Fecking 3D glasses: Why don't they work they way I expected?
August 2, 2010 3:05 PM   Subscribe

I thought I knew how the current crop of 3D glasses worked, but an experiment with my glasses and my laptop screen proved that something weird is going on. Why does the direction I'm looking through the glasses change what I see, and why do the polarising filters appear to be oriented the same way?

I've recently been to see a couple of 3D films in a UK, non-IMAX cinema. I was given a fairly boring-looking pair of shades to make the 3D magic work. My assumption, based on A-level physics and some fiddling during the boring bits of Avatar* is that the 3D works by simultaneously projecting two images using light polarised into different planes, then bog-standard polarising filters in the glasses ensure that each eye only sees one image. Simple.

However, I was playing with the glasses when I got home, looking through them at my laptop screen and the face of my LCD watch. From a vague understanding of optics and liquid crystals, and from trying the same thing with an old pair of polarising sunglasses, I was expecting each lens of the glasses to allow normal viewing when held at the "correct" angle, but black out the screen and watch face when held at the "wrong" angle.**

Instead, something really weird is going on.

(a) Holding the glasses between me and the laptop screen, looking through them in the normal direction, slowly rotating them clockwise:

0, 90, 180 and 270 degrees rotation: No obvious change.
45, 225 degrees rotation: strong blue tint to screen (but red areas of screen still look red)
135, 315 degrees rotation: strong red tint to screen (but blue areas of screen still look blue)

(b) Holding glasses between me and the laptop screen, looking through them backwards, slowly rotating them clockwise:

0, 90, 180 and 270 degrees: Slightly dimmed
45, 225 degrees: Screen completely blacked out!
135, 315 degrees: completely clear

(c) For the conditions in (a) and (b), both lenses behave identically. So when one is giving a red tint, the other is too. And when one is blacking out the screen, so is the other.

So, to summarise:
(a) Looking through the lenses forwards changes the colour tint but not light intensity;
(b) Looking through the lenses backwards can black out the screen but not change the colour tint;
(c) Whatever effect is causing this, the lenses are oriented the same way.

Optics wonks of metafilter: Srsly, WTF? Why does light's ability to travel through the glasses depend on the direction it's going? And how can the 3D effect possibly work if the two lenses are aligned the same way?

To pre-empt an obvious question: I have used this pair of glasses in the cinema and they seemed to render the 3D effect very well. Either these glasses have been built correctly (otherwise my first assumption would be that a manufacturing defect had alignined the lenses incorrectly) or I'm susceptible to a 3D-placebo effect to an astonishing degree.

*Almost the entire film. Still, it was very pretty.
**I can't reproduce this now to test this memory, because that old pair of polarising sunglasses is currently sitting at the bottom of the Solent. I had to remove them to read my diving computer's screen without dislocating my wrist and a stray bit of spray washed them overboard. Damn physics.
posted by metaBugs to Science & Nature (7 answers total) 2 users marked this as a favorite
According to Wikipedia, the glasses are circular polarizers. I'll don't exactly understand circular polarization myself, so I'll let someone else try to explain it.
posted by mbrubeck at 3:19 PM on August 2, 2010

Is it possible that you are seeing the effects of circular polarization? Sony's RealXLS glasses use this, and it reacts a little differently than typical polarized lenses.
posted by quin at 3:21 PM on August 2, 2010

Wired on Avatar glasses

Suite 101 article

I'm going to guess it has something to do with light polarization, but I'm not an optic wonk so I'm probably wrong. The glasses are probably trying to do their job of filtering light for just a particular eye to process.

I'm blind in my left eye. When I put on a pair of the glasses and looked in the mirror, I could see my blind eye clearly. I had to make friends cover one of their eyes to get the same transparent effect.
posted by beardlace at 3:21 PM on August 2, 2010

Those links on circular polarisation are very useful, thanks. Based on those, I think I understand some of what's going on.

My new hypothesis is that these glasses have two layers. The outer layer is a quarter wave plate (converting circularly polarised light to linearly polarised), and the inner is a linear polarisation filter.

So two images are projected at the screen, one using left-handed circular polarisation and one right-handed. Then e.g. the outer layer of the left lens converts LH circularly polarised light to linearly polarised light in an orientation that can pass through the inner (linearly polarising) layer. The same outer filter converts RH circularly polarised light to linearly polarised light in an orientation that cannot pass through the inner polariod layer. The right eye works just the same, but has an outer wave plate that converts RH-polarised light into linear light that will get through the inner layer, while converting LH-polarised light into linearly polarised light that won't get through the inner layer.

So, when I look through the glasses the right way:
(i) Linearly polarised light comes out of the screen; and
(ii) passes through the quarter wave plate at the front of the lens and become circularly polarised; then
(iii) some proportion of this circularly polarised light passes through the linearly polarised filer unimpeded. Because the polarisation is now circular, the orientation doesn't affect the proportion of light that makes it through (although there seems to be some wavelength-dependent effect).

When I look through the glasses backward:
(i) Linearly polarised light comes out of the screen; and
(ii) hits the linear polarisation filter first; which
(iii) blocks the light unless the glasses are held at the correct angle. All light that makes it through the inner layer then passes through the quarter wave plate and becomes circularly polarised, but that doesn't matter because it's not an effect that I can perceive.

And the glasses appear to be aligned the same way because the only effects I can see are results of the linear filters, which are aligned the same way because they both need to be in the orientation to correctly allow/block light from RH- and LH-quarter wave plates.

Am I close? I can't be dead on, because my hypothesis doesn't explain the colour tint effect.
posted by metaBugs at 4:06 PM on August 2, 2010 [1 favorite]

Good article from Wired but for whatever reason inaccurately describes red/green lens 3D, which is really so simple -- the old red/green gasses are for black-and-white movies, while the gray polarized lenses are for color. In the 1950s, movies were made in both formats but the red/green is so obviously second-rate, it's fallen out of favor, except for the occasional novelty. Originally those polarized glasses just had vertical lines on one lens and horizontal lines on the other, now things are much more complicated with circular polarization, but you're right -- one lens is CW polarized and the other, CCW (counter-cockwise).
posted by Rash at 4:28 PM on August 2, 2010

Tangential, but for a bit of fun put on your circularly polarized glasses and look at a mirror. The effect is caused by the mirror flipping the direction of polarization (from CW to CCW and vice versa). So neither eye can see itself but can see the opposite eye.
posted by 6550 at 9:43 PM on August 2, 2010

I figured this out with a couple pairs of 3D glasses from some movie theaters in the southeastern US last summer. Each lens of my glasses had two layers: on the outside, a circular polarizing layer (different circular polarizations for each eye) and on the eye-side, a horizontally polarizing layer. So when you wear the glasses, circularly polarized light is only admitted to one eye or the other, with the same polarization selected regardless of how your head is tilted. (This is the advantage over 3D glasses with perpendicular linear polarizations, which I used in some (IMAX?) theater probably ten years ago: you can mix or swap linear polarizations by tilting your head.) But because of the second layer, the light that actually reaches your eyes is horizontally polarized. And if you wear the glasses backward, they select for horizontally polarized light, but the light that comes out the front of the glasses is circularly polarized.

So, your comment suggests we have the same sort of glasses.

The reflection off a shiny floor or a body of water is horizontally polarized, if you want something more reliable to test on. Garden-variety polarizing sunglasses admit only vertical polarized light (the "cut the glare"), so if you wore your 3D glasses on top of a pair of sunglasses they would combine to make an opaque layer. I don't know of any kitchen-physics source of circular polarized light.

The light from a light-emitting diode is strongly linearly polarized. Laptop displays are usually oriented with the polarization at 45° one way or the other. I suppose this is so no sunglasses-wearing customer complains that setting their display sideways shuts it off.

The color tint happens because a quarter-wave plate is wavelength-sensitive device: all other things equal, a quarter-wave plate for 700 nm (near-infrared) light is a half-wave plate for 350 nm (near UV) light. The circular polarizing films (Edmund Optics sells them as "retarders", much more expensive than their polarizers but still cheap at $2/ft2) produce light with roughly uniform circular polarization over the whole optical spectrum. I don't quite understand how that works but I'm prepared to believe there is some trick, where another optical property of the material changes with wavelength in such a way that the same material thickness serves fairly well for all colors. Maybe that's described in this 2006 patent, or maybe it's just not as big an effect over the visible spectrum when 90% polarization is enough to be useful. Last year I found polarization-vs.-wavelength plots from a couple distributors, but I'm not finding them tonight.

6550 points out a neat effect, also illustrated by Anchor Optics in their catalog. Circular polarized light carries angular momentum (in quantum mechanics, it's "photon spin"). Reflection from a plane mirror changes the direction of the light but not the direction of its spin, so the circular polarization is reversed on reflection. If you wear your 3D glasses in front of your bathroom mirror and close one eye, your open eye sees the closed eye in the mirror but the reflection of the open eye is blacked out. Switch eyes and of course the blacked out lens switches.

Now if your bathroom has two mirrors, you can arrange them at 90° to each other and look at your reflection's reflection -- the image in the center of the two mirrors that's not left-right swapped. The second reflection switches the circular polarization again, so in that image you'll see only your open eye.

Notice that this suggests if you send very bright circularly polarized light into a rigid mirror construction that reflects it twice, or to a device made out of quarter- or half-wave plates, you're transferring angular momentum to the mirror and it should eventually start to rotate. This was first described in a classic paper by Richard Beth in 1935, if you like that sort of thing. Unfortunately I haven't thought of a kitchen-physics way to do that experiment, but it would be fun to try.
posted by fantabulous timewaster at 1:53 AM on August 3, 2010 [4 favorites]

« Older Can't talk -huge rubber thingy balanced in my ear.   |   Refrigerator help Newer »
This thread is closed to new comments.