Inventor claims he can make things invisible

news observer ^ | 4.29.03 | MARTHA QUILLIN

[Skywalk]

Yeah this is great and all, but can you fire while cloaked?

"Cry havoc and let slip the dogs of war...Our revels now are ended, Kirk"

[AmericanAge]

That is precicely the problem with a fiberoptic concept - the loss of intensity. Now, the human eye will compensate for that somewhat because we perceive light on a logarithmic, not a linear, scale; however, it would still be quite noticable; the more directions you include, the worse it will get, as more and more light hits tubes not designed for its direction. Furthermore, channelling *all* light through is physically impossible - even if you just consider one direction, if you're acceping all light into the fibers, that means that the fibers have to take up the entire width of the object (no space inside). If they need to curve outwards to allow for space, the area they curved into won't be invisible.

And, again, you have the "exponential" problem - when you double the density of sensors/fibers in one axis (let alone both x, y, and the xz and yz angular possibilities - a 4th power increase if you double them all), you double the number of fibers (or wires, for a sensor/LED concept) needed total, but their length doesn't shrink any - they still need the full length, so you don't gain anything by that miniturization. When you factor in the other axes necessary to have a 1-order visual improvement, you get a 4-order manufacture difficulty increase.

[AmericanAge]

1. No - pixels do not go on chips. Period. Pixel is short for "Picture Element". A pixel is a point of light. You don't put points of light on silicon wafers that route electric currents (a computer chip). Pixels, in this case, would be generated by LEDs. LEDs are far, far, far, far larger than the size of wires on a modern computer chip.

2. 5 million sample points (more specifically, CdS cells - Cadmium Sulfide) will not fit on the size of a chip. You don't print cadmium sulfide sensors on a computer chip, at least not in the normal sense. ;) You're using a utterly inaccurate analogy. Light sensors != computer chips. They can be *accessed* by a computer, they can even be placed on a silicon wafer (why you'd want to do that in the case of a suit is beyond me), but they are completely separate, much more expensive (per "element", I.e. compared to a gate on a chip), and much larger (again, same comparison). And again, you have to set up a networking system for communication between the CdS cells and the LEDs, which is an exponential problem.

If you want 5 million light sensors, 5 million LEDs, and the immense networking capability to route that, it would cost an utter fortune, if it's even possible with today's technology, which I doubt.

If you want high enough res for an urban situation, you're really going to have to rely on "natural" light phenomena, such as a fixed 3d picture on the suit that compensates for curvature and shading (a holography approach), or some sort of downconversion/transmissions approach. Anything that relies on individual elements is an economic nightmare, even as technology gets cheaper. Tech gets cheaper at a roughly O(N^2) rate (with CPU speed, ram size, and hard drive size, I know it halves once every 1.5 years; ram speed and hard drive speed are much, much slower in terms of cost efficiency). The sort of suit you advocate requires an O(N^4) rate for every single aspect of every single component. It's just not going to happen any time soon.

[lepton]

Please explain my 5 Megapixel digital camera. It is made by Olympus.

[AmericanAge]

That means that is has 5 million pixels. They are not stored on a "chip", in the sense that the word was used before (in a reference to the low-cost and tiny size that computer chips - i.e., microscopic switches - are manufactured to via lithography). There are light sensors (I didn't design the camera, so I can't tell you how many or what they're structured like) (unidirectional, of course). They might at best be implanted on a silicon wafer, but the process - and cost,size,etc - are nothing at all like designing and producing a CPU. The number of sensors very likely isn't anywhere remotely close to 5 million; there are a number of methods to have a set of sensors of a much smaller number read "in between points"; in fact, it is possible to design the camera with only one sensor which takes 5 million readings while the projected image is shifted around it. This saves a fortune in development cost; however, you have to have such a mechanism for each "camera" position if you want to apply this technology to the suit. Furthermore, you have to be able to network all of the data around it for the display. And you have to take directions into account, which exponentially increases the difficulty of the problem.

Again, while this is technically a feasible solution, don't expect anything good enough to work in a city with a cost that even the US could afford for actually deployable units, within the next 10 years. I would be surprised to see it in 15. Perhaps 20 would be reasonable, for a high-priced one.

And, again, even such a mechanism could be easily detected by other means, the simplest of these being a mere polarization filter.