All aviable Laser Scanned tracks @ rF2 ! ! !
- Thread starter memoNo1
- Start date
- Status
Lazza
Registered
How are you seeing the realroad mesh in the game? (I know of certain situations where the realroad mesh could become apparent, or has in the past, but I haven't looked into it recently)
I find discussions on laser accuracy (or point density) interesting - not just because there are differences and limitations in how the game engines use the data (rate, then the tyre model, and how that feeds back into the vehicle physics and FFB) but that people overstate how much difference numbers make.
It's like when people overclock their CPU from 4.0 to 4.2GHz and then rave about how much faster windows loads. Or plug pedals into a USB game controller instead of into a wheel and get 16 bit resolution instead of 10; 65536 steps instead of 1024!!! I can drive so much better!!!!!!
Meanwhile, the "just noticeable difference" (JND, though there are other equivalent terms) for weight - that is, the smallest weight change that can be detected 50% of the time, in large tests - bearing in mind weight is one of the more accurate when it comes to sensing changes (moreso than light or sound) is 2%. 5kg weight? You'll need a change of 100g to reliably feel it. I bet most people think they can feel a lot finer than that. In fact in this sort of thing there's always people who decide they're an aberration (reflex response time is another good one for that).
Apply that 2% margin to a real life driver going around a track and feeling "weight" on the wheel as they drive over bumps. How much could you raise a particular bump to stay within 2% and make that change invisible to the driver the very next lap? Now factor in the suspension compliance, slightly different inputs, slightly different line or car attitude at that point. I think you could make a substantial (in numbers) change without them being any the wiser. What about an hour later? A day? A week?
Now you take that same driver, put them on a sim with very different sensations (even if you have a full motion rig and VR or whatever provides the best immersion) - how much extra margin does that introduce?
Keep going down the line and you get to a guy sitting in his $3000 setup judging the accuracy of the track by how it feels, with some videos - maybe limited to only on- or off-board - as reference. I see a lot of scope for error there. But many will compare tracks and games (ha! More variables!) and give their expert opinion on which is more correct.
(On pedal resolution: I'd be interested to see some proper testing with someone applying pedal movement incrementally, by the smallest possible movements (without looking at their own feet, of course) and have someone watching the data and tallying up how many steps were needed to reach full travel - and how many movements weren't movements, or ended up going slightly backwards - and comparing to the resolution people strive for. Might make some people less happy about their purchases though)
vittorio
Registered
For me it was very certain that rF2 uses an irregular mesh for Nordschleife too even before release. But its already such a long time that i can't remember exactly anymore why I was so certain. I've seen those Nords "wireframe screenshots" of regular meshes of course too, shown in rF2 Roadmaps long before release. IIRC I had a discussion with Marcel at discord and he said rF2 uses irregular meshes too (maybe search at discord for "Marrs and irregular"). At actual Nordschleife rubbered RealRoad polygons are way too dark, looking at the dark polygon mesh for me it was clear too it is an irregular mesh ( too dark RealRoad is corrected at coming Nords looking at screenshots of latest Roadmap). But as I said, I can be wrong too (but I don't think so)
EDIT: BTW, looking at that regular mesh screenshot (showing the corner). I count 64 quads on track width. Lets say the track width is 8 meter. That would mean a regular mesh resolution width of 12.5 cm (and yes, max. few mm in height). Polygon count not optimized via irregual mesh. I just think that is unlikely, thats a lot of polygons (even if you assume 32 or 16 quads at width regular mesh for straigths).
EDIT2: and as AC to rF2 Nords conversion showed, irregular meshes worked without a problem, even no code change needed. Why shouldn't S397 use that optimization pass, which even looks better on RealRoad?
Last edited:
vittorio
Registered
Thats why I added to my latest post talking about numbers and calculating:
"BTW, don't take this too serious, i just did some calculations".
But the differences between NON laser scanned, aerial LIDAR/LASER scanned and ground LASER scanned should be clear right now.
Last edited:
Mitch9
Registered
@Lazza Laser scan is to simracing what Hi-Fi to audio; sooner or later its more perception placebo than performance (even when there are measurable diferences between the content and equipment being used)
@vittorio "and as AC to rF2 Nords conversion showed, irregular meshes worked without a problem, even no code change needed. Why shouldn't S397 use that optimization pass, which even looks better on RealRoad?"
Well, according to svictor here (https://forum.studio-397.com/index....ow-they-interact-with-real-road-effect.66953/):
"3. Physical mesh not created with rectangle shape, such as some tracks from AC, use large amount irregular mesh shape and varied vertices density. These can create visual artifacts for real road groove or marble, such as black patches here and there, or distorted groove texture."
Never noticed those, because I haven´t even used those tracks long enough, but maybe that´s why you can see the mesh through the realroad? Here´s a screenshot of nordschleife´s mesh from discord:
...manual quotes because the regular ones are broken for me
@vittorio "and as AC to rF2 Nords conversion showed, irregular meshes worked without a problem, even no code change needed. Why shouldn't S397 use that optimization pass, which even looks better on RealRoad?"
Well, according to svictor here (https://forum.studio-397.com/index....ow-they-interact-with-real-road-effect.66953/):
"3. Physical mesh not created with rectangle shape, such as some tracks from AC, use large amount irregular mesh shape and varied vertices density. These can create visual artifacts for real road groove or marble, such as black patches here and there, or distorted groove texture."
Never noticed those, because I haven´t even used those tracks long enough, but maybe that´s why you can see the mesh through the realroad? Here´s a screenshot of nordschleife´s mesh from discord:
...manual quotes because the regular ones are broken for me
vittorio
Registered
Thats a really good comparison. If I have the choice to get my favorite song in 32, 64, 128 or 192 kbit/s MP3. I will take the 192 kbit/s MP3, even if there is not much of hearable difference between 128 and 192 kbit/s.
And another (or same) "psychological" effect (me included): If i tell my racing interested friends coming to my flat and race in rF2. They only drive Nordschleife all the time cause they drive it in real every year too (and they drive it in rF2 for preparation to the real track) . If I tell them the real track was laser scanned. The raw data has mm accuracy. It is optimized to be computable for personal computers though, but the layout for sure is correct as every slope and camber. Even tarmac is detailed too using real data from 2018 laser scan -> they say WOW thats awesome. Let's go.
If I tell them an artist using track layout drawings and all kind of free information available, adding fake tarmac details to feel a bit more realistic -> they say MEH. maybe we should drive it in another sim? (they already know AC and iRacing)
Last edited:
OdeFinn
Registered
What I'm counting wrong, I get 2.3cm.
Lazza
Registered
No, you've got it right. Goes to show doesn't it? I read what vittorio said and didn't question it lol
2ndLastJedi
Registered
I was supposed to be in bed by midnight 
- Status