"Project Dot" the laser scan project

Hi Marc

re.equipment

flypt's thread caught my eye re.scan possibilities,
http://isiforums.net/f/showthread.php/17327-Cheap-alternative-to-laser-scanning
I may be able to improve the technique slightly, although I know it will not have a high detail resolution for small track surface bumps I'm hoping it could be good enough for track surface cambers etc as well as correct elevation & layout.

re.track

each winter I usually acquire a car to use for 1 or 2 track days
(the car is usually someone else's cast off)

I have a car in storage and hope to revisit Donnington Park Circuit ( just 30mins from my house ) to destroy it perhaps during late summer
but do a few slow laps first to collate data & take a stack of photos too

although I intend to test/familiarise myself first with roads near my house.

very interested in how the track surface (rf2 track mesh) is laminated/sealed to the point data cloud
or maybe it's a case (not so good option) of doing it manually based on visual info of the point cloud

I'm not really concerned right now about trees and stuff like that, it's the track surface and how accurate a track can be using methods like this that really my motivation for this
(as I'm always moaning about track surfaces in rf2, I'm kind of spoilt by Iracing when it comes to tracks)

also awaiting answers from bobs track builder -the laser version concerning his software and the equipment he used for
Eastern creek ( an rfactor 1 "laser scanned" track )

thanks for your interest, it's early days yet but hopefully project dot will become an interesting project, if I can achieve an modelled track surface to a decent quality that's working within rf2 , I'm more than happy for others to add surrounding visual stuff I'll have lots of photos and I have some colour swatches so things can match up etc

there's a 10k road running race that I'll be doing there too so I'll be running around the track soon, it's a great track and only a quality version should be in rfactor 2
(my garmin gps watch provides elevation data but I'm hoping a point cloud will be far high resolution )

p.s mallory park is very close to me also, another good track
if I can get this scanner thing working and optimise it as good as it can be
then I don't think there's a problem getting the point clouds as these are not for commercial use etc, these places are happy for people to take video / photos etc
all I'll be doing is capturing point clouds as oppose to flat images
 
Last edited by a moderator:
hi all,

some minor progression here, firstly I'm starting to consider using a static professional grade scanner
I'm going to work out how long it would take to scan a track the size of Donington using this method based on point scans 25 to 40 meters apart & 15 minutes per point scan.
There will be two of us, one on the scanner and one moving the reference spheres/assistant etc.

Due to this type of hardware producing the point cloud takes much longer i would obviously need exclusive access to the circuit so Ive been in communication with Donington,
Ive made it clear this is not a commercial profit making project,
they have expressed some concern that a license could be needed to be in a commercial piece of software,
ISI seem disconnected from 3rd party mods or users free share personal projects I'm hope I'm correct in informing them that there is no issue.

the scans can be done through the night if need be providing conditions are dry,
photography & colour matches can be done during the latest morning hours possible.

The scanner I have access to is a Faro 3D S1 20

I'm not familiar with Faro Scene software
( the software used for stitching the scans into one point cloud )
but once the scans are complete theres plenty of time to play with the raw files & software after,
.... anyone familiar with Faro 3d ? ,

once the point cloud is complete it can then be exported into the build software
perhaps 3d max.

I'm fairly confident using the scan hardware/reference spheres etc, the Faro 3D software dosn't scare me at all, the thing I'm not liking is sealing/encapsulating the track surface mesh to that of the point cloud
(the whole target here is achieving a quality laser scanned track surface ), iracing has software that does this for them apparently ( see their cota track project vid )

I think there may be a script/plugin for 3d max for this? does anyone have info of how accurately it seals to the point cloud?
this is such a key thing otherwise a high quality laser scan would be wasted.

p.s if Donington dosn't approve of this project & allow access
( Ive explained the promotional advantage etc )
then I'm considering backups (different tracks) that are fairly local to me.

My objective here is to get this project to a level of a "tech track" eg quality laser scanned road surface with barriers and track edges as a first stage ( & release ),

Im also assuming that the rfactor 2 software can handle high quality laser scanned tracks just as good as other titles. (anyone know any different ? )

the trees/buildings etc could be done in a second stage & I may ask/need a joint community collaboration at this point
 
Last edited by a moderator:
I don't know if it would be of any use to you, but I have been looking into this....

http://www.interstudio.net/highroade.html

maybe get the road layout done and then map the scanning images on top, then build the objects and buildings and stuff.


cheers adrian

I'm not sure if this has the "mesh to point cloud" facility that I'm most interested in, but I'll check it out thanks.

I will not be actually designing anything, I'm simply laying down textures within a point cloud.

I'm geting hold of a random sample point cloud
(scanned by the actual model of scanner I'm considering using)
and test with software to apply a mesh and visually look close up at the detail etc
 
Last edited by a moderator:
I like
dots.jpg
 
Well, I did see the original post and was very interested and excited at your plans. I didn't have time to comment at the time, because I was at work. But I am super pleased to see my most local track, as well as being one of my favourite tracks, coming to rf2! All the best with this project!

Cheers
Justin

Sent from my GT-I9305 using Tapatalk
 
How are you planning to get track access for a static scanner? Surely Donington isn't going to let anyone just hang around the track, taking point cloud data for free. Most circuits licence that data, don't they?
 
How are you planning to get track access for a static scanner? Surely Donington isn't going to let anyone just hang around the track, taking point cloud data for free. Most circuits licence that data, don't they?

your right, sadly if a scan project happens using this method, exclusive access is needed,
theres now about 8 emails gone back & forth between myself & Donington,
I'll update if go ahead is given but the worse case is I'll have to find another option (track), shame as I really like
Donington
 
Ive also worked out ( roughly for now ) scan time :

Donington 2.5 miles approx

= 67 point scans approx

based on a point scan every 60 meters and allowing 12 mins for scan/reposition (approx 8 to 9 mins for actual scan)

total 13.5hrs approx + few more scans off track if time +any small delays

this is based on 2 people & 1 scanner & working quickly without a break
( would love to be able to get hold of a 2nd scanner! )

2 people = 1 on scanner & one assistant quickly moving reference spheres etc

I'm limited to only a small choice of scanners, might look into see if theres one that can scan each point faster or could safely do a wider scan area
( I think 60mtrs is pushing a bit with this model)

iracing averaged 86 meters ( scan steps moving around track)on cota
but right now not sure which actual scanner they used, cota approx 3.5 miles and it looks as if they did approx 64 point scans for main track
 
Last edited by a moderator:
p.s re.Donington

I cant really offer them much more :

Agreements in writing that the point cloud will not be used in a commercial project in anyway & be only shared for free
"shareware" and not connected with any financial gain.

A file containing the finished Donington park race circuit point cloud together with raw files to be used if needed as they wish or sold by Donington as their property
( & without any commission/charges from me )

Any advice from me re.simulators or use of the finished shareware track if required
( have mentioned that ISI do charge for a licence for use of their software which in this case would between Donington & ISI )

no charge for getting to know me


(last one joke)
 
Last edited by a moderator:
That scanner looks proper cool not sure which exact one you got but just watched a demo of it, and if it works how they advertise it push a button and your good to go it said something about 120 meters too and about colouring the scan as well. Is that in theory all you would need then to make a track is the scan of it or is there a load more too it? I dont live far from Donnington if you needed any help for the day would be willing to give you a hand would be cool just to see how it worked hope everything goes to plan anyway. I can imagine there being a problem with donnington though bet they cant get there head around it being free and not for profit but fingers crossed.
 
hi all,

some minor progression here, firstly I'm starting to consider using a static professional grade scanner
I'm going to work out how long it would take to scan a track the size of Donington using this method based on point scans 25 to 40 meters apart & 15 minutes per point scan.
There will be two of us, one on the scanner and one moving the reference spheres/assistant etc.

Due to this type of hardware producing the point cloud takes much longer i would obviously need exclusive access to the circuit so Ive been in communication with Donington,
Ive made it clear this is not a commercial profit making project,
they have expressed some concern that a license could be needed to be in a commercial piece of software,
ISI seem disconnected from 3rd party mods or users free share personal projects I'm hope I'm correct in informing them that there is no issue.

the scans can be done through the night if need be providing conditions are dry,
photography & colour matches can be done during the latest morning hours possible.

The scanner I have access to is a Faro 3D S1 20

I'm not familiar with Faro Scene software
( the software used for stitching the scans into one point cloud )
but once the scans are complete theres plenty of time to play with the raw files & software after,
.... anyone familiar with Faro 3d ? ,

once the point cloud is complete it can then be exported into the build software
perhaps 3d max.

I'm fairly confident using the scan hardware/reference spheres etc, the Faro 3D software dosn't scare me at all, the thing I'm not liking is sealing/encapsulating the track surface mesh to that of the point cloud
(the whole target here is achieving a quality laser scanned track surface ), iracing has software that does this for them apparently ( see their cota track project vid )

I think there may be a script/plugin for 3d max for this? does anyone have info of how accurately it seals to the point cloud?
this is such a key thing otherwise a high quality laser scan would be wasted.

p.s if Donington dosn't approve of this project & allow access
( Ive explained the promotional advantage etc )
then I'm considering backups (different tracks) that are fairly local to me.

My objective here is to get this project to a level of a "tech track" eg quality laser scanned road surface with barriers and track edges as a first stage ( & release ),

Im also assuming that the rfactor 2 software can handle high quality laser scanned tracks just as good as other titles. (anyone know any different ? )

the trees/buildings etc could be done in a second stage & I may ask/need a joint community collaboration at this point

Correct me if I'm wrong, but it sounds to me as if you expect to be able to use the native high definition scan results in a direct convert into a mesh, prettt much without loss of data - this will not be possible. No sims can do this, but iRacing are using a technique where tiny variations of the track surface are stored as seperate data for physics calculations (not visually rendered) - rfactor2 don't, but IMHO still has a superior immersion of 'connectivity' with the track.

No matter the source for x,y,z data you'll always need to manually model "low resolution" meshes for a track - which ofcourse can be "glued" to match the scanned data.

Hope I did not burst your bubble, getting track layout, dips, bumps and cambers right is still possible at a very high degree though - and let's face it, this is what's important, smaller cracks and bumps change from season to season anyways.

Good luck with the project! :D
 
Back
Top