Modelling curtainwall.

[re:FORM]

I'm modelling up some curtainwall at the moment, and was wondering if having glass and the mullions co-planar would be a problem.

Here's a funky ascii diagram showing the glazing in plan view.
(ignore the dots... I needed to add them to space the diagram out.)
___ ____
___||____ <<Glass
....___..... <<< In my model this gap doesnt exist (coplanar)
....|||||.....<<Mullion
....|||||.....

The top two pieces are the glass, with a gap for the silicon joint.
The certical piece is the mullion. The way I model this usually is to snap the mullion to the glazing so the edge of the mullion is flush (coplanar) with the back face of the glass. Will this cause AA problems with the coplanar faces (or sampling slowdown) meaning I should model it more like the ascii diagram above with a gap between the glass and mullion. If I do need a gap, how does the minimum size worth using relate to any of my vray parameters?

Thanks for any advice

[dapeter]

Hmm, not too sure about a slowdown, but I'd say it's not great to have the coplanar faces. Most of the time when I model curtain walls I'll just make it out of one object and poly model the mullions, and set it up with a multi-sub material.

[re:FORM]

Hi Dapeter,
Thats what I thought.... just wanted to hear if anyone had any suggestions on the matter.
I'm writing a script at the moment that builds a curtain-wall (frameless-on-the -exterior type curtainwall) from a base mesh. ie

-make three reference copies of the base mesh (glass, mullion, silicon)
-edit poly, Inset polys by joint thickness, delete the strips between panes, add shell modifier. This leaves you with individual panes of glass in the one polymesh.
- same procedure for the mullions and silicon with a few variations.

So from a mesh that is subdivided where you want the framing to be, its one click of the script and you have your curtain wall ready for mapping. You can also adjust the position of the framing without rebuilding the model by editing the base mesh... handy for adjustments.

I'll add a safety distance to my script so none of the elements are coplanar. Any suggestions to the minimum gap between elements to avoid them appearing to vray as being coplanar?

The reason I mentioned a slowdown, is that I noticed that with a shell modifier with heights of 0 and 0, ie coplanar surface, the render slows down to a crawl. I assume coplanar surfaces cause vray to use loads of samples.

[DaForce]

co-planar faces are really bad especially for glass/transparent objects.

I think as long as you can see them not touching (the faces that is) then thats fine.

[Clifton Santiago]

Why can't you just pull the mullions away from the glazing slightly? That's what I do.

[re:FORM]

No problem with doing that... I just like to keep my meshes nice and accurate, so having the glass flush with the mullion would be the best option. But given the problems with coplanar face calculations, pulling them apart is the best way to do it.
Check out the script I'm working on to automate curtainwall construction... I'll have a value that sets the seperation between glass and the other elements to avoid coplanar faces

http://www.chaosgroup.com/forum/phpB...=147279#147279

[dapeter]

Well, I also have to admit that I cheat it a lot of times also - we use Arch Desktop here at work which has a Curtain Wall creator. A little tricky to work with sometimes, but it ends up creating each individual pane of glass as a seperate item, so adding spandrel or opaque areas is pretty easy.

[teabag]

depending on the distance of the camera i have used renderable splines between the panes made from aluminum....this fixes problem of modelling as well as ads a bit of needed definition and its fast