Don't really like that analogy. To me it's more like another band making their own arrangement, or a pastiche. Laguna in rF(com8) has quite different and much less severe elevations and cambers than the iR LS version. The Corkscrew itself should feel like falling off the edge of the world but it's a gentle dip of no note in rF. You don't get that sense of balancing on the edge of oversteer in the downhill sections. The trouble is they all (rF, iR and RL) look surprisingly similar on video. And unless a driver is on the limit and gets it wrong, you can't really see what he's doing (e.g. backing off of the throttle), from the outside. It's like the proverbial swan paddling away furiously underneath the water while looking serene on top. And since sim drivers will push everything to the nth degree, that's when inaccuracies manifest as behaviour that's well away from reality; true of both tracks and physics generally. That's a good analogy IMO. The character (sometimes including mistakes) remains. And regarding bumps, some of those and patches/repairs have effects for years.
No one said that LSed data are turned into mesh. iR uses those data for physics engine. Meshes are flat like hell. However, iR can do that. rF - can't. I'm curious if rF2 will can.
In addition to its sight-picture visuals, iRacing.com replicates the precise physical features of each track’s racing surface. Our laser-scanning technology produces a mathematical ‘bump map’ of the track’s camber, cracks, undulations and patches – recording every millimeter of the surface. A series of “point clouds” capture the three-dimensional profile of the track surface and adjacent curbing. Combine this mathematically-precise surface mapping with iRacing.com’s hyper-accurate sight-pictures and proprietary mapping software and you have a powerful tool that allows even drivers at the highest levels of professional motorsport to use virtual seat time to hone their skills and improve their real-world performance. That was taken right off of their website. So yeah. I was wrong. They don't claim any mm or cm accuracy. They just state that the scanning records the track surface and curbing to that accuracy. After that really it's open to what the builder wants to use. As far as off track objects they take pictures. This LiDar method does look pretty interesting. I'd be curious if a developer would decide to go that way.
Within rF I would love to use laser data to create a seperate driving mesh (different from the visible mesh that most tracks atm use for physics), and majorly increase the data used to really show what's possible with laser scanning in rfactor. Thinking along those lines, if you were to laser scan a track, then you could have a free version with just the normal mesh, and a paid version with a high-quality physics mesh... hmmm ideas, now get me a scanner
The vid showing iRacing Scanning part of Mid Ohio has the surroundings in the point cloud. Obviously, you need photos too. Even on the road, a change in texture picked up by the point cloud needs to be interpreted - graffiti/skid mark/yellow or white line/patch or whatever. I'd guess they use the point cloud to put some main surrounding features in the right places but photos for a lot of it to save time/cost in processing(?). Certain details used by RL drivers as reference points are probably right e.g. Corkscrew at Laguna, people use a tree to find the exit line.
Yeah and that's the real issue with GPS. If you can get 5 metres of error, that really translates to a lot more. If at the start of a turn you are 5m out, and the end of aturn you are 5m out, and you built a curve between the two, that curve might not be very accurate at all, as your 5m could be in opposite directions. You might actually be better off using a satelite photo... Same with elevation. If you imagine taking a reading at the bottom of Eau Rouge, what if the GPS says you are 5m higher, than you take another GPS reading at the top of Eau Rouge and the GPS says you are 5m lower. You then have an Eau Rouge that is 10m less tall than it should be. Scanning does have it's benefits, but even if you are using a laser scanned track to prep for driving it in real life, you still have to adapt to the real track and the car when you get there. All I really want (personally) is something which drives the same. I think this is what I like the Monaco vid for so much. As driving track, and then watching the movie Grand Prix, it's eery how similar it seems to be. That's all I guess I personally want. I know some folks want different things though, and scanned tracks would be welcome in rF2 if it was possible, but for the pricing level we're looking at it just isn't feasible. In the end, someone has to pay for it. If we did it, that would be you.
What about an altimeter? Lets say you are fortunate to complete a lap without any air pressure changes, it should be precise yes? Where I am going with my Q's is this http://www.youtube.com/watch?v=5_UJKA8AwlA&feature=youtu.be&hd=1 I spent a long time working on elevations, probably 80% of the building time was spent on RnD and tweaking the elevations, from a mixture of existing rf versions of this track, some data found off the net and this video, I was able to get a base model set up, then by liaising with a friend who has access to a ls version of the track, he drove that track and my track comparing the 2 and advising me of tweaks needed until he could drive the track as comfortably as the other one. In the end, I had some people who drive the track irl test it out and they were satisfied that the elevations were quite precise. Although I don't think I will put this much effort into track elevations in the future, but it was interesting to attack the challenge and see how close it can come. Here is a video of my track if anyone wants to compare the 2 http://www.youtube.com/watch?v=pGzRDSsG-cE
As I stated earlier...civilian GPS CAN attain sub centimeter accuracy in x,y and z...we do it here every day. In fact...if we were contracted to construct a race track...it would most likey ALL be surveyed using GPS...civilian GPS...albeit expensive GPS. Now if you are talking handheld consumer based GPS...well then yes...you get what you pay for. The typical RTK system with base station/cellular data access would cost you about 50 to 60K...and we currently employ 6 of those such setups. Aerial based Lidar can deliver pretty decent accuracy...perhaps not as accurate as land based lidar..ie laser scanning...but would definately give you very good elevation relief...most likey more than enough to get the "feel" of proper elevation changes in just about any circumstance. Some of the Lidar I have seen here comes in a 100mm grid format...thats right...4 inch grid. I would think that with that kind of point cloud one would be able to model the **** out of just about any track/location.....if there was good Lidar data available. Combine that with good photos and video and I think most track could be built vey close to as they are in RL. BUT....this all costs money...and perhaps lots of it so as I said before this is most likely not an option for "joe Modder" but for accuracy..obviously the best choice.
Yep. As funny as it sounds, even laser scanning relies on GPS data. All laser scanning does is gives you points relative to other points, but it doesn't give you those points relative to the earth as a whole. GPS data is still used to tell you what angle to position the point data at. You can have a totally accurate pointcloud and still positon the data at an angle if your GPS data is wrong, essentially making the whole track incorrect if it's sloping wrongly. But that's why multiple GPS readings are important, too, in that usage.
Thanks you so much for saying this. I haven't read all the posts,. Much has been said about laser scanning, but there is another side that track accuracy alone is not a limiting factor. I was going to draw a diagram to explain, but anyway, I will try and put it in words. Imagine what you have ingame, picture a view of a tyre hitting a kerb. You will see a mesh of finite elements with infinitely sharp edges . Somewhere along the line one has to make decisions about the phyiscs of this situation, how to interpret terrain, resulting forces etc, and to whatever degree of complexity you may wish to do this as long a you can keep it in realtime of course It could be a long story, but will keep it short, for example, the tyre deformation, what happens on impact, the springiness if you like, If those are things are only approximate, you have lots of room in many areas that can throw things off. Anyway, my point, good car physics and tracks go hand in hand. Perhaps decent car physics on a slightly less accurte kerb may actually result in more accurate suspension movement and repsonse when looking at the telemetry as opposed to the other way round. As the OP said, everything is relative, the healthiest balance one would hope for is to consider each part carefully, and not overly investing in one part over anohter, when one part is no longer a limiting factor. Good thread, good read, got to read the rest of it some time
Nice to see lots of different opinions about Personally i remain with my position..laserscanned tracks have surely better precision and accuracy respect standard built tracks, specially in terms of banking, elevations ecc. These points are really difficult to reproduce with big accuracy without a good surveying; even with .dwg CAD files is difficult. That's the main difference IMHO. BUT there are several "standard built" tracks really well done. What's the "problem" in rF? difference in term of quality between tracks (and mods). Comparing physics and tracks accuracy is nice but, imho this is not the true point here in this discussion; physics is already a big important side in a simulation but...here we're (am i?) assuming the "perfect physics" on cars; so..we're trying to analyzing: "perfect physics" car vs "standard built" track "perfect physics" car vs "laserscanned" track In this thread imho discussion object is just between different ways to build a track
If all tracks will be built with such attention to detail and artistic verve like Monaco from promo video, then I can live without lasser scanned tracks. I'm trying to imagine how Monaco looks at sunset or in the rain....and I can't
Yes, really depends about attention to detail; like posted before there are several tracks built without laserscanning but really good..like for example Virtua_LM tracks. For rF2 would have liked a sort of "quality and legality control" (tracks and mods) but..probably this one will remain in my dreams.
Hell of a good read so far. I believe as well like many already stated that laser scanned tracks (if done correctly) would be a really nice thing for all of us. I do think so myself. But is it absolutely needed? From my point of view, no. Since I will (most probably) never drive on real race tracks, or just occasionally, I don't care that much if a track is ALMOST (laser scanned) as real as the real thing in terms of elevation, position or bumps or "just" really CLOSE (traditional modelling) to the real thing. But I agree, it has to be CLOSE. Some of you pointed out tracks which are several meters off compared to the real layout. Some tracks which are barely recognizable as the ones they're pretending to be. This is of course something we should not be targeting at, but to be honest, we are far away from such things today. Most tracks created by modders and sim (or game) studios are fairly accurate in my opinion. There are a couple of curbs which can be slightly off, but the overall quality increases steadily. I am amazed how close for example Virtua_LM's MidOhio or Sebring come to iRacing's versions (just using it as a reference here for laser scanned tracks generally). From a personal experience I would say that, for example, VLM's Sebring is more than sufficient. As a fan of mixed GT and LMP class series I love the ALMS, and I race on tracks which are used by the ALMS with the Enduracers mod most of the time. Due to that, we had a race at Sebring every season. We raced a version which is really different from iRacing's laser scanned version. All turns were different, some to a really great extent (especially the last one back to the start/finish line). This season we're using VLM's version and even though there are obviously still some differences, I would have no problems switching from VLM's version to iR's version on the fly. This is probably due to one simple fact: If you want to compete at the very top, you'll need to be able to ADAPT very quickly. This is probably the most important thing for a driver. If I can beat a driver on a track which I race on for the first time and the other driver is one of the fastest on a different version of the same track (where he did thousands of laps) which is just a little bit different from the one I beat him on - then he can't adapt to new conditions very well and lacks a very crucial ingredient of being a very good sim racer. Like stated, real tracks change over time. Not only the surface, sometimes there are new types of curbs added, others are removed. Run-off areas are added to improve safety, and sometimes those will change the lines real race drivers will take on the track. They change in a minimalistic way, but they do change. As long as the overall layout and racing line doesn't change too drastically, you will have no problems at all to 'feel at home' again after a couple of laps. Another thing will be rain (hello rF2!). Let's just pretend you did thousands of laps on a specific track in the dry, and then all of a sudden you're supposed to race in the rain. Not only the braking points will change, even the racing line could be completely different now. Of course VLM's or Com8's tracks are still pretty intensive resource-wise (time is a very expensive resource!) as there are months or years spent for retrieving suitable reference data and for the modelling process itself compared to "simply" scanning (texturing, logics etc. have to be done on both afterwards so that's not taken into consideration), but I'm impressed every time groups and developers release such amazing works. As for that, I would be totally fine with tracks which are really close to a scanned track.
Yes you are right I mean I didn't read the whole dicsussion at the time, but thought would throw that thought in there anyway, to make for interesting discussion and to look at the value in the bigger picture. Sometimes people can get sidetracked by how important the value of one mehod is over another when in fact there are many cases where it is not the most important factor overall. I would go as far as to say, as long as a track is reaonably accurate, you know, not just something made from a pamphlet but done well enough to capture the essence of the track within the limitations of the sim in question. For example when a tyre hits a kerb the question is not only. This kerb is X cm but we wil ignore the fact the car bounces off it like crazy ( common thing in many rF1 mods for example ). It is the sum of the parts that need to work accurately as possible. All these expensive methods are being pushed into our faces a bit, they don't hurt for sure to use them, the more accurate the merrier if you can justify it but IMO a lot of it is marketting hype too. For example iracing can say. "Look we have laser scanned our tracks" it sounds good, but when you put it in context it is no the be all and end all. I bet, If you got Mr Dave Keamer drunk enough, Tim may know how to do this Get him down to the pub and had an open converation, and asked him what the bigger fish to fry are in the sim world. I highly doubt he would say, "Oh it is this bump on that track that throws everything off". I bet there are still far greater uncertainties in the tyre model itself potentially, be it in the formulation itself or due to limited data. Okay I will stop talking about this aspect now, back to the true topic as you say. All I hope that some take away from my post to have these other things in the back of your mind, and what it means in the bigger picture
hey, i didn't want to be rude, i really love this thread and you did a good post talking about physics, i'm sorry if my last post was a bit "rude", was not my intention, sorry! that's why i'm not so good with english and..i use just few words for explain my concepts. I'm not a pro with english language. Anyway you're right, physics is absolutely an important side of a simulation (the most?) so..when you simulate a certain car, you need accurate tracks, but..yes for sure an accurate physics and sounds too. Regarding tracks, Laserscanning limit (big? ) is just the cost in term of money; i'm not an expert in this world but..building a track using "standard" methods (not just using CADs, most of us don't have possibility to work with it) is a lot difficult..and it takes an huge amount of time compared to laserscanning method and, yes, using "just" google maps, tons of photos and onboard videos, you can't reach laserscanning quality. I don't want to be rude, of course, i really respect all modders work (i'm rFactor user since 2007) but..we can see lot of quality difference in all tracks released I'm just curious to know if Virtua_LM guys usually uses CAD reference files or just google maps + videos + photos. Anyway..even with laserscanning data you have to do a lot of work before see the track done BUT, you have a lot of reference data. I'm surveyor (an almost architectural engineer...that's why i LOVE this thread) and..i remember how time it took to make -and report in Autocad (R14 lol)- a survey using old theodolites (6-7 years ago)..now, with modern instruments, you "just" have to do survey and..using a specific software you have ready entire .dxf with all reference points..ready for Autocad. That's the progress
Just as a illustration for what is being discussed above, I'm very familiar to Interlagos circuit (Brazil) and I can assure you the most accurated in terms of elevations, cambers and objects along side the track is from Codemasters F1-2011. Looking into the 3dModel with proper tools, I can see that track surface is much more detailed and have many more points than any Rfactor version I've ever seen. HOWEVER, even being the better and more realistic 3d Model of the track, that's also the worst imersion experience. The cause of that is the bad physics. (The worst track version + better physcis) > (better track version + worst physics)
your post was not rude at all and I did bring it offtopic, a habit I have sometimes but points taken. When i said a good track I was referring to some cad data of sorts or good reference, even if it is a bit out of date, even if the track been resurfaced as already discussed, but cm accuracy in every area on a freshly laser scanned track from yesterday, no need. couple of months, heavy winter, few new cracks appear, data accuracy is already gone out the window to replicate all the bumps and what not. IMO, kerbs for sure could be done well with good photos and some measurement tape, some modders have in fact made some very good tracks with limited info, but yes ,they will never be as good. A glimpse to the future. 2030, Gjon has a grey beard, the others are on a similar path. Satelite scans track every hour, start up sim rFactor 5, download latest data from server. Generate track surface on the fly based on this data on track loading. Now I got to stop
