Frosted glass

[Peter Milner]

The frosted glass material here is created solely by adding a fine sand texture.

The glass partition is modelled as solid.

The background image shows through the frosting clearly without being broken up and also shows through in areas where it shouldn't be visible.

I shall see what happens if I use blurry transparency instead.

[Peter Milner]

Blurry transparency gives me exactly the same problem.
It also makes no diference which rendering engine I use.

[Roy Hirshkowitz]

I'm not 100% sure of what you mean, but planar backgrounds have significant limitations. The biggest one is that they do not refract. What you get for background information is always the pixel coordinate-- regardless of the direction the material sent the refraction ray. Spherical bgs will work better for these.

[Peter Milner]

I was trying to set this up as a HDRI planar background.
If you look on the left side of the image you can faintly see the trees through the frosted glass even though there is a wall beyond the glass.

[Roy Hirshkowitz]

Like I said it's a limitation of these. What's probably happening is that a ray which is striking the translucent surface is transmitted off in a funny direction and ends up striking the background. Once that happens-- the background pixel is located by its original pixel number-- which is clearly wrong due to the refractive change in direction.

If you were to place a glass sphere somewhere in the scene you can absolutely get the same error-- any refraction rays that happen to strike the background might make it look as if there is a "hole" in the image through to the background.

[Peter Milner]

So it looks like I'll have to resort to adding backgrounds in Photoshop and then blurring the bit behind the frosted glass.
Shame.

[Roy Hirshkowitz]

It might be a shame-- but there isn't really a way to do this. These backgrounds do not exist in 3d space and it's meaningless to ask what portion of it a refracted ray strikes. Cylindrical backgrounds might work for you-- provided the ray strikes somewhere on the bg-- I think it's black if it misses.

BTW-- alpha mapping will have some problems as well WRT frosted glass-- not as bad though.

[Peter Milner]

Here's another problem.
I've frosted all the glass in this image, but notice that the reflection in the top of the storage and desk is ignoring the frosting and showing the window mullions beyond.
This is the same using either texture or blurry reflection.

[Roy Hirshkowitz]

Not sure I get this-- you may need to point out the area in question more precisely. It's very unlikely that reflections are "ignoring" either bump or blurry trans-- these are just ray tracing fx and will apply equally to eye rays or reflections rays unless there's some very weird bug in there. However, there could easily be some perceptual difference between reflection and transmission if you're using planar bgs-- these two types of rays will see different backgrounds for sure.

[Peter Milner]

This has nothing to do with background images.
Here is the same model rendered with clear glass. The reflections circled here should not appear in the first image as they are obscured by the frosting.
In case it's not clear there is another room behind the glass.

[Roy Hirshkowitz]

I'll try to put together a similar setup here to see if anything's wrong. You may be able to nail down whether there's a problem or not at your end by substituting a mirror for the desk material....

[Peter Milner]

I've nailed the problem down as you can see from the attached images. Everything works fine with a perfect mirror, but reduce the sharpness of the mirror reflection and it goes wrong. Notice also the difference in brightness of the reflection in the 3rd image.

[Peter Milner]

Following on from this, have a look at the two attached images.
The first has a mirror with sharpness set to 1.0. The second has a mirror with sharpness set to 0.99.
Notice how the brightness of the object reflections remains the same but the brightness of the sky reflection changes significantly.
Decreasing the sharpness further doesn't further affect the brightness of the reflected sky.