How to project drawings (TechDraw, Drawing, Export DXF, svg)

[Roland]

Hello Assemblers,

This topic rose from a topic on TechDraw. There, it was suggested that TechDraw needs "valid" shapes in order to proceed properly. (Hopefully, TechDraw will answer that question.)

The shapes that I am making are invalid (says the Parts Geometry Check). Perhaps that they are discontinuous? But they are intended the way they are. Please check the attached, as an example.

GeometryTest.FCStd is an assembly made by A2+. It contains only 1 part now: CentralBody-900. In my use case, further elements would be added using A2+. The resulting unit would be further assembled into higher levels of complexity.

For the purpose of the further analysis, I am not increasing that level of complexity now in GeometryTest.FCStd, but rather I run Parts Geometry Check. It tells me that CentralBody-900 is invalid.

The original of CentralBody-900 is a file called CentralBody-900_ass.FCStd. It contains 3 parts. Of each of those parts Parts Geometry Check tells me that they are valid. So, A2+ produces an invalid thing out of valid parts.

Questions:
What is validity?
Should we bother, in view of further model elaboration (e.g. the preparation of drawings using TechDRaw)?

Please let me know your views.

[chrisb]

My shapes are invalid. Says Geometry Check.

I split the old topic from TechDraw forum and moved this post to Assembly, where it may get the attention it deserves. Feel free to edit the subject, albeit to remove the final part in brackets.

[Roland]

Apologies for what may seem a repetition.

I would be curious to hear the view of Vocx and Wanderfan on the following:

• Is validity important for TechDraw? (why)

• If so, either I am doing something entirely wrong, or, hopefully, the TechDraw approach could be adapted (?).

It was suggested that TechDraw needs "valid" shapes in order to proceed properly.

The shapes that I am making are invalid (says the Parts Geometry Check). Perhaps that they are discontinuous? But they are intended the way they are. Please check the attached, as an example.

GeometryTest.FCStd is an assembly made by A2+. It contains only 1 part now: CentralBody-900. In my use case, further elements would be added using A2+. The resulting unit would be further assembled into higher levels of complexity.

For the purpose of the further analysis, I am not increasing that level of complexity now in GeometryTest.FCStd, but rather I run Parts Geometry Check. It tells me that CentralBody-900 is invalid.

The original of CentralBody-900 is a file called CentralBody-900_ass.FCStd. It contains 3 parts. Of each of those parts Parts Geometry Check tells me that they are valid. So, A2+ produces an invalid thing out of valid parts.

Questions: See above

Best regards

Roland

[chrisb]

I moved it back. Splitting and merging across forums is a pain. Please edit your post and create new ones to your liking.

[Roland]

@Vocx and Wanderfan:

Thanks for your suggestions. I shall test them.

Further, please consider my questions about validity of shapes.

Best regards,

Roland

[wandererfan]

• Is validity important for TechDraw? (why)

• If so, either I am doing something entirely wrong, or, hopefully, the TechDraw approach could be adapted (?).

Is validity import? Yes, if you want to do any further OCC operations on the Shape. OCC algorithms expect a well formed shape(s) as input and don't behave well with ill formed shapes. Since TD has to do OCC operations (fuse/make compound, cut and HLR) it needs the input to be well formed.

I don't believe you are doing anything "wrong". The problem is just in the nature of assemblies. A quick example may help.

Make 2 default Part Cubes, then move 1 of them 10mm along X. We now have a minimal assembly that is aligned the way we want it. Select the 2 Cubes and use Part.makeCompound on them. Our assembly is now a single "shape". If we use Check geometry, we will see that our compound is an invalid shape, because the mating face appears twice in the Compound. In a correct shape these two contiguous Faces should be replaced by a single Face that is shared by both Cubes.

OCC has individual algorithms (ShapeFix, Sewing, etc) that will fix various conditions, but AFAICT, it doesn't have a big one-stop algorithm of the "fixEverythingWrongWithMyShape" variety. So the program has to be able to a) figure out what's wrong using OCC's ShapeAnalysis tools (this is what Check geometry does), and b) apply the appropriate strategy to correct the problem.

So long term the fix is either a) have the assembly routines produce valid OCC shapes, or b) develop a shape fixer routine to apply to input shapes before trying to work with them.

[wandererfan]

Related discussion in Assembly Forum: https://forum.freecadweb.org/viewtopic.php?f=20&t=38892

[wandererfan]

Note that this simplistic approach doesn't help. Starting with our compound of 2 Cubes,

Code: Select all

>>> c = App.ActiveDocument.Compound
>>> cs = c.Shape
>>> cf = cs.Faces
>>> shell = Part.makeShell(cf)
>>> solid = Part.makeSolid(shell)
>>> Part.show(solid)
>>> Gui.Selection.clearSelection()
>>> Gui.Selection.addSelection('_CubeCompound','Shape')
>>> Gui.runCommand('Part_CheckGeometry',0)
>>> Gui.Selection.clearSelection()

The resulting solid contains 12 Faces and is invalid. It should only contain 11 Faces. The duplicate mating Faces should be replaced by a single Face shared by the 2 Cubes.

[Roland]

Dear Wanderfan,

I cannot reproduce your analysis of the 2 cubes.

Two facing cubes, compounded, yield 1 valid shape for me. File attached.

(Note this is before you gave the code. I shall check that as well.)

Greetz

Roland

[Roland]

Stupidly enough, Wanderfan, I do not know how to use your code. I pasted it in the python console. Just got plenty of errors.

Sorry,

Roland