Nordschleife - laserscanned?

[Simon Kaminski]

Hello, I have seen in the launcher files, that we get teh Nords.
(X:\Games\AC_TechPreview\launcher\themes\default\graphics\64\track-ring_icon.png)

Laserscanned?

Ps: I posted this in the AC Forums but it got closed. Trying my luck here now hehe.
http://www.assettocorsa.net/forum/index.php?threads/nordschleife-laserscanned.506/

Feel free to post presumptions

 

[3dnooblet]

Theres no doubt that laser scanning is the way to go. Whoever said its a waste of resources does not care for authenticity, in which case im perplexed as to why that poster is here and not on some GT5\Forza\trackmania\mario kart forum.

 

[BaldApe]

Theres no doubt that laser scanning is the way to go. Whoever said its a waste of resources does not care for authenticity, in which case im perplexed as to why that poster is here and not on some GT5\Forza\trackmania\mario kart forum.

Whoever understands that there are other differences between AC and mario kart than just laserscanning is not perplexed.

 

[Mr Whippy]

Exactly Gamer Muscle.

You can get good results with and without laser scanning.

All laser scanning does is make sure that artists don't make things up, they work off exact data, so there is no scope for them to make bad decisions or choices off incomplete data sets.

If you need a good track surface and you are paying people by the day, then laser scanning is the quickest and cheapest way to get it right, AND you also get the bonus of capturing a lot of the landscape around the track thrown in for free. Great... just what we need.

There are other routes but all considered they just cost more in the long run, despite being cheaper on kit to get the data, more man hours are spent getting it feeling and looking right.



Hampus Andersson, those images are interesting.

Note how all the AC images are low angles so you can't see all the areas the scanner isn't hitting behind objects. How the iRacing images are high up and leave no area un-scanned.

No they don't go out of the bounds of the track, but by definition of the scanning they cover any given point on the course from at least TWO points (that is how they overlap with static scanning)... thus any given object is likely to be nicely covered with points from both sides... useful when building from the data.


That tower has a curved profile from how it looks, so as the planar profile has travelled along it's scanned along it's curved face and given some points.
Had the building been square to the track and square sided, you'd not get any data for anything except the face facing the track.
You can see an example of that missing data on the pit tyre wall where the data off the bits facing us are partially missing.

However, in the same image (top one) the armco to the left seems to have data behind it. I wonder if they also take pano-scans at certain locations and combine the data... it certainly looks like it from the images.

And if that is the case, then it makes you wonder how the time scales are impacted.


I'd happily work from either data set, but personally I'd prefer to work off the fuller coverage that iRacing type scanning offers.
The AC type scanning is still perfectly fine, but I know I'd still have to refer back to photos/videos more often to just make checks on certain details than I would if I worked off static tower scans!


Just an observation that is interesting that is all.

But as Gamer Muscle said, in the end it's about the love and attention that the authors put in, not the 'data collection' method used.

AC tracks are probably just as good as iRacing ones... just the artists might spend a bit longer using reference materials but save £££ on scanning

Dave

 

[Hampus Andersson]

That tower has a curved profile from how it looks, so as the planar profile has travelled along it's scanned along it's curved face and given some points.
Had the building been square to the track and square sided, you'd not get any data for anything except the face facing the track.

But as long as you scan part of it, that´s enough.
Photos will tell you the visual aspect of it.

At minimum if the tower was a building that was a square it would have covered at least two sides of it. Well enough to understand where the building sits on the track.

You can get both laser and non laser scanned tracks that are fantastic

That´s subjective.


Can you get a non-laserscanned track as accurate as a laserscanned one? Not even if you spent 300 years with various technology or one billion photographs.

 

[GamerMuscle]

That´s subjective.


Can you get a non-laserscanned track as accurate as a laserscanned one? Not even if you spent 300 years with various technology or one billion photographs.

Obviously properly done laser scan track will line up with a real world track better than a non laser scan track , I don't think anyone has ever made an argument against that.

In terms of experiences derived from a track , baring in mind most sim racers wont ever race the real track , Its possible that a top level non laser track can produce as meaningful or more than good enough driving experiences as what would be gotten from a laser track.

But that only happens if the track artist is ridiculously good and has the time to dedicate to the track , Which most wont especially in a traditional development environment.

So In the end if you can give a moderate to good track artist laser data they will probably if not always produce something better than what a pro artist could produce without laser data.

Id just be careful not to outright dismiss a track simply because it was not laser scanned , but its certainly the case that for outright quality and consistency laser scanning makes the most sense and is the future for "serous" driving simulators.

On a separate point , outside of overall layout and corner accuracy. One thing that I think could be a nice compromise for fictional tracks or tracks which are real but teams don't have permission / rights to scan is to use laser data from other tracks / road surfaces just to get the nature of the macro road texture feeling believable.

For me one of the biggest benefits of laser scanning outside of track accuracy is just getting a believable texture to the small bumps and frequency at which you tend to get divots or points that unsettle the car.

 

[Mr Whippy]

That is my problem, the dismissal of non-laser scanned tracks.

They can still be very good, very accurate, and give a fantastic level of immersion for both casual gamer and professional driver alike!


Just because a lot of games have crap tracks in them and laser scanned ones are always good, doesn't mean that non-laser scanned tracks can't be excellent.


As laser scanning gets cheaper (although does it get more expensive in the hands of copyright owners who can retail rights to all the game makers these days?!), the ability to just use it for most of your work becomes possible if you have a big budget for a main release title.





Hampus, how would a square building have two sides scanned if the mobile scanner is perpendicular to the building?
The scanner can't look forward or backwards, it just looks to the sides in a 2d plane.

OK if there is the slightest deviation in angles then you'll get some points, and I'm being a bit pedantic on the point, but you get my drift I hope.

Scanning from static locations with overlap does give an artist a whole load more confidence in modelling from.

I'd argue that from the images AC have shown of their scans that they also combine some panoramic static scans into their mobile scan data... or they do a few passes or drive different routes to get overlap or coverage in different places.

Suddenly I wonder if the times allowed for the mobile scanning process are not a bit too optimistic. Do they really just do one pass? Or many passes?

I'm sure it's still cheaper, but as much as we assume it is?

Dave

 

[Hampus Andersson]

Hampus, how would a square building have two sides scanned if the mobile scanner is perpendicular to the building?
The scanner can't look forward or backwards, it just looks to the sides in a 2d plane.

That doesn´t make much sense as it would never be able to capture the track if that were the case.

Car is driving in the middle of the road so to capture the sides and the track it just passed something like this is more probable.

If it were pointed left and right of the car then it would have to have a 180 degree field of view otherwise it would miss the most important part, namely the racetrack.

Suddenly I wonder if the times allowed for the mobile scanning process are not a bit too optimistic. Do they really just do one pass? Or many passes?

I´m positive they run once. Perhaps some places where it´s very wide, like main stretch and pitlane they run once in every lane.
Militaries or autonomous vehicles uses laserscanners to get an understanding of what´s around it for example.

I believe some military missiles also use a form of laserscanning.

 

[Hampus Andersson]

 

[Mr Whippy]

It's hard to tell on those videos how they scan.

From a picture I saw of a van that had been scanned (it had vertical stripes), it's clear the laser is making a helical path through space (corkscrew along direction of travel).
It also looks (from the last video) like the scanner is indeed scanning in a cone, as all the shadows are pointing forwards and away from the vehicle.

However, this still leaves gaps, ie, what about the back of buildings if you use the cone system? Or even any surface that is shadowed because it sits further back behind an object than the cone angle allows you to cover?

One of the scanning vehicles had two heads (one looking backwards?) and that seemed to fill in the empty areas better because it was in essence like having a scanning pair with one covering the blind spot of the other (90deg to it I assume)

But are they all double headed?

I assume a lot of time is spent processing too... suspension movements/bumps in the road will all impact the scan, and without other data to check against (remember the scan is a single long helix and each point is taken at a fraction different time to every other), how do they get it smooth?
Maybe all mobile scanners are double headed to check both ways and get a denser cloud and a second set of points to correlate against and reduce error?


All very interesting stuff... but all considered I do wonder how cheap GOOD mobile scanning is for wide-area coverage (ie, not just the surface which you'd drive on in a game)

There must be a reason iRacing went with the wider coverage and fixed scans?!


I'm genuinely not fussed, just interested in why one developer would choose one type and another a different type.

So far it seems iRacing just wasted £££ and time on data they didn't need? Or might it be reasonable to say that the mobile scanning isn't any cheaper in the long run. More expensive gear and data management (getting it accurate) processing time, offsets the time for a man to walk around with a fixed scanner?!


Hmmm

Dave

 

[Hampus Andersson]

There must be a reason iRacing went with the wider coverage and fixed scans?!

Because it´s the most accurate way of doing things.

What AC does is exactly the same thing as Formula One teams.
If it´s good for Formula One simulators and their drivers it´s good for a simracing game.

 

[Hannes-Joshua Thor Barthel]

you are all just talking about fixed scanners? what about a rotating one? would be the thing I would put on the roof of a drivable scanning vehicle..

 

[Chris James]

I say if AC has the ability to laserscan without breaking the bank or taking too long, then by all means, I would prefer it. But if they don’t have the means to laserscan the Ring, I will definitely take a conventional version. I just hope they put realistic bumps that actually provide a challenge. I’ve driven too many conventional versions of tracks that are too flat and plain, and I’m not just talking about elevation changes, but track imperfections and oddities that provide challenge. Look at this video. See how the driver has to constantly fight the wheel over the bumps? I want that in my sims more than almost anything else. AC seems to have the physics and FFB capabilities to provide this sensation more powerfully than any sim before. Here’s hoping.

 

[Mr Whippy]

Because it´s the most accurate way of doing things.

What AC does is exactly the same thing as Formula One teams.
If it´s good for Formula One simulators and their drivers it´s good for a simracing game.

iRacing do static because it's more accurate, but AC do the same as F1 teams so that is good enough?

But that doesn't really answer my question. Why are iRacing doing static scans if the mobile are faster, cheaper and just as good?


As per F1 teams, all they care about is the surface topography though? When game developers scan they are looking for as much data as they can get while they are there I would guess, to maximise the value of the scanning process?


So are we saying fixed scanning is better overall, but mobile scanning is just as good for the surface scanning of the track itself?


Dave

 

[Hampus Andersson]

Edited: i seem to just repeat myself so i´ll advise you to re-read my posts instead.

 

[Mr Whippy]

So iRacing went with fixed scans because it's the most accurate way of doing things, as per your post.

But AC do what F1 teams do, which is 'good enough'

But F1 teams only want the surfaces... not all the stuff around the surface?!

F1 teams are not making 3D models of tracks for game/realtime graphics engines... maybe someone else does on their behalf...

So are those making the tracks on F1 teams behalf thinking mobile scanning is good enough for both surface details AND track visuals production?

Still wonder if that is the case, why iRacing went with that technique then. It just makes no sense and having re-read your posts you don't explain why.


Thanks

Dave

 

[Hampus Andersson]

But F1 teams only want the surfaces... not all the stuff around the surface?!

Where did i state this?
I think that´s more your thinking that mobile scanners can´t pick up anything other then the track.

 

[Marcel Pfister]

I think F1 teams want the stuff around the surface too for finding the braking point and something like that.

 

[Sebastian Wolff]

I myself am really NOT interested what iracing does...

 

[Hampus Andersson]

 

[Mr Whippy]

Where did i state this?
I think that´s more your thinking that mobile scanners can´t pick up anything other then the track.

Hi Hampus,

They were questions, hence the question mark. Sorry if my reply was a bit confusing.

My main thinking is why then did iRacing invest more time and money on needless scanning if F1 teams, and others, already know that mobile scanning is cheaper, just as good, faster etc etc?

I'm merely asking people who appear to have some knowledge of scanning etc from their responses so far, why that might be? I'm interested in the choices that teams might make in this regard.




Marcel, yeah I gather that is the case. Mobile scanning does seem to be good enough out to the barriers/fences, all the areas drivers look a great deal of the time I would assume.

But then beyond that the mobile scanning technique would become less capable vs the static scan techniques. Simple geometry and basic understanding of the kit enables you to make that observation... even looking at some of the scan point cloud images and you can see gaps in the mobile scan data that seem to have good coverage from static ones.

That doesn't mean you can't make a good track from it, it's still perfectly good data to build from, it's just less good vs statics I think. I'd have to start referencing photos or videos at times to fill in some of those gaps.



And that was the whole point of my enquiry. Is that where the difference in methods of scanning comes in between iRacing and AC? Did their graphics TD make a preference of one type of data or the other? Was it just a coin-toss choice?

Ultimate budget was higher for iRacing and lower for AC? End user expectations?

Hmmm

Dave