Alireza - Five great questions.

Rob - Five great responses.

I've done these simulations with and without spacing. There are probably

examples I've produced of both ways floating around. So I'll try to explain

why you'd want to do it one way or the other.

The basic idea is to make sure that you account for the window thickness

and edge condition once and only once in your simulation. The genBSDF

tutorial touches on the difference between a system specific BSDF, which

includes edge conditions, and an infinite layer BSDF, which has no edge

conditions. If your BSDF does not include the edge condition, then you want

your sending and receiving surfaces to be coincident, so that the edge

condition is accounted for in the view and daylight matrices. If your BSDF

does include the edge conditions, then you want a space between your

sending and receiving surfaces so that the edge condition is not duplicated

in the view and daylight matrix.

So how do you know whether you have an infinite layer BSDF or a BSDF with

edge conditions?

If your BSDF is generated by Window then it is an infinite layer, without

edge conditions.

If your BSDF is generated by genBSDF then it contains edge conditions,

unless you took care to minimize sampling of edges (see genBSDF tutorial).

If you're using proxied geometry, than your sender and receiver geometry

needs to be spaced (the thickness of the geometry) so you should make sure

that your BSDFs contain the edge effects.

For the 5-phase method, your strategy starts with what you want to do for

the direct sun coefficient matrix.

If you want to use proxied geometry than the sender and receiver must be

spaced apart and the BSDF should include edge effects specific to your

model.

If you want to use a tensor tree BSDF, then it depends on what you did to

create the BSDF. If it is a typical genBSDF tensor tree run, then edge

conditions are included in the BSDF and you should include a space between

your sending and receiving surfaces. If however you took care to create an

infinite layer BSDF (tensor tree) with genBSDF then your sender and

receiver surfaces can be coincident.

Hopefully this gives a little bit of backgroud to the whole space or no

space between sending and receiving surfaces.

Best,

Andy

## ···

On Sun, Jan 31, 2016 at 8:40 AM, Rob Guglielmetti < rob.guglielmetti@gmail.com> wrote:

Lots of questions, I think I can answer some of them…

>

> 1- In the tutorial, a single polygon (relative to the outline of the

glazing surface) is modeled in the scene at the exterior side of the

window. This polygon is referenced to generate the daylight matrix. A

second polygon is modeled in the scene on the interior side of the window

and it is referenced to calculate the view matrix.

>

> I was wondering if I could generate two identical co-planar polygons

where the normal vector of one polygon is looking at the interior side and

the normal vector of the other polygon looks at the exterior side to be

referenced, respectively, for the view matrix and daylight matrix

calculations? if so, would it still be necessary to model the thickness of

the walls hosting fenestration systems in the simulation scene?You could do that, just make sure that your octrees for the view and

daylight matrices only contain one the other polygon. If you have a

coincident pair of polys in the same octree, it’s up to chance which one is

hit by each ray traced. It’s already critical that every ray be counted

toward something meaningful (this is the reason for the phases in the first

place, with complex fenestration systems).> 2- What is the "proxy geometry" in the five-phase method? is the "proxy

geometry" exclusively a reference to the polygon at the interior side of

the window system that is assigned a BSDF material?

> or is the "proxy geometry" a reference to the 3D geometry of the

venetian blinds (or any other type of shading layer of the fenestration

system) in the scene?It’s the latter. That geometry serves to block the direct light, in that

phase of the workflow.>

> 3- what is the "thickness" as the first parameter in the Radiance BSDF

material type referring to? is it measured as the distance between the

daylight matrix polygon on the exterior side of the window and the view

matrix polygon at the interior side?Yes.

>

> or does "thickness" refer to the depth of the entire fenestration system

including the glazing panes and any shading layers?

>

> 4- If I model the mullions defining the edge condition of any individual

glazing pane in 3D scene, what would be the difference between using LBNL's

Window to generate a BSDF file and using genBSDF?Hmm, I don’t think there would be a difference in terms of the Radiance

BSDF and calculations.>

> 5- What would be the difference between using a BSDF file based on

Klem's division scheme with higher resolution (where each Klem's patch is

subdivided to 4 new patches) and using a tensor tree BSDF?The difference would be higher accuracy in the case of the tensor tree

BSDFs, at the expense of longer stimulation times, much larger files (both

the input and output), and longer BSDF generation times.- Rob

_______________________________________________

Radiance-general mailing list

Radiance-general@radiance-online.org

http://www.radiance-online.org/mailman/listinfo/radiance-general