Tracks Any dedicated people interested in collaborating on a track together?

[Emery]

Part of the trick with Brainerd is to realize it also was called Donnybrooke back when it was built. I have not built up much of a photo library because it's not bubbled to the top.

However... here's a 2016 TransAm race:

 

[Emery]

Some good detail shots in this 15 minute home movie:

 

[Emery]

And excellent use of drone to provide us with some good views:

 

[Emery]

Looks good to me. Nice long sweepers to get your attention in the old cars!

From back in the day...

 

[BluePotato00]

A couple issues with the LIDAR:

1. The imagery was taken before they added the new layout, so we'd only have LIDAR data for the 3.1mi "Donnybrooke" circuit. We'd have to superimpose some more current imagery (Google Earth) and add the remaining stretches for the 2.5mi "Competition" circuit without LIDAR.
2. The resolution (is that the word when used with LIDAR data?) may be too low. Here is turn 3 zoomed in:

 

[elloLeo Kinnunen]

From back in the day...
View attachment 334199
View attachment 334201

Great details and what a joker! "If you go off at 10 you'll have plenty of room for sliding (and meditation) without hitting anything"

A couple issues with the LIDAR:

1. The imagery was taken before they added the new layout, so we'd only have LIDAR data for the 3.1mi "Donnybrooke" circuit. We'd have to superimpose some more current imagery (Google Earth) and add the remaining stretches for the 2.5mi "Competition" circuit without LIDAR.
2. The resolution (is that the word when used with LIDAR data?) may be too low. Here is turn 3 zoomed in:

I don't think that will be a problem, particularly if we do the Donnybrooke layout with a 60s flavour - which allows a bit more 'artistic interpretation'. The intensity image shown is only really used to line everything up... so once you get the lidar elevation data in blender as a mesh it will look very different and we should be able to see the all important track features. As Gary said earlier elevations and corner cambers are what is crucial to get from lidar... Even if they are a bit low resolution. Then we insert the smaller details ourselves.

Getting excited!

 

[Emery]

1. The imagery was taken before they added the new layout, so we'd only have LIDAR data for the 3.1mi "Donnybrooke" circuit. We'd have to superimpose some more current imagery (Google Earth) and add the remaining stretches for the 2.5mi "Competition" circuit without LIDAR.

Hah! Usually, as a party interested in historic tracks, it's the other way around and it's necessary to interpolate bulldozed ground and/or scrape off unwanted bits. Quick check of Google Earth historic images shows they added the circuit after May 2006, but before May 2008. Camping around the lake & widening of T9-T10 was added between 1991 and 2003.

There are several distinct periods in the track's history. 1968, when it opened, it had the painted & buried tires inside the corners in addition to some concrete rumble strips just inside the tires. The tires were removed pretty soon (like in a year or less) and it wasn't until you get to about 1987 before more significant changes start happening. In 1987, a new tower was built. The new circuit around 2007-8. T1-T8 repaved in 2013.

Two videos help show the early changes. In the first video, Donnybrooke begins at 2:20 and there are in-car shots beginning at 5:34 from a Porsche parade lap. The second video shows the new tower still under construction and you can see the rumble strips are still on the inside of the corners. Incidentally, Mid-Ohio and Road America also used rumble strips like those.

So you can pick & choose which era to model. Without having the shorter circuit in lidar, I'd propose a pre-2007 version.

 

[Emery]

Hmm, as far as image quality prior to 2008, it looks like the 1973 image has the best quality. Advantage of 2006 image is that it is in color.

 

[BluePotato00]

Okay! I have the LIDAR data exported to an FBX and in Blender. I'm at the 11 minute mark in @LilSki 's first video at this point. One concern is that my object is showing as dimensions of 16.7m x 17.7m, which works just fine if we're going to race hot wheels (or RC cars!) on the track. Why are my dimensions so small? I'm afraid to just apply an arbitrary scale to scale it up that I won't be accurate.

Also, who's ready to be added to the Google Drive shared folder to collaborate? It sounds like @Emery and @elloLeo Kinnunen are interested. Can you guys send me a PM with your Google ID so I can add you? Thanks.

Next steps (I think) are to (1) fix the scaling, (2) get the background reference image on a plane, (3) correct any issues in the track - although I don't see any glaring peaks on the track like on the oval in @LilSki 's video (4) lay down splines and for track boundaries, (5) build a high-poly mesh for the track, and (6) "shrink wrap" that track to the FBX.

 

[LilSki]

Import the raw cloud in ply format and that should be the correct scale. Then if you need to scale the mesh to the cloud size.

 

[Ryno917]

Is it save to assume that it's just 1% the scale it should be in this case? 1.67km across sounds reasonable for that plot to me. I mean, I'd assume the scale issue was just one of units; mesh exported in km but Blender imported it as meters or something.

 

[BluePotato00]

Import the raw cloud in ply format and that should be the correct scale. Then if you need to scale the mesh to the cloud size.

That worked!

Is it save to assume that it's just 1% the scale it should be in this case? 1.67km across sounds reasonable for that plot to me. I mean, I'd assume the scale issue was just one of units; mesh exported in km but Blender imported it as meters or something.

That worked too! Correct, my .fbx when imported came in at a scale of 0.01 but the .ply came in at a scale of 1.000. Both are imported now in the .blend file and perfectly matched. My dimensions are below:

 

[BluePotato00]

Blender scene now has 4 aligned objects:

• Mesh from CloudCompare (imported from .fbx)
• Point cloud from CloudCompare (imported from .ply)
• Screenshot from CloudCompare to use as reference (I used the Import-Image as Plane built-in plugin)
• Screenshot from Google Maps to see the second track layout when needed (using the same plugin as above)

Unfortunately, the .blend file is 798MB. I assume (hope?) once I get my track and terrain laid down, I will be able to remove the CloudCompare mesh and point cloud for a massive file size reduction.

 

[BluePotato00]

And I've locked the location, rotation, and scale of these 4 objects so I don't clumsily alter anything!

 

[Emery]

(3) correct any issues in the track - although I don't see any glaring peaks on the track like on the oval in LilSki's video

There's the area under the bridge to fix because aerial lidar can't see under overpasses. In retrospect, it's probably best to remove bridges that go over the track while in cloud compare and bring them into Blender as a separate mesh layer?

 

[BluePotato00]

There's the area under the bridge to fix because aerial lidar can't see under overpasses. In retrospect, it's probably best to remove bridges that go over the track while in cloud compare and bring them into Blender as a separate mesh layer?

Right. I had noticed that too. Should be a fun area to model.

From the LIDAR mesh:

Here's the underpass from an onboard lap:

By the way, that sign on the bridge says:

 

[Eric Rowland]

the good old days....

 

[Emery]

Major automobile racing series and when they ran at Donnybrooke/Brainerd
1969 USAC two heat Indy car race
1969-72 TransAm
1970-72 CanAm
1971-72 IMSA
1975-78 TransAm
1977-82 IMSA
1979-80 CanAm
1980-89 TransAm
1984 CanAm
2010-17 TransAm

 

[BluePotato00]

Bezier curves laid out for outside and inside edge of track. The curves are completely flat and not yet shrink wrapped to the contour of the LIDAR data.

By the way, I read that straightaway is the longest on a race track in North America. It doubles as a drag strip. You can see in the curves that this stretch of track is wider than the rest..

It's difficult to determine where the track surface starts and ends since I based their placement on the LIDAR intensity map screenshot, which has a poor resolution, even when I increased the zoom level on the screenshot in CloudCompare. Here's what an example corner (turn 3 on the southeast corner of the map) looks like zoomed in with my bezier curve on it:

I'm concerned to go further if I won't be able to re-edit these bezier handles after stitching a mesh between these two curves. Anybody else is welcome to take a look at it.

 

[LilSki]

Duplicate them and experiment.