Point cloud reduction

(Ripvannik) #1

I'm having difficulty reducing the volume of points from a laser scan to within 500mb limit.

Have others been through a similar process ?

A typical scan with a faro TX5 scanner, of an interior location will generate about 900Mb. Once these are registered together..it can be a massive amount of data depending on the number of scan positions.

I'm going to look at cloud compare, but interested to know if there are other workflows out there.
Interested to hear of any other data capture methods too.


(Nebulousflynn) #3

hey @ripvannik - one possibility using a program called Meshlab. Import your point cloud data and go to the top menu Filters > Sampling. There are various solving methods for the reduction process.

This video shows the results of using the possion-disk sampling method:

Hope that helps!

(Ripvannik) #4

Thanks for the video link Thomas !
I was very impressed by the pointcloud upload of Hintze hall on your page. . I'll try to get my head round meshlab and upload the points i have here incase you are interested.

Do you find the 500mB upload limit restrictive ? The smallest data sample i have is 900Mb for an individual scan, before registering the others i have together.

I've been using veesus data studio for point cloud editing. Its an excellent piece of software but has no sub sample feature.

Kind Regards.

(Tom Goskar) #5

Another option is to download CloudCompare (open source) and use the Edit>Subsample tool.

(Ripvannik) #6

Thanks ! I was aware of cloudcompare too, but haven't looked at any options yet.

My workflow is currently:
Scan and register in Realworks using the 30 day licence I get when I hire a scanner.
Export to Arena Data studio to build video animations: https://youtu.be/WI1W2tKrYGc
Import to Thinkbox Sequoia to mesh.
Import to 3DS max to clean up and optimize.
Upload to sketchfab or export to 3D PDF:

This is all done on trial software for now, and I'm sure everything can be streamlined but that's where I'm at for now.
Next mission is to reduce the volume of points and see how this improves matters.


(Ripvannik) #7

Hi. I managed to get a point cloud up. Its still very heavy but by introducing a 20mm filter i have something that loads on firefox on the lap top. https://skfb.ly/69ovB

I'm going to keep editing and reducing points untill i get something that works on the mobile, with a view to using cardboard with it.

Advantages: no need to mesh / no need to cut polys in 3Ds max..potentially saving on software licence cost.
Disadvantages ? not sure yet !

Overall goal is to have a mobile friendly 3D visualisation without a huge drop in quality using vertex colours only.

(Ripvannik) #8

40mm radius, takes a 48m point cloud down to 1.8m. https://skfb.ly/69pF7
big drop in quality, but loads on the mobile with ease.

still need to find the optimum range.