The Last Garage: May Dev Blog And Q&A With Marcel Offermans

The-Last-Garage-May-Blog-Q&A.jpg
Images: The Last Garage
Ex-rFactor 2 developer Marcel Offermans unveiled his new project in early 2024. Now, the May dev blog for The Last Garage highlights its progress - additionally, Offermans gives more insight in a Q&A.

After its January announcement and hands-on trial at Sim Formula Europe 2024, @Marcel Offermans' The Last Garage project has advanced quietly, but steadily. The former Managing Director of rFactor 2 developers Studio 397 has set out to create a highly-detailed engine that simulates even unconventional suspension layouts and car configurations accurately.

More on that in the May dev blog, however - over to Marcel! For more info on the project, head over to the The Last Garage website.

The Last Garage - May Dev Blog​

As the days are getting longer and warmer, the Sim Formula event last January seems like a long time ago. A good memory. The positive feedback there has led to quite a few people getting in touch. Some expressed an interest in somehow using or licensing the engine, others just to learn about its progress. From those discussions, the idea arose to provide a quarterly update to keep everybody who is interested in the loop on the current developments.

The-Last-Garage-Night.jpg


Vertical Slice​

What was shown at Sim Formula is what, in software development, is often called a “vertical slice”. It’s a technique to develop software where you build the essential and minimal parts of all the subsystems to end up with something you can run and evaluate. The primary aim was to get feedback on the “driving feeling” as ultimately every racing simulation is as good as the feeling you get when behind the wheel. Given the positive and constructive feedback we got, we can definitely conclude it was a success.

For the continued development that meant that the next phase is to look at all these different subsystems and “finish them”. That’s still a lot of work, and not every part of it is as visible or exciting as that first time when you can start driving the simulation, but it is crucial and a modular physics engine consists of many small building blocks that give you an almost infinite amount of elements you can add and refine over time.

So what have we been up to?​

One area where a lot of progress was made was in the modeling of vehicles and how those different physics elements can be linked to various parts of the 3D model. Due to the modular nature of the physics, it is possible to model all kinds of non-standard vehicles, such as a historic F1 car with six wheels that works and
steers exactly like in real life.

Taking that car as an example, you end up with a vehicle that is difficult to model as you suddenly need to handle non-standard steering and suspension setups and that is something that trickles through to all different areas of the simulation.

The-Last-Garage-Night-2.jpg


For example, your telemetry now needs to be able to handle a vehicle with six wheels, and in multiplayer you need to position all of them. Same goes for sounds, any data overlays you might show on a broadcast, and many more things you would not even think about initially.

Starting from scratch means there is an opportunity to properly deal with such variations and this is one of the things that has been worked on over the last few months. The development is not done yet, but there is a framework in place that allows us to deal with the variations per vehicle.

To test this system, we are modeling a couple of completely different types of vehicles to push different aspects of the engine. From high-downforce modern prototypes to karts to classic cars.

Physics Modules​

Another area where things moved ahead is the different available physics modules. As mentioned before, this is a huge topic as there are a lot of different things you can and need to model.

Since the first demo, the whole drivetrain has been significantly upgraded. The manual gearbox received some updates, improving amongst other things the clutch handling. The engine itself gained a couple of features that make it behave more like an internal combustion engine. It can now stall and you can start it using either a starter motor or by simply letting the vehicle roll and engaging the
clutch. It will also correctly run idle.

The vehicle brakes also gained a few new features as they can now heat up and cool down. Their stopping power is now dependent on their temperature, a relationship that can be specified as a bespoke graph based on available real-world data. Brake cooling ducts are modeled and even affected by wind direction and strength, so if you park your car with its nose in a strong headwind, those brakes will cool down quicker.

The-Last-Garage-Bridgehampton.jpg


We’ve also added different steering mechanisms, supporting either a rack and pinion setup or a pitman arm, both with force feedback that is calculated by measuring the torque response on the steering column as calculated by the physics engine. The steering angle can be configured to match the real car, using software-based end stops, provided of course your wheelbase can actually turn
that far. If not, we’ll just use whatever maximum is available.

Suspension Kinematics​

A substantial part of vehicle simulation is the correct recreation of suspension kinematics. This directly influences the rotation and translation of the wheels relative to the chassis and therefore also the road surface while driving. As the suspension is “working” values for camber and toe are constantly changing. So, to simulate a vehicle correctly it is necessary to simulate the kinematic correlations as correctly as possible.

In recent months we spent a lot of time creating different types of suspensions to make sure that we can simulate all of them. Those who’ve been at Sim Formula Europe 2024 or watched the videos from that event have probably already seen the two cars there. The classic red “tintop” was using a solid leaf spring rear axle, while the 70’s F1 car used a traditional double wishbone suspension.

The-Last-Garage-Double-Wishbone.png

A double wishbone suspension setup.

For pre-war Grand Prix cars, we’ve modeled the De Dion, Swing Axle and Parallel Trailing Arm suspensions. On the other side of the spectrum, for modern prototype, formula and GT cars, we have both Push and Pull Rod suspensions, optionally including a third spring.

The McPerson suspension was added to be used in compact cars, and for the modern road and sports cars we have Multi-Link suspensions.

The-Last-Garage-McPherson.jpg

A McPherson suspension layout.

Besides all those, we have modeled a typical Solid suspension as seen on karts and finally a Solid Three Link Axle suspension as seen on off-road race cars.

Of course this list does not cover everything, so fundamentally we have the ability to assemble any suspension from individual parts using linear or torsional springs and dampers and connecting them to different types of joints and arms! We also added the ability to optionally hook up and configure anti-roll
bars.

The-Last-Garage-Third-Spring.jpg

A third spring suspension setup.

Wrapping It Up​

Some of the physics features described above have been donated back to Chrono::Vehicle and a few days ago a new major release of the whole Chrono project, version 9, was published with an impressive set of changes ranging from adding new visualizations to updated tire models, a generic template allowing you to freely model arbitrary suspensions and various other smaller physics improvements.

Similarly, a few of the graphics features and bug fixes were donated back to Godot. From a simracing point of view Godot is a great engine as it uses a Forward+ renderer which is well suited for “fast moving” games. It uses Vulkan as a graphics API, which makes it platform independent, keeping the door open for builds that run on Linux, the Steam Deck or Mac OS X. An upcoming new feature in the next Godot release will be support for DirectX 12, which will open the door to deployments on the XBox console platform.

That wraps up the blog for now, but not before answering a question we get asked a lot: What simulation will you be building? What features will be in it, and what content? The honest answer right now is “We have not decided yet”. There are many ideas floating around and we’re definitely looking to provide some unique experiences, but for now the focus is on the core technology. We’ll provide another update near the end of summer, and by then we hope to also have news about Sim Formula. From our little garage, we wish you all a great summer and we hope to meet you again soon!
About author
Yannik Haustein
Lifelong motorsport enthusiast and sim racing aficionado, walking racing history encyclopedia.

Sim racing editor, streamer and one half of the SimRacing Buddies podcast (warning, German!).

Heel & Toe Gang 4 life :D

Comments

Thanks Marcel for the dev blogs/update.
The wait for the next one will be very long and exhausting lol.

The more I read about it more I appreciate the fine details you guys are looking at (like how brakes, suspension and clutch work).

Die the open source nature of the project would be cool a demo to play with...not only for testing it but also to work on top of it, I am thinking to modders and developers, probably this would help creating a community ready for the release date and able to deliver content and upgrades/plugins as soon as released.

Btw, good work so far keep it up!
 
Last edited:
Premium
Interesting reading that the engine covers many suspension solutions, and btw, de-dion is not a pre-war only solution, it was used by Alfa Romeo in the 116 chassie in the late 90's with double wish bone in-front :)
Good point, and I was not trying to imply that was a pre-war only solution.

Even if this seems to be a modular approach, but how is this different from current engines? Double wishbone and mc person most be one of the most common solutions together with multi-link on modern cars. There is a few different schools how to calculate the behavior of these solutions but more of how this solution is different from current sims would be interesting read as well.
Obviously, in the end, all physics engines simulate physics, so from that point of view they are all the same. You can probably even say that most of them use a "multibody system" as their basis, coupled with a tire model. That said, there are certainly differences in how that is implemented, what choices have been made and what specific algorithms are used to calculate specific elements.

One thing that I have been focusing on is keeping the whole system modular. For a large part, that ends up being irrelevant for the end user (if it works, it works, right, it does not matter how), except maybe for the fact that it allows me to model, for example, a 6 wheel F1 car. I'm not really in the best position to start comparing physics engines (I've obviously seen a few up close, mostly ones based on the rFactor engine) but this specific example is something I have not seen others do.

Most high-end engines "claims" to be accurate and provide "real" forces only, and well, anyone who been with a car on a track can confirm it's not an accurate description in most cases, some more close than others.
You only have to look at recent events in F1, specifically in this example Red Bull, to know that everybody struggles to correlate what works in the sim with what works in real life. Therefore I won't claim here that the simulation I am working on is accurate. It makes a model of the real world, and a model always simplifies certain things. Of course I do want to end up with one that, when properly parameterized and balanced, gives a realistic feeling of what it is to drive a car at the limit.

All of this said though, it would be interesting to have a reference car at a reference track under standard conditions that you can drive on all of the different sims and then compare how they handle and feel.
 
OverTake
Premium
Good point, and I was not trying to imply that was a pre-war only solution.


Obviously, in the end, all physics engines simulate physics, so from that point of view they are all the same. You can probably even say that most of them use a "multibody system" as their basis, coupled with a tire model. That said, there are certainly differences in how that is implemented, what choices have been made and what specific algorithms are used to calculate specific elements.

One thing that I have been focusing on is keeping the whole system modular. For a large part, that ends up being irrelevant for the end user (if it works, it works, right, it does not matter how), except maybe for the fact that it allows me to model, for example, a 6 wheel F1 car. I'm not really in the best position to start comparing physics engines (I've obviously seen a few up close, mostly ones based on the rFactor engine) but this specific example is something I have not seen others do.


You only have to look at recent events in F1, specifically in this example Red Bull, to know that everybody struggles to correlate what works in the sim with what works in real life. Therefore I won't claim here that the simulation I am working on is accurate. It makes a model of the real world, and a model always simplifies certain things. Of course I do want to end up with one that, when properly parameterized and balanced, gives a realistic feeling of what it is to drive a car at the limit.

All of this said though, it would be interesting to have a reference car at a reference track under standard conditions that you can drive on all of the different sims and then compare how they handle and feel.
should we define track/car and go for it? ;) I would have a few available haha
 
Premium
Doesn't every game have Brands Hatch and the same field of GT3's?
 
maybe ask for raceroom and KW studio to switch to your 3D engine. I don't know any other dev who needs an engine asap. I'm interested in this project.
 
Last edited:
Dont know about the sound engine used but try maybe to simulate a modern Murrays T50s track car.

Having a Murray T50s track car in your garage, that really the last garage you ever need. So should be on the box cover for sure when released :) .
 
Last edited:
Marcel left S397 2 years ago.

A better analogy would have been, that Stephen Hood has basically done diddly for sim-racing his whole life - and point stuff in his direction. Oh wait, the same Stephen Hood whose team who in 10 years - only came up with 2 or 3 "ok" versions of the F1 license.

He's the Head of Motorsport Games, so he's the one getting the pass here. The book stops with him alone. Stephen hasn't just failed S397, he's caused major issues with several MG projects on top. Of course now, rF2 advancements have stalled - whilst resources refine the engine for LMU. Which oddly enough is also not the savior of sim-racing either.

If Stephen doesn't extract his finger, there will be no more rF2! The company is in a dire position and isn't out of the woods yet. It's a guess on my part, that LMU won't save MG from the impending doom they created.

I appreciate rF2 wasn't perfect, and still isn't - but at the point when Marcel stopped being involved (more than 2 years ago) - you had a much healthier game than the state it was handed over to him in.

Also not personal, also the truth.
Wut?

I hope you are aware that Marcel was in charge when some very questionable desicions were made and community feedback was completely ignored for ages. rF2 development stalled completely in critical areas or went in wierd directions at times. I am not trying to have a dig at Marcel here because he has allways been very passionate about this hobby and allways a nice bloke. And it's easier to judge something from the comfy home chair and forget from where he started when he took over rF2, but there were times when I really wondered where the studio was going with rF2. Fact is, rF2 has never been in a better place than it is now in terms of features and content wich is a result of the work of many people who worked on it over the years, even after Marcel left the team. Anyway, we are all human and people make various desicions for different reasons. And I am pretty sure that there will be desicions made in the process of development of this engine that we question in the future. Such is life.
 
Wut?

I hope you are aware that Marcel was in charge when some very questionable desicions were made and community feedback was completely ignored for ages. rF2 development stalled completely in critical areas or went in wierd directions at times. I am not trying to have a dig at Marcel here because he has allways been very passionate about this hobby and allways a nice bloke. And it's easier to judge something from the comfy home chair and forget from where he started when he took over rF2, but there were times when I really wondered where the studio was going with rF2. Fact is, rF2 has never been in a better place than it is now in terms of features and content wich is a result of the work of many people who worked on it over the years, even after Marcel left the team. Anyway, we are all human and people make various desicions for different reasons. And I am pretty sure that there will be desicions made in the process of development of this engine that we question in the future. Such is life.
Of course I am aware.

I think it's easy to forget that all of these projects are huge now. Sims are really complex, and not just in regards to handling physics. As a pad user, rF2 was always dire compared to some of the competition to me. No one solution we buy is perfect, and never will be. Dev's have plans set in stone, their roadmaps, and they don't have unlimited resources. They have to balance everything, and earn money from their platform.

The reality is, rF2 had issues, but nothing more than every other titles before and after. Four years after release AM2 isn't as good as I know Reiza can do, and it's still not installed on my PC. These issues also affected AC, PC1 & 2 et al. They are broke, we are unhappy, they try to fix it. It then goes forwards, sideways and sometimes seemingly backwards. It is what it is.

We all have opinions and are all entitled to them. I think most people actually really appreciate the work nd gamble Marcel is making to benefit modern sim-racing for the coming years. These huge ground-breaking long term developments ain't coming via Codemasters or EA that's for sure.
 
Sounds awesome, and if the community can get behind and invest in Project cars then there should be plenty of support for this engine.

Something small and focused like the above mentioned Goodwood Revival: The Sim! would allow us to provide support without the baggage of being investors.
The Goodwood Revival thing would be a good platform, though I don't know about licensing and royalties and how it would all work, I'd like to see all/many the original cars and their drivers that run in the revival, that could be massaged along by supporting the sports past and current drivers through a profit percentage going to a Racing Driver Benevolent fund to help drivers and the Marshals injured or lost their lives doing what they love... Motor racing, the Marshals give to the public free because they love the sport as much as the drivers,
Goodwood has a history of honoring and those drivers who's past has entertained many a fan and perhaps it could be seen as a way for the Sim Race community to give back to the sport,
 
The fact that Marcel Offermans and Niels Heusinkveld are still actively participating in this community, and are willing to listen, is such a breath of fresh air.
AMS1 and Nordschleife in rFactor2, thank you for everything you've done, I will never forget it.
You guys are legends and you deserve all the praise.
 
Of course I am aware.

I think it's easy to forget that all of these projects are huge now. Sims are really complex, and not just in regards to handling physics. As a pad user, rF2 was always dire compared to some of the competition to me. No one solution we buy is perfect, and never will be. Dev's have plans set in stone, their roadmaps, and they don't have unlimited resources. They have to balance everything, and earn money from their platform.

The reality is, rF2 had issues, but nothing more than every other titles before and after. Four years after release AM2 isn't as good as I know Reiza can do, and it's still not installed on my PC. These issues also affected AC, PC1 & 2 et al. They are broke, we are unhappy, they try to fix it. It then goes forwards, sideways and sometimes seemingly backwards. It is what it is.

We all have opinions and are all entitled to them. I think most people actually really appreciate the work nd gamble Marcel is making to benefit modern sim-racing for the coming years. These huge ground-breaking long term developments ain't coming via Codemasters or EA that's for sure.
What we forget is that Marcel, with that rf2 experience, is perfectly aware of how difficilt it is to work in an environmznent (engine) you haven't created yourself. On the contrary, with that in mind, he is in a perfect position to create a relatively easy to use plateform for future external developpers. Easy to say yes :D

it seems the game uses a step by step approach to a ovoid letting any model's parameter left untested, which is absolutely great. Because this is were many issues seem to come from in the (10 years old) modern sims : placeholder values in many parameters , because no real.data are available, which lead to try empirical new values until the driving feels better (which seem minor improvements but which are not). I assume there's a balance to find between complexity and easy of use, especially for a project of a simracing plateform. I'm glad we are able.to follow the developments with Marcel's devblog, it is really interesting. And the idea of using open sourced engines is another step forward, imo, which I assume will avoid explaining from scratch to future users (developpers) how everything works but, on the contrary, will make Marcel's team to focus on bringing (selling) its heavy expertise to these users and help them making the best sim possible in short times. Well, I can only assume all of this, it just seems logical.
 
Premium
The Goodwood Revival thing would be a good platform, though I don't know about licensing and royalties and how it would all work, I'd like to see all/many the original cars and their drivers that run in the revival, that could be massaged along by supporting the sports past and current drivers through a profit percentage going to a Racing Driver Benevolent fund to help drivers and the Marshals injured or lost their lives doing what they love... Motor racing, the Marshals give to the public free because they love the sport as much as the drivers,
Goodwood has a history of honoring and those drivers who's past has entertained many a fan and perhaps it could be seen as a way for the Sim Race community to give back to the sport,
When I referenced the Goodwood Revival earlier in this thread, it was more a nod to Marcel’s description of the varying drivetrain/suspension configurations that he is aiming to simulate rather than an actual request for an actual licensing agreement with the Duke of Richmond.

To be taken at all seriously, in fact, such an arrangement must not only simulate the Settrington Cup, but also include a comprehensive Goodwoof DLC and an early-access bonus Valentino Rossi-style tour of the estate:


All this may prove to be more world building than Marcel can duplicate on his own, but if he delivers the right sim and the right tools, modders will do the rest….
 
Last edited:
I'd leave the J40's and motorcycle estate tour to the modders, as no one would really expect to have to join the que at LeMans Unlimited for a go on one of the Big Wheels or one of the hot dog stands.

Reality says if it's left to modders to do the majority of the work... unpaid, and having to work around a RL schedule it would likely be a protracted effort that wore out it's time, better if the sim were made complete by a studio and allow the modders some tools to do their bit without leaving the sim empty and needing
 
Meanwhile, rf2 is still a broken mess with missing/broken features and Marco gets a pass while people are still bashing Ian Bell (deservedly or not).

Not personal, just an observation (AKA the truth)
the Ian bell spirit is still on the Madness Engine and AMS2, each update is "the best of the best changing everything" ;)
 

Latest News

Article information

Author
Yannik Haustein
Article read time
11 min read
Views
4,855
Comments
37
Last update
Back
Top