emulating a UV polarizer through computation

[Fandyus]

Haven't posted in a while, but I have discovered an interesting method for emulating a polarizer in any wavelength by stacking several images, so I thought I'd share it. I'll demonstrate it on this banana cross section.

 

All images were taken with a ZWB2+QB39 Tangsinuo stack, lit with 365nm LEDs.

 

 

As you can see, the surface of the banana is extremely glossy, and that is very distraction when it comes to the texture and intricate details of the cross section. You could get around it a little bit by either using a diffused light or having a long exposure and moving the lightsource around when the camera is exposing. Neither technique helps this very much though. Another way, that I use in the visible, is to use a white LED spotlight with a polarizer mounted on it, and then using another polarizer on the camera itself. Turning the polarizers in opposite directions then causes all specular reflections to be eliminated. I have demonstrated this on a lemon cross section in visible.

 

I however, do not own two UV polarizers, nor can I afford them. Even if I did, I am not sure how I would mount them.

As a workaround, I have figured out that taking four pictures with the light positioned in a different spot each time, then aligning them and stacking them in Photoshop with stackmode set to "minimum" will successfully fake the effect.

 

PIC 1

 

PIC 2

 

PIC 3

 

PIC 4

 

Stacked:

 

You are then free to edit the image as you wish:

 

For comparison, here's the same cross section in visible made with the analogue technique of using crossed polarizers.

 

This obviously works for any wavelength, so long as it has specular reflecitons.

 

[colinbm]

Lots of good experiments here.

[Fandyus]

7 hours ago, colinbm said:

Lots of good experiments here.

Thanks!

[Andy Perrin]

Yep, I mean it’s not really simulating all the effects of a polarizer (no pretty colors for example) but it is a good way to get rid of specular reflections.