[Plugin] Triangulate Points

[TIG]

This is a version of Didier's points_cloud_triangulation.rb but with a protecting class and a new delauney3 file to avoid clashes with other scripts.
Extract both .rb files from the zip into ../Plugins - the Plugins menu item is 'Triangulate Points [TIG]'
All triangles now face 'up'
Colinear vertical cpoint errors trapped
Triangulated mesh is now grouped
Layer dialog asks for layer if you cancel = Layer0 - type in Layer to use... that if not exists its made.
The ZIP contains TIG's 'delauney3.rb' that can also be used by other tools quite independently...

TIG_points_cloud_triangulation.zip

[simon le bon]

Thank you TIG,
All these old good plugins actually revisited and enforced by the gold ruby team is very appreciated
Well-done,
+++simon

PS:for who don't know, this release come from a very interesting request from Solo: [REQ] drop and drape

[gring]

Thank you TIG,

[pav180]

TIG,

It's a great plugin. Does it work with SU 8? I have installed it fine into SU 8, but I do not think it works. I could have done something wrong and I am not sure. Please clarify.

[TIG]

TIG,

It's a great plugin. Does it work with SU 8? I have installed it fine into SU 8, but I do not think it works. I could have done something wrong and I am not sure. Please clarify.

It works fine in v8 [as it should in v6&7].
Is the .rb file place directly into the v8 Plugins folder?
Do you get a item in the Plugins menu?
Do you preselect faces first?

[pav180]

My answer is positive to all your questions. But it is negative if you ask me if I was sucessful. I have attached a sample point cloud I am working with.

[pav180]

And one additional question please: your plugin works for selected spoints which is great, but it doesn't work for a group or whichever it's called in SU. After having imported a point cloud into SU with say Point_V8 it is a component or a group, but I have to unglue/ungroup it and than select it all again if I want to triangulate it as a whole with TIG plugin. Is it on purpose or I am doing something wrong as usually?

[Gaieus]

Why don't you just edit the group and triangulate inside?
Any new geometry created with ruby is safest inside a group anyway. You never know what other, loose geometry it encounters (and ultimately merges) with otherwise.

[pav180]

Right, Gaieus, this is what I have to do: triangulate portions of a group, and it is in fact very safe. I have experienced bad triangulation results of the whole group if it has complex organic shape. I was wondering if TIG has made this limitation on purpose to avoid inaccurate meshing of a compex shape, or I haven't figured something out. It's not a big deal anyway. Thanks for your suggestion!

[Rich O Brien]

I have experienced bad triangulation results of the whole group if it has complex organic shape

What is the bad result?

SU has problems skinning faces less than 1mm. If your points are closer than 1mm try scaling the point cloud up by 10 the Triangulating. Afterwards scaled it down by 0.1.

[TIG]

If you examine the points cloud it occupies barely 1 sq mm - so some are very close!
I scaled it up x100 and it faced fine...
As Rich says you can make very small facets in Sketchup <~0.5mm will usually fail.
I suspect the points were perhaps let's say in 'm' and you imported in 'mm' ?

Capture.PNG

[pav180]

Wow! While I was working on my reply TIG has triangulated my point cloud. Great! But it was a planar model and I did the same in SU 7.1 (I can still not manage TIG's package in SU 8), let's do something more difficult.

I scan with my3dscanner.com, I receive and download a .ply point cloud, than I import it into SU as CSV via Cloud_V7.1 (6 or 8).

I am testing SU plugins to triangulate. I get bad results if I try to triangulate the whole model either by Cloud_V8 or great TIG package.

I have scaled the model, but it didn't work. I am a little confused about scaling (my3dscanner HAS NO SCALE AT ALL which is bad), but scaling proposed by Rich did not work for Lion model.

My point is to illustrate bad results (this is what Rich has asked for) if you mesh the model as a whole and if it is complex, not planar.

This is a workflow: .ply dense point cloud, csv sparse point cloud imported into SU, bad triangulation results.

LionSUNotGood.jpg

I do know that if I triangulate a model in portions it will work. Little by little. I also suspect model orientation is a key. You are welcomed to play with the file I tried to triangulate (unfortunately CSV format is not supported here).

Please be forgiving to my questions and comments. Sadly I know very little about SU and I am here to learn, not to promote my3dscanner which I used to be one of the developers. I feel awkward discussing these matters with gurus of SU.

[TIG]

You might have made it too big by x10 but that's better than too small...
The use of a points cloud import is typically from surveyed land, so it's not expecting a 3D object - more a countoured surface.
I suggest you slice you points in something like 'half' horizontally and then merge the two surfaces together later... More like making a mold or a face-mask than a full 3D object straight off...

[pav180]

TIG,

I see your point. Size limitation comment was very helpull as well. Thank you, guys!

[pav180]

And TIG, your triangulation plugin works in SU8 beautifully. I was a dumb head...

[thomthom]

It does not respect "Cancel" as a response to the inputbox...

[TIG]

It never did - even in Didier's original version!
If you cancel the layer naming dialog then it'll use layer 'nil' - e.g. 'Layer0'.
Once you run the tool it runs the processing to the end - the layer dialog only lets you determine a layer for the mesh-group if desired, with cancel canceling the layer naming but not the main tool's operation, which carries on relentlessly. Frankly the whole layer part could be omitted, it's results are all grouped and you could easily re-layer that as desired later...