Laser scanned or built with different reference materials can BOTH offer exceptional quality.
The problem is that by the time you invest in techniques to get very accurate data you approach the cost of laser data collection so you may as well just get point cloud data and build/surface from that.
There is a good chance that because of the way ownership rights and all that jazz works, that data is owned by many tracks these days, or at least is licensed by scanning businesses/providers to end users under agreement with the track owners.
So basically, the AC team probably just pay for data then build with it.
Having laser data doesn't guarantee anything though as the big step is clouds > surface for what you drive on and ultimately that is a statistical process. Even iRacing will 'lose' some surface detail beyond their track mesh density size.
If you compare a really well made track that didn't use scan data with one that did, they are both often exceptionally good. Half the issue is that many games ran single cross-section tracks so adding bumps was impossible.
I'll agree with Hampus though, getting top-down is the easy bit. Getting vertical data is the really really tough bit.
I have logs from good GPS and in top-down the averaged plots match up with my land-line vector data, satellite imagery, and are within a few percent of measured data.
BUT, vertical accuracy isn't so hot... 10cm accuracy with 2m CEP from what I remember on my device, so polling every 50cm means my vertical trace through space wasn't so good. Averaging it out removed all the details.
However, walk along with the same device and get 3cm density, or about 30 points per 1m, and suddenly you'll be getting much nearer that 10cm on vertical accuracy.
With a decent GPS base station and logging kit on a smooth race course I think you could easily see 2cm vertical accuracy. You could walk around each side of the course and down the centreline and have a pretty accurate trio of profiles to build a very very accurate track from!
But as said, when you have laser you get the kerbs, run-off areas, armco, fences, grandstands, etc etc, all in with the same data set. That is useful for the artists as well as the people making an accurate driving surface, so I guess that is why they go laser data.
Not sure what my point is, but laser scanned doesn't guarantee anything vs non laser scanned...
Obviously it tends to mean artists get things in the right places, and looking the right shape/size, and the track surface comes with enough detail... but that can ALL be done without laser scanning. It just takes time and generally we don't see many 'games' invest that much in their tracks.
I could build any track extremely well without laser data, it'd just take an absolute age to get data sets good enough to match laser quality.
I'm looking forward to the next generation of xbox kinect type scanning tied in with GPS and photogrammetry type point clouds from motion!
Imo in 5yrs for these types of jobs (ie, non real life super-accurate construction types), we'll be able to scan a track and surroundings for not much at all vs current laser scan systems!
Dave
