Near infrared polarizer has arrived!

[Andy Perrin]

I got a near infrared polarizer! It was on eBay for a very cheap price (it is quite old) so I snapped it up and tried it out today. The wavelength range is 780-1250nm, the extinction ratio is 1:1000, and the peak transmission is 20% or so (bleh). But it doesn't have much wavefront error so it does okay for imaging, unlike the Thorlabs ones.

 

 

Here we have a favorite test subject, St. John's Seminary and Chandler pond. The filter is the 980BP10 filter for the dark water effect. I stacked it with the IR polarizer and stuck the whole thing on an old Tiffen polarizer ring, to allow rotation. The camera was the converted Sony A7S, the lens was the 80mm/5.6 EL-Nikkor (metal). Exposure was F/11, 2.5", ISO100.

 

I took three images at 0, 45, and 90 degrees (approximately) and put them in red green and blue channels after alignment in Photoshop. I white balanced the result and adjusted the saturation a bit by transforming to the L*a*b color space. Then I transformed back to RGB.

 

"Red"

 

"Green"

 

"Blue"

 

False color

[Stefano]

I guess normal polarizers don't work that well in IR. I have some 3D goggles, I never tested them in IR but they are opaque in UV (are the "lenses" made of polycarbonate?). So for example in this shot we can see that foliage reflects light of all polarizations evenly?

[Andy Perrin]

Yes, that was what I observed, the foliage was unaffected by the angle of the polarizer. The roof, on the other hand, changed dramatically. I was surprised the water didn't change more, but I think that's because the wind was making the surface diffuse, and randomizing the polarization somewhat. I will have to try again on a calm day.

 

I guess normal polarizers don't work that well in IR.

Most don't work at all.

[bvf]

So how do you interpret the final image?

 

Does white mean no polarisation?

 

Does red indicate polarisation in a particular plane?

 

What is the roof made of?

 

Why are the clouds blue? I wouldn't have expected them to exhibit any polarisation.

[Andy Perrin]

Ignore the clouds. They were moving rapidly so the color is irrelevant.

 

Red/blue does probably encode information about the plane of polarization relative to the viewing direction. Gray/white means that it was independent of how much I turned the polarizer, so therefore must be unpolarized.

 

I have no idea what that roof is made of but it seems almost metallic.

[bobfriedman]

shouldn't a regular (normally visible light) polarizer work in IR?..

[Andy Perrin]

Nope, they do not. If you have one you can easily verify this. (Keep in mind I’m using this at 980nm, so quite far from the visible spectrum. It’s possible they do work in the 700’s.)

[ulf]

Nope, they do not. If you have one you can easily verify this. (Keep in mind I’m using this at 980nm, so quite far from the visible spectrum. It’s possible they do work in the 700’s.)

That is my experience too.

I would suspect that most modern polarisers for VIS are based on similar manufacturing processes and see the same functional limitations, regardless if they are made for camera filters or sun-glasses.

The later type sometimes also have a layer of added tint, do give them a brownish tone.

 

I measured my brown-tinted polarising over-goggles and got this transmission:

 

I interpret the different wavelength ranges like this:

• <400nm, offset due to internal crosstalk in the spectrometer, no real transmission.
  
• 400nm-450nm, increase due to low extinction of blocked polarisation.
  
• 450nm-650nm, normal working-range, with reasonably good extinction. Upward-slope is due to a brown tint filter, (sun-glasses).
  
• 650nm-800nm, increasingly worsening extinction. At the end almost none left.
  
• >800nm, losses mainly due to surface reflection. No meaningful polarisation.
  

[dabateman]

Most polarizers don't work in UV or IR. There are some rare ones that do.

 

Andy, you said to ignore the clouds. But that was the only thing that surprised me in your image. Look at the "Red" image. You have a black cloud, top center. Then in "Blue", or 90 degrees polarization, its bright. Thats really odd.

 

As liquid water will be black or dark with your main filter. But ice water will be transparent or clear. I think with your polarizer you are able to catch these two phases of water in clouds.

[Andy Perrin]

I said ignore the clouds because they moved a LOT between shots. No guarantees it’s the same cloud even! Also, pay attention to how I processed the image - I did an L*a*b colorspace transform and then transformed back later. I think image subtracting is giving rise to the weird darkness in red/blue. It’s not meaningful and you should not try to interpret it.

[eye4invisible]

I really like the colour composite, and it's a really nice scene in and of itself.

 

Might be interesting to shoot some video whilst turning the polariser to see the results real-time.

 

There isn't much foliage to really say for certain would be affected by the turn of the polariser. Perhaps in the summer, you can re-run that test?

[bostwickenator]

If you are repeating this I suggest you take an image at 135degrees offset as well. With that additional frame you can reconstruct the stokes parameters and generate a true angle + degree of linear polarization image. I can post examples and the scripts to do so if you are interested.

[Andy Perrin]

It's a good idea. I can write my own scripts. I prefer MATLAB to Python. Thanks, though!

[dabateman]

You may also have better luck with old linear polarizers in IR.

I ordered an old cheap linear polarizer to take apart to use as a double ring to mount on my microscope. I quickly scanned it last week and was surprised that it might work through the full IR range.

I will need to rescan and test to be sure. Now I may need to order another cheap ring for my microscope.

[Andy Perrin]

dabateman, my polarizer is actually designed for IR, it isn't an "accidentally working in IR" kind of thing. (Also, it is linear.)

[Stefano]

Basically you can simulate every image taken at a certain angle of polarization? You can "fill in the gaps"?

[Andy Perrin]

Yes that’s his idea, I think.