What is Digg?81 Comments
- dsignr, on 10/12/2007, -4/+43Am I the only one confused about this? Wouldn't having a single-pixel camera just show one color? Is the intent to have multiple single-pixel cameras in one jumbo camera?
- Zippo, on 10/12/2007, -4/+40Sounds promising... I have to agree that a lot of cameras are just stuffing the megapixels in... and it doesn't really do much for quality. A good lens and proper settings make all the difference.
- JesusFaction, on 10/12/2007, -1/+31do they plan on using smoke too?
I wonder - SkippyDoorknob, on 10/12/2007, -1/+25Better hope your camera doesn't have a dead pixel!
- imtigger2, on 10/12/2007, -2/+18I agree... I have one super simple rule I tell friends and family when they ask "which one??". I always say the quality of the lens is very important, and the LARGER the glass, the better your photos will be (allowing more light and depth).
Even when I was shooting all film, I don't know how many times I'd recommend a quality film to someone, then look at their pics that were messed up because of poor focus, crappy dirty lenses and bad lighting.
Oh, and USE one of these.. will ya?: www.lenspen.com :) Then again, who cares about a good lens now days... everyone is shooting pics with cell phones that have 1/8" or smaller lenses, and they think the quality is acceptable. I guess it's the same people who think the video quality on youTube is fine as well. - Necoras, on 10/12/2007, -0/+15While there are zero details in the article (Kudos once again to the scifi blogs for their in depth research :P ) It actually does make quite a bit of sense. The expensive part in a camera is the sensor because it either a) has silver as it's main active component (like traditional film) or b) it has photovoltaics which change light into electrical data (difficult because most photovoltaics are only sensitive to IR)
Using mirrors one could bounce the light from an entire panorama onto a single sensor. Essentially you'd be scanning the picture... the reverse of a typical CRT television. Since light is so fast (weeee) you'd end up with a great photo (theoretically) and only need one sensor. - masamunecyrus, on 10/12/2007, -1/+14Basically, A normal camera takes in all of the light in through a lens and focuses it onto an array of, let's say 10 million pixels. However, because of all of the noise, compression, and other miscellany, it throws away 80-90% of the data it takes in.
A single-pixel camera like this, instead of taking one picture that is composed of 10 million pixels, would take 10 million pictures that are each composed of one pixel. This will create a much more accurate picture and won't have the same inefficiency problems noted above. - inactive, on 10/12/2007, -0/+12This should've been linked to the BBC article: http://news.bbc.co.uk/2/hi/technology/6263551.stm
It explains it A LOT more. The blog post confuses the terminology; these guys are creating a camera with a single-pixel SENSOR; obviously the resulting image will have millions of pixels.
'"Instead of using a million really expensive sensors, we can use one really expensive sensor and still give you a million-pixel image," said Dr Kelly. '
'As the light passes through the device, the millions of tiny mirrors are turned on and off at random in rapid succession.
Complex mathematics then interprets the signals assembling a high resolution image from the thousands of sequential single-pixel snapshots. '
EDIT: Darmichar figured it out before I posted this. - meepus, on 10/12/2007, -4/+16I got dugg down, but a single pixel camera IS science fiction. A single sensor camera however, is not. A pixel is an output device, not an input... unless you believe that old internets joke about monitors being able to take pictures.
- Darmichar, on 10/12/2007, -0/+10Just a wild guess, but imagine creating the image one pixel at a time.
The mirrors reflect one pixel of information for the camera to capture, shift one pixel, capture, shift ad nauseum. (This is done insanely fast.)
Actually a decent idea if they can actually get it to work.
That's a lot of faith to put into the mechanism controlling the mirrors though. - inactive, on 10/12/2007, -1/+10But a Pixel is just a picture element, not exactly a pixel as base on a camera...so the statement has some falsehoods to explain first.
As for digital cameras today, I work at Best Buy as some might know and I rarely ever sell people point and shoot cameras with more than 6+ MP due to the fact that many of them use tiny 1/2.5" sensors that are noisy and often horrid in low light.
I'm surprised that my Canon A80 is only 4 MP but it's a 'whopping' 1/1.8" sensor that some 10 MP cameras have.
Some of the best digitals out today are definitely Canon and Fuji, because they don't make large megapixel cameras without including an appropriate sensor to go along with it.
Sensors are packed with photosites which work like film's emulsion in the same sense that the larger these are, the more effective they can be.
Mostly it goes like this:
1. Image focused through lens
2. Image processed through sensor
3. Any type of noise reduction or colour control managed by sensor
4. Output of resolution is then created.
It's rather sad how Sony has the W50 and W70 which are 6 MP and 7 MP but on the same exact sensor.
My opinion of a great camera would be a Canon SD80 IS with a Fujifilm 1/1.6" sensor as found on the F30 - Sneedly, on 10/12/2007, -1/+7I agree that the article is not very clear about the actual implementation.
I think the idea is two "time-multiplex" a single sensor. Normally, the light comes through the lens and is sampled all at once in parallel. This device may serialize the sampling using mirrors to essentially slow the light until it can be sampled? Kind of like "scanning" the light a using a single pixel. Just guessing. - inactive, on 10/12/2007, -0/+6I know that was a joke, but the reason "dead pixels" are such a problem is that traditionally, in a million-pixel sensor, if you have one pixel that's dead, you can't just replace it; you have to replace the WHOLE sensor, which is not very cost-effective. Since this only has a one-pixel sensor, it would be MORE cost-effective to simply discard the rejects, meaning dead pixels would be LESS of a problem (probably non-existent).
- pikestrider, on 10/12/2007, -0/+6Here are some single pixel snapshots from my last vacation...
At the beach
.
Bicycle built for two
.
At the airport
.
Also, my wife naked (WARNING - NSFW)
. - BESTenemy, on 10/12/2007, -0/+6DLP technology in video projectors replaces pixel arrays with mirror arrays and the result is more cost-effective. Half the scanners on the market today already use technology as the one mentioned, having either a sliding stripe of light sensitive diodes or one diode with a mechanically adjustable prism. The problem, for the most part, has been the ability to rapidly re-direct light sources into the receiver in order to capture the required frame. Since the picture scanned does not change throughout the scanning process, the device can take its time processing. With digital camera exposure time is critical, so a mirror or prism array would have to work significantly faster.
So, what the researchers suggesting is going back to scanline capturing from progressive. Interesting idea and it might, indeed, help miniaturize the devices even further, while reducing power consumption. - inactive, on 10/12/2007, -0/+5"hey check out my new camera! it's one of those new 3 GHz models!"
- inactive, on 10/12/2007, -0/+5Yeah, drop it and millions of years of bad luck ensue
- bjohns, on 10/12/2007, -0/+5And a picture of your penis
. - blahtastic, on 10/12/2007, -0/+4Soo now I get to look forward to buying a 7 Meg-minimirror camera as opposed to a 7 megapixel one...
bob555, you know, I thought of that thing in the digg article about a month ago where they're trying to gather electricity from radio waves. Lucky for the inventors of each of these products that we lack movitation (and 100% willing to bet neither of us have the brains to pull it off either.) - Wyzard, on 10/12/2007, -0/+4The "point of megapixels" is to make your images larger in terms of pixel dimensions -- 1024x768 vs. 2048x1536, for example. Whether it actually improves picture *quality* depends a lot on the lens and how sharply it focuses the picture onto the sensor; there's no benefit in making the sensor's resolution higher than the acuity of the lens, because where you would've gotten one pixel from a lower-resolution sensor, you just get several pixels that are all the same color instead.
Anyway, the name "single-pixel camera" is misleading. The camera can actually record as many pixels as you want; it's just limited to recording one pixel *at a time*. It manipulates its micromirrors to "look at" each tiny piece of the scene in rapid succession and record a pixel for each. - froman98, on 10/12/2007, -0/+4You know, everyone thinks that the more megapixels you have the better the image is going to come out. The biggest thing to get great looking images is a good quality lens. I could totally see how a one pixel camera could work, it'd have to capture the image extremely fast though as Darmichar pointed out. Doing that with tiny mirrors is totally feasible... but I'd be totally pissed if I dropped the camera.
- feinest, on 10/12/2007, -0/+4wow, your wife is hot.
- etruscan, on 10/12/2007, -0/+3So now the amount of mega-mirrors you have will be the important factor.
- noamsml, on 10/12/2007, -0/+3Hey, those are all the same color!
- bob555, on 10/12/2007, -1/+4I swear I thought of this just the other day. Lucky for them I procrastinate and lack motivation.
- yoda133113, on 10/12/2007, -0/+3Actually a pixel is not a display only element, in fact it stands for picture element, meaning it's an element of a picture, and kinda like our periodic table of elements, it's the simplest unique part of a picture, a pixel is only one color, but it can be either CMYB, or RGB meaning that it has one color, but the next one can have another. I would guess that the concept here is that you would take say 8 million single pixel pictures (each having only one basic color and a certain intensity of that color) in series (meaning not at the same time, but rather exceedingly fast) with a one pixel sensor instead of todays technology where you take 8 million single pixel pictures in parallel (meaning at the same time) with a multi pixel sensor. This could be a fantastic development, or it could be crap, but it sounds similar to DLP technology, just in reverse.
- S1ngular1ty1, on 10/12/2007, -0/+3Ok, it seems not many understand the concept. The image would be rapidly scanned passed a single pixel sensor by action of mirrors. This idea is sort of a CRT screen in reverse. In a CRT a single electron gun paints the screen. In a single pixel camera, the screen paints the sensor and the image is reconstructed in software. All this must happen really fast (much less than 1 second).
Like someone else said, the mirrors would have to work extremely fast so there would be no shutter lag to compete with current multi pixel (megapixel) digital cameras. - carve, on 10/12/2007, -0/+3Gxdesign: I have an SD800IS, and I got the F30 for my parents. I've wished for the exact same camera as you! I want the sensor and battery from the Fuji, and everything else from the Canon (especially the wide angle lens). If Canon would start sourcing their sensors from Fuji instead of Sony I think I'd get wood :)
- inactive, on 10/12/2007, -0/+2You know that's actually what most digital sensors do now, right? Each sensor only senses one colour, and then they're merged together. It's effective because they use twice as many green sensors as red and blue, green being the colour that the eye is most sensitive to. This is called a Bayer filter mosaic, and there's a competing technology called the Foveon X3 sensor that actually does sense all colours simultaneously in each sensor, but it's not very cost-effective for consumer cameras (it's being introduced in a few Sigma cameras).
- Wyzard, on 10/12/2007, -0/+2It's basically a DLP in reverse. DLP televisions use a single light source (a lamp with a spinning color filter wheel in front of it) and a micromirror array to reflect specific colors (chosen by the timing of the spinning color filter wheel) to arbitrary positions on a screen (chosen by the positions of the mirrors) to form a picture. The single-sensor camera would take *in* a picture and reflect each tiny piece of it onto the sensor, one "pixel" at a time.
If you're doubting whether it's possible to actually manipulate an array of micromirrors that fast, go to an electronics store and take a look at the DLP HDTVs. :-)
http://en.wikipedia.org/wiki/DLP
However, televisions have a much longer "frame time" than a camera needs. They may need to make the mirrors faster after all, to support fast shutter times like 1/10000 sec. - vajra918, on 10/12/2007, -0/+2I see, so it would draw the image, pixel by pixel using mirrors and one sensor, as opposed to a lens and an army of sensors. Now how fast would those mirrors need to be able to actuate in order for this to be functional?
- endobur, on 10/12/2007, -0/+2This is actually pretty cool, but there are a few things that should be highlighted.
1. Current Digital Micromirror Devices (DMDs) are designed such that they work within the visible spectrum. They are not optimized for IR, Near-IR, UV, Deep-UV, or Extreme-UV. IR may not behave predictably with current designs. All DLP projectors have a hot mirror that keeps IR off of the chips.
2. All digital cameras, and all cameras for that matter are a compromise. No camera records the full spectrum of light that it receives. Charge-Coupled Device Sensors in digital cameras have a narrow range in which they work (~700nm to ~400nm, although some have Near IR sensitivity to about 900nm), and most modern digital cameras work by taking 3 pictures in different wavelengths simultaneously. This is accomplished by having Red, Green, and Blue bandpass filters in a pattern over individual pixels. The red, green, and blue images are combined to "guess" at the wavelengths in between. Of course each of the filters lets in a range of wavelengths, so the blue filter at ~475nm is actually biased a little towards purple, in order to get a cleaner peak at the ~510nm Green Filter. Human eyes are most responsive to colors between 500-550nm, so good reproduction in this range is critical. The red filter at ~650nm is typically little wider.
Film cameras work similarly. Specific chemicals in the film react more strongly to certain wavelengths of light. So again, there is an approximation of the color.
3. This device is not meant for visible light photography. It's really optimized for monochrome photography, where a set of filters is put in front of the device, and the device then samples pixels of the image in enough different ways to get the desired level of detail for the image. In this way, the device can achieve a higher true resolution than the DMD itself. (Sample multiple pixels at the same time, and in different combinations). The upshot of all of this is that this type of camera is capable of very high dynamic range. (details in low light portions of the image are not washed out by areas of high intensity light).
4. So the way this device is set up is with a photo bucket that can either count photons or produce an analog signal based on flux. A CCD type device would not really be appropriate. We'd want something with a bigger aperture, and the ability to amplify the light coming in without noise. I'm thinking a photomultiplier tube (PMT) is probably what these guys have in mind (not cheap). They react much, much (orders of magnitude) faster than CCDs. In order to get the scan speeds needed for the operation of a device like this, it would be necessary. Assuming that a DMD device could be made that operates across all practical spectrums, all that would need to be done is ensure the inside of the camera has a coating that is "black" across all wavelengths we'd be interested in, and put the appropriate filters in front of the camera to isolate the wavelength we're interested in.
Sorry, but I feel I have a professional obligation to clarify on things such as this. Way too much spurious engineering speculation on this thread. - inactive, on 10/12/2007, -0/+2It doesn't need to do 4 million exposures to get a 4-megapixel photo. It's not breaking the field of view into pixels and then scanning it one-by-one; it's much more complicated than that. I found the actual project page, linked from the BBC article: http://www.dsp.ece.rice.edu/cs/cscamera/. It shows actual results, with the number of measurements and the number of pixels, and there are links to publications in PDF format.
- BuzzLightyear, on 10/12/2007, -0/+2So replace millions of electro-optical sensors with millions of electro-mechanical mirrors.
We can already create mega-pixel cameras cheap enough to fit in mobile phones, do we need this?
Progress? - mikeazorin, on 10/12/2007, -1/+3Actually the girl in that damn ad calls them 'me-ers' whatever the hell that is.
- ellimist, on 10/12/2007, -0/+2I suppose the mirrors would be moving in small angles so as to scan whatever it is you are photographing.
Each time it moves the solid angle equivalent of one pixel, the single pixel camera records it. This way, it can go whatever resolution you want really easily.
Of course, this is happening really, really fast.
Edit: Heh, Darmichar said it, too. - zbeast, on 10/12/2007, -1/+3Here's a company that already makes a single pixel camera.
http://www.pixim.com/html/tech_about.htm - BESTenemy, on 10/12/2007, -0/+2 Simple implementation - a convex lens, 2 rectangular prisms and a singe light sensing diode beneath, rotating perpendicularly one doing so 60 times faster than the other. For every degree of rotation around one axis you get 60 degrees of coverage around the perpendicular axis. Using the scanline methodology you move the image over pixel in a type-writer carriage manner, allowing it to capture the picture in its entirety. However considering the resolution of the image, the sensor would have be to able to respond to exposures faster.
You'd get a camera with infinite resolution, constrained only to exposure timing. The more virtual pixels you scan, the less time you can afford to examine each one and the less light you're sending to the CCD sensor. Sure, 1 pixel costs less than 10 megapixels, but that single element has to process light 10 million times faster, and will also be more expensive. - Jjcd7, on 10/12/2007, -2/+4It's the mirrors!
But, really, I don't get this. Isn't the whole point of megapixels to supply us with color and clarity? - kaffein, on 10/12/2007, -0/+1Reminds me of DSF and WSD 1-bit audio formats:
http://www.korg.com/gear/info.asp?a_prod_no=MR1000&category_id=3
"1-bit recording is the latest advancement in audio and has been adopted for use in the SACD recording format. Its uncompromising fidelity, low noise floor, extended dynamic range, lifelike imaging and analog quality depth have been praised by top experts. Another benefit of the 1-bit format is that it can be converted to any other bit depth and sample rate without error or degradation." - dylanrush, on 10/12/2007, -0/+1It would work 4 times faster if you had 4 light sensors instead of just one. Eventually you would have a megapixel sensor in its place, and then we're just talking about a different lens instead of a different camera altogether.
- imtigger2, on 10/12/2007, -0/+1There's an excellent book out on Procrastination called "do it now!", and it looks like an easy read, and people swear by it.
I got it for Christmas a couple years ago, but I keep putting off reading it. - noamsml, on 10/12/2007, -1/+2No, it's nothing like that. It uses moving (I think) mirrors to ensure that light from different spots of the picture gets to the sensor at a different time. It has nothing, and I repeat, NOTHING, to do with myspace kids taking pics of themselves in the bathroom mirror.
- dpcamp, on 10/12/2007, -1/+2that article did a crap job of explaining the technology, but then again this is a had concept for me to grasp so I guess will just have to wait till it comes out 10-15 years from now..
i checked out this article too but it just went over my head maybe someone can read it an maybe explain it into dumb people speakage.
http://www.physorg.com/news79019816.html - kaffein, on 10/12/2007, -0/+1All depends on the resolution/noise/dynamic range and color quality combined of course...
The Canon EOS-1Ds has a 1.16 % resolution relative to 35mm, and the Kodak DCS Pro 14n is 1.42% the resolution relative to 35mm. These SLR cameras were released in 2002. - KiDD420, on 10/12/2007, -1/+2It's like a DLP or Laser projector backwards
- LogicallyGenius, on 10/12/2007, -0/+1Trust me, that camera will be use less for taking videos that my 6MP photo Camera can do, ie. 640x480 30fps high quality video.
Loosers, u will have to flip that mirror a billion times to match that video quality. - xxMarka, on 10/12/2007, -0/+1would it be accurate to say that an LCD monitor is to a CRT monitor as a "single-pixel" camera is to a mega-pixel camera?
- greghunt, on 10/12/2007, -0/+1That's why it will use a bigger more sensitive sensor that has higher accuracy.
With this technology you can use a single big sensors rather than millions of tiny ones. - TD_Birds, on 10/12/2007, -0/+1Mirrors are the new Lasers!
-
Show 51 - 80 of 80 discussions
The Digg Toolbar for Firefox lets you Digg, submit content, and keep track of Digg even when you're not on the Digg site. Download the official 
© Digg Inc. 2009
— Content created and posted by Digg users is