Kart Track - Lidar/Could point data vs Laser scan

[morvio100]

A general enquiry ... ..how accurate do you think the Lidar method would recreate a kart race track...as compared to a laser scan of a kart track...would the Linkster method be able to pick up surface detail close to/or to closely rival a laser scan

[Colin Jones]

No idea. But in the current Linkster-DEM course I am working on, I have needed to level out a few inground pools in rear yards of adjacent houses.

[Jimbo]

If it can do the Indy, you could do the Brickyard.

[morvio100]

No idea. But in the current Linkster-DEM course I am working on, I have needed to level out a few inground pools in rear yards of adjacent houses.

hahaha

[linkster]

I've been working on combining DEM's and Lidar point clouds and QGIS and the good ol' APCD for about 1.5 years now and I have learned a few things along the way.

Lidar based DEM's are great but they are not perfect, especially when trees, bushes, and grasses get involved. The Lidar point clouds contain a whole mixture of points at different elevations and someone, or something, has to sort out which is a ground point, which is a vegetation point, or if a point is an outlier or noise. Some surveys are sorted really well and some are not. Lidar scans work well for golf courses because the fairways and greens are generally not over topped by vegetation. When creating the bare earth DEM the areas without ground points get surface interpolation based on the surrounding points of the "hole". 1m raster DEM's are created by taking all the ground lidar points that fit inside that 1m x 1m square and averaging the elevation.

The "averaging" of the points to create the DEM "smooths" out the surface model. For a flat green this may be a good thing, for a steep bunker edge the averaging of points likely is a bad thing. I have noticed about 10 cm / 4 inches in variation in lidar points just within a 1m DEM point on a green. Most surveys are better but 10 cm / 4 inches of error in lidar points from an aerial survey is possible.

Then you get into the way that APCD renders the terrain between the vertices. Elevations I import into the APCD are nearly exact but APCD renders the surface as "best fit curves" or maybe splines that take into account the adjacent verts. This leads to a surface that is further "smoothed" more than it is in real life. More verts dramatically helps the APCD capture the terrain but we are limited to 527k verts in a course file. So then the consideration is how much area of a golf course can be covered to keep under 527k verts.

I started doing almost all 2m vert grids for the APCD, then I would delete areas of the course to get under the 527k vert limit. Lately I have been doing 3m vert grids as my default import and sometimes 5m grids when the course terrain looks smooth to start with. My most detailed DEM import is of St Andrews Old Course at 1.5m vert spacing but the amount of work in the APCD to get the course ready is excruciatingly slow! It is a balance of vert count, course area, and amount of work to be done in the APCD.

That all being said about the limitations of the APCD, I think the DEM imports with geo-referenced Google Earth images can produce accurate results that exceed the newest golf sims. I have watched gamers play on "lidar' courses in PGA 2k23 and the new EA sports game and I think the APCD terrain is better than these courses provide. I have no idea on what the vert count they work with for the course terrain but I think the newer games put most of the verts into moving trees and 3d objects and not on re-creating the golf course terrain. I watched someone play Pinehurst No 2 on PGA 2k23 and I was disappointed on the accuracy of the green contours based on what I have completed of the course in the APCD so far.

So, back to the original question, the APCD version of a kart track would have it's limitations. It would likely be off by about 15 cm / 6 inches at most with a high vert count. Yes a laser scan / survey would be more accurate in newer rendering software, but how much accuracy do you need and would it matter that much? All this stuff is just a sim of the real life version.

[morvio100]

Cheers for that Linkster....plentiful info there...not just in relation to a track surface accuracy...but for all Links designers....I thought maybe QGis may have had some color coding facility wehereby you could isolate the track surface itself...so as to FOCUS and finetune the accuracy of said isolation...thus... just basically EXTRACT the particular area....and then possibly convert the DEM for any 3 d software app for vert manipulation....anyways great information...good luck with St Andrews you have a few competitors now regards your Lidar system....re ..the new releases like Bradens The National and Croocked,,,by Wells...(thats the kicker)..lol...Agusta of course is a fine example of were the sim is heading...kudos