Slashdot Mirror

Transparent sensors. Every picture is a 3d array instead of a 2d array.

Foveon

Ummm ... the guy created the first digital image, on the "only programmable computer" in the US at the time. I would say yes, that's an invention.

From "A Brief History of the Pixel":

Paul Nipkow had filed a German patent application on his mechanical-scanning TV or Elektrisches Teleskop1 in 1884, in which he referred to Bildpunkte--literally picture points but now universally translated as pixels...Alfred Dinsdale had written the very first English book on Television in 1926, but instead of picture element he had used there lots of other colorful language: a mosaic of selenium cells, a great number of small parts, thousands of little squares...

There's much more, but it suffices to say that Russell Kirsch is only a minor footnote in terms of the history of the pixel. He may have invented something, but it wasn't square pixels. No doubt, someone colored in blocks on a sheet of graph paper to make an image before pixels were ever used in conjunction with an electronic device. And using square pixels on a computer connected raster scanned display is just common sense, not an invention - it makes the math simpler.

To be even more pedantic: In terms of raw CCD sensor data, the number of bits per pixel is actually likely to be 12 indeed. With the exception of devices like the Foveon CCDs, most have some kind of colour filter mask placed over the individual pixels, either the well-known bayer pattern or a variant such as the CYMK grids used in some Sony CCDs (probably others too, can't be asked to dig it up). Each output colour pixel is interpolated from 4 adjacent, but spatially distinct physical pixels. Simply assuming the bits per colour pixel to be 3*12 is therefore not entirely correct.

I wonder how this is going to compare to the Foveon sensors. They capture RGB data at all pixels - filtered based on depth rather than location. Now if only those babies cost less.

I agree that its too little too late. To me...the bayer filter should be gone altogether since the foveon x3 sensor came out. If other camera makers would use this technology the price would come down.

I just gotta wonder whether or not Micron is licensing the CMOS technology from http://foveon.com/.

Nope, MU has been developing CMOS image sensors for years. It's an in-house product. I know because the firmware programmers are right around the corner from my office.

I've gotta friend that has the Sigma SD10 (a DLSR) that uses the X3 chip. He giggles like a school girl when he talks about it. It's kind of disturbing. Of course, the pics he takes with it are phenomenal.

in reality it has 4M green pixels and 2M of each red and blue

I read about this 3-4 years ago in Discover Magazine.

X3 is a CCD technology by Foveon Inc. that captures all three colours instead of one per pixel as traditional CCDs do.

What this means (in my mind) is that traditional CCDs throw away two thirds of the image data and the software makes up that missing data. I.E. Take three pixels; one captured the red, the next blue, and the third green. The SW looks at the blue and green to determine how much of those colours should be in the red pixel and likewise for the blue pixel and the green pixel. So yeah, while the pics made with digital cameras look good they're only one third real.

But the X3 CCD by being able to image all three colours (red, green, blue) in each pixel creates a sharper image and one truer to the original scene.

Wikipedia knows about X3

Sigma Corp. makes two camera with X3 CCDs. When I finally go digital in photography I'm getting one of these.
The SD9 and the SD10

Of course, quality will jump tremendously when we switch over from the RGRB CCDs to tri-color CCD's. Slightly offtopic, does anyone know the progress of this? When will we be able to get true 3-color CCD cameras? About two years ago I had heard this would be in about a year...

You are probably thinking of Foveon's X3 sensor. The first production camera with this sensor was the Sigma SD9 Digital SLR, way back in November 2002, but as of writing the sensor is only available in a handful of cameras.

The X3 sensor works with three photodetectors in a column at each pixel location, rather than a conventional sensor which uses a flat RGBG matrix and only measures a single colour at each pixel location.

Consensus is that X3 is definately better than a conventional sensor with the same pixel resolution, but not as good as a conventional CCD with the same number of photodetectors. So the 3x3 megapixel sensor in the SD9 (marketed by Foveon/Sigma as 9 megapixel) is unquestionably better than a conventional 3 megapixel sensor but not as good as a conventional 9 megapixel sensor, losing in particular in the resolution stakes.

It has also arguably been somewhat hobbled by a lack of mainstream support and a number of annoying quirks on the Sigma cameras (the SD9, for example, was RAW only.) This lack of support is understandable many of the big manufacturers have significant R&D invested in their (conventional) in-house sensor technology - and at the moment this tech seems to be advancing faster than Foveon's; I think we could be seeing another Transmeta situation here.

Okay, this is a longshot, but could something like the foveon sensor be applied to LCDs? How long before we get real square pixels from RGB or RGBE stacked LCDs?

Looking at scans of 35mm negatives, my estimate was 8MP to get similar resolution, at least for relatively inexpensive film. You could go significantly higher if you use more expensive slower speed film. What I'm really looking forward to is the Foveon X3 technology becoming more widespread and available in higher resolution.

if you have a decent eye, you'll see quit a subtle 'bad quality' in all of non-X3 CCD's ... due to it not beeing *of course* able to capture the some Red Green Blue wavelenghts per pixel/amount of 'photons', but instead utilizes a 'mosaic matrix' to compose the image *which is uurrrkkk, urruk, 'bad quality'* - I mean, just look at this interactive tutoral (java), see how many 'photons' a non-X3 misses? ... then think about that a camera then tries to compose a picture with that amount of 'phonon' count loss... *iirrk*

*crawls back to his geek-hideout :)*

PS. BTW $0.03 that this post is just one of those 'slashdot Post Ads'

if you have a decent eye, you'll see quit a subtle 'bad quality' in all of non-X3 CCD's ... due to it not beeing *of course* able to capture the some Red Green Blue wavelenghts per pixel/amount of 'photons', but instead utilizes a 'mosaic matrix' to compose the image *which is uurrrkkk, urruk, 'bad quality'* - I mean, just look at this interactive tutoral (java), see how many 'photons' a non-X3 misses? ... then think about that a camera then tries to compose a picture with that amount of 'phonon' count loss... *iirrk*

*crawls back to his geek-hideout :)*

PS. BTW $0.03 that this post is just one of those 'slashdot Post Ads'

if you have a decent eye, you'll see quit a subtle 'bad quality' in all of non-X3 CCD's ... due to it not beeing *of course* able to capture the some Red Green Blue wavelenghts per pixel/amount of 'photons', but instead utilizes a 'mosaic matrix' to compose the image *which is uurrrkkk, urruk, 'bad quality'* - I mean, just look at this interactive tutoral (java), see how many 'photons' a non-X3 misses? ... then think about that a camera then tries to compose a picture with that amount of 'phonon' count loss... *iirrk*

*crawls back to his geek-hideout :)*

PS. BTW $0.03 that this post is just one of those 'slashdot Post Ads'

Interesting. So 16MP is ~ same as 100 ISO slide film.
But then there is the whole issue of digital camera interpolation. Which is why I really love the Foveon X3. But it's only used in a few very expensive cameras.
By inference, when digital cameras have 48MP, or Foveon goes 16MP, I will stop using my RDP III.

So far, analog neuromorphic VLSI has hit a dead-end in terms of real applications. Also digital signal processing has been speeding up to the point where it can go almost as fast as a lot of the parallel analog models.

The one exception is that the work on analog retina models lead to the development of the Foveon X3 technology, which is just packing R,G, and B CMOS sensors into a single vertical column on a chip. But again, the neuromorphic part of the retina model is not the X3 technology, the X3 technology is stacking CMOS sensors.

Analog neuromorphic VLSI did have one big result, the electrical engineers managed to teach the biologists a lot about signal processing, and the cross-pollination of this knowledge has lead to discoveries such as ripple analysis in auditory cortex.

http://foveon.com/SD10_info.html

I'm thinking we are nearing the end of the major advances in digital cameras.

Then I'm thinking that you are really, really wrong.

There are so many limits with film that can be overcome with digital. Already we have seen huge leaps in sensor quality at higher "film speed" settings. Even on my EOS D60, ISO 400 is pretty grainy. The 10D improved that a lot. Now the 1D M2 has moved very high clarity to ISO 800.

Over time we will see further improvements to this. Eventually you'll be able to shoot at ISO 3200 with no loss of quality compared to ISO 100. Maybe we'll even get to 6400. Why would you want this? How about the ability to stop action in very dim light? How about to gain depth while hand-holding in macro photography?

Colour accuracy is another area that will improve over time. I think the similar tech to the Foveon sensor will become standard. Instead of having one sensor with a mix of pixels picking up RGB, we'll have layers of 3 sensors in a camera. This will also allow for more pixels without having to make the individual pixel sensors as small as now. Larger pixel sensors = better quality images.

Battery life will improve. Memory will get faster. Memory will continue to increase in density. LCD viewers will get brighter and larger while using less power. Shutters will get faster. Memory buffers larger. Prices will come down. I could go on and on -- yet I am sure there are dozens of things that I've never thought of that will also appear.

Major advances in digital cameras are just beginning. The end is nowhere in site.

Sorry to say this, but the parent in NOT "Informative". The only sensors with 3 photosites per pixel are Foveon's. The vast majority of digital cameras has ONE photosite per pixel, and a Bayer mask (RGB filter) layered on top of it. Pixel color in the final image is then interpolated from the measured intensity of the three adjacent photosites. Yes, this means that digital cameras have higher Luma resoultion than Chroma. No, it does not matter much, because the eye is much more attracted to Luminance detail.
Almost all of the manufactured sensors are black and white; only Foveon's are 3-color, and they're expensive for the resoultion and the first generation software had color clipping problems (overexposed areas of images went abruptly to white). This has apparently been fixed.
A monochrome sensor with external filters is much more flexible than the single-duty Foveon, so I guess that's why they chose it. Also, NASA doesn't usually buy space-faring hardware off-the-shelf two weeks before launch, and this full-color sensor simply did not exist a couple of years ago.

They could have used a Foveon Sensor if they didn't want Bayer interpolation.

Digital cameras are great for consumers who like to take pictures for the web and to print cheap but passable 4x6 prints on inkjets, but digital sensors won't be able to replace film where you'd like to print high-quality enlargements.

I recall reading somewhere that 35mm film has the equivalent of 24 megapixels. Right now consumer SLRs such as the Nikon D100 and the Canon 10D produce only 6MP images. Moreover they produce these images using software interpolation; sensors nowadays commonly contain patterns of red blue and green photodiodes whose image must be demosaiced in order to create a full color image. For an example of a RAW image have a look halfway down this page.

Digital cameras won't full replace film until (1) sensor resolutions become high enough to approximate both 35mm and format cameras and until (2) full-color sensors (such as Foveon's X3 Sensor) become ubiquitous.

If you add up all the Red, Blue and Green elements in a digital camera's CCD, you end up with the number advertised... 5 or 6 megapixels or whatever. (In this example I will refer to a 5 megapixel camera.)

However this does not REALLY equate to that many pixels as we would normally think of pixels with other devices.

If your LCD monitor can support a maximum of 1280 x 1024 resolution, that multiplies out to be 1.31 megapixels.
But if we were to do the math the way digital camera manufacturers' marketing departments do their math, that same screen would be 3.93 megapixels... which is basically a lie.

What happens inside digital cameras is a certain bit of deception. They use the luminance factor from each of the 5+ million CCD sensors to achieve a semblance of the resolution advertised. However the color value for each of those so-called "pixels" is not independant, but rather is derived from the values of the surrounding pixels.

Therefore we have the baffling paradox of saving a RAW file at full resolution on a 5 MP camera and getting a 7.5 megabyte file; but strangely a TIFF file of the same resolution saves out at 15 megabytes in size. How can this be possible, you rightly ask? Just what is the camera adding to the raw sensor data to create a full resolution file which is somehow twice the size of the raw data? Here's what happens...

In the RAW file there may be 5 million 12-bit samples, half of which are green elements, with the other half evenly split between red and blue elements. Or, there may be 2.5 million 8-bit red, blue and green values each, with not all of them corrresponding to actual CCD elements.

In producing a 15 megabyte TIFF file from 7.5 megabytes of RAW sensor data, the camera's firmware defines a virtual 5 million simulated pixels, each of which has its 24-bit color values derived from the other adjacent physical "pixels." Then once 8 bits each of Red, Green and Blue data are derived for each virtual pixel om the memory array, the whole simulated thing saves out at 15 megabytes.

The ONLY cameras available which do not deceive you in this way are those new ones incorporating the Foveon CCD sensors, which are novel 3D arrays of elements, each element of which produces its own true RGB color values. With the Foveon CCD, each pixel is a true pixel, and the color definition is superior.

To be fair... the color interpolation firmware in standard CCD cameras has gotten so good, that it probably is worth putting up with the marketing deceptions and these artificially puffed up file sizes (200% of what they should be), at least for now.

Anyone know what's going on with affordable cameras using the Foveon chip? I heard (from seriously reliable sources) about a year and a half ago that it was stupid to buy any traditional one-color-per-pixel camera because the coming Foveon was going to be a million times better and would render everything else obsolete. The technology does seriously look awesome and legitimate. I know there's a $3000 camera which is supposed to be awesome that uses the Foveon X3, but anyone know when something more like $1000 will be available?

There may be 8 million pixels in the camera, but that only equates to 2 million of each color!!! 3/4 of the data is interpolated in that 22.8mb picture!!!

So, when people say that Fuji Velvia (35mm) is equivalent to 10 megapixels, that's not equivalent to a 10 megapixel camera.

The reason is that there's a filter grid over a monochromatic sensor... only every third (or fourth with this camera) pixel is of each color, and the camera internally "guesses" at what the values in between would be.

The only current exception to this are cameras based on the Foveon chip (either the X3 chip, or their older prism-based 3-CCD camera), like the Sigma SLR. See www.foveon.com

So... until cameras get to ~30 MP, they won't be "equivalent" to the resolution of 35mm film. (that's not THAT far off). 10MP is tons for most uses, but it's not equivalent.

MadCow.

nick

Foveon's technology looks really good, although there is only one camera using it at the moment. Price is about the same, but doesn't include the lens.

Hopefully someone will introduce a Foveon "consumer" model in the $500 range before too long.

That's because it's been interpolated from a RGBG (or whatever) matrix of original colour-filtered pixel values. Only Foveon have made a true per-pixel RGB single-chip sensor, and it's only available in one rather weird camera so far; every other current single-sensor camera uses filtered pixels, which is why their RAW format output is different from their TIFF format. RAW preserves the original pixel values, without interpolating the colour from surrounding pixels.

That said, the filtered pixels do not give you a cleanly arithmetically reduced resolution. They give reduced chrominance resolution, and some reduction in luminance resolution as well that's determined by the image and the filter colours and pattern. If you're shooting an all-red or all-blue scene with an RGBG-filtered camera, you'll get one-quarter resolution. If you're shooting an all-green scene, you'll get half resolution. If you're shooting a normal scene, you'll get something approaching the resolution you'd expect from the raw pixel number.

The subject matter and photographic situation may make it impossible to capture a full resolution image anyway, mind you. My slightly old EOS D60 review here has more to say about effective resolutions and what they mean.

When comparing pixels to film, the actual pixel resolution is only part of the equation. Yes, standard color CCD arrays use an offset-overlay technique to interpolate more resolution in the final image than any of the single color channels has. The exception to this is the Foevon chip, which has full color in every pixel, and the very high end systems you mention above.

The huge, HUGE advantage of digital imaging that you have not mentioned is grain. The spatial resolution (or how much detail is actually in the picture content) is actually very poor in 35mm, especially in less-exposed areas. If I accidentally underexpose my digital image by one or two stops, I can use a level adjustment to recover a near-perfect image with very little grain, and plenty of detail in even the darkest areas. If I try that with a 35mm film scan, it will be extremely grainy, even from a very low ISO film. The reason 35mm gets by is that at full frame from a reasonable viewing distance and at a correct exposure, the softness, gamma, grain and falloff present a nice pleasing picture.

In every day practical use, I find that a 6 megapixel standard CCD (not foevon) producing a 3k file has better detail than the average 35mm image. Downsampled to 2k and it's an extremely sharp, excellent 2k image. Right now I even have a 3 megapixel (2k) image from an older camera on a billboard just outside of town, it's about 15 feet across, looks really nice! Average viewing distance is a big factor as well.

Most digital visual effects for 35mm and features finished to anamorphic 35 are rendered at 2k resolution. A few years ago I did most of the animation on a 35mm film spot for American Express. It was rendered at 2k and transferred to 35 and it looked gorgeous. If you have very sharp spatial resolution in your 2k image (such as computer generated imagery where every pixel is sharp and perfect) you will not gain much of an advantage going to 3k or above. The only thing that kind of resolution is useful for right now is IMAX. I dispute the idea that 35mm has 4k of useful pixels. After about 3k you won't percieve any practical difference.

CCD technology will not be able to replace film (35mm) for at least another 5 years, if ever.

5 years for widespread distribution is practical. "ever" is ridiculous. :)

Remember, when talking technology, think about practical application and end results. pixels don't exist in a vacuum. (but when they're on a CRT they exist in a vacuum tube! :)

Personally, I'd like to see variable frame-rate 2k to 3k systems for regular movies, and 4k - 5k digital systems for IMAX sized projections, using a format that can be created and previewed on desktop PCs with very fast disk arrays and hires monitors. (check out IRIDAS for an excellent digital cinema and desktop hires playback system, including 3D!)

One word. Foveon.

More info, and more yet. Combine this with high resolution and good lenses - digital is about to surpass film in quality.

Could it be that the effect in question has been patented for some other use? I'm not familiar with the patent quagmire, but multiple similar uses for the same physical phenomenon (light absorbtion into silicon) might be the issue...

I almost bought a digital camera near the beginning of this year, and then Foveon announced "immediate availability" of their X3 sensor (Feb 11 press release). I was going to nominate them just now, but checking their site I see they announced on Nov 17 that the Sigma SD9 is actually available for purchase.

Still I had to check that Froogle had some listings for this thing before I let off the trigger finger. Now I just wish they'd put out some consumer-level cameras with the thing, like they claimed should have been available soon.

I almost bought a digital camera near the beginning of this year, and then Foveon announced "immediate availability" of their X3 sensor (Feb 11 press release). I was going to nominate them just now, but checking their site I see they announced on Nov 17 that the Sigma SD9 is actually available for purchase.

Still I had to check that Froogle had some listings for this thing before I let off the trigger finger. Now I just wish they'd put out some consumer-level cameras with the thing, like they claimed should have been available soon.

For most purposes, yes, no contest. Until the first camera that uses an 11 megapixel foveon chip.

You'll be getting far better sharpness and color representation at that point. Only then will it simply be a megapixel race. We may have high megapixels nowadays but that isn't the whole battle. The interpolation algorithms are still sitting between reality and digital storage.

Foveon is solving that cheaply. Their sample images are amazing:

• All Images
• Boxer
• Cat
• Pool Table

Unfortunately, you're looking at the theoretical maximums for film. The reality of most shots is far different. The contrast range isn't an issue for 99.9% of the population either, since prints only have a contrast range of around 100:1.

The biggest difference right now is color accuracy. Until the Foveon chip reaches 11 Megapixels, we won't have anywhere NEAR the color range of film.

For further reading: here's and excellent summary of the topic.

Pixel pitch and dots per inch are different measures, camera pixels are used like display pixels; that is 100dpi monitor looks great, and a camera that can fill that 100dpi monitor 1:1 will also look great, both on screen and in a continuous tone print.

100 dpi binary printers would look pretty horrible. Most inkjet printers and most laser printers are binary, some are continuous tone to various degrees. The easy way to think about it is that each pixel on screen (typically) shows 8 bits of information, whereas each "dot" from a binary printer shows 1 bit (or 24bits per pixel vs. 3 bits per dot since each dot can be C/M/Y (black doesn't really add in this analysis, it's used to replace equal values of CMY which look grey, but should theoretically be black). A 400 dpi continuous tone laser printer makes vastly better looking prints than a 2400 dpi binary color laser printer.

Screen printing combines groups of dots into patterns called rosettes (because of the way they look, except random screening which looks smooth) with various complicated mathematical functions starting with what's known as a screen frequency. The screen frequency is far less than the resolution of the printer: typically a 2400dpi linotype machine can produce a 150lpi screen (but milage varies). Roughly you have a 150 pixel per inch printed color image (that's about what National Geographic uses, for example). One can trade off screen frequency (spatial resolution) for the screen detail (color resolution) in a fashion analogous to having to set a lower bit depth to get a larger frame buffer in a memory limited display card.

So if you want a good continuous tone print, 150 pixels per inch is a good number, meaning roughly 1500x1200 pixels for that 8x10... but wait! It ain't so simple.

The math to construct the screen craps out if the screen frequency is the same as the pixel density, it needs to be far more. The conventional wisdom is to use a pixel pitch between 1.33 and 1.5x higher than the screen frequency, or 3k X 2.4k for the 8x10 print to come out nice and sharp...

BUT that's not all - that's right there's more! Digital cameras (and consumer camcorders) have one detector pixel per output pixel (at best). But each detector pixel is EITHER R, G, or B (maybe a Y thrown in on some). That means that even the best digital camera (except the Foveon) is interpolating data at the very least in the color space, and depending on the image this can be painfully obvious. To eliminate the artifacts thus introduced, one must frequently downsample the image. If the RGB (or RGBY) pattern is approximately 2x2 (it's not usually, but close enough) this requires a 2x downsample to clean up the image.

So that means if you want a print that's (as far as spatial resolution goes) equivalent to ASA 64 Ektachrome shot in a really good camera and printed out at 8x10, you'd need roughly 6k x 5k starting pixels (the ektachrome has about 8k equivalent pixels, though that's debatable: it's definitely more than 4k and almost certainly less than 16k equivalent, the MTF of film decays gracefully without aliasing).

It's still not quite that simple because of the response curve of film vs. the response curve of solid state detectors. Silicon responds almost linearly from the first photo-electron until the detector well is filled to saturation - quite unlike the human eye. Whereas chemical film is more stoichiometric, and responds gradually to increasing light, with increasing response around the optimal reaction rate, and then slowly decreasing to a soft saturation. The result is that details are captured in the shadows, the properly exposed section has good contrast, and there are details captured in the highlights. With digital photography the shadows are filled with electron noise and the highlights are blown out.

So not only do you really want about a 6k+ camera to get good prints, you want it to capture at least 30 bits per pixel. The Kodak captures 36 bits per pixel, 4.6k pixels across, which (remember there's a trade off between color fidelity and spatial fidelity) means that it should print a very nice 8x10, to all but the most arbitrarily anal, as good as any 35mm film.

You might want to look into this. The three layer technology will give you 2(4) times the resolution for your greens (reds and blues) that your usual CCD will give. Your CCD uses only one layer that is partitioned into a checkerboard of separate color elements meaning that you will have to take 2 times less resolution than your promised 11 megapixels. With this new Foveon detector, the three layers will give you a less modified image (CCD cameras pass through a DSP which inevitably destroys information) that is 2 times the resolution (if they make a 5.5 megapixel detector, thats 11 megapixels in terms of what we call it).

Of course, you have to divide those pixel numbers by 3 or 4 to get a useful pixel count. Camera makers like to count each color as a separate pixel. Tacky.

I'm waiting for Foveon technology to go mainstream. All the colors for each pixel are sensed at the same location, so you don't get color artifacts on sharp edges like you do with other digital cameras. So far, they only make super high end cameras, but I went to a talk by their CTO, and the device isn't inherently expensive if made in volume.

I agree, less than VGA rez 640x480 isn't very useful for pictures (though it's adequate for video, i.e. VCDs). Said Logitech camera is actually based on technology from Smal Camera, and is also employed in the Fujifilm Axia. The tech is small enough (and fairly advanced - it has automatic saturation control so details still come through) and could be integrated into a cell phone. Sony also came up with a memory-stick using camera that was about the size of a pack of gum. Not to mention the recent X3 technology which should give big improvements in color and sharpness and smaller sensors per rez, particularly for low-rez imagers which have the most trouble with color artifacts.

In any case, the technology is already there now and will get significantly better in the near future. A tiny camera you have with you at all times can still be very useful.

Although I could not find any information pertaining to the Fuji Finepix 6900 on their website I have been doing research in this area ever since Slashdot ran an article talking about Foveon's new CCD chip that is suppost to revolutionize the industry. Although Sigma's SD9 hasn't hit the market yet, other manufacturers have been lowering the prices on their cameras, binging 5 and 6+ Megapixel cameras closer to reach.

For example, Nikon's Coolpix 5000 is a 5Megapixel camera that retails for $1K US, but can be found on the Internet for closer to $700. It has the ability to add and remove lenses, but it is nothing like the bayonet mounts that you may be used to with a standard 35MM SLR. Canon recently came out with the EOS D60 digital camera with a 6.3Megapixel CCD chip , that retails for approximately twice as much as the Nikon. It is suppost to handle almost exactly like a 35MM SLR, including the ability to switch lenses, which is what a professional photographer would want to work with.

But if you are like me, then you will be waiting with baited breath for the Sigma SD9 and its revolutionary CCD chip. Even though the chip itself is only a 3Megapixel chip, the quality of the photographs taken are said to be comparible with 9+Megapixel cameras. Or you could stick with the tried and true 21+Megapixel analog film and emulsion camera.

It is inevitable. Film is dead...in 2-5 years this question itself will be reduntant...like Jack Valenti shouting about Boston stranglers and VCRs .

Soon cameras using CCDs from companies like Foveon will make 'film' redundant.

And why do you need film?