glass with frosted film on one side

[SgWRX]

hello. am i right in thinking that i should flip the normal on the "inside" of a .5" thick box that represents clear glass on one side and frosted film applied to the back side? thanks.

[SgWRX]

thanks. i've done some testing and it appears that i can make one whole poly of the glass frosted and i do not have to flip the normals. they can all face "out". render's pretty much look the same.

 

i ran into something intersting though, i'll post later with screen shots. basically it's how MR can't handle photometric visible area lights without horrible sampling artifacts (grain) no matter what render, local material, or any other sampling setting. BUT drop a visible MR light in there and it is 100x better (literally i measured 100 times lol).

[SgWRX]

[ATTACH=CONFIG]53411[/ATTACH][ATTACH=CONFIG]53412[/ATTACH]

 

i've tested with physical camera exposure control, physical units disabled with mi light set to 1500 units/physical units enabled scale 1500 with mi light set to 1.

 

the RGB "real" numbers using the color picker in the 32bit rendered frame window are 7 for the mi light and much higher (like 30?) for the photometric light.

 

mr just does not treat these two lights the same nor does there seem to be a way to get one light to behave like the other, except that in appearance the mi light can be adjusted in this case to have an inverse square fall starting at 2' (tested smaller and larger) to get closest in appearance to how the scene is lit with the photometric light.

 

final gather is off, no gi. unified sampling.

 

so notice how much better the mi light is at being seen through the frosted glass polygon on the back side of this "thin walled" glass material.

 

specular on/off with the photometric light doesn't make any improvements. there's just simply less grain in how MR handles it's own lights.

 

woops, i think i saved the wrong frame window with the mi light - it appears a little darker than the photometric light. must be one where the fall off started sooner than 2ft.

[SgWRX]

no it wasn't.

 

another thing i've just noticed is that if i set my refraction glossy samples in on the frosted material, higher and higher, the glass overall gets darker. for example 64 samples it looks pretty bright overall, or at least what you see through it. if i set to 256 samples, what you see gets a lot more subdued.

 

EDIT: you can see this effect in the material editor. create a "solid glass" material, then change it's refraction glossiness to 0.8. then try 64 samples and then change to 256 samples. it'll appear darker.

Edited August 30, 2015 by SgWRX

[Peter M. Gruhn]

I notice you used the word "thin." If your light path through the glass hits two polys, it is not a candidate for the thin glass preset. That expects only a single poly window. I don't know if that would make a difference in your model.

 

Another question is do you need a thick material? The tests you post make it seem more like you will have a relatively distant plane of glass. That could be modeled as a single plane with a material that creates the sum total effect of the glass. If you aren't taking luxurious tours illustrating the thickness of the glass and the front face reflections offset from the delicate sandblasted grain of the back face... why bother creating it? If you do need one side to be glossy and one side no, then use a double sided material on a single plane. Both sides identical but one glossy, one not.

 

Bear in mind... this is all coming from a guy who has yet to make a version of the light wall at Barcelona Pavilion that he likes.

[SgWRX]

yeah, that seems to work as you'd think! glossy reflection on one side of the plane and blurry refraction. right the only time you'd need to do this is thick glass hero shot. but even then, it's still going to have separate glossy reflection/refraction values. if i were some type of animation where you swing around from one side to the other i'd have to do blurry refraction and blurry glossiness as another material.