Slashdot Mirror
Polymer 'Muscle' Changes How we Look at Color
New Scientist is reporting that in the not-so-distant future computer monitors, and televisions may utilize a color changing polymer that responds to a current instead of existing techniques. From the article: "Aschwanden and colleagues built arrays of 10 pixels, each 80 micrometers across. The pixels consist of a piece of polymer covered with ridges tipped with gold. When white light is shone at the polymer from one side it reflects out of the screen and is also split into different wavelengths by this 'diffraction grating'. However, a slit above the polymer ensures that only one wavelength of light escapes, giving the pixel its color. The pieces of polymer also contract in response to current, like simple muscles. As they do so, the fan of light-waves is moved, changing the color that is fed through the slits above and out of the screen. Cutting the current causes the muscle to return to its original state."
I like the idea of reducing our current RGB model to a "true pixel" technology, because it will make displays smaller, sharper and more. But as far as I understand our vision system is itself based on a sort of RGB sensor and the human eye is not really capable of seeing e.g. orange, which is why the whole RGB (and CMY) display technology works in the first place. There are some high range displays (at least in research facilities) giving you a larger dynamic per color than the 256 scales of traditional 24 bit images, so the lack of "true colors" mentioned in the article might be solved by conventional technology.
But what about the use for data transfer over fiber? One of the nice things about fiber is that you can send several "colors" in parallel which will not disturb each other, something impossible with copper. Up till now they use laser diodes with a fixed wavelength, so the number of diodes determines how many parallel signals you can send.
Now there is a technology that can create any wavelength. Combined with matching optics, could one not use one of those polymer displays to create multiple wavelength signals and send them through one fiber, in theory allowing an indefinite number of signals? Still limited by the number of pixels on the display and the accuracy of the sensors on the other side, but much easier than to arrange several thousand laser diodes.
[Just speculating, no real clue about optics.]
memomo: free web based language trainer DE-EN-ES-FR-IT
"Man - the color on your monitor packs a real punch!"
For certain applications. It's my understanding that usually the synthetic muscle stuff isn't particularly speedy in changing shape. My first question is how many flips per second can you get? Are we aiming for TVs or variable paintings? My second question is about power requirements. 300 volts, sure, but are we talking amps or microamps?
Neat, as most science is, but possibly not terribly useful.
So we have a "pixel" that can be truely any color. Does it mean "any" color, as in Hue, or can it truely be of anything (ie. full spectrum output; Image of fluorescent light would have spiky spectrum, etc.). If the former, instead of RGB we can simply transmit HSV (Hue-Saturation-Value(Brightness)), but if it's a continuous spectrum...
Instead of transmitting just RGB values from 0-255 (24 bits) per pixel, instead you have to somehow convey the entire spectrum. At what resolution do you get? Instead of three values (R, G and B) do you get 400 (one per nanometer, from 300 to 700 nm?) - or 4000? What kind of format do spectrograms use?
Anyway, consider transmitting data from a spectrogram - times some standard monitor resolution - for multiple frames per second. That's a lot of uncompressed data.
It sounds in some ways like those variable mirrors (TI I believe) some rear projection TV's use. Combined with a prism. Certainly an interesting approach to the problem. As for your question I believe it would have to change at least at 60 Hz or better.
...Sweet! That's almost as good as the camera on my cell phone!
It reflects white light. It works like a glass prism. Do you see people working with prisms getting microwaved to death or skin cancer from UV? There's your answer.
At the time you put it on a real product, it makes no difference (maybe outside the price) if you have an array of leds or a device capable of emiting any frequency. You receiver won't be able to read on a perfectly sharp spectrum, and light will scatter on the fiber, adding noise to the frequencies close to the ones you are using.
At nature, you never have infinite precision, so anything you do can be discretized.
Rethinking email
Instead of transmitting just RGB values from 0-255 (24 bits) per pixel, instead you have to somehow convey the entire spectrum.
.red, blue or green.
It's just a tunable filter with a default value. That default value could be. .
The filter is "tweaked" by sending it another value, say, one between 1 and 255.
KFG
Wouldn't a display built like this be able to display fewer colors than a regular one? In particular, I don't see any way to control brightness or saturation with this setup.
it brings a new meaning to the term "Muscle TV".
The higher the technology, the sharper that two-edged sword.
how much can your monitor bench maaan~!
*ducks*
VLC FOR MAC IS DYING! IF YOU DEVELOP, PLEASE SAVE IT!!
Cutting the current causes the muscle to return to its original state.
Depending on what you're watching, that's a lot like regular TV.
Push Button, Receive Bacon
The ability to generate any visible light frequency would not only extend the gamut to the full human range (unlike other schemes, like the 6-color Iridori system presented at SIGGRAPH 2004), but it would also allow tetrachromats to enjoy television and computers much better (this issue was discussed previously on slashdot).
Of course, as the article suggests, they will still have to use multiple emitters per pixel, as it can only generate colors on the edge of the CIE Color Space (warning, you can't see what colors they are, because your monitor cannot display anything outside the RGB Triangle). And of course tetrachromats are rare but have been found.
I For one Welcome our new Polymer Muscle Changing TV overlords
People often ask what's the point of 64-bit processing - who benefits? Well, the one thing that would probably be worth having is system wide 64-bit color - as long as the display technology could handle it. Maybe this is one way of achieving that.
Any technology dependent on gold is always going to be expensive even if a finished screen only requires small quantities. A gram of gold is around $20 today - at least it's cheaper than coke but it's still expensive.
spoonerize "magic trackpad"
But how does that produce any more colors than a traditional screen? I thought the entire point was to escape the limitations imposed by the RGB standard - with non-integer red green and blue values between 0 and 255 the number of possible colors increases to infinity, assuming any decimal value is fair game, as far as I can tell. Is this correct? Someone please explain. I don't understand your fancy modern displays or your color definition spectrum, I'm just a caveman!
They're taking Albert Michelson's work to a whole new level.
Of course we can only see in four colours (Red, Green, Blue, and "night vision"), but our sensors have broad, overlapping peaks of sensitivity, thus green will also evoke blue and red. However, blue-green will evoke considerably less red (less than half of what green does), so a real blue-green emitter will give us a colour not deliverable with existing technology. Screens should be sharper as a result, through less polluting cross-colour; not only will black be very black, but Indigo will be very Indigo...
Wikileaks, no DNS
Until it can do World of Warcraft, does anyone really care?
</sarcasm>
800 volts?!? It certainly sounds like they've got a long way to go before we see this tech bloom.
... will it show "Snakes on a Plane"?
...it make for a display that would probably be too dark most of the time and have a narrow viewing angle, the sort of thing they had to address with LCD displays (and they've already succeeded, so...). Passive displays (kind of like paper) are a neat idea though.
Wouldn't this be a good solution to a lot of the problems currently faced with active camouflage?
"My only suspicion is they design them to burn-out on purpose in order to achieve residual income. Standard light bulbs illustrate this perfectly. Some of the earliest light bulbs ever made are still running. There are light bulbs that have been running for 100 years and still going strong, so why can't manufacturers produce that level of quality today? Because they don't want to"
Um, no! Now do I have to spoon-feed you the answer, or are you going to do the research you were suppose to?
This is definitely an interesting technology, but I still think that display technology really has a long way to go before it can be used for general purpose things such as books and the like. The one thing that bugs myself about monitors/lcd's is that they always require a backlight or some active light source to function -- which IMHO really can bug the crap out of your eyes. What if they could invent a technology that didn't require a light source to be seen, but just reflected whatever existing light.. like a normal surface. Not only would this sort of thing definitely bring down the strain factor, but it would seem like you're reading a book or paper or something. Just a thought.
Obliterate advertising!
Sources?
Sounds to me like this system could be a potentially useful miniaturization step for anything with a diffraction grating in it--OPOs spring to my mind. (Oh well. I got out of optics so I could _stop_ thinking about crap like this.)
I'm *assuming* you've experienced a prism. White light into an infinite gradient of spectrum.* Keeping your position still and moving the prism sweeps the colors across your vision. The "muscle" in their prototype is doing essentially the same thing. The "slits" act like filters. Kind of like what's in present displays. Each filter is essentially going to see the gradient for it's particular color (the intensity should be equal for all pixels). A color LCD essentially sees variable intensity (controlled by the twist) which adds up to a particular color.
Wikipedia has a good section on color.
*Not strictly true for some kinds of light sources, but let's keep this simple.
It doesn't produce more colors, but it would, in theory, produce qualitatively better colors and picture quality. The reason is that a single pixel on your rgb monitor mixes the three primary colors which your eye dithers (for lack of a better term, at least one that I'm aware of) into the intended color. This technology, however, would have each individual pixel actually turn into the intended color. If these new pixels can be manufactured at the same size as rgb subpixels, then you can essentially triple the resolution with the same number of logically addressible picture elements.
I didn't mean to suggest that would be the way the technology would be used. There are all sorts of ways to combine the elements to get intereting results. I just hoped to give you a simple idea of the technology.
One of the most obvious ways to use it is to send each pixel a number between 0 and 16.8 million and instantly triple the resolution, since you no longer need three pixels per color.
Want 256 times 16.8 million colors? Add 8 bits of memory for each pixel (which means you'll need four times the video memory, because each pixel is now addressed by all 32 bits, instead of 8 apiece).
Your digital computer is, well, digital. It doesn't actually know about decimals. Deep down it's just beads layed out on a board. Beads are integer. If you want numbers with more places you have to add more beads to represent that number. And that's the way it is.
256; 2.56; 25.6; All three place numbers. All eight bits.
KFG
The end of the "paper bag" era is coming.
This new tech could allow you to "paint" on the face of your favorite celeb to the girl you bring home; when beer goggles arent enough.
"*Not strictly true for some kinds of light sources"
:-p
Depending on exactly what you mean by "infinite gradient of spectrum", it's not true for any kinds of light sources, as photons only exist at discrete frequencies, the spectrum will be stepped into a finite number of colours... but yes the number's still big and this isn't sticking to the "let's keep this simple" thing either
The revolution will not be televised... but it will have a page on Wikipedia
You probably do not need a continuously variable spectrum for each pixel. A simple set of red, green, and blue primaries cannot reproduce the stimulus of all the spectral colours, yet they give a good enough representation of most scenes. This works because the eye-brain system transmits brightness, colour, motion detection, and other signals as firing rates in nerves. The nerves will typically have a significant background firing rate even for zero signal, so the system has to continuously try to calibrate itself, and work out what the zero and scale signals are. This is why we can look at printed images with a typical contrast ratio of 100:1 and a white point as set by the ambient light, and recognize a scene without worrying that the blacks look grey or the whites look coloured. Many illusions depend on fooling this feedback process. For example, if you look at a slowly moving object for some time and the look at a still scene, it may seem to rotate in the opposide direction because your motion sensors have adapted. Well, the same happens with your sense of colour contrast - that will adapt to compensate for the variations due to intensity. If you look at a dimmer version of an image, the colour difference signals are weaker but colours you see will look much the same (until you get down to mesopic light levels, and the adaption system begins to pack up altogether). If you are looking at an image in a darkened room, and the colours are 10% desaturated, you will probably not notice unless there is some other stimulus (such as a red power LED on your monitor) to act as an independent reference. It many seem that a three-component display can only get at about half the colour space within the spectral locus, but under typical viewing conditions, we are poor judges of colour contrast. If you want to make an image look more colourful, make it brighter. Get a slide projector and move it close to the screen so the image is small but really bright - you know the colours have stayed the same and only the intensity has changed, but you will probably find the colours a lot more satisfying.
There are other reasons for wanting to go for more primaries. You eye does not have uniform colour sensitivity: it will detect colours differently in the centre and in the periphery. The brain tries to remove this variation, as it is part of the eye not part of the image. You do not see this variation directly, but you can get to see it if you look at a large white patch on a screen where the left and right halves have different spectra. If you have an RGB projector with broad spectral primaries, this will give you a similar stimulus to a general reflection scene in the central and the peripheral vision, but you will not be able to get the saturated colors. If you have narrow band primaries, you will be able to get the deep reds, peacock blues, and violets you cannot get with the broad primaries, but you may have strange side-effects because your central and peripheral vision no longer match. make a projector with six primaries, and you could get the best of both.
But, is the extra effort really worth it? It is a bit like 3D - twice as much technology giving you a bit of extra stimulus that can startle, but can also detract from the nett visual experience. I would love one of these variable filters as a research tool, but I don't expect fully spectral displays any time soon.
It seems like top athletes will soon have another reason for possessing EPO, testosterone,THG,... ;) Sick mothers-in-law or dogs that for some reason need EPO don't cut it anymore these days. "No sir, I only used it to brighten up my TV".
How can a simple "Slit" allow for only one wavelength to pass through?
Wouldn't this pass all wavelengths shorter than a given wavelength which is proportional to the slit's width?
(This is more like a low-pass filter than a band-pass)
Sigs are for the weak.
The red, green, and blue light coming from my computer screen are chosen to give a broad range of colors for a normal human eye. Unfortunately, these colors suck for us color-blind dudes. A significant benefit of this technology may be improved color rendering for color-blind folks like me.
In my own eyes, the cones sensitive to red have a slight defect that shifts their sensitivity towards green. In the real world, some reds look black to me, since they don't fire the cones. Others (closer to green) show up nice and bright as red to me. However, TV screens and computer monitors always use a very deep spectrum red. If they did not, they would not be able to display that color, since higher frequency reds also fire some green cones in a normal human eye.
So, even though my world has red in it, my TV and computer never display any. This new technology could fix that, which would be very cool.
Beer is proof that God loves us, and wants us to be happy.
"In time, it may be more economical for publishers to publish e-paper based books than paper books (any arbitrarily large number of pages in the book being condensed to a single viewable e-paper display that can display any one of them), but I don't see that happening anytime soon."
If the movie/music/software industry taught them anything. They'll leave it in dead-tree form. Still not completely immune, but as people have demonstrated with the "see no problem with copying" story. They have no problem with abusing technology, and technology can never be a substitute for moral or ethics.
Continuous phenomena can be approximated, meaning that you can represent some useful set of values with finite precision. Currently your graphics hardware approximates the colour spectrum with three integer components ranging between 0 and 255. It could be better, but obviously it's functional enough and you don't need infinite storage or transfer bandwidth.
Seeing as HDR techniques are pretty much all the rage in graphics right now, I wouldn't be surprised if the pipeline were to go entirely floating point in the near future. That's to say, you would be able to represent a finite set of decimal values within very reasonable amounts of storage space. Current real-time HDR renderers use 16-bit floating-point components, which already provides quite nice quality.
if you can flick a single wavelength a million of times a second
or 10,000 wavelengths 1,000 times a second.. what has higher bandwidth?
every day http://en.wikipedia.org/wiki/Special:Random
In 2003, a group of researchers at the U of Toronto unveiled a prototype photonic crystal gel technology for electronic paper. I looked it up again, but nothing has been published on it since.
Read a preview of my novel CYBERCHILD at www.smartalix.com/cyberchild
what is this 'diffraction grating' of which you speak?
And pray, explain once again, how may I 'post to this internet' on my 'digital computing device'?
10000*1000 > 1000000 (by a factor of 10)
But the question is really whether your data can be decorrelated and correlated (or scattered across the wavelengths and gathered up on the receive side) across that many channels.
The important question in packet-based networking is when the last bit of a packet arrives, not when the first one does.
Let's make it 10 000 channels of 1000 modulations per second versus one single channel of 10 000 000 modulations per second, so the aggregate rate for the serial versus parallel link is identical (10 mbps).
If all your L3 traffic is expressable as 10 000 bits per conceptual L2 frame, then in both cases you get 1000 pps; one thousandth of a packet per "blink" on the serial link, one packet per "blink" on the parallel one.
If your L3 traffic is less than 10000 bits (and the link is not a bottleneck) then it takes one blink to send the L3 traffic across the parallel link no matter what the L3 packet size is. This means that your minimum link latency is one millisecond on the parallel link. The serial link's latency is a function of packet size; smaller packets take less time for the last bit to arrive; you save a microsecond per bit.
For packets that are larger, you get into situations where you have, for instance, 10001 bits to send across an otherwise empty link. On the serial channel, this takes you 0.1 microseconds more than a frame of 10000 bits. On the parallel channel, 10001 bits takes a whole millisecond (10 000 microseconds) longer.
This is certainly an effect which can be discovered end-to-end (akin to MTU exploration), and could be adapted to by the transmitter, however there are probably no in-use transfer protocols which deliberately look for step function increases in packet size to delay.
Moreover, when the link is a bottleneck (which is expected at every link across which TCP bulk transfers happen) and variable sized packets, scheduling becomes a problem, if you want to avoid the step-function from being a serious source of jitter. If you have multiple packet flows occupying the queue, you inevitably end up with some packets which will suffer from millisecond jumps in delay for the final bit at the receiver, compared to the serial link. While this is not an intractable problem, minimizing the incidence of partial-packet delay across streams of variable-sized packets is a difficult one, particularly as you scale from megabits/second aggregates to gigabits/second aggregates (one 10GHz channel vs ten thousand megabit/second channels), and as you increase the number (also the variability of number) and end-to-end throughput (also the burstiness) of independent flows.
This will be an important problem for Internet Engineering in the future, as there are difficult price challenges in 40Gbps/optical channel now, and difficult price and physics challenges in anything faster than that. Slashdot readers familiar with megahertz versus cores will probably notice an echo.
Also of interest (and stuck in the realm of routing and control plane research): with your serial channel, a failure anywhere in the path disables the whole 10 Mbps channel. With the parallel channel, you introduce a whole range of partial failures that wipe out one (or a handful) of parallel colours. It may be better for a network's users to continue to use even 1/2 of the 10 000 parallel channels (depending on things like traffic load, how a failure of the whole channel is dealt with at various restoration or routing levels, and so on).
Pseudo-parallel connections are even worse; there are a variety of situations in which the receive side detects simultaneously transmitted pulses of two frequencies at different times...
In short, massively parallel links in theory should be no worse than equivalent serial (or even "lightly" parallel) ones when it comes to goodput; however, we don't know how to do this in practice yet.
So, sadly, for real world Internet traffic, I would not be surprised if a 1 000 000 blink/second serial channel performs better than 10 000 parallel channels of 1000 blink/second each.
Filtering a narrow wavelength colour from a wideband source is never going to be power-efficient.
Most (currently-available) white LEDs for backlights have a dip in the green-cyan (500nm) region of the spectrum, and fall off in the deep red too. You can't reproduce colours which aren't in the backlight. Making wider band backlights would reduce efficiency further...
For people with normal colour vision, three-channel encoding is all that is needed, "negative" RGB values can be used to define colours outside the colour-space defined by the standard sRGB/Rec.709 colour triangle. These extended-gamut colours can already be displayed by putting R/G/B LED backlights behind existing LCD panels; going for 4- or five primaries can extend the gamut further still... but increases costs and decreases efficiency.
Sounds to me like the development may be novel and useful, but the display angle is mostly just some flimsy spin for the benefit of journalists and funding bodies!