Convert 3D CAD Solid model to 3D Point Data

[Anonymous]

Hello,

I have a 3D Model of a drone which was taken from a 3D Scanner by converting images into 3D Point Cloud.

Now I have a solid model and is there anyway to convert this solid model into 3D point Data?

Thank you.

[Xun.Zhang]

Hello @Anonymous,

Not sure if there is a way to produce a 3D solid model to 3D point, but 3D model can be converted to Mesh file.

What's the reason behind? Because you already have the precise 3D model, why you still need the 3D points? 

Mesh file is also a way to make it lightweight and easy for rendering purpose.

Hope it helps!

[Anonymous]

Thank you for the response.

I already have a precise model but I need the 3D Data points to simplify the model further. Also its easier to stream the point clouds over the local network and I would like to use it on other applications where the file isn't supported.

I was wondering if it is possible to save the file as IGES/STL, which in a way represents the mesh and this can in someway be used to extract point clouds. 

Thanks again.

[Xun.Zhang]

Base on your requirement, Stl file (mesh) should be the solution.

Hope it helps!

[Anonymous]

Thank you again.

Could you please elaborate on how I could proceed with an STL format?

[Xun.Zhang]

Simply select File -> Export -> 3D format and select the stl file format from drop down list.

https://knowledge.autodesk.com/support/inventor-products/learn-explore/caas/screencast/Main/Details/...

Thanks!

[ToddHarris7556]

If you follow the process through, in my mind, converting back into a point cloud is a little bit of an 'unnatural act':

- when a point cloud is generated, it consists of a 'sampling' of data points. Yes, there are a LOT of them, but there's not actually a data point that represents every precise location in space. There is some 'resolution'... i.e. space between the points. 

- when you process a point cloud (convert from PC to mesh to SOLID), the software is doing it's level best to come up with surfaces/curves/equations that best describe the shape outlined by the point cloud. This is exactly like collecting a bunch of data points on an X-Y graph, and then trying to fit a mathematical curve that describes them. The whole point is to go from a million data points, to a much simpler/cleaner curve. The benefits are 1) the curve/surface is much easier to manipulate and 2) the data points/point cloud has 'noise' involved - i.e. a little bit of error with every point. When you convert that to a surface, you're intentionally 'washing away' the error. (OK, this is technically inaccurate - the triangulated surface that you start with still has all of the raw data points and error built in, but typically the end goal is to simplify/approximate/clean up the surface)

- when converted to a solid, all connection to the original point cloud data points is lost. You now have a clean approximation of the original data set. That shape could be described by an infinite number of data sets. If you go back to the XY graph of data points, it would be like plotting data points, fitting the curve, erasing the data points, and then trying to guess what the original data points were that resulted in that curve. 

So while going back to point cloud might not make sense, the most meaningful thing you could do would be to go with a mesh that fits the solid that you currently have. i.e. STL, as has been suggested. That's going to give you a finite number of elements (faces and vertices) that come very close to the solid model. 

At least, that's my thought process.