All aviable Laser Scanned tracks @ rF2 ! ! !

Discussion in 'General Discussion' started by memoNo1, Jul 28, 2020.

Thread Status:
Not open for further replies.
  1. kashif

    kashif Registered

    Joined:
    Dec 23, 2017
    Messages:
    42
    Likes Received:
    20
    "Definition-A laser is a device that crops a monochromatic intelligible beam of light while a LIDAR uses penetrating pulses of laser light to facilitate the process of measuring the arrangement and structure of the atmosphere. This is a huge difference between the two."

    https://lidarradar.com/info/differences-between-lidar-and-laser

    They are not the same thing and as matter of fact produce very different results.
     
  2. LilSki

    LilSki Registered

    Joined:
    Oct 7, 2015
    Messages:
    57
    Likes Received:
    364
    I don't believe you read the article currectly...

    It's basically just saying a laser is different than lidar but lidar uses laser technology...

    "Laser scanning" is just a cheap marketing term for lidar because it sounds cooler than "lidar scanning"
     
    OdeFinn, Emery, ApexModding and 2 others like this.
  3. TheMenace

    TheMenace Registered

    Joined:
    Dec 27, 2019
    Messages:
    10
    Likes Received:
    13
    True: http://www.rfpro.com/digital-road-models/lidar-survey/
    Guess they don't need the fancy terms and the deceptive marketing at their level.
     
  4. kashif

    kashif Registered

    Joined:
    Dec 23, 2017
    Messages:
    42
    Likes Received:
    20
    Sorry about that, after reading it more thoroughly, it seems your are right.

    Although, there still seem to be different types of Lidar.
    https://new.certainty3d.com/blog/4-types-of-lidar-remote-sensing-explained/

    Do you know which methods are used in simracing?
     
    Bidle likes this.
  5. vittorio

    vittorio Registered

    Joined:
    Jan 11, 2012
    Messages:
    1,029
    Likes Received:
    464
    I don't care that much between laser scanned or LIDAR scanned (basically the same). Important is the accuracy, so is it scanned at ground on the track (usually higher accuracy) or scanned from air (usually lower accuracy). And what accuracy is even needed, cause laser scanned tracks on the ground with its high accuracy will get processed anyway to a polygon mesh with a lot lower count of polygons. Otherwise not computable with todays personal computers.

    (Though with good algorithms that reduce the polygon count, a high accuracy scan still can be very important. Think of a crack on tarmac surface. Even a low count polygon mesh can be optimized to model that crack exactly. Not possible if the scanned data source doesnt even include that crack due to low accuracy. Sebring for example is a track full of cracks. Not saying S397 algorithm is that good to model all major cracks at Sebring. Maybe it does, or maybe later)
     
    Last edited: Sep 9, 2020
    felirrari likes this.
  6. TTM

    TTM Registered

    Joined:
    Dec 20, 2012
    Messages:
    32
    Likes Received:
    14
    Discussion about laser scanned tracks in any game is gone ridiculous. Why we need 1mm accuracy? Many gamers not even notice if accuracy is between 1-5cm.

    I have made some drone ”scanned” rallycross tracks for AMS with 1-5cm accuracy and many of the players says that track is as it is in a real life. More important thing is that track and it’s surroundings looks real.

    Luckily we have today technolygy and tools, so we can make quite accurate tracks to games than few years ago...
     
    memoNo1 likes this.
  7. memoNo1

    memoNo1 Registered

    Joined:
    Apr 13, 2019
    Messages:
    960
    Likes Received:
    1,155
    Some projects planned for rf2 as well?
     
  8. TTM

    TTM Registered

    Joined:
    Dec 20, 2012
    Messages:
    32
    Likes Received:
    14
    I have plans to move rF2, but first i have to learn to make tracks to game and new shaders stuff. Also many things will depend on what S397 will do with rF2.

    Sorry about off topic..
     
  9. mantasisg

    mantasisg Registered

    Joined:
    Aug 17, 2016
    Messages:
    1,666
    Likes Received:
    2,399
    OK just joking. But actually not joking :D If for example we'd take Nordschleife with 100% accuracy laser lidar nano-accuracy scanned, but then increase it's scale by 0.0024% its overall length would increase by 50 meters, and I guarantee no one would notice without measuring it.

    In my personal opinion there are critical parts of tracks such as apexes, cambers and humps that are super important to get right, I suppose in relative accuracy (relatively to objects in ~30-50meters radius) it should be possible to eyeball stuff within 20-40cm accuracy even by just referencing to aearial pictures such as like those from Google Earth and pictures + videos.

    If you have used some data that is 1-5cm accurate, I guess it is fair to say that your end result must have been more accurate than that. If I understand LiDAR, the error that is specified for the pointcloud must be meant for individual points. And the more dense the cloud is going to be in given area, the better errors will cancel each other out, and after smoothing the actual error for surface should be very low.

    Another aspect is the accuracy of features. I'd suppose, that will depend on the execution, the way point cloud is going to be used. I think it is fair to say that Nordschleife of Kunos and S397 are completely same, and completely different at the same time. All the differences are in the features of road, Kunos Nords seems smoother, has less pronounced undulations. The Nords of S397 is a lot more rich in road features. For example these two are very aparent in S397 version:

    The "double top" of Shwedenkreuz, the bump right after the top of the hill, it is well pronounced in rF2. You can see it very well in this video at 0:38


    The dip at 1:19 in this video never felt right in AC to me, but it could also be related to aero, front height sensitivity and similar stuff that would take part in unloadign the car or even lifting it up from the tarmac at undulation like this. This also tells that car physics working together with track model has an impact of perception of a track. Laserscanned car physics when ?
     
    Mitch9 likes this.
  10. kashif

    kashif Registered

    Joined:
    Dec 23, 2017
    Messages:
    42
    Likes Received:
    20
    If you can't appreciate the difference in laser scan and non-laser scan, don't deprive those who can of the luxury.

    1mm difference can make the difference of a bump being there or not being there. Very small nuances in the road get picked up in laser scan that make a very notable difference. Especially in racing, very small things make big differences. There's a reason iRacing only exclusively uses laser scan, there is a clear quality difference between laser scan and non-laser scan. It's the reason the Nords laser scan and Le Mans laser scan was a big deal for S397.

    Don't downplay the difference.

    Having said that, I can still appreciate non-laser scanned tracks within a certain level of accuracy.
     
    vittorio likes this.
  11. TTM

    TTM Registered

    Joined:
    Dec 20, 2012
    Messages:
    32
    Likes Received:
    14
    Of course there is hardcore racers who want maximum accuracy to tracks. There is still huge amount of players, who plays racing games only for fun or they want to participate to online races to race with friends and they not care if track is not 100% accurate... But we all have our opinion, but most important thing is that we enjoy our time with racing games.
     
  12. mantasisg

    mantasisg Registered

    Joined:
    Aug 17, 2016
    Messages:
    1,666
    Likes Received:
    2,399
    1mm might make slight physical difference depending on the speed of vehicle and what the 1mm relates to. If it is one milimeter height difference to five meters of distance, then perhaps nothing changes for average racing speeds and average chassis + tires, but if it is one milimeter of height to three hundred milimeters of distance then it is probably little bit of difference. At low speed larger legth undulations wouldn't even feel as bumps, just as incline, at higher higher speed it could feel more as a bump. That also depends on how fast the suspension is, suspension travel, tires compliance... So it is not only geometry of the road surface that makes up for a bump that ends up perceived by a driver of a vehicle. 1mm might make a difference, but no one will know if it is one mm too little or too much or just exact. Range of 1-5cm differences sounds much more realstic to be noticed. But it depends on things.

    I would also speculate that all sims physics would probably struggle to read physics to the accuracy of one milimeter. High refresh rates means that physics are updated in small distances travelled. But I suppose it might not be end of a story. Systems might have different capabilities to be totaly accurate in the first place. For example I know that AC is pushed to the limits with Nordschleife, because it is very big map for their 32bit system (if I recall it correctly, please don't judge me for not being dead certain on things), this is right on the verge of starting to cause inaccuracies for physics engine, some stuff can be perceived as shaking slightly when stationary, cars that migth be sharper might appear slightly more difficult than somewhere else in my experience.
     
  13. vittorio

    vittorio Registered

    Joined:
    Jan 11, 2012
    Messages:
    1,029
    Likes Received:
    464
    Come on, its not about 1mm accuracy at todays computers and todays laser scanned tracks we get in any high quality sim. Maybe in 10 years simracing.

    The difference is:

    1. take non laser scanned tracks like free rF2 Nordschleife. Some corners (for example the last Nordschleife corner before T13) are wrong by 10 meter and I am only talking about layout. Not talking about tarmac detail, which is completely wrong anyway.

    2. What accuracy do we get from free LIDAR scanned areas? When I worked with LIDAR and satellite scanned data from USGS (though 20 years ago) the accuracy was 1 meter at best. So you got a pointcloud with 1 meter resolution. Not because higher res wasn't possible, but because of military and strategic decisions. I am sure today you get higher res scans. What res by the way, someone knows? (for example for Nordschleife as i used that track in this post?)

    3. What accuracy we get from paid laser scanned tracks? Probably in mm accuracy. Is that accuracy used/needed? Not sure. I took a look at AC Nordschleife polygon mesh. The smallest polygons are somewhere 5-10 cm in length with few mm in height. So looking at AC Nordschleife mesh, yes that high accuracy at least was used.

    Is that high accuracy needed for simracing? A more difficult question.
    Lets take a look at AC tire model with 400Hz (if i am correct) . When racing with 200 km/h you get new tire/physics calculations every 1.39 meter. So that resolution (5-10 cm polygons) probably is too high.
    Lets take a look at rF2 tire model with 2400Hz. When racing with 200 km/h you get new tire calculations every 23 cm. So that resolution probably still is too high, but you need some kind of oversampling too. this is only theoretically. but looking at rF2 polygon mesh is less dense than AC mesh, so theoretically it seems appropriate. But that is a difficult question anyway.


    Is that high accuracy needed for simracing? A more difficult question.
    Lets take a look at AC tire model with 400Hz (if i am correct) . When racing with 200 km/h you get new tire/physics calculations every 13.9 cm.
    Lets take a look at rF2 tire model with 2400Hz. When racing with 200 km/h you get new tire calculations every 2.3 cm.

    BTW, don't take this too serious, i just did some calculations.

    EDIT: as @OdeFinn correctly noted, i miscalculated -> corrected the post
     
    Last edited: Sep 19, 2020
    OdeFinn likes this.
  14. lagg

    lagg Registered

    Joined:
    Oct 1, 2012
    Messages:
    1,699
    Likes Received:
    1,078
    In fact you can test it with the conversions of these tracks to rF2. The % of use of the CPU is so big, that many people, that uses rF2 normally, has CPU problems in these tracks.
     
  15. Mitch9

    Mitch9 Registered

    Joined:
    Jul 26, 2020
    Messages:
    27
    Likes Received:
    27
    Pretty sure rf2 uses cm accurate scans, while pro simulators use mm resolution if they need to work on car physics/setup. Also, the AC tracks are dense, but could some of that cpu load be because of bad optimization on the conversion?
     
  16. vittorio

    vittorio Registered

    Joined:
    Jan 11, 2012
    Messages:
    1,029
    Likes Received:
    464
    I think the data base (scans) that rF2, AC ... and pro simulators use are the same (in case of Nordschleife again, bought from Nürburgring GmbH). The difference is how those data are used. rF2 uses personal computers for the sim (and even only using 4 CPU cores at best). rfPro uses multi-CPU systems, so a lot more optimized for parallel computing.

    More important is, while rF2 probably has IMO the most advanced simulation model for sim racing using personal computers for private users. rFpro is aimed at professional market with its simulation model, which means to provide interfaces to standard simulation tools like ModSim and others.
     
    Last edited: Sep 13, 2020
    Mitch9 likes this.
  17. stonec

    stonec Member

    Joined:
    Jun 19, 2012
    Messages:
    3,224
    Likes Received:
    1,342
    Someone mentioned mm, we aren't even close to that in consumer sims. Below is an example of AC Nordschleife mesh (I was able to open it since it wasn't encrypted). The new circuit section is a bit denser scanned, but most of the old circuit looks like this. Let's say the road width is at most 10 meters in this section and we take a slice horizontally, there are about 15 vertices in it, which would give a resolution of 1000/15 = 67 cm if my calculations make sense.

    Nords.PNG

    The tracks that are converted to rF2 don't even use this type of asymmetrical mesh, they use a bit lower resolution variant, probably for mentioned performance reasons. rF2 scanned original tracks are all encrypted, but I'd be very surprised if they had order of magnitude different accuracy than AC tracks.
     
  18. vittorio

    vittorio Registered

    Joined:
    Jan 11, 2012
    Messages:
    1,029
    Likes Received:
    464
    regarding AC mesh. It depends on what section you are looking and i found 5-10 cm polygons as smallest polygons. and i was looking at AC to rF2 conversion which used an asymmetrical mesh. i looked at it years ago when it got released.
     
    Last edited: Sep 13, 2020
  19. mantasisg

    mantasisg Registered

    Joined:
    Aug 17, 2016
    Messages:
    1,666
    Likes Received:
    2,399
    rF2 meshes has to be composed out of regular quads because they need to look neat for real road. Though if their resolution would be much higher, they would look neat with irregular mesh too.

    I have been wondering for a while if AC meshes use actual laserscan pointcloud meshes, or they just wrap around new meshes over pointclouds to follow their shapes and then modifying those meshes to use irregular trinagulated patterns to look more like laserscan mesh, and thus feel more irregular when driving. I think it is fair guess, mostly because they are Italians :D Thats fair technique, I would use it now, but it will inevitably loose some of accuracy depending on mesh resolution that is being wrapped up.

    And lol @ 1mm resolution meshes. That sounds a bit low to me, not real enough. Resolution of 0.01mm would be ok, of course with every pebble in 3D, normal maps will be ancient tech one day, tires will work at 100000HZ and will use 3D data of surface to work out mechanical keying if you know what it means. And we will be able to heat our houses purely by simracing one day, it is never going to be cold again in winter.
     
  20. vittorio

    vittorio Registered

    Joined:
    Jan 11, 2012
    Messages:
    1,029
    Likes Received:
    464
    @mantasisg ok, Im only answering the first serious section of your post, if even is. (and ignoring the other sections, whats the use of that clown sections in your post BTW? this is not discord).

    While it is true that all ISI and S397 tracks before laser scanned tracks used regular meshes. New rF2 laser scanned tracks by S397 use irregular meshes. As you could see on posted work in progress wireframe meshes by S397. You can even see that when looking at RealRoad on S397 laser scanned tracks.
     
    Last edited: Sep 13, 2020
Thread Status:
Not open for further replies.

Share This Page