livescience.com— Imagine a digital camera that not only takes up less power and space, but can also take night-vision pictures as well.
Oct 2, 2006View in Crawl 4
Wont it like be able to see through clothes during the daytime?Didn't sony do a recall on a night-vis video camera lens because it saw through clothes?
This article is BS. First most CCDs pick up infrared light (night vision) theres a filter infront of the sensor though that blocks out most of the IR. Its underserable to have the snesor pick up these spectrums. Second, the only way the system could work is if either the sensor physicaly shifted around or the lens moved to focus diffrent points onto the 1 pixel sensor. seriously people this is outragious. The camera would still have (assuming this magic pixel picks up a large area of the information comming from the lens) to compress the image data. Raster images are raster images regardless, unless they intend to store this as a raw.Lastly, they never say how this 1 magix pixel works...
You should read the article.The idea is that the most important part of the sensor is changing all the time. The image projected into the matrix of thousands micromirrors, and that random reflected image is directed into single pixel sensor, which get average intensity form all those directions. However every n millicecond the direction of mirrors in the matrix is changing at random, and the new signal sample sent into pixel. Thus, information the sensor got is in fact form different directions, but all mixed together. Special statistical algorithm extract image information from the distribution of those samples. Cool mathematics here.
This is complete bulls**t! 128 pixel image and it takes one minute to take the picture, YAY! OMFG, common guys this thing is way off from being useful or interesting, additionally the night vision they talk of is bull s**t. Night vision and IR cameras are to different things.
From the article:[quote]Essentially, if cameras scan in a fraction of pixels from throughout a scene?say, 3 or 4 percent, or just thousands of pixels?new algorithms can take such data and extrapolate out what the original image appeared like.[/quote]So it is guessing what the pixel inbetween would be. Say it captures a pure white pixel and next to it a pure black one, what would be the pixel inbetween? Grey? Might have been red or another black pixel. Guessing what could have been there in the photo is never a good way to do things imo.
lazydrumheadOct 3, 2006
I didn't believe it until I read the other article TWO STORIES DOWN on the front page....much better description: <a class="user" href="http://www.physorg.com/news79019816.html">http://www.physorg.com/news79019816.html</a>
dangerouslehOct 3, 2006
sounds great... but when does it come out? (if ever)
Closed AccountOct 3, 2006
Wont it like be able to see through clothes during the daytime?Didn't sony do a recall on a night-vis video camera lens because it saw through clothes?
lostspyderOct 3, 2006
This article is BS. First most CCDs pick up infrared light (night vision) theres a filter infront of the sensor though that blocks out most of the IR. Its underserable to have the snesor pick up these spectrums. Second, the only way the system could work is if either the sensor physicaly shifted around or the lens moved to focus diffrent points onto the 1 pixel sensor. seriously people this is outragious. The camera would still have (assuming this magic pixel picks up a large area of the information comming from the lens) to compress the image data. Raster images are raster images regardless, unless they intend to store this as a raw.Lastly, they never say how this 1 magix pixel works...
strdOct 3, 2006
You should read the article.The idea is that the most important part of the sensor is changing all the time. The image projected into the matrix of thousands micromirrors, and that random reflected image is directed into single pixel sensor, which get average intensity form all those directions. However every n millicecond the direction of mirrors in the matrix is changing at random, and the new signal sample sent into pixel. Thus, information the sensor got is in fact form different directions, but all mixed together. Special statistical algorithm extract image information from the distribution of those samples. Cool mathematics here.
digitalunltdOct 3, 2006
This is complete bulls**t! 128 pixel image and it takes one minute to take the picture, YAY! OMFG, common guys this thing is way off from being useful or interesting, additionally the night vision they talk of is bull s**t. Night vision and IR cameras are to different things.
inveterateOct 3, 2006
Direct link to the actual article at Rice U.<a class="user" href="http://www.dsp.ece.rice.edu/cs/cscamera/">http://www.dsp.ece.rice.edu/cs/cscamera/</a>
weaklingOct 3, 2006
From the article:[quote]Essentially, if cameras scan in a fraction of pixels from throughout a scene?say, 3 or 4 percent, or just thousands of pixels?new algorithms can take such data and extrapolate out what the original image appeared like.[/quote]So it is guessing what the pixel inbetween would be. Say it captures a pure white pixel and next to it a pure black one, what would be the pixel inbetween? Grey? Might have been red or another black pixel. Guessing what could have been there in the photo is never a good way to do things imo.