backtop

Print E-mail del.icio.us 16 comment(s) - last by DeepBlue1975.. on Feb 8 at 2:44 PM
New Kodak camera chip could make camera phones much more useful

Kodak cut its teeth in the world of print photography and film based cameras with a history going back for decades. With the advent of the digital photography revolution, Kodak has been reinventing itself as a player in the digital photography industry rather than a maker of film and film-based cameras.

Kodak leveraged its vast patent library to devise a new camera chip that could make camera phones much more usable. Anyone who has tried to use a camera phone knows that typically the images turn out dark and blurry because the camera portion of most phones is something added on as a secondary function.

Kodak sees its new camera chip as making the camera portion of the phone one of the main features in the future. According to the Associated Press (AP) the new Kodak sensor is the world’s first 1.4 micron, 5 megapixel camera chip.

The AP quotes Fas Mosleh, a marketing director in Kodak’s image sensor business, as saying, “[the new camera chip] is at least doubly sensitive to light than current devices. It produces crisper images even when light conditions are not ideal - such as at a candlelit birthday party - or when shooting a moving target.”

Kodak plans to sample the new chip to camera makers in the spring of 2008 and says that using the new chip won’t add cost to the cameras. The new chip will sell for $5 each in bulk lots of one million chips or more. New cameras using the Kodak chip are expected to hit market in a year.

Kodak was in the news last year when it introduced a new line of all-in-one printers that gave consumers about a 50% cost reduction in ink costs.

Comments     Threshold


This article is over a month old, voting and posting comments is disabled
meh
By goku on 2/7/2008 3:20:33 PM , Rating: 1
Great more megapixels, so we can see more finely detailed unintelligble pixels when showing off our pictures! How about they put higher quality sensors instead of simply adding more megapixels. My phone's camera takes much better pictures than my friend's phone (1 megapixel) and mine is only .3 megapixels! (640X480)




RE: meh
By noirsoft on 2/7/2008 4:07:40 PM , Rating: 3
Read the article, stupid
quote:
“[the new camera chip] is at least doubly sensitive to light than current devices. It produces crisper images even when light conditions are not ideal..."

It is more than just a higher MP count.


RE: meh
By inperfectdarkness on 2/8/2008 10:11:59 AM , Rating: 2
you seem to be a novice at photography.

more pixels means more raw data to manipulate. it wasn't until medium formats started hitting the 30+meg range that the benefits of digital actually started to outweigh traditional film. film is limited in resolution.

you can always save an image as a compressed, shrunk format. but you cannot expertly manipulate data that wasn't there to begin with. a photoshopped 10Mp picture will always be inferior to the same picture photoshopped on a 20Mp camera--provided both pictures were taken with the same expertise.


RE: meh
By s12033722 on 2/8/2008 10:45:28 AM , Rating: 3
Not true. You assume that both the 10 MP and the 20 MP sensor are equally capable of collecting data at a pixel. This is false. Every time you double the pixel count in the same format, you halve the pixel size. That reduction in pixel size causes an unavoidable increase in noise because your maximum signal from the pixel goes down, while the noise doesn't go down nearly as fast (see my other post in this thread for more math). In addition, as pixels get small, they approach the sizes where they diffract light rather than capturing it as incident on the sensor. This effect is not constant across wavelengths, so not only do you get a loss in true resolution from the bending of the light beams by the sensor, you also get chromatic abberation because the light is not all bent equally. It's entirely possible for a 20 MP imager to have LESS resolution than a 10 MP imager depending on the pixel sizes.


RE: meh
By maven81 on 2/8/2008 10:56:11 AM , Rating: 3
Well that's a riot... you don't seem to understand photography OR electronics and you're trying to educate people?

The big advantage that film held wasn't just limited to resolution. It can capture a higher dynamic range (millions of shades of gray), and doesn't have effects present in chips, such as relatively quick saturation (once a pixel absorbs a certain number of electrons it can no longer absorb more, and worse yet, it can spill over into neighboring pixels, creating spikes around very bright objects). In addition color film has higher chromatic resolution as it's a 3D medium, rather then the 2D Bayer arrays found in chips.
We've found ways around this, using 3 ccd sensors on some cameras (one for red, green and blue) and there's that foveon 3D sensor that basically replicates film. There are also sensors that have extra surface area to improve dynamic range.
The most interesting advances in digital imaging definitely aren't about the megapixel count.

Frankly what kodak is describing here sounds impossible... tiny pixels have tiny surface area, and tiny distances between them, allowing for "dark current" to flow, creating high noise. Now they may have come up with some sort of major innovation here, more sensitive pixels, microlenses, who knows, it's not clear from the article.
But all things being equal, a low res sensor with larger pixels, with more surface area will produce better images then a sensor with tiny pixels and loads of noise.
Not only that you don't seem to understand sampling at all, meaning that lenses are a huge part of the equation, and that depending on what the field of view is, two sensors with a different pixel count could be producing images of identical resolution (because a large blurry image doesn't have any more detail then a small but sharp image)
Furthermore you don't understand how image processing works. For example, I could have a small sensor stack 30 images to produce one image, which would significantly boost the signal to noise ratio and produce a better image then a higher res sensor that took only 1 image.


That will be a truly useless sensor.
By s12033722 on 2/7/2008 6:18:41 PM , Rating: 2
With a pixel size that small, boundary effects are going to completely eliminate any supposed gain in resolution due to the pixel count. In addition, they can't have a full well of more than 10,000 electrons on a pixel that small. That means that they have an absolute maximum dynamic range of 6.7 bits in a perfectly noise-free camera. All the increased sensitivity (with a 1.5 um active area? Who are they kidding?!?!) would do is saturate their tiny pixel faster.




RE: That will be a truly useless sensor.
By DeepBlue1975 on 2/7/2008 9:11:33 PM , Rating: 3
Wait!!!

How did you calculate the dynamic range off the electron count per pixel?
And more, how do you estimate the electron count per pixel?
I guess you're estimating the space between pixels as well, but would really like to know how.

It is a serious question, I really like stuff like these but don't easily find formulas. What the hell, I didn't even know you could calculate dyn range by just knowing electron count per pixel!!

What is the relationship? would you care to explain?

Thanks!


RE: That will be a truly useless sensor.
By s12033722 on 2/8/2008 10:38:58 AM , Rating: 2
I am guessing at the full well capacity based on the capacities of larger Kodak sensors with released data. Their 7.4 um pixel has a full well of 30k, and their newer 5.5 um pixel has about the same, so I figured 10k for a pixel with an area less than 1/10th the size would be a generous number, even accoutning for advanced tech.

As for calculating SNR from that, I am assuming that the only noise source within the camera is the photon noise inherent within the arrival time of a stream of photons. The distribution of the arrival times of photons at a sensor follows the Poisson distribution and is a fundamental physical limit. It cannot be improved upon by better sensors, etc. (it's the photon equivalent of shot noise or Johnson noise for any EE types). Once the photon hits the detector, it has a percentage chance of creating an electron within the pixel well. That chance is called the quantum efficiency of the sensor and is a percentage less than 100% that varies across the wavelengths of light.

That percentage relates the incoming photon noise to the electron noise within the pixel and thus limits the noise in electrons to the same distribution. Poisson noise is given by the square root of the number of items being considered, in this case 10,000 electrons, so the fundamental noise floor is 100 electrons. If you have a max signal of 10000 electrons and a noise of 100 electrons, the SNR is then (10000/100) = 100. TO express that in dB would be 20 log (SNR) = 40 dB. Effective number of bits is (log(SNR)/log(2)) = 6.64 bits. An easier rule of thumb is that ENOB ~= (SNR in dB/6), which is where 6.67 bits came from. Hope that helps!


By DeepBlue1975 on 2/8/2008 2:44:14 PM , Rating: 2
Great insight!!
Gives me lots of concepts to start reasearching.
In fact the only thing I'm familiar with there is the poison distribution :D

Thanks a lot!


By ElFenix on 2/7/2008 2:59:10 PM , Rating: 3
these little sensors out resolve the lenses on cell phone cameras by 3 MP or so. higher sensitivity will be very much appreciated in this market, what with everyone trying to use their cell phone cameras to take pictures in dim lighting at restaurants, clubs, etc.

i wonder what effect the reverse method of counting electrons has on power consumption?




By Oregonian2 on 2/7/2008 3:08:45 PM , Rating: 2
Shouldn't think power consumption should be an issue as little as the camera would likely be active (although perhaps not if it is used for video).

What REALLY is needed is one of the optical stabilization technologies somehow applied (meeting the other goals). In my experience anyway.


By Oregonian2 on 2/7/2008 3:10:16 PM , Rating: 2
Forgot to mention -- I mostly wonder how much noise is present. Even my current 2MP cell phone camera has probably 0.1MP of resolution after processing to remove the noise (which is still there some).


Kudos to Kodak
By Mitch101 on 2/7/2008 4:20:55 PM , Rating: 2
You know you have to give a pat on the back to Kodak. When the first digital cameras came out I figured that Kodak would be history. At that point you knew film would eventually be dead to all but the super serious pro's. But they redefined themselves and are moving with the times. Guess the film cash cow they had for ages allowed them to redefine the company. However today they must compete in technology. There really isn't any area I can come up with that they really have exclusively any more like they did with Film. Fuji had a run at it but a lot of places wouldn't take their film.

Try finding a film developing store today.




RE: Kudos to Kodak
By Darkskypoet on 2/7/2008 6:11:57 PM , Rating: 3
Walmart.


how does this compare?
By inperfectdarkness on 2/7/2008 3:04:48 PM , Rating: 2
how does this size compare to what's already in existence?

granted (it sounds like) this same technology could allow point and shoot cameras to reach 15Mp resolutions or more.

at last, resolutions that were reserved for professionals using medium-format cameras could nearly be reached by your average SLR small-format.




RE: how does this compare?
By ElFenix on 2/7/2008 4:10:56 PM , Rating: 2
i imagine this is the same size as existing cell phone camera chips. kodak is smart enough to know that it's easier to sell chips if the cell phone maker has to do as little as possible.


"If they're going to pirate somebody, we want it to be us rather than somebody else." -- Microsoft Business Group President Jeff Raikes











botimage
Copyright 2009 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki