LiDAR circuit models

Great post Revvin, specially about this point.

+1
we need quality, not quantity.

 

I used to work for a Formula 1 tyre manufacturer and I can categorically state that we needed to collect the same data every year from the same circuits purely because of the changes year to year, even if there wasn't any track resurfacing.

If the professionals who rely on this data for their job can't do it, no one can. You can only get close, and even then, it's only close for a short period of time. It will soon become obsolete.

Close enough is near enough, as stated earlier, it's just a representation. You, me or anyone else won't know the difference of a few metres unless you're out there banging out the laps in real life. What's it matter if it's only 95% accurate anyway? As long as there aren't obvious flaws like elevation changes that don't exist, I don't see how it changes your experience.

 

I'm sitting at a desk I would consider to be 'flat'. If I actually took a good look at it I might find out it's not; further, if I acquired some proper tools I could probably not only confirm that to be the case, but come up with a specific measurement of it's 'unflatness' and in comparing that to various other everyday objects (and flatness itself), I might come to the conclusion that my desk is decidedly nowhere near being flat. Perhaps I would even consider its manufacture to be shoddy.

And yet here I am every day, working and putting stuff on it in the way you would expect of something flat, and its very likely imperfection has never been an issue.

We can talk about 'obvious flaws' in tracks, and that's fair enough. But if the track is made sufficiently well that you can't (or at least, don't) notice that anything is wrong - which would obviously require it to be relatively correct - I don't see what there is to be gained from having a laser scanned version. Of course the laser-scanned version will be 'better', as in more correct, but if the manually created version is that close to start with you probably wouldn't notice the difference between the two anyway. Knowing the second one is laser-scanned, though, might give you some peace of mind if imperfections really bother you.

And yet the one thing that comes up more than any other when it comes to laser-scanned tracks is the bumps. Take a look around at these sorts of discussions, at least half the responses extolling the virtues of laser-scanning refer to how bumpy the track surface is compared to unrealistically smooth manual creations - but a few hours spent watching footage from the track will very easily give you an excellent 'map' of where the bumps are, so there's no reason a track maker can't get bumps in most of the right places. Exactly right? No, but would you notice the difference? Probably not.

Meanwhile, the real bumps on the real track that you would get from laser-scanning change from year to year with track resurfacing, weather conditions, effects from race cars (in braking zones and apexes, for example), and whatever else might affect them - and in arguing that laser-scanning is the only way to accurately reproduce all those bumps you're making the assumption that what makes it to the game contains the same level of detail as the scan itself, which would actually seem quite unlikely.

In the end, what's being done with it? Are we using these tracks to conduct flooding simulations and evaluations for the purpose of formulating disaster response plans? Are we using our super-accurate track reproduction for something much less important in the scheme of things, such as developing a car (and setup) for a real team using the combination of our flawless track and world-first perfect physics engine?

No, we're going racing with approximately correct car parameters on a current-technology physics engine with (mostly) consumer-grade input hardware, viewing our progress (and the track) via an on-screen representation of what we would actually be seeing... and we can do all that even if there are 'obvious flaws' because everyone still has the same track to drive around, which is all we actually need.

Is the increased cost involved in laser-scanning worthwhile, given that so many areas of a sim will unavoidably be a compromise because of the limitations of modeling combined with the limitations of current hardware capabilities and the need to have everything running in real time? I'd argue that its benefits over a very well made track are not enough to warrant the effort and money. In comparison to a lesser-quality track, and especially a converted track that was put together at a time when hardware further limited the fidelity, sure, but that doesn't need to be the level of manually created tracks.

If laser-scanning seems so attractive, and I can see the appeal, let me suggest this: perhaps it's not quite so accurate as you might believe. Maybe at times there's a bit of fudging of data to make things fit together, and quite likely some 'lost' data as things are simplified to be able to run on a home PC. It's still based on laser-scanning, and it'll be called a laser-scanned track, but is it perfect? Is it even as good as the version that'll be created from a new scan (or the existing data) in several years time? No. Of course it's close enough for you to not notice what's 'wrong' about it, and you have the assurance from it being labeled "laser-scanned", but it's still an imperfect representation of the real thing. And in that sense it's not really any different to a quality manual version

 

Excellent post Lazza. Laser scanning, like most things iRacing, is mostly hype.

 

What he said


I was watching a NASCAR program that was a day before a race at a track called Dover. The show was NASCAR performance and it's a technical show with two or sometimes 3 crew chiefs discussing the next race from a setup/technical point of view. Anyways! They were talking about the expansion joints that are cut into the concrete surface of the oval track which creates alot of "chatter" in the suspension of the car. These expansion joints are cut every two or three feet. In the laser scanned version of the Dover surface you CANNOT feel these expansion joints. So there is something that is very obvious. I'm sure there are tonnes of other stuff that isn't so obvious when it comes to the accuracy of the track or something that would be alot easier to spot if you've been there or race at a specific place in real life.

 

Even if you had a perfectly representative simulated circuit, it becomes redundant if car physics are rubbish... The latter should be the primary focus of our interest. If car physics are bang on, you won't care so much about a circuit being 'only' 95% accurate.

It's all relative!

 

If car has physics defined in wrong why, you may always think that you are driving some car which has not been produced yet. But if you drive on straight named circuit - you cannot say that it is some theoretically existing track, you assume that it is Spa, not some kinda-like Spa replica.

 

And what if the car is named aswell?

 

+1. I remain with my opinion, laserscanned tracks are better than "standard" ones. Ok, you need to know what you're doing but, IF you know it, with laserscanning data you can do an almost perfect job. Simply you have lots of reference data on which you can work.That's the fact imho.

 

I knew someone will ask for that
That's mean you have been cheated by a car salesman

 

I think we can all agree that with good data a person can create a good representation of a track. GPS, photo's, video footage are all well and good but there is still a large margin for error by the track creator’s interpretation of that data. Does that camera angle make the corner look tighter? Does that GPS properly show the elevation change and was that a piece of debris bumping the car in the video or an actual bump on the track? With laser scanned data there is still margin for error but that’s far smaller than the GPS/photo’s or video data. If we're happy to accept such imperfections then why is the internet awash with sim racers arguing over things like Simbin's Macau or the numerous video's showing the virtual and real track side by side proudly boasting of its accuracy? if we didn't care for track fidelity why are we not still playing the original Grand Prix but just with better graphics? Laser scanning is a step forward, is it perfect? No, is force feedback? no but we still buy force feedback wheels, still buy newer wheels with better effects, better ways of providing the feedback but if it’s not perfect why not just go back to my first PC wheel - a Thrustmaster T1 which had a bungee cord to centre it. Motion pits are not perfect but it keeps improving, people are still buying the units and enhancing their experience. They will all improve as has the track technology and physics in these sims.

We as sim racers keep demanding improvement, to be more accurate, to be more realistic with every new release, we hold new games up against our existing ones and pick them apart for the sake of proclaiming our new game is better than sim X, Y or Z. If the track is not as close as it can be then it also renders other interesting concepts like iOpener a little futile, why pit yourself against a real driver on a real track if the camber on your track is not right, that the kerb he rides is too high in your electronic version etc etc. We keep moving forward but is the argument against laser scanning here more about because it may be that rFactor 2 won't have it at launch, maybe never have official laser scanned content and another competing sim does? I remember many critics or iRacing's use of laser scanning technology doing backflips when the first laser scanned track arrived for rFactor. If you could download just one track and you had two guys release their version of Spa. One announced his as being "A pretty good representation, it may not be totally accurate and I think some of the cambers, radius of corner and elevation changes are not right but I did my best from the GPS/photo/video data I had" and the other guy's release was announced as "An extremely close rendition of Spa using laser scanning technology to make sure the cambers, corner radius and elevations are as close to being realistic as possible" I know which one you guys would download.

 

You can say iRacing is crap, in all sorts of ways, and I wouldn't argue with you. But saying laser scanning doesn't make a real difference isn't one of them. It's the tragedy - that is iRacing - that only they can afford to do it, and are probably keeping people there just because of it. Some of whom are in denial about the other shortcomings. And I was one who originally thought laser scanning wouldn't make any difference to me before I'd tried them.

The shape of the surface - the contour/camber and local dips and rises makes all the difference even if you don't care about RL accuracy (and there are plenty of pros saying they're spot on and useful if you do care about that). You have to budget grip-wise for all this variation and make constant corrections. Regular tracks feel flat in comparison even though there are obviously rises and falls as you go around. But to drive on, they're pretty much the same as each other but just a different sequence of rights and lefts - which is probably why some folk want so many of them for a sense of variety. But as I've said, I don't see why this has to be the case. But someone's got to recognise the facts before it'll improve.

BTW, I'm only talking about road tracks. I know nothing about ovals.

 

Give me a road give me a car and I'm gone..... Laser scanned? Doesn't matter it still looks like a cartoon... Until we can render Real time video, what's the point.. So you refuse to drive on a fictitious track? I have just as much fun on well made tracks. Just get the physics right so we can have some side by side battles!! Like two Cartoon Cars going after one another.... That's where we are at in 2011 and I still love it... I'm not getting pushed off the road by a 50mph cross wind either.. Lets just have some fun till 2020 when the good stuff comes out;o)

 

laser scanning has not proven to be any better than good old fashioned hard work. Unfortunately there are alot of devs and track builders who don't put in the hard work. I think that there is a misconception as well that unless it's laser scanned the track surface is completely flat. I think that's false as that has more to do with FFB, tire and suspension model in alot of cases. Kudos to iracing for brainwashing some of the sim racing community. It all comes down to personal choice. The fact is really if you want laser scanned tracks you're going to have to put up with crappy physics.

 

I wouldn't bet on "better looking"

EDIT:
I was just simply pointing out that laser scanning has nothing to do with how tracks look

 

I'm certainly not saying iRacing tracks look better. More detailed, possibly. And some people like that 'clean' look. Whereas I think they look highly synthetic/computer generated because of it. Same problem with the sound. Kills immersion for me. In fact, iRacing is probably the worst game I've tried out of RBR, rFactor, GTL, GTR2, P&G, LFS,NKPro for over all feeling you're still sitting in front of a PC.

Regarding wheel FFB: A while ago I turned it off to see how much of the feel I like of the laser track was due to that. Got used to it surprisingly quickly and was just as fast, which supports my suspicion that it's not that useful informationally but more about immersion. Anyway, I decided it's not the FFB because the 'shape' of the iRacing track surface ( camber etc.) is still having it's influence and you still have to deal with it just the same. The FFB is a bit of conundrum because sometimes it gives you useful information but other times misleads. I put back on just for immersion really.

 

First of all, LS tracks (in iR) are not "completely flat". Only micro-bumps are moved to physics layer data.
Otherwise it would kneel-down computers due to number of polys.

Depending what you mean saying "good". LS is more accurate, it hasn't to be proven. It is a fact.+++

Of course it has to do with FFB. but only because physics has data to work with - LS data, in opposite to flat or faked-wavy surface in rF.