I'm wondering how Steven Segal would get hoisted into the ring? I'm also wondering how many of them would collapse from exhaustion during the walk from the corner to the center of the ring? A middle weight or larger MMA fighter would have no problem taking any of those guys out.
The issue is that reality got in the way of whatever simplifying assumptions he made when creating the original tire model. The failures are occuring in "extreme situations", that turned out not to be that rare.
My guess is that the failures that occur in "extreme situations" exposed a weakness in the core mathematical model, and having to re-implement a new model from scratch is why it's taking so long. Plus there's the issue that a new model may suffer from similar weaknesses at the extremes, so you get stuck in a trial and error cycle with variations on the mathematical models, which is very time consuming.
The only other individual I'm aware of trying to do a similar thing is Todd Wasson, and I'm not sure if he ever achieved what he was trying to accomplish with his tire model.
Most complex mathematical models involve tables of data obtained from experiments, to be used as coefficients in equations (typically differential equations) but I don't know if any such data would be available to a small game developer, and if any data exists for the problematic "extreme situations".
Anyway, I hope he finally gets something to work and is able to continue.
This thread is mis-labeled. A "tyre physics progress report" thread should only have 1 post in it. This thread should be "rant while we're waiting thread".
Screenshots showed up on youtube around March, 2009. The Asian (Taiwan) beta for World Online resulted in many youtube videos around October 2009. I'm not sure how the information about videos was handled back then, but at the offical EA forum website, they stated it was OK to post videos of the game play from the March 5 - 8 beta test.
One thing was the lack of any serious bugs in the game, so there wasn't anything to hide there. The overall map and tracks are retextured and visually enhanced versions from a combination of previous NFS games (Most Wanted and Carbon), so nothing significantly new was being revealed either.
The videos are essentiallly free advertising for the game.
This beta had a bit less than 1/2 the areas unlocked, and it's my guess that the retexturing and enhancement of tracks and the overall map are still being done in the locked areas.
One potential and monumental task would be to also include the map and tracks from Underground 2, which had a huge map. One issue is that part of Underground 2's area and tracks are located on a fairly high elevated area that overlooks the rest of the area, which requires several more sets of textures to deal with far away views. Example video:
Maybe not average, but similar to Underground 2, you do use up most or all of the steering. This is why the software for wheels usually includes a sensitivity setting, so 60 degree throw to either side would give you full lock for these games. I'm using a joystick to steer with about 40 degrees throw to either side for both arcade and sim oriented games. There were some really fast GPL players that used joysticks, so it is possible to deal with this.
I was also in the english world online beta (there was a prior beta test in Pacific Rim asian countries). It's a fun game, but not a sim. The AI in this game are much better than typical racing games, although the traffic cars are an annoyance. If you're interested, there should be a ton of videos online. Some example videos I made, starting with the non racing stuff:
For virtually every racing game ever made, including LFS (and iRacing), the expert players figure out setups and driving techniques that exploit the physics of those racing games, and most of the time it involves setups and techniques that would never work in the real world.
The replays in NFS Shift can't be saved, so not many players watch them, but they do provide a nice view of the cars, so I made a video of some driver duels and some 1 versus 1 races using various cars:
In the case of hockey, instead of naming the teams after countries such as USA or Canada, why not give them more truthful names, such as National Hockey League Team #1, National Hockey League Team #2, ...
Who came up with the idea of short track speed skating? It's not like they have short track sprinting in the summer olympics. Then again, the USA does have arena football in converted hockey stadiums. Apparently a place for those not competitive in real speed skating or real football venues.
Then there's biathlon, guns and skiing? Some tv announcer made a comment about having guns in hockey games. Now that could be interesting, using rubber bullets though, and using the rifles instead of hockey sticks, I might watch that.
And then there's curling. Will that mean that summer olympics will end up having shuffleboard, both full size and bar size events as well as lawn bowling and croquet someday? Curling is to winter sport as drifting is to motorsport.
True, but if one of the wheels is spinning, then the rate of angular acceleration of the entire drivetrain, engine, flywheel, transmission, driveshaft, differential, rear axle, and spinning tire, will end up much being much greater than if neither tire was spinning, and the torque will end up being used to overcome the angular inertia of the entire drivetrain and that spinning tire at the rapid rate of drivetrain acceleration, and the torque delivered to the wheels will be much less.
This can be demonstrated by trying to accelerate a car with an open differential with one tire on wet pavement and the other tire on dry. Most of the torque will end up just spinning the drivetrain and spinning tire. With a limited slip differential, the rate of angular acceleration of the spinning tire is limited to about the same rate of angular acceleration of the non spinning tire, the otherwise rapid rate of angular acceration of the spinning tire and drivetrain is prevented, and most of the torque will end up going to the tire on dry pavement, and the car will accelerate much faster.
I'm in the USA, and very few cars sold here, will oversteer, regardless of AWD, FWD, or RWD, except from excessive throttle on RWD cars.
Whether or not to countersteer, depends on what caused the oversteer to occur. If it's because of excessive throttle input on a RWD car, then you ease off the throttle and countersteer. If oversteer occurs while braking, then countersteering would make the situation worse, and you'd be better off leaving the steering where it is or inducing understeer by steering inwards. In a street situation, you'd want to slow down enough before releasing the brakes that any pendulum effect will be small enough that you don't initiate another oversteer condition. In a racing situation, you only use enough input to recover from the oversteer, but even expert racing drivers sometimes overcorrect.
Most cars sold in the USA have abs, leaving just motorcycles as the main vehicles sold here that don't have abs. If the streets are wet, then I drive more cautiously (and I avoid driving the motorcycle when it's wet).
In the case of real racing or racing game with RWD cars, depending on the car, it's common to use more steering input (semi-induced understeer) in a turn while slowing, and less steering input (semi-countersteering) in a turn while accelerating. Example video, showing in car view above, and chase view below allowing you to see the steering inputs, tire orientation, and car's response:
I'm not aware of any racing game that properly models how to launch a car with a clutch, mostly because you have no real sense of movement of the car in response throttle and clutch response.
In real life, for a manual transmission, the key is to use partial throttle, with clutch fullty depressed, and slowly release the clutch pedal until the clutch starts to engage but still slip, and the car starts to move, then to hold the clutch pedal still in that position, allowing the clutch to slip until the car has gained enough speed that it's clear that the clutch is no longer slipping, and then you fully release the clutch pedal.
The key point here is learing to hold the clutch pedal still while the car is gaining speed. Sometimes this is badly described as releasing the clutch slowly, but what you're really doing is holding the clutch pedal at a fixed position while the car gains speed.
Eventually you'll learn to modulate throttle and clutch inputs so they are coordinated with how fast you are trying to launch, but the actual movements will vary from car to car.
In real life and in most racing games, you probably want a bit of toe in at both ends.
During a corner, when at the limits, the optimal slip angle for the inside tires is less than the outside tires, so you need toe in at the front and back.
When not at the limits of cornering, but at the limits of acceleration, toe in at the rear of a rear wheel drive car helps stabilize the car. If the car gets a bit sideways due to slippage, the "outside" tire points more inwards than the inside tire, relative to the direction of travel, and you get a bit of self correction effect to steer the rear end back behind the front end.
Front end stablity is mostly related to caster, although most passenger cars use a small amout of toe in.
For racing games, depending on tire wear model, tire heat model, and the amount of grip decrease once the slip angle of maximum force is reached, how lateral and longitundal forces are dealt with in a particular racing game, using toe out can help stablize a racing game car under heavy braking. This was a common exploit used in Grand Prix Legends. What happens is that the tires are at the limits due to braking, so the critical slip angles are very small, if the front end starts to slide outwards, the "outwards" tire gets a higher slip angle and less grip, and the increased grip on the "inwards" tire tends to steer the car back. On some cars in GPL, such as the Lotus, and with braking bias set to a narrow region 52% to 54%, you could end up with a "reverse steering" effect under heavy braking:
Except for an endurance race (6 hours or longer), racing transmissions don't require usage of the clutch. An earlier post already showed an example of left foot braking, no lift upshifts, and throttle blip downshifts, I'm not sure of the shifter type on this car.
To put this in perspecitve, how would the poll be much different if it was about any racing game, old or new? I'm wondering how many players have diminished interest in racing games, or gaming in general (other than first person shooters which continues to attract new players).
A bit off topic, but since not much else is going on these days at LFS:
For a lot of older players like myself LFS has become a living legacy, and it should have been obvious that progress was going to be slow with a 3 person team. 2009 was particularly bad because of the tire physics redo. I've always thought that 3 people was just too small for something as complicated at LFS in it's current state.
The next smallest racing game development team I'm aware of is Synetic, just 8 people, but it's just large enough that they managed to make a series of games, with the two most popular being Mercedes Benz World Racing, and the sequel World Racing 2. I show 664 downloads for a WR2 savegame during January 2010, so there are some significant numbers of fans for these older and niche market games. Since WR2, the series is mostly popular in Germany, related to some television series called Alarm for Cobra.
Manual clutch via a button should be faster than auto-clutch. On some cars in LFS, you can hold down the shift up button (or paddle) and nothing happens, but then tap the clutch button, which cause the actual shift to occur, which results in a very fast shift. IIRC, this works on the Formula BMW.
Realistic depends on the car being modeled. Formula 1 cars use relatively small steering movements compared larger and heavier race cars. Generally in real life, the steering wheel forces are much greater and the movement much smaller than the wheels that you clamp onto a desktop. Short of a chassis and a high end force feedback wheel, you're not going to duplicate the feel of the wheel of a real car.
In the case of left foot braking, it's possible to do this on many racing cars. It's done even in Nascar by some drivers on road courses. Racing transmissions can be shifted without using the clutch, and just a blip of the throttle is enough for a downshift, while left foot braking. Heel and toe might be nice to do, but it's no longer needed in most racing cars. The upshifts just require a bit of a lift. Some cars have no lift shifters which limit or cut fuel during shifts, both for H pattern and sequenial shifters.
In addition, the sedans modeled in LFS are antiquated by todays standards. Most modern cars have ABS, some have stability control, and some now have DCT (dual clutch transmissions), that offer clutchless, no lift shifts, or fully automatic shifting, with what is otherwise a manual transmission.
Instead of modeling flat spots with a 1 or more fixed sized segments, a flat spot would be a calculated surface area of the tread, essentially the size of the affected contact patch on the tire. This area would then become one of the dynamic segments on the tire. If the tire was flat spotted in another part, then another dynamic segment would be created. If the segments overlapped, they could be merged into a single segment if appropriate. There would need to be some reasonable limit on this, perhaps 3 to 6 segments would be enough.
For the hot spotting, or at least to deal with camber issues, the hot section would be a cylinder all the way around the tread, but only partly across the tread, with an average temperature and an temperature gradient parameter for outside to inside of that hot "cylinder". 3 cylinders zones around the tire would probably be enough.
In real life, unless the flat spotting is extremely minor, you genernally have to pit and get a tire change.
I think my offsetting weight method would produce similar results, the tire would tend to lock up at a previous flat spot. I don't know if or how the current segmented model deals with flat spots that are longer than one tire segment, so it's hard to say just how realistic it is.
A dynamic (variable size) segmented model would be more accurate, to prevent it from getting overly complex, there would need to be some arbritrary limit on the number of dynamic segements on a tire.
I recall a comment from some post at RSC. SR seems like a honorary award. It's like the iRacing developers are thinking, OK, we know that most of the players at iRacing aren't competitive, so we'll promote them for being safe drivers instead, and allow them to run qualify sessions to increase their SR in case they can't increase it in races. Sort of like having a Nascar award for "cleanest car" at the end of each race. I guess this stops at "pro" level though, where I assume there's some minimal iRating requirement?
On a side note, is the number of active players continuing to increase and about how many are there now? With the reduced pricing, I would expect a pretty good jump in the number of players.
This is one case where the option of turning collision off in such situations might be a good idea, even if it's not realistic. Some arcade type racing games do this (collision off if not on the same lap or if a player is driving backwards).
What I referred to was the current segmented tire model, which isn't realistic either, since real tires are analog not digital.
Flat spotting could be emulated instead of simulated, perhaps by offsetting the weight on a tire to make it bounce instead of trying to simulate an actual flat spot. The end result would probably be close enough to the current model to accomplish the same goal as the current model.
The current hot spotting model isn't that realistic anyway, so I'm not sure that it needs to be re-implemented in it's current state with the new tire physics.
My concern here was that the current segemented model may conflict with the new tire physics, as I suspect that the model needs to consider the contact patch and the parts of the tire affected by the contact patch as a whole rather than as a bunch of somewhat independent segments.
A more realistic approach would be to have dynamically sized segments, but that seems overly complex.
The main point here is that since the tire physics is being redone, perhaps now would be a good time to do a rethink of the segemented model.
Just wondering if most players would be willing to accept the new tire physics without the hot and flat spot segmented tires, if it meant that the new tire physics could be released sooner.