Scaling scanned geo to real world measurements

Hello Alex,

as you point is correct, you should first tell us what is the format of the 3D scan object, as well as what software / tool you are using to scan. This is essential to know the possible 3D modeling software compatibility. I see three possible approaches. Last useful information is knowing the average size of the object. You mention caliper so I suppose that real world objects are average small scale but it is good to know some parameter.

Also consider that the match approach of 3D scanners is usually based on absolute coordinates just depending on the scanner absolute scaling then the metric is applied further but the math model post processing.

Scaling models based on units it is just normal and it is  good methodology applied in telemetry and photogrammetry. For example when you 3D scan a geo surface in meters or tenth of meters then it is easier to manage the model for mapping or any other application (e.g. archaeology virtual reconstruction of sites) considering 1m=1cm

The first is seeing if your 3D scanning software has some settings to configure it:

• Does it support unit settings ? Maybe you are scanning with millimeters resolution but for some reason these are considered  centimeters or millimeters.
• Is possible that you are thinking decimal units while the scanner is configured to consider units in Imperial system? In this case every 1 units you expect (no matter if are cm or mm or m) these are about 2.5 bigger
• As far as I read you have done more than a single test; can you experimentally verify if the 3D object is proportionally resized or the scaling factors are different on the three axis?
• Did you verified if the scaling factor is a constant or it is random? If it is not a constant and it is not possible to set up it before scanning maybe there is some bug in the software (again, good to know what is the scan software you are using).

Consider that the 3D scanning software it is expected it think like a CAD: the scanned area is a sort of bix voxel, a virtual 3D space you are managing and your object is expected to be inscribed inside a cube of absolute dimensions (the number of 3D pixels determined by the 3D scanner resolution) to which you assign a natural real-world size, maybe corresponding to the real object size or bigger or smaller (this is the advantage of the 3D scanners, creating different sizes 3D models than the real object).

About the CAD to be used, as a matter of fact you can use Maya or Rhino (if you have knowledge using this or other commercial CAD) as well as free 3D CAD like the Autodesk but in my opinion as this is more a math problem it is not the worth to manually process every scanned image. You risk to be inaccurate and spend a lot of time making post-processing measures.

In this case I suggest using a non-interactive design tool like SCAD that makes very easy creating automated scripts (I don't know this application very good but I think they can be launched directly from terminal or from inside a Python script for post-processing the 3D scanned image) as just a math operation should be applied. I have done something similar in past with ArtCAM software (that I discourage you tu use but it is efficient for this role).

Enrico

Thankyou for the reply!

You know, this is the first time I have used laser scanned geo. I assumed the scale would be arbitrary like with photogrammetry. However, after breaking out the calipers, it turns out to be scaled up by a factor of exactly 10. So nice and easy Thank you so much for your message as I would have gone down a huge rabbit hole before discovering this if it weren't for you! Is it to be expected with laser scanning that the scale of the geo will be faithful to real life?

However I do use photogrammetry for other things and I would still like to know the best way to go about that problem. IE, when you can be sure that the scale is just random, but you have a set of measurements from the real world that you can apply, is there a smart way to punch in those values like you would with constraints in a CAD software?

Since there is margin for error when corresponding real world measurement points with their equivalent in your scanned geo, ideally you would enter all the values you have and it calculate an average between the inevitable discrepancies. I know that there is no out-of-the-box solution for this in Maya or C4D.

If this doesn't exist, it does sound like something that could be achieved relatively easily in something that could be scripted, as you suggested.

Once again, thanks for the response!

Alex

PS I should mention - I didn't do the scan myself nor do I have access to the scanner or the software used, hence I did not answer your questions regarding those things.

PS I should mention - I didn't do the scan myself nor do I have access to the scanner or the software used, hence I did not answer your questions regarding those things.