CHECK THIS OUT!! Future RF2 addition ??

Someone else on this forum had posted some great rigs of rods videos a while back, that's how I first knew about (sorry forgot who it was).

Even the GTA4 damage model is quite good even if not as refined as RoR.

 

Heck, even the Viper Racing damage wasn't bad, better than most racing games.

 

I'd love this damage model to be in RF2, but surely the point of a sim to to give you the best handling characteristics and I'm really glad that ISI are working on that. If you are heavily into crashing and not that much into handling you might like to try Flat Out 2.

 

I know I had a screenshot of NASCAR Heat damage somewhere !!

 

Having said that, damage is an important part of racing. The damage model as it stands now is nowhere near good enough in my opinion. I understand it may not be a priority but it is definitely something that needs LOTS of improvement.

If I go off at Spa and hit a fence, I expect the car to be damaged beyond repair with many pieces missing. Being able to barrel roll, cartwheel and shoot 20 metres into the air landing sideways on my wheels and then driving off with effectively just some bent steering is not realistic obviously.

In the Formula cars I would expect bits of wings to be breakable (upper flaps, end-plates etc), the rear wing able to break/collapse (as de la Rosa demonstrated at Monaco courtesy of Pastor!) etc., barge-boards to be broken off etc.

I won't mind as much if the damage doesn't look fantastic (like dented panels in a touring car) but missing parts and being breakable is definitely important.

 

Don't forget that visual damage and physics-based model damage are not always mutually dependent.

I don't think rF2's aero model is calculated dynamically by some form of real-time wind-tunnel/aerodynamic-particle simulation (which is likely to be heavy on the CPU especially for something as complex as the geometry of a modern formula car). It's more likely to be based on predetermined data that is taken from something like a CFD (computational fluid dynamics) simulation of the entire body or individual parts and then they produce coefficient values for the various setup changes you can make to your cars wing angles, ride height (for ground effect), etc, that affect the dynamic download.

So even though you cannot see more detailed levels of damage (such as bits of wings) flying off the visual model, does not mean conclusively that they haven't taken damage into account for the aero-model. So as you hit your car on something and would expect at least a little damage to it (for example the side of the wing), just because you can't see it doesn't mean the aero-model hasn't affected the available lift to that area/side of the wing.

Ofc, i don't know if this is the case and even if it is so, it would be nice to see more dynamic visual damage so that we have a more visual cue for how much progressive damage has been sustained so we can make quick on-the-spot judgements as to the possible severity of performance lost (more prudent for the formula cars that rely so heavily on wing performing).

 

My point was aimed at visual representation of the geometrical model rather than the physics impact of missing bits.

The latter is too complicated to calculate real time. To simplify, I'd imagine plugging a few values into a simple equation to calculate your Cd value for the car giving 'x' amount of force at 'y' ground speed. And then for example missing the entire front wing will decrease Cd by a nominal (canned) value 'z' giving 'x' at 'y'. That same 'z' can be defined for all other bits and pieces that can come off too.

I have no idea how the aero model works now. I do know that it does not interact with ambient conditions (I did a temperature/humidity test posted here: http://isiforums.net/f/showthread.php/4839-Temperature-effects-on-Aerodynamics?highlight=).

 

Good initiative on testing the effects of temperature on wing down-force production. I'm going to guess that the ambient temperature only affects road and tyre temperatures perhaps? I hope ISI fixes this because it shouldn't be complicated to fix.

Regarding the damaging model, i don't know if i would want the visual representation of dynamic/detailed-component damage unless it proportionately affected the aero model. I think if it were only visual that would/could leave the impression that you should repair your car as you'll have performance loss.

I suppose it would be good to test to perform incremental tests of the down-force produced after impacting at various speeds at a wall and repeating it up until the point of visual damage (i.e. no wing). Would love to know what results would yield. That way we would know if the physics model is more complex than the visual.

 

Perfect example with this Pic! See the Damage that all around the car, when we wreck in rfactor and hit the pits we get a brand new car- I know if you get that much damage there is noway for it to come out perfect!

 

http://youtu.be/YelodRFABH0?t=1m14s

 

The physics interaction with the environment is very important as we've discussed (more here: http://isiforums.net/f/showthread.php/5812-Elevation-Altitude). If car performance (aero and engine) is not dependant on environmental conditions (temp, humidity, air density) then I believe it's a big flaw on the physics model.

Ideally yes the geometric model should be replicated in physics terms. Right now it seems it's the physics which respond to damage to some degree but no (or limited) geometric representation of it. If I bend my suspension, is it geometrically modelled? I'm not sure it is or if so I haven't noticed it.

 

There is no way aerodynamics can be calculated real-time (even in the assumption that the car would be stationary). You'd have to take into account: boundary conditions (temperature, wind velocity etc.), interaction between all the complex body parts, etc. Even realtime boundary layer calculations for a flat plate would be difficult (and obviously that is not a valid representation to say the least). F1 teams still use a lot of wind tunnel test, because it is still the way to go to test the effects of complex part geometry. CFD is much cheaper of course (since you don't need an expensive windtunnel), but is far too intensive to apply for real-time game simulations.
I think the Cd value approach you suggest would indeed be a decent and cost-effective way to go about, but of course this will only affect the drag the car experiences at any given moment. However, Formula cars are all about airflow management (downforce, cooling, etc) and the drag coefficient doesn't affect that in the slightest.

 

Maybe not, but it seems that it is somewhat probing
Here you can see the effect with the flat bottom where air seeps underneath on a car who exploit ground effect. Like real effects.

 

Yeah, I know exactly what you mean, but that's the problem with mods or servers running decreased damage levels.
In the league I race in, even harmlessly looking contact may end up in tire puncture - all it takes is some bad luck.


As for the video from first post - always look at such things with some reserve. This wideo shows two things - flexible car bodies and damage model. Just imagine how much impact would flexible car bodies have on these new tyre models in modern simulators. They could ruin entire simulation. In a general-purpose game engine lack of accuracy and realism is common and acceptable. Things only need to look believable and that's all.
If the same approach was used in simulators, you could put a donut physics instead of tyre model and show people a nice video, of how "tyre" is deformed while cornering - many would be amazed

Or if you look at water effects in modern FPS or RPG games - when character walks through water you can see nice waves. This doesn't mean that the same technology can be used for a motorboat racing game, because it's just a visual effect - not true hydrodynamics.


Another thing that a non-simulation software can do is to run things on GPU, which for many reasons is not very suitable for a realtime simulation that needs to use GPU for graphics at the same time. There's only one GPU and it would mean that update rate of physics would be affected by update rate of graphics.


First step to put this kind of damage model into rF2 would be to make collision detection / response run at different and variable rate than the rest of simulation. This would allow it to be handled by GPU. Now imagine over 40 cars performing bump-drafting on a superspeedway with physics and collisions running at different rates...

It's a challenge. Yes it can be done, but it will require some research and a lot of time before we will see this kind of damage models in simulators. A time that can be spent on other things that are more important for racing. This doesn't mean we can give up on it My point is that even if you see something cool in a game, it doesn't mean that there's a straight way into simulation for such feature.

 

Indy Car racing was on windows 3.1, when windows 95 was released- indycar 2 appeared - in 1998 GPL reinvented sim racing. i bought a 12 mb voodoo card to run it! ..its been a little more than 10 years- try 17 years just since Indy car 2 - I gave up GPL after 10 years and that was 2008. Been a long time since anyone tried to properly simulate late 60's F1 cars-

 

I need to correct one thing: IndyCar Racing (both of them) were MS-DOS games. Otherwise you're good. =)

 

Fire is a HUGE no-no in NASCAR licensing these days. Interesting to see it and remember it.

Damage obviously is something we will tackle, but it is very important to state that this is a racing simulation. We're never going to have (or try to build) a product where the crashing is more impressive than the racing experience.

 

Thank you! Don't feed the crash kids!

 

I'd like to mention Banger........oh nevermind

 

Thats why if we do NASCAR content we have to do it without licenses, just some general stock car content like in rF1. Then we can have fire and what ever we want. Modders make sure that we get real manufacturers, drivers, brands etc