitors cars. The fastest solution could be using a single car model as generic and creating the teams with that car model with different hdv (where you input the numbers) files.

[LVDH]

Maybe I could just write an email to the teams behind the best games. I was able to play a bit with my "little" cousin during the holidays but I am not really up to date when it comes to games. Who should I try? I know rFactor, Project Cars, Assetto Corsa.

[CAEdevice]

I had good modding experience with Assetto Corsa (I have been in contact by Linkedin with the developers). The graphics are very very good and the physics (including sospension design, aero, kers, ...) can be edited with text files.

The AI is getting better and better, but I am not completely happy. The simulator with the best AI is TORCS, but the graphics and the modding are not very good...

The biggest problems (fornall the driving simulators) are:

- Very much time needed to import a car (many ours for me)
- Do we want a "pure" aerodynamics challenge or can we accept a deep impact of AI and random events on the results?

By the way: I can't wait for the new solver settings and the new rulebook proposals.

[Ft5fTL]

As far as i know pCars doesnt support moding. People who tried Asetto and rF2 says that AC has the looks and rF2 has the feels. I agree with that assessment but it doesnt feel that too much different when you drive. AI is pretty good on rF2 since there was a singleplayer only module. Ai quality is mod depended though. I dont have any idea of file structure of AC for moding but on rF2 its pretty detailed. Anyone interested can check these:

http://wiki.rfactor.net/index.php?title=Hdv
http://yoss.free.fr/rfactornfr/Modding_ ... .shtml#HDV

These are for rFactor 1 but its mostly the same on the second game.

Besides that rF2 has a realroad technology that creates rubber on the racing line dynamicly.

- Very much time needed to import a car (many ours for me)
- Do we want a "pure" aerodynamics challenge or can we accept a deep impact of AI and random events on the results?

I think if we decide to use a simulator, it should be just for show. Ai can definetly mess things up.

By the way: I can't wait for the new solver settings and the new rulebook proposals.

How about changing the intro class to GTE class style cars? One chassis without bumpers, wings, engine cover, diffuser etc.

[etsmc]

my opinion is that this is a CFD Challenge and the way the races are done now is ok...
don't get me wrong what Ft5fTL has done is fantastic and maybe a separate challenge would be good.

How I see it running would be
Normal CFD challenge is run as it is now
At the end of the CFD challenge, you nominate the version of your car that you think has the best chance at an Ai Challenge
The car you nominate is put into a game (model and CFD numbers)
3 races are simulated with the Ai on 3 different tracks (high, medium, low downforce) or however many you want really.

This means the cars only have to be created once to go into the game, if the Ai is scriptable we could choose a driver or they get picked depending on the standings in the CFD Challenge.

doing it this way I feel would be less time intensive even the suggestion of a generic model with just the CFD information would be acceptable.