Quote - just read my posts above and look at the graph.

 

That addresses something slightly different from what I was looking at.

BB's shader uses a formula for working out the attentuation for spherical light sources. They behave (well, the way the math works for determing how light emits from them behaves) in a particular way within 3d space. A shop window, as in your example, is not spherical, so I would expect it to behave in a slightly different way. And it does.

The shape of a light source is important as well as its size. Since spherical light sources are identical from the perspective of any target object then they follow a fairly simple function. When you bring in shop windows, and all other kinds of light-emitting objects that can have a variable profile presented to a target object, things get a little more complicated. To compare a light-emitting shop window kind of set-up with an equivalent Poser light displaying the appropriate attenuation function, you would have to create a shader to apply to the light just like BB's but it would also have to implement a different formula that doesn't merely rely on radius or, in the case of the built-in Poser ISF option, an even simpler function, and which takes into account the profile of the object as presented to the target object.

(I also think that the results your graphs and experiments are showing could easily lead someone to interpret them incorrectly, into thinking that the fall-off of a non-spherical light source can change in its underlying function of attentuation from one extreme (inverse square) to another (constant; no change) as a result of its distance from the source, but that discussion will probably drop into deep mathematics and be of no relevance to anything one would be willing to do in Poser.)