OSArch Community

The Seed. :)

  1. T
    
        /*#19= IFCPRODUCTDEFINITIONSHAPE($,$,(#15));*/
    
          /*#15= IFCSHAPEREPRESENTATION(#9,'Body','SweptSolid',(#18));*/
    
            /*#9= IFCGEOMETRICREPRESENTATIONSUBCONTEXT('Body','Model',0,$,$,$,#3,$,.MODEL_VIEW.,$);*/
    
              /*#3= IFCGEOMETRICREPRESENTATIONCONTEXT($,'Model',3,1.0E-05,#8,$);*/
    
                /*#8= IFCAXIS2PLACEMENT3D(#14,$,$);*/
    
                  /*#14= IFCCARTESIANPOINT((0.,0.,0.));*/
    
            /*#18= IFCEXTRUDEDAREASOLID(#21,$,#22,2000.);*/
    
              /*#21= IFCRECTANGLEPROFILEDEF(.AREA.,'1m x 1m rectangle',$,1000.,1000.);*/
    
              /*#22= IFCDIRECTION((0.,0.,1.));*/
    

    @yorik , @Moult

    Possible to roundtrip just this between Freecad and Blender, and maintain all the parameters through strict implementation?

    We'll call it the 'The Seed!' :)

    Thoughts?

  1. M

    @theoryshaw Blender currently doesn't read the rectangle profile def. It'll convert it to an arbitrary polygon (which may happen to be identical to a rectangle). Perhaps this is something to fix first?

  1. Y

    FreeCAD can export exactly that above, but last time I checked, Revit had problems reading the IfcRectangleProfileDef entity. So there is even an option to disable it in FreeCAD (all profiles are then exported as IfcClosedArbitraryProfileDef). What happens at import, at the moment, extrusions are recreated, but the base profile is not a rectangle anymore, but a polyline. We could actually add a check to turn it into a rectangle if applicable... I'll have a look at that. Then I think we'll have 100% back-and-forth compatibility with the above.

  1. T

    Very cool.

    Yeah, i don't think we should worry about trying to accomodate a strict roundtrip with Revit anymore--unless someone could do a ground up converter through Dynamo or Revit's API where we could more easily maintain a strict implementation.

    Thoughts?

  1. M

    It sounds like @yorik's implementation in FreeCAD, my implementation in Blender, and Revit's implementation all ignore the rectangle profile def and use a regular closed profile. Maybe we can round-trip that instead?

  1. T

    Yes, IfcClosedArbitraryProfileDef sounds a like a good seed as well.

    Any thoughts on what we can call this strict implementation initiative? FreeMVD?

    Eitherway, I suggest not trying to bend over backwards to get Revit to comply as well. Unless we can find someone to implement a ground up converter through Dynamo or Revit's API. In my experience, feature requests to IFC for Revit, fall on deaf ears--i've tried a lot over the years and nothing.

  1. Y
  1. T

    This is great!

    @yorik , can you put (2) cubes in your FreeCAD file, one extruding in the Z and one extruding in the y or x?

    I just know Revit won't obey extrusion direction, so I can show that it FAILS. ;)

    Also, can you add a 3rd freecad file, where you import your IFC back into Freecad? I think that's valuable documentation to see, as well--as that demonstrates the 'roundtrip'.

    Excited.

  1. R
  1. M

    @theoryshaw I think Revit does obey extrusion direction, and I've implemented it - but I'll wait for @yorik's file first.

  1. Y
  1. T

    I reorganized a bit

    https://github.com/OpeningDesign/FreeMVD_WorkFlow/tree/master/Specifications_Test_Files/Extrusions

    Proposing we show pass fail on the file itself. Would require less upkeep on the 'table'.

    With files, we can show the branching of various import/export 'paths' in and out of various programs.

    ...

    I also deleted the simple extrusion test, as that would be tested for in the 'multiple extrusion' test... just less stuff to maintain.

    ...

    @Moult, yes Revit imports extrusion directions correctly, but does not export them correctly. :\

    ...

    Please let me know if you have any push back on this proposed way of organizing... i'm sure the structure of how we organize these tests will change quite a bit over time anyway.

    ...

    Yes, agree with @yorik circles, and arcs, would be a nice next test.

  1. T

    Also, forgot to mention, Using the IFC-commenter I added corresponding 'commented' IFC to the tests as well. It helps me to understand the file's structure and run trail/error scenarios on the file. Let me know if that's helpful to others as well.

  1. T
  1. T
  1. T
  1. Y

    Looks good! But since this is about import and export, i would expect every app to produce its "seed file".. would that fit in your naming scheme?

  1. T

    Hi @yorik, sorry for the delay, it's been a crazy week.

    i would expect every app to produce its "seed file"

    Are you thinking each application, would recreate, from scratch, their own native test file? If so, I can see that getting complicated. Would be hard to verify and assure that the originating author creates their particular native file exactly as someone from another native application.


    Having said that however, as we get into this, it just might be necessary to do this. If so, perhaps at that point we can use folders for each native file seed.

  1. Y

    Yes, i would think creating a file from scratch might give different results than importing the ifc file, and then it could be interesting to compare.. but yeah then we'd need to describe better what needs to be modelled.

  1. T
  1. T

    @yorik possible to add your 'Extrusions_03.FCStd' file?

    That is...

    Extrusions_01.FCStd --> Extrusions_02_PASSED.ifc --> Extrusions_03_PASSED.FCStd

  1. M

    I will try BricsCAD IFC BIM :)

    Extrusions_01.FCStd --> Extrusions_02_PASSED.ifc -->

    Cannot import file: C:\Users\Admin\Documents\IFC\Testing SEED\Extrusions_02_PASSED.ifc (Failed to read IfcModel from C:\Users\Admin\Documents\IFC\Testing SEED\Extrusions_02_PASSED.ifc.

    See C:\Users\Admin\Documents\IFC\Testing SEED\Extrusions_02_PASSED.ifc.in.log for details)

    It just says..

    Parser Error at 7.11: syntax error

  1. M

    Yeah ... I do need to improve the geometric representation round-tripping in Blender ...

  1. T

    When importing this file into blender28-bim-200525-win I get the following error...

    !

  1. M

    @theoryshaw the reason for that is because the IfcBuilding does not have any object placement. Theoretically, the IfcBuilding does not need an object placement and the specification says that it is optional, unless it has a representation.

    However, there is an implementer agreement required for certification that contradicts this, that states that vendors should create a relative placement equivalent to the spatial tree. It looks like FreeCAD does not honour this implementer agreement, resulting in that error.

    I could accommodate it, or maybe FreeCAD should fix it? Not sure, buildingSMART has left this a bit ambiguous. Thoughts, @yorik?

Go to page:

  1. 1
  2. 2
  3. 3

Login or Register to reply.