CAD software discussion forum > 3D CAD/CAM > Point Cloud Sectioning Example

Point Cloud Sectioning Example

Rank: 1

Kyonghun

Newbie

posts: 0

Registered: 2002-8-28

Message 1 of 7

 Point Cloud Sectioning Example
07-11-2003 10:25 . am | View his/her posts only
There have been a few inquiries about this topic, so I figure I'd do a simple tutorial.

We can currently use point block tools to divide point data into sections. Each section can then be fitted by with a surface.
This is not as convenient as automatic sectioning through boundary edge generation method, but can be still preferable over the existing "single surface" method for manu reasons. here is a trivial example illustrating the general procedure.

BTW, if you haven't guessed already, future enhancements will also revlove around making these basic procedures easier and faster.

Unzip the file and re-play the history.

Rank: 1

ChrisWard2k2

Newbie

posts: 2

Registered: 2011-11-22

Message 2 of 7

04-12-2003 04:17 . am | View his/her posts only
Oops! We can't replay the history because there is an import of an un-known file at the top of it

Can you publish the point cloud file too?

Rank: 1

Kyonghun

Newbie

posts: 0

Registered: 2002-8-28

Message 3 of 7

04-12-2003 09:14 . am | View his/her posts only
D'oh!

Here it is.

Rank: 1

ChrisWard2k2

Newbie

posts: 2

Registered: 2011-11-22

Message 4 of 7

09-12-2003 08:59 . am | View his/her posts only
Thank you for the file. I have stepped-through the history (as far as I could using v9FT, the history fails approx 50% through). What I see though is not quite what I had hoped. The process is rather long-winded and section curve positioning requires an expert eye to judge. The task of inserting the curves is I think enough to drive most people a little mad Also, the shape created from the points using the FEM surface tool looks rather "lumpy", but most modification to improve this results in the face being off the section curves.

I have an example where the faces are a closed shape, originally produced by a sweep in VX. These shapes seem to be even more stubborn than the Rabbit example. I also wanted to produce manufacturing quality shapes from a set of 5 (customer) Point Cloud files. These describe shapes to produce a model aircraft kit which would be an injection moulding. I don't think our tools are a solution for this task as yet, though obviously progress has been very good.

It may be that some form of surface tessellation of the points would be a good solution, especially for those customers whose aim is reverse engineering. This is the reason why (in the CAM Forum), I asked about a Point Cloud to STL translation. VX's import of STL data is very good indeed. For many customers, no further work is required, it can go straight into a CAM Plan. Hey, wouldn't it be great if we could machine a Point Cloud..........



Rank: 1

Kyonghun

Newbie

posts: 0

Registered: 2002-8-28

Message 5 of 7

09-12-2003 01:42 . pm | View his/her posts only
Well, as mentioned earlier, this is not a general solution. This will help some cases where you'd actually want those little lumpy details (as in embossing, etc) and/or if the "single" surface solution becomes too complex to fit in a reasonable amount of time.
Machine-able surface will have to come from some sort of fairing, afterwards.
--

For the "lumpy" surfaces, try lowering the degree of the surface via the option form settings.
The entier purpose of sectioning is also this. To have a number of lower degree surfaces rather than a huge complex one.

--
The section curves are projected to the final surface to creat the final trimmed edges, so, for the boundary of a neighbor face, say, the edge that was created off the section curves from the frst surface should be used, not the section curve. This will help maintain continuity across that edge for both surfaces. (tangency option can be of further help, if the data points are really bad). This also means the section curves do not really have to be precise...all it needs to do define a trim boundary.

--
With tessellation implemented, much of the proecure will be the same, too, but minus the clumsy boundary curve definition part.

Rank: 1

Steve

Newbie

posts: 0

Registered: 2004-3-1

Message 6 of 7

11-03-2004 11:25 . am | View his/her posts only
Chris,

There are a few systems that can machine on a point cloud. You're right, it is very useful for reverse engineering. The trick is to triangulate (tesselate) the cloud and machine on the resulting polygons.

As far as lumpiness in fitted surfaces, it would be nice if we could specify that a fitted surface is convex everywhere (any line drawn between 2 points on the surface lies entirely inside the surface). This would help a lot when reverse engineering convex shapes. I don't know if this is mathematically possible, though. I submitted a PCR on this issue a few weeks ago when I was reverse engineering a wing fuel tank. I spent many hours working the lumps out of the surface.

Steve F.
Ace Clearwater

Rank: 1

Kyonghun

Newbie

posts: 0

Registered: 2002-8-28

Message 7 of 7

01-04-2004 01:35 . pm | View his/her posts only
Can you clearify the "convex" idea a bit more? Ar you talking about convex hulls for the point data?
Might be something worth looking into.

While we are on this topic:

There is no "feature" detection of any kind yet in the current version. The "base face" option sort of
goes around this problem since the user may specify the general shape of the surface by creating
one. The "iteration" method just helps the user do so without first having to parametrically design
a complex base face. But, if feasible, one can still design a base face first.

Tesselllated point data, on the other hand, can facilitate these things better, for obvious reasons
and that is the route we are taking. Should be obvious by now, right?
See also