BMW M4 GT3 Coming to rFactor 2

reshaderacer said:

drive and feel twice as good as AC.
Rf2 has the ier car beat handily.
rf2 simply drives way better.
Rf2s latest cars are the pinnacle of sim racing.
AC is rubbish in terms of driving compared to the latest RF2..

Click to expand...

Trava said:

So how much were you paid?

Click to expand...

reshaderacer said:

The LMP. But at this point don't know why the AC guy tried so hard. Because rf2 and it's tyres drive and feel twice as good as AC.

I bought the ier thing, dude. It's good but I think you'd agree numbers are not feel. They can also be arbitrary.

The mod though was constantly overstated. Rf2 has the ier car beat handily.

At this point rf2 simply drives way better.

I think the entire discussion was excruciating and pointless. Rubbish is rubbish. Rf2s latest cars are the pinnacle of sim racing. It leaves the ier thing in the dust. It's AC, mate, that can't do your numbers justice. AC is rubbish in terms of driving compared to the latest RF2..

Click to expand...

tr1v1um said:

Something that rF2 does out of the box without any form of hacking.

Click to expand...

kart1 said:

@mclarenfpapa
you are not the only engineer working in the field it would be nice to say that you speak for yourself and not the whole profession, for example Andrea Quintarelli said: "It is no surprise to me either that most of these professional simulators (or centres) use rFactor. As I have had the opportunity to prove myself, when you know how to build your models correctly, rFactor produces very accurate results, at a very low cost" and Petros Mak, who is well known in the sim world and who has also worked with pro teams, said: "All of our race team customers and even our series customers, they don't just use rFactor to get familiar with the track. They also use it to test setup data before they go to an event, they use it to test data for potential new parts, developing new brakes or new engine performance and testing them in-game before committing to building them in real life. rF1 and rF2 provide much higher accuracy for these things than any other simulation out there... The physics engine in rFactor 2 is by far the most simulation-based physics engine using real-world aerodynamic and physics data that any other title has come close to."
but maybe the pro teams like ferrari redbul etc... who use or have used the rFpro software for their work are goats...

Click to expand...

mclarenf1papa said:

I'm friends with the engineer who did sim work on RFactor for Core Autosport and he's expressed interest a number of times in switching to AC due to its current feature set. But yes, your argument is spot on and it's why many hesitate to switch platforms. Depending on the software, the learning curve can be quite arduous, and that's actually the main reason he hasn't switched (though he's off writing his own lap sim now anyway and has a full time job in the automotive industry).

Click to expand...

NobodyKnowsReally said:

Unless I stated it wrongly, those are my thoughts as well, an empirical or semi-empirical model is easier to play with. I did not say whether rF2's tool has flaws or not, nor I am interested in it, neither whether or not empirical is better than physical or viceversa, because what I wanted to discuss and reply to was this statement. But that seems to have slipped past.
@mclarenf1papa what do you think of the following statement?

Click to expand...

tr1v1um said:

What amazes me a bit in this whole physics discussion that we compare a product that is still in development with a platform that was so bad that a guy had to rewrite the physics engine to make his few cars work. I am sure, if S397 was purely focused on maximizing what the engine is capable in terms of physics features and completely ignoring the rest of the product and the fact that you have to update the content along with the features and garantue some sort of backwards compatiblity they would most likely be able to give you more or similar options. People are comparing a heavily modded product, that the community is hacking to make their stuff work with an in development software.

On another note: I would be interested in how you guys have updated AC so that brake ware, brake duct tuning and radiators and their influence on drag work within AC? Something that rF2 does out of the box without any form of hacking.

Click to expand...

mclarenf1papa said:

A bit half and half. It was so delayed on the model side that the physics side didn't get much attention. Mechanically it's alright, but aero and tires leave some to be desired. P13 much more representative.

Click to expand...

Wayne Kerr said:

It might very well drive better. But at the time when we (the team) needed the solution (both for driver training tool as well as a rudimentary setup check), what rf2 brought to the table was insufficient despite already having both Norma and Oreca in their lineup. Hence the approach to IER. I was happy using chassis sim until I realized a driver in loop can somewhat kill both birds with one stone, but when the "in game" setup menu doesn't reflect the actual changes I can make to the car, no amount of tire modelling accuracy is enough to sway us to say "yup, let's use that".

At this point in time after twenty months of collaboration, the ship for us has sailed to start from scratch using rf2 again for those two cars. If someone in the modding community think they can do a better job using the rf2 engine, feel free to ping me directly with a proposal. There is a secondary program in gt cars I just got involved in that I will be looking for some assistance in this area.

Feel free to post questions here, if I can answer without giving out too much info I'll try. if you are skeptical and plan to attend an IMSA race in person, PM me the race you plan to attend and if you want a f2f convo we can set something up.

Click to expand...

bozakn said:

Real drivers don't care about the FFB/driving feel accuracy as much as we do.

have a word or two with the engineers about the setup/handling issues.

Click to expand...

Trava said:

Wow, an ENDURANCE CAR that's supposed to be raced for 100s of laps without accidents is easy to drive, who would have thought!
But I'm sure that you have more experience than both the engineers and drivers of said car to come to this conclusion and it's perfectly valid. I wonder why they don't just hire you instead? Would've saved them a ton of money and time.

Click to expand...

Kyuubeey said:

So why exactly are amateurs with no simulation experience lecturing a professional simulation engineer about what drivers and engineers want from his car and how it should drive? Something just seems off about this premise to me.

Click to expand...

v4v474 said:

Ok, so I am a man of word and yesterday afternoon/ early night I bought your P13C and gave it a reasonably comprehensive spin.

I used a simple combination/test, which was to race it at Spa in dry weather.

It should be noted that I am a slow and unskilled driver.

Not completely deprived of ability to get a car around track after getting acquainted to it, but under no circumstances, "a natural".

1. The car does feel reasonably nice as in some points it seems to have more meat under the bones that the average AC car.

And I will not even argue that the model may react logically to set up changes on aero, suspension, etc. (note: I made no setup changes)

2. The car exhibits some interesting and pleasing behaviors which I have rarely felt on AC cars with this degree of believability, like, for instance, understeer/front lightness under braking if going too hot into a turn, the whip effect that the tyres give when we abuse throttle with TC low, etc... (other cars have this as well, I am just commenting that in this case, it seems better done).

3. So, I have little doubt that your work has some degree of quality, making it, at first sight, a very valid addition to the Assetto Corsa roster...

4. However, on my first lap, with stone cold tyres, I went up Eau/Radillion in 6th gear, full throttle, without even trying to be precise and, from memory, I only grazed the stripped curbs on the left hand side of Radillion.

5. Not a single time on the 30+ laps I did around Spa, I had ANY trouble getting Eau/Radillion done with a ZERO level of concentration and ZERO attempt to be precise (always in 6th gear and always at full throttle).

-- I felt zero compression at the bottom of Radillion and zero lift/lightness at the top --

6. So, I have to say: nice car, but just like the Porsche GT America, there is clearly something glaringly missing on the physics that make it feel very fake and arcade-like under some circumstances.

7. I am sure that your professional work is an upgrade to this, as this car would never, ever, under any circumstance, pass the smell test of a pro-driver being 100% frank in his/her opinion.

A couple of final notes:

A) you mentioned that your professional work involves more code beyond the CSP, so, there is no real theoretical difference between what you do and what teams do on rF2Pro, adding their physics packages, ie. you are no longer using AC, you are using your frankenstein version of AC, so there is no point in comparing that with rF2..

B) And, more important of all, I will not debate physics with you on technical level.

When I say that rF2's tyre model is the best, I mean from the point of view of providing an illusion of driving a car, a simulation of driving. That is what I look for on a sim, whilst others might prefer to juggle setup changes and role-play the race engineer - not me.

This means that, driving around Spa on your P13C might have been 100% correct from a raw physics point of view - and as mentioned above, it wasn't, by a country mile - that in reality it would not have mattered to me at all, as I DID NOT FEEL that I was driving.

There was just a faint illusion of driving a prototype in some sections of the circuit, but the overall feeling out of the lap was that I was fighting just an old spring tensioned wheel.

PS: Tried the VRS's LMP cars (somewhat different generation/car specs, I know) and the tyre feel was more numb but... surprise surprise, getting Eau/Radillion done at full throttle required actually trying hard.

Click to expand...

bozakn said:

"It might very well drive better. "
That's a crucial statement right there, and IMO it should end this pointless discussion.
It's fantastic to have some real insight from the race team owners and sim engineers, and I find many of the info I've read in this thread very educational.
But it should be made clear - most of us are not engineers, we are amateurs/hobbyists/enthusiasts.
And as such, most of us, including myself, don't care about real race car development and professional implementation issues consumer sims have.
All I care about is how it feels.
Does it resembles the real deal in the handling department?

And now we have a race team owner saying rF2 "might very well drive better".

I believe everything that's been said in this post regarding rF2's problems under the hood.
It has its issues.
But it's the only sim available that can convince me I'm driving an actual car.

I've really tried to like Asseto, but couldn't come to terms with it handling wise.

And that is the only thing that matters to us amateurs.

We are not real racing drivers.
Real drivers don't care about the FFB/driving feel accuracy as much as we do.
They have all the realism at their disposal when they get behind the wheel of their race cars.
In a sim, they want to memorize a track, keep their reflexes sharp and afterward have a word or two with the engineers about the setup/handling issues.

When I'm in a sim, I want my brain to be tricked.
I want the illusion to become reality.

Now would some engineer step up and tell me why the hell can't I get that feeling anywhere else apart from rF2?

Click to expand...

mclarenf1papa said:

He's the engineer, not the owner, and he means that the car itself would drive unrealistically better. i.e. it would feel "better" to drive but would not be at all accurate to the real car. i.e. an idyllic fantasy version of the car. Same could be said for a lot of the Kunos cars in AC.

Click to expand...

mclarenf1papa said:

I've also done work in AC with Mak Corp, though it was quite a while ago (circa 2015) and under NDA. Also worth noting that Petros is the project manager and does not do the physics for their cars.

And RFPro is only used as a graphics platform for those teams; they run custom physics engines. Dallara uses AC, but again, same deal.


I can maybe see where you were headed with your point then, in any case it was less a response to your post and more a clarification on Arch's behalf. But yes, I would agree that garbage in is garbage in in any model. I think it would be harder to do what Arch is saying in a physical model than empirical.


I think the question is, does it really matter? A programming team of primarily two people were able to add night, rain, graphical improvements, performance improvements, and physics improvements in their spare time. Not to mention that RF2 has been in development for longer than AC, so if you discount the continued development of AC, you're then comparing a product under continuous development (RF2) to one that hasn't been touched since 2016, which would make even less sense. Regardless, as Wayne alluded to above, the bulk of these physics changes were updates to the setup UI/options, not fundamental changes to the physics engine (there were certainly many improvements to the engine, but you could still make a pretty strong model with "vanilla" AC, it's just much less usable for a serious team, in the same way that RF2's setup options are).

I'll also note that the very first iterations of each AC model (P3 and P2) were declared better than their respective RF2 models by every driver involved in the programs (and the engineers, otherwise I wouldn't have a job).

Click to expand...