Well, I wanted to spice up my last render with some nice bokeh effects, so I set up a polygonal, slightly biased (0.3) aperture in the VRayCam. Turned out a bit funky in the end.
I’ve never seen bokeh refracted. Like the calculation order of the dof rays was done before the refraction, but this is all too odd. I really don’t know whats going on here. Take a look at the attached image, it is an overexposed part of my klein bottle, where you can see the out of focus bubbles bokeh refracted, and distorted.
Adaptive DMZ with adaptation value of 1, and a 1-50 min max rate was used to deal with all the sampling.
Not sure what the problem is. Are you saying that the bokeh effects through refraction look wrong? The only thing I see is maybe the refrected bokeh is too dim in relation to the non-refracted. Maybe render element the refraction, and use it to pump up the refraction bokeh a bit.
I don’t think the “shape” of the bokeh should be distorted if it comes from a refracted object, makes no sense imo. Can’t reproduce this also with my camera irl.
after a bit of research this does seem correct to me. As an example :
Check the Green circled area. Looks to me like it´s the same bokeh highlight. Once Original and once refracted. For more pictures using different lenses check:
It looks quite correct to me as well; for example, imagine that you have a transparent (but not refractive, i.e. IOR of 1.0) object, which itself produces bokeh, and through it you see another object which also produces bokeh effects. You would expect to see bokeh from both object, right? Now make the transparent object refractive… same thing.
You guys were ritgh! This is how it’s supposed to look like irl. I’ve had some time, and reproduced pretty much the same setup on my desk. As you can see, the bokeh in the glass behaves exactly like the one in the render!
The shape of the glass compensates for horizontal focus, and this is why I see the bokeh “distorted”. Cool.
This was not the first time, when a rendered phenomenon inspired me to look a bit deeper into reality itself. Fascinating.
Like the calculation order of the dof rays was done before the refraction, but this is all too odd. I really don’t know whats going on here. 
