h a l f b a k e r y
A riddle wrapped in a mystery inside a rich, flaky crust

meta:

account: browse anonymously, or get an account and write.

 user: pass:
register,

Better invisibility suit.

---> <----
 (+3) [vote for, against]

So, the problem is to make an invisibility suit, based on the long-known idea of cameras on one side linked to displays on the opposite side.

For the sake of simplicity, we could aim first for an invisibility cube. You want to be able to put this down somewhere, and have it "invisible" no matter what direction it was viewed from.

The first problem with the camera/screen system is that the cameras themselves will be visible, meaning that you can have front-to-back visibility but not vice versa.

However, modern cameras can be almost arbitrarily small, so that you could design a screen with a camera peering through an almost-invisible dot in the middle of it. (The cameras on the back are linked to the screens on the front, and vice versa).

Nevertheless, this still leaves the invisible cube appearing like a collection of small black dots (the camera holes), which is less than ideal.

BUT

(this is the important bit) LEDs act as photodiodes as well (just not all that efficiently). You can try this yourself by putting a voltmeter on a single LED and shining bright light on it.

So.

It should be possible to create an LED-based display which can also work as a camera. The LEDs would be activated (emitting light) for perhaps 1 millisecond, and would then be turned off and instead used to detect incoming light for a further millisecond. The eye will see continuous illumination. (end of the important bit)

However.

This won't work, because the individual LED elements will detect light coming from all directions - so they won't work as a camera.

But however.

The answer is to construct the device like an insect eye, only more so. Put each pixel (each LED) at the bottom end of a tiny tube (the tubes are orthogonal to the screen - ie, they point outward). Now it will only see light coming at it square-on to the screen. Collectively, all the pixels on the screen act (when used in "sensing" mode) as camera, with a "parallel focus" looking straight out of the screen.

If you now link two of these compound-eye- cameras/displays back to back, you will have a cube which is completely invisible when viewed from directly in front or directly behind (another two pairs of screens make it invisible from the sides, top and bottom also).

But however nevertheless.

All of this will only work for viewers standing squarely in front of (or behind etc) the cube. Anyone looking from an angle will see a black cube: the LEDs are buried at the bottoms of their tubes. Even if this were not the case, the image they saw would be "off" - at the wrong angle.

But so nevertheless however.

Now we make it a bit more complex. So far, each of our pixels is just a small tube with an LED at the bottom (acting as both camera element and display element).

Now imagine that each pixel is actually a little cluster of tubes, all radiating out like the spines on a sea-urchin. Call each of these tubes a "sub-pixel". If each pixel consists of 20 subpixels (pointing in 20 different directions), then the "invisibility" will work perfectly from each of those 20 angles. If the viewer is standing at, say 45 degrees to the face of the cube, then they will see the image which is seen by the sub-pixels which are looking at 45 degrees away from the opposite face. Which means the cube will be properly invisible. The more sub-pixels there are to each pixel, the more angles the cube can be viewed from and remain perfectly invisible.

But.

I hear you say "if each pixel is like a sea-urchin, you won't be able to pack them together, and the image will be dotty" (did you just say that? I think you did).

Aha! Say I. Not so. If you design your sea-urchins cleverly enough, they will pack very closely, with their spines intermeshing nicely.

Of course, if each pixel has 20 sub-pixels, this means that light from any one direction will only be detected with 1/20th the efficiency; but no problem - that's good enough (the electronics are sensitive). More worryingly, the projected image will only be 1/20th as strong, and appear on only 1/20th of the surface. But again this is not a problem. Each sub-pixel can be made arbitrarily small (so you don't notice the dottiness), and the brightness can be racked up 20-fold to compensate. After all, the monitor you're reading this on has significant black gaps between the pixels, and you are not aware of those.

(You could also do the same sort of thing with a very dense flat array of LEDs and a series of micro-lenses, as I think are used in some displays which give an illusion of 3D, or like those embossed plastic cards which show a moving image as they're tilted; but the sea-urchin business is perhaps easier to understand. Either way, the resolution of the image will be 1/20th the pixel density, if each pixel looks in 20 directions. Current pixel densities are maybe 100 per inch, giving a displayed resolution of 5 per inch - good enough to hide a big box when looked at from a few yards away.)

So - Gadulka! - a video-based all-round unvisibility shield is doable.

 — MaxwellBuchanan, May 28 2011

 Interesting idea. There will be issues regarding spectral sensitivity of LEDs used as light sensors, and color gamut transmission for arbitrary scenes.

 But could be lots of fun at parties.

I want to see the closest-packing math for sea urchins!
 — csea, May 28 2011

I can see you wearing this, [MB].
 — infidel, May 28 2011

 [csea] colour sensitivity (in 'camera mode') will be correctable to an extent by the electronics. However, it would probably be necessary to develop LEDs that had better sensitivity across the spectrum; doable. (Crudely speaking, the colour sensitivity of an LED acting as a photodiode is related to the colour it emits.)

 Colour representation (in 'display mode') should be good enough in most cases - a popular trick is to display, on your monitor, a photo of the scene behind it, to give the illusion of a "window" - the illusion works quite well.

[infidel] you only _think_ you can see me. I'm actually over here.
 — MaxwellBuchanan, May 29 2011

No, I can see YOU wearing this... whatever.
 — infidel, May 29 2011

[MB] Good enough re LED sensitivity, but you didn't address the critical issue of sea urchin packing.
 — csea, May 29 2011

 Ask a sea urchin. But, it's not that critical. Suppose that the LEDs can be made small and bright. Then, they don't need to be packed very densely in order to create an image that looks contiguous.

As an indicator of how brightness can overcome packing density, remember that your monitor can display a "solid" blue colour, despite the fact that only one third (at most) of the screen is emitting blue light.
 — MaxwellBuchanan, May 29 2011

Brilliant. [+] But, what about ambient light issues? Will LED displays be bright enough in daylight, outdoors (where you presumably would use it)?
 — Boomershine, May 29 2011

That's certainly a factor, but LEDs are getting brighter all the time. I think it would be difficult but perhaps possible. You might have problems if the object to be invisibulated were between you and the sun.
 — MaxwellBuchanan, May 29 2011

 LEDs are already much brighter than a perfectly white object in full sunshine. Some of them now come with warnings of the danger of eye damage.

 //LEDs act as photodiodes as well (just not all that efficiently). You can try this yourself by putting a voltmeter on a single LED and shining bright light on it.//

 I finally got that to work. It seems you need a good meter; the output is a few microvolts. You can get a much better response by using the 'diode' setting, with the LED reversed. The voltage reading is then rather responsive to light levels. It seems to work best with a modern, superbright LED. The response is supposed to be highly linear, which means you can have a high precision quantum photometer for about 10 cents!

 There exist highly directional LEDs, with an emitting angle of a few degrees. The best ones have such precise focus that they project an image of the LED chip itself onto a surface. These would be more efficient than putting LEDs at the bottom of tubes.

The neat thing about using these is that they ought to have precisely the same angular distribution of detector sensitivity as emission brightness.
 — spidermother, May 30 2011

 [annotate]

back: main index