Actually, I think a dynamically changing track is needed to refine tyre physics. Because the feel of the car comes from how the tyres interact with the surface they are on, so basically half of what's needed to do reasonable tyre physics is right now pretty much an afterthought.

I also agree that sims on the whole would benefit if there would be more attention put on the "ambience" of motorsports, as becky describes. Allthough I do think that many of those suggested things are not really within the performance range of todays computers.

Quote from Zen321 :

Yes, but then it would not be *that* realistic. To have the most advanced simulation of track heating, the patches of tracks would have to be as small as possible, otherwise you might find yourself driving in a patch of 50x50cm that is heated well, then take the corner too wide, and put your tire on a patch of 50x50cm that is too cold, resulting in a very shapr loss of traction

Not really. Instead of defining 50cmx50cm patches, there would be points defined. The temperature of the track's side of the contact patch would be calculated by interpolation.

If you want all of these things that all of you are suggesting, go buy a car and do track-days. Simulators can only simulate what the computers that they run on can process, and the technology that is availible. If the technology and the processing power for these features was availible, you think we wouldn't have them?

Besides, this is a game, at the end of the day, as is every other sim out there. People keep pushing and pushing for more things that are unrealistic in terms of chances of getting it, so perhaps it's maybe why new sims are not exactly 10 a penny, and why the Dev's of this dear game don't mention anything.

Quote from S14 DRIFT :

Simulators can only simulate what the computers that they run on can process, and the technology that is available. If the technology and the processing power for these features was available, you think we wouldn't have them?

I think you greatly underestimate the speed of today's standard home CPUs. Maintaining the surface track temperature wouldn't be that big of a deal, processing wise. The current physics calculations aren't limiting the CPU either, otherwise the fast forwarding in single player replays wouldn't be possible (there it's doing the calcs as fast as it can). The graphics are much more of a performance killer than all the physics and environment calculations you can throw at it.

Also, there is lots of space for improvements which don't require much CPU, like the execution of racing and flag rules. How many poeple in here care about the yellow flag for example?

Quote from AndroidXP :

I think you greatly underestimate the speed of today's standard home CPUs. Maintaining the surface track temperature wouldn't be that big of a deal, processing wise. The current physics calculations aren't limiting the CPU either, otherwise the fast forwarding in single player replays wouldn't be possible (there it's doing the calcs as fast as it can). The graphics are much more of a performance killer than all the physics and environment calculations you can throw at it.

As true as your point may be, since you're not playing, do playing replays even use the physics engine? Since it's normally either watching a video of the car, or just a big piece of code that 'drives' the car exactly as you did

A replay is a series of driver inputs, so those require the physics engine to translate into vehicle motion. Bearing in mind the replays are NOT a movie, how could they NOT need a physics engine?

There is plenty of capacity for further use of the CPU in LFS in particular, which still only uses 1 CPU core when dual core has been standard for some considerable time.

Not that limitations of hardware are an issue as entry level systems can cut back on eye candy to accomodate increases in basic overhead, and LFS has very low overall system requirements by the hardware standards of the installed userbase anyway.

This is all moot though, as the open letter's purpose was to highlight areas that are not considered for simulation as being what actually defines what a sim racing simulator should be.

Quote from tristancliffe :

A replay is a series of driver inputs, so those require the physics engine to translate into vehicle motion. Bearing in mind the replays are NOT a movie, how could they NOT need a physics engine?

Ahh, cheers for explaining

So, the temperature of where the tire is contacting the ground could be calculated like in the attached image. All of the values within a certain distance of the corner of the contact patch would be needed. Using those values, and their distance to the corner of the contact patch, the temperature of that corner would be calculated. Then from those four values the average temperature of the whole part of the road could be calculated. It's enough that there would be a definite difference between the racing line and the outside of the track. In fact, perhaps the dots could be variable. Like, using the ideal line, the dots could be re-arranged in such a way that there is a higher resolution of temperature dots near the ideal line after many people have gone around the track.

A more LFS way would probably be to read the temperature at each of the 3 cross-sections of the tyre, but in principle yes you've got it in one.

For clarity though i'm not just talking about track temperature, but dust/sand and dirt being dragged onto the track, standing water in bumps, sap under overhanging trees, marbles from wearing tyres, small peieces of debris from contact. All these things and more besides are constant evolving factors in the daily life of a race circuit.

One doesn't truly work without the other. A dynamic track is only as realistic as the physics behind the car/tyres rolling over it, whilst a tyre can never be perfect if its only operation is on a single unchanging surface.
LFS's tyres are good enough for LFS to be fun, but not good enough to be ignored whilst other things are explored. If LFS's tyres can be judged by their most bizarre behaviour, then they aren't very good, and certainly not good enough. Things like the temp/pressure/grip relationship, the way a road tyre behaves on "gravel", the way an abused tyre becomes ultra-high-performance again if you abuse it for long enough.

Quote :

For clarity though i'm not just talking about track temperature, but dust/sand and dirt being dragged onto the track

good point there. up until now we only have dirt as a very individual hinderance on the car's tyres -> quite simply a temporary loss of peak-traction capability. The surface itself is utterly static.

However thinking about how limited the potential sources of dirt are on a track I think that there can be something done using a good amount of approximation. We already have visual dust spread when a car's tyre goes slightly off the tarmac. So I suppose the information on where the dust falls down is already there. Now one would have to somehow develop a kind of "real-time dust-capacity per square meter" + the aforementioned temperature for the ground to introduce some new values for grip-limitation on the ground-surface side of things in addition to the tyres.

Oooops: here we go again: unfinished tyre model. So without changing this we don't get the other. You guys got me thinking

Plus this would present the need to be communicated throughout the connected clients and the host: so do I sense a major rework in netcode, here? Is it at all possible to achieve sufficient speed at communicating these values when racing online?

Anyway - although just losely connected with LFS - there are some interesting thoughts in this thread.

Disclaimer: no I am in no way a skilled programmer -> so please do not hesitate telling me if I'm way off the point here.

representing dust & dirt on the track in the physics is pretty easy and just involves an indexable list of surface-'particles', these particles being created by a hook into the existing particle system. The surface-particles themself being linked in some way to the node map of the track for portalisation.

It's not the easiest thing to representing it visually with consistency to the physics and keeping it all working to a reasonable performance level on entry level systems (as lists are inherrently infinite by definition), but it is well and truly within Scawens ability to do so.

That's just one possible implementation method, i'm absolutely confident that with more thought than this forum post has taken, and with a mind like Scawens, a better solution could easily be devised, and likely one that would include supporting other forms of debris on the circuit too.

Quote :

do I sense a major rework in netcode, here? Is it at all possible to achieve sufficient speed at communicating these values when racing online?

Yes it would need communicating as the result effects the physics and therefore must be in sync, but it's just an additional packet type, no biggy really.

Quote from DrBen :

Plus this would present the need to be communicated throughout the connected clients and the host: so do I sense a major rework in netcode, here?

Not really. Communicating state changes of the environment is done in many online games. LFS already does it, to some extent: moving track objects like cones and tire stacks. Your car can collide with a loose tire that was moved by another car.

The most difficult part is probably integrating all those influences into a believable physics model.

Quote :

Is it at all possible to achieve sufficient speed at communicating these values when racing online?

In most cases, the changes will be fairly slow: rain, marbles and dust build up and disperse slowly. It may also be feasible to spin out on oil spilled by the car that is right in front of you. The only case that seems too hard (computationally) is hitting flying debris.

It's an interesting opening post. I find I agree with some of it, but not all. Perhaps an open letter to all sim developers isn't really going to make any headway. The currently developed sims are all already rather disparate in their focus.

It's inconceivable, at least to me, that sim developers like iRacing, LFS, ISI et al, have not already considered in their own fields all of the points raised. They have forged ahead with their own lists of priorities and areas of focus. Unlike the receiving market - the consumer - it's very specifically their business, their income and their working passion. Few of us spend 8hrs+ per day pondering the development direction of a given sim but I'm sure that the actual racing sim developers do. And then some.

As to the importance of the tyre/suspension models, I'm afraid I'm with the sim developers. Whether they like it or not (and they possibly are perhaps even irked by it), the fact is that WE (said customers) prioritize tyre, suspension and aero physics as definitive elements in determining a sim's worth. Rightly or wrongly, we will NOT buy a sim (sim, SIM, SIM!!) that doesn't model these things at least well, and at least with these things as points of focus. Sim developers have little/no choice but to maximise on the simulation of these aspects of vehicle physics for the sake of credibility. And like it or not, the racing-sim-buying public give massive worth to these aspects.

So, to gameplay.. well, all sim developers have different ideas about gameplay. All the current sims operate in different ways. Some are primarily offline stuff with online bolt-on, LFS and iRacing are focused on the multiplayer side of sim racing - and even these are fundamentally different orientations to the same goal; rF focuses clearly on the community modding aspects and less on the application of real-world physics simulation. There are already fundamental differences in direction for all the existing sims. An open letter to them all is, tbh, not really going to crater the surface of their businesses or seize their attention, much less stop them in their tracks and make them realise their direction and focus has been wrong. Sorry but I just don't see it happening.. they're all far too successful at doing what they're doing to hear a single voice in the consumer crowd

The point of a forum isnt to get people to listen Sam, the point of a forum is to let people vent whilst not having to listen! Whilst the stated recipient of the open letter is the developers, the anticipated consequences of it are none. It's not the first time i've written an open letter, I think you might recall my lambasting of Peter Molyneux and that other guy I forget who in my Sensible Soccer rant... That had no effect either, at least, outside of indi circles.

I do think that a sim which focuses on the simulation of the environment whilst providing acceptable (or better) levels of vehicle simulation will find a healthy place in the market though, and whether existing developers take it on board or we have to wait for the next generation to see any influence of environmental modelling isn't the point. This isn't a direct feature request, simply some opinions expressed on a direction that i'd like to see taken within the genre.

As you know I did ponder writing such a simulation myself some time back, I decided not to go down that route personally as i'm more focused on other things these days. That's not to say that I or some other indi developer couldn't do it in the future though.

I dont state my views just for the sake of it, a sim which gives equal or at least 'adequate' attention to environment simulation is something I would like to see and i'm sure enjoy playing. It is however, in my mind as I envision it, a drastic change of direction in terms of existing sims. I cant even see how to do a lot of the things I would like to achieve myself with the 3D media from any of the current titles.

When I make a game i'm a great believer in investing in the gameplay dynamic and in sparring much attention for small superfluous details. LFS has it's fair share of both, and i'm in no way criticising it.

The problem with tire physics is that there isn't a good allround model to use. Basically every tire model is based on a solid static ground. Pacejka is very common for example, it performs well on a solid ground, but only if you feed it with a large number of correct tire parameters. One wrong number and the results are horrible. And that's just on static solid ground. Imagine the amount of parameters you'll need to model a tire on different surfaces. And then deformable surfaces...

Another problem with simulating a surface that acts like a fluid, such as water or even gravel, is that the tread has alot to do with it. Because currently no racing sim that I know of simulates the geometry of the tread. Tire manufacturers must have some way of calculating how to make a tire tread better in rain. But rain is alot different then dry pavement. Gravel/sand is pretty simmilar to water. Except that sand has more mass then water. So it requires more force for the tire tread to push sand/gravel out of the way then it does water. The problem with using a particle system to simulate tires on a wet road, or in a gravel trap, is that the particles would have to be realistically sized. Unfortunately, for a whole track of gravel, a huge amount of 3-dimensional particles would be required. At least then we'd have a deformable surface as well! Even though you'd need about 50 computers with core i7 processors just to get 30fps.

You dont need each physical particle to exist in physics or real-world space. Clustering/clouds are more than sufficient, and the effects of tread can be dealt with in general terms and does not need ultra-detailed modelling, that was the point of my original post.

Effects such as partical exhaust like water spray or dust can be dealt with either with the existing LFS particle system, or perhaps in the future with things like nVidias onboard physics system.

If a sim was to recreate displayed particles of gravel on an individual basis it is possible with todays high end technology, where a displayed particle plays no further part in the physics system and expires they can be created in large quantities locally on each client on an as needed basis for rendering purposes. The physics can then be much more general and deal with spray in more general cloud/cluster terms.

There are lots of ways to go about achieving better environmental simulations, the point however is not what is possible and what isnt, but that the era of the static environment needs to come to an end for sim racing to move forward. Vehicle simulations (especially those with no or little in the way of aerodynamic modelling) can only take sim racing so far and the ramifications on gameplay deficiences from the lack of environmental simulation are huge.

Did any of you catch the interveiw with the dev. of rFactor? He was talking about this very thing in that interveiw. Episode 32 or 31 of Inside Sim Racing. Sorry can't remember which.

Quote from kingfag :

The problem with tire physics is that there isn't a good allround model to use. Basically every tire model is based on a solid static ground. Pacejka is very common for example, it performs well on a solid ground, but only if you feed it with a large number of correct tire parameters. One wrong number and the results are horrible. And that's just on static solid ground. Imagine the amount of parameters you'll need to model a tire on different surfaces. And then deformable surfaces...

True, the Pajecka formula is a gross oversimplification, but it's only dependent on three variables. The difficult bit is to "compress" the x variables present in real life into three input variables in the formula.

I can't remember where I saw this, but here is an interview with a physics guy explaining how physics are done in racing simulations. He focuses on tyre physics. It's a bit long, but the bit about the formula is around 16 minutes in.

This thread has been an interesting read, and I think many good points have been made. My money is on iRacing to get it done before any other sims. They've got the tire physics, car physics and static surfaces right already, and their develpment carries a lot more momentum than LFS (unfortunately). Dynamic surfaces should be a logical next step for iRacing.

Quote from Toddshooter :

Did any of you catch the interveiw with the dev. of rFactor? He was talking about this very thing in that interveiw. Episode 32 or 31 of Inside Sim Racing. Sorry can't remember which.

Yep, that was a good interview. Looks like rFactor 2 is planning to do this too. The future looks bright!

Your post is very interesting, but I will pick up on something few people have. Quoted post contains removed parts.

Quote from Becky Rose :

An Open Letter to Sim Racing Developers : A report on the accuracy and short comings of current sim racing titles and recommendations for areas of improvement.

Trackside objects dissappear from view and are reset, stricken cars removed from the circuit. These are atmospheric killers which in real racing would be dealt with under yellow flag conditions or a safety car.

There currently is not enough focus on gameplay dynamics, with better handling of environmental repairs (trackside objects and stricken cars) will come a change in driving mentality. With better modelling of environmental conditions will come a change in racing mentality. Both of which will bring sim racers closers to racing drivers.

We must see the inclusion of better retirement handling - stricken cars sitting out a race by the side of the race track and their competitors bound to the pits, the introduction of a safety car (with automatic and manual control), trackside repairs being handled properly, yellow flags not being waved so wantonly that their effect counts for nothing and people do slow up. When, as a sim racer, i'm forced to react to the same things that a real racing driver must react too, then I am playing a simulation.

For as long as sim racing focuses on tyre and suspension modelling it is nothing more than a vehicle simulator.

LFS could improve a great deal with simulating motorsport better. By simulating motorsport, I'm not reffering to car physics. I'm talking about the rules of motorsport.

LFS would have much better immersion if the flag marshalling would change. I don't think it would require much more of a graphics card hit to have GPL style 2D flaggies on the other side of the barrier. No-one would really notice any 3Dness whilst driving past at a reasonable speed. As far as I know there is also infinate resolution where flags come on. It would be nice to have actual predefined flag points.

There is a second part to this. There are more flags in motorsport than yellow and blue. No simulator I know of uses the white flag (as in the International Sporting Code flagging), and use of the meatball flag would be helpful for online racing to inform drivers that their car is almost destroyed.

In addition, a properly simulated Safety Car would add greatly to the realism, as would red flags, which no sim has had the balls to do. Rolling starts would also be a good option, too

Repairs should be handled better - Richard Burns Rally has a good system, it wouldn't need the time acceleration in LFS apart from between sessions. It's a bit like the Colin McRae Rally 2 system, for those not so au fait with RBR.

Very good post Becky.

+1 To all of that, there are 100s of fine details which may seem insignificant but are very important.

Also things like drivers spinning off and throwing dirt onto the track, standing water puddles and aquaplaning are also important because often these things change every lap.

Dynamic weather also, its very annoying when you go out on a dry setup and it then pisses it down, but you have to bear it, also agree about the marbles being the best place to be when racing in the rain, kerbs are also ultra-mega slippery.

Marshals out in the middle of the track frantically waving their yellow flags at drivers refusing to slow down would be a nice touch.

Another thing although not ncesseary is sound, when cars pass at first they sound relatively quiet, but then you get a blast of exhaust as they pass and move away, I can't think of a sim that does this properly (well GPL had a stab at it, but it was either no sound or full engine sound.).

Well if it were as easy as writing it, why aren't these ideas (which would be nice, but aren't 'needed', since the line has to be drawn between a sim and RL..) in production if it were this easy? It's all very well to sit here and say "yes i want this and that and those as well", but it's not as easy as that. And Becky, being in programming, you should know that better than most

I mean come on, nothing's ever good enough for anyone.

Edit : I mean, yes. In the future etc, but people just push and push for more stuff that, ultimatly, doesn't matter. So you run off track and get dirt on your tyres, but oh no that's not good enough - you want dirt on the track. and the list goes on.