[Discussion] A different approach to Arch modelling (Arch Assembly?)

A forum dedicated to the Draft, Arch and BIM workbenches development.
Forum rules
Be nice to others! Respect the FreeCAD code of conduct!
carlopav
Veteran
Posts: 2062
Joined: Mon Dec 31, 2018 1:49 pm
Location: Venice, Italy

Re: [Discussion] A different approach to Arch modelling (Arch Assembly?)

Post by carlopav »

vanuan wrote: Fri Aug 14, 2020 10:31 pm At the very least, you should be able to produce standardized representations of reality. Those are IFC models and 2D drawings.

To produce 2D drawings, FreeCAD Arch currently uses section plane cuts. It means that wall joins would look exactly like the 3D geometry.
Sure, but the level of detail really depend on the design phase: you simply do not need to have layers represented and correctly joined from the beginnig, so - at least here in italy - the design is always organized roughly in those steps:
- preliminary design (no layers) you need to quickly change the layout, not much detail needed;
- final design (no layers) you still make many small modification of the layout;
- execution design (layers) the layout is less or more unchanged, but you detail everything. In this phase many times you just do not represent layers in the whole plans, but detail the joints and the other important parts in separated drawings. That's when you generally swear an yell to the automatic joining system. And makes you divide the wall into it's layers or, worse, draw each layer as a separated wall. Awful, believe me.

So we could take this into account to make something that changes according to which phase of the design you are: starting really simple, but being able to complexify things during the design process...

What do you think?
follow my experiments on BIM modelling for architecture design
User avatar
vanuan
Posts: 539
Joined: Wed Oct 24, 2018 9:49 pm

Re: [Discussion] A different approach to Arch modelling (Arch Assembly?)

Post by vanuan »

carlopav wrote: Sat Aug 15, 2020 8:31 am
So we could take this into account to make something that changes according to which phase of the design you are: starting really simple, but being able to complexify things during the design process...

What do you think?
I think this can be supported at the object level using ViewProvider. Some global override is needed. Maybe somewhere close to "View -> Draw style" functionality.

The question is, should the different levels of detail properties be stored in the document or be hardcoded in the FreeCAD. That is, are phases of design part of the document or part of the software?
carlopav
Veteran
Posts: 2062
Joined: Mon Dec 31, 2018 1:49 pm
Location: Venice, Italy

Re: [Discussion] A different approach to Arch modelling (Arch Assembly?)

Post by carlopav »

vanuan wrote: Sat Aug 15, 2020 10:41 am I think this can be supported at the object level using ViewProvider. Some global override is needed. Maybe somewhere close to "View -> Draw style" functionality.
I think this can be tricky. We should check better the DisplayMode property, that controls how the ViewProvider display the shape. At the moment the wall ViewProvider python wrapper supplied by @realthunder overrides the FlatLines display mode to show the wall shape and its children at the same time. You can try to play with ViewObject.DisplayMode properties to explore it.

But the fundamental problem is that the ViewProvider always display Object Shape, so should we have 2 different shapes, one for monolitich "bounding wall" and one with layers divided? Should we sort them in some way so we know that every solid correspond to a different material?
I'm absolutely in favour to tackle this topic...
I think it's just overkill to write a custom ViewProvider from scratch... also it's not useful to unbound it from the obj Shape, because we would lose the OCC representation feature...
follow my experiments on BIM modelling for architecture design
User avatar
vanuan
Posts: 539
Joined: Wed Oct 24, 2018 9:49 pm

Re: [Discussion] A different approach to Arch modelling (Arch Assembly?)

Post by vanuan »

carlopav wrote: Sat Aug 15, 2020 11:25 am But the fundamental problem is that the ViewProvider always display Object Shape, so should we have 2 different shapes, one for monolitich "bounding wall" and one with layers divided?
Why should we have two? Are you thinking of a case when multiple detail levels are shown at the same time?
carlopav wrote: Sat Aug 15, 2020 11:25 am Should we sort them in some way so we know that every solid correspond to a different material?
Aren't layers modeled by a single multi-material object?
carlopav wrote: Sat Aug 15, 2020 11:25 am I think it's just overkill to write a custom ViewProvider from scratch... also it's not useful to unbound it from the obj Shape, because we would lose the OCC representation feature...
Not sure I follow. You can reuse the functionality of an existing provider if needed. There are many patterns you can use: composition, inheritance, decorator, etc.
carlopav wrote: Sat Aug 15, 2020 11:25 am because we would lose the OCC representation feature...
Could you elaborate? Are you talking about topological shapes, like points, edges, faces? Aren't those ephemeral shapes that are recalculated when you change the object properties? Should these be persisted with their own properties? The only use case I can think of is IFC topological representation and maybe assigning textures of wall finish layers.
Bojan_Bekic
Posts: 26
Joined: Tue Feb 05, 2019 9:01 am

Re: [Discussion] A different approach to Arch modelling (Arch Assembly?)

Post by Bojan_Bekic »

carlopav wrote: Sat Aug 15, 2020 8:19 am
No idea how to do that and not my priority: if the line correctly disappears in the svg output of techdraw, I can live with that line in the 3d view. :)
- Support for materials so one could output drawings at least with color, if not hatching, and doing a simple bill of quantities would be possible
- Fixing the section view output
Yes, that's something high priority I believe.
- Adding more options for windows
Mind that you can design your custom windows through PartDesign and embed them into the opening tool. You can even embed a current window into the opening! (extending the window presets is also easy, if you can write some simple python).
That way we could already have a thing that works. We could actually produce drawings :) And only then tackle the complicated world of multilayered walls.
Yes, this should be the short therm goal to attract more people to Arch and BIM workbenches.
For the joints, fair enough :)
Designing custom windows sounds great, didn't really look into it yet, thanks!

Ah, I forgot about the centerlines and orientations of walls, dunno if they are a problem to be added.

My point is that if arch fuse is doable and these things (Section view, centerlines and materials) are not a lot of work for you, then we are really close to be able to do some real drawings. I think THAT IS VERY IMPORTANT, without that, not enough users testing stuff, etc...

As for the multilayer walls, until they are possible, for me as a user it's still ok to just draw/model them all as separate elements. It's still BIM, still way faster and more intelligent then AutoCAD :)
carlopav
Veteran
Posts: 2062
Joined: Mon Dec 31, 2018 1:49 pm
Location: Venice, Italy

Re: [Discussion] A different approach to Arch modelling (Arch Assembly?)

Post by carlopav »

vanuan wrote: Sat Aug 15, 2020 12:54 pm Why should we have two? Are you thinking of a case when multiple detail levels are shown at the same time?
hmmm... yes maybe you are right... I was thinking to 2 to be able to chose which shape to display with a property in the ViewProvider (being always able to access the 2 without having to recompute the object)...
carlopav wrote: Sat Aug 15, 2020 11:25 am Should we sort them in some way so we know that every solid correspond to a different material?
Aren't layers modeled by a single multi-material object?
Yes they are, I was thinking to the usecase when the user model each layer separately and group them under the wall object. To keep being able to have the materials correctly assigned.

carlopav wrote: Sat Aug 15, 2020 11:25 am I think it's just overkill to write a custom ViewProvider from scratch... also it's not useful to unbound it from the obj Shape, because we would lose the OCC representation feature...
Not sure I follow. You can reuse the functionality of an existing provider if needed. There are many patterns you can use: composition, inheritance, decorator, etc.
Probably you are right, we should investigate more in this...
Could you elaborate? Are you talking about topological shapes, like points, edges, faces? Aren't those ephemeral shapes that are recalculated when you change the object properties? Should these be persisted with their own properties? The only use case I can think of is IFC topological representation and maybe assigning textures of wall finish layers.
Well, the Shape itself is persistent untill you recompute the object, and it's also stored in the FCStd file (at least I think so). The different representation context of IFC are indeed something that could fit having 2 shapes of the object.

We should also think about 2d views: a window in represented in 2d in a very different way... dont' know if we should host this representation in the object itself, or having it computed when we create a SectionPlane and a Shape2dView... Same goes with different wall fill or hatching...


BTW, if you could have a deeper look at the code i'd love if you could give some suggestion to better setup the wall object from the beginning :)
follow my experiments on BIM modelling for architecture design
carlopav
Veteran
Posts: 2062
Joined: Mon Dec 31, 2018 1:49 pm
Location: Venice, Italy

Re: [Discussion] A different approach to Arch modelling (Arch Assembly?)

Post by carlopav »

BTW, that's what i had in mind to easily provide a preliminary supporto for wall layers:


phpBB [video]



What it does is simply take the generated base shape of the wall and slice it in layers according to the multi-material selected. Slicing is done through Part.generalFuse, that is really quick!

https://github.com/yorikvanhavre/BIM_Wo ... ef14cfce33
follow my experiments on BIM modelling for architecture design
Cyril
Posts: 133
Joined: Wed Aug 23, 2017 5:04 pm
Location: Geneva (Switzerland)
Contact:

Re: [Discussion] A different approach to Arch modelling (Arch Assembly?)

Post by Cyril »

carlopav wrote: Mon Aug 17, 2020 7:39 am BTW, that's what i had in mind to easily provide a preliminary supporto for wall layers:
I am very enthusiast about your progress. Nice work. It looks more and more user friendly and functional over time.
I blog about HVAC / BIM / Energy : pythoncvc.net. If you like you can follow the RSS feed.
carlopav
Veteran
Posts: 2062
Joined: Mon Dec 31, 2018 1:49 pm
Location: Venice, Italy

Re: [Discussion] A different approach to Arch modelling (Arch Assembly?)

Post by carlopav »

Thanks Cyril!
follow my experiments on BIM modelling for architecture design
carlopav
Veteran
Posts: 2062
Joined: Mon Dec 31, 2018 1:49 pm
Location: Venice, Italy

Re: [Discussion] A different approach to Arch modelling (Arch Assembly?)

Post by carlopav »

Hello and sorry for the long absence. It is rather complicated to have some freetime in the last period, and this is going to last some other months probably.
I'm continuing anyway to follow the forum and Osarch where I discovered TAD. Even If I didnt use it extensively, I think the concepts around which it is built are really strong and worth exploring. This made me think I was following a too mechanical way of conceptualize architecture that could fit better the construction phase rather than the preliminary design. And I'd like to give some attention to that too... What do you think?
cheers!
follow my experiments on BIM modelling for architecture design
Post Reply