Have we all been duped?

[luca_stramare2]

<p>Ken Rockwell does what all the people do, he uses the system that best suits his needs. The difference is that he writes about it in the internet, and, as per many things, it is not because it is written in the internet that something becomes an absolute truth, that applies to everybody worldwide.<br>

IMHO the whole debate film vs. digital needs to be shifted from the absolute level to the individual one: is film (or digital) better, more fun, more practical, ... FOR ME? I find those debates about pixels vs. lines per millimeter a little pointless. And even assuming that film has some kind of superiority, since new digital cameras come out at warp speed, one new model after the other... how long can a statement hold true? Film was for sure better in quality than yesteryear digital cameras and for sure sooner or later digital cameras will be better than any existing film camera. It is just a matter of time, considering the incredible amount of money that is invested in digital camera development and the fierce competition, you need something new every 2 years to convince shooters to trash their current camera and buy a new, better, one. And, besides, the quality of current films and prosumer digital cameras is more than adequate for a any application for which film was used in 1999.<br>

So, at the end of the day, the story becomes: for me, now, do I feel better using a digital or a film camera? I have my answer: film. If I were a wedding photographer or a photojournalist, or even a casual shooter willing to take some pictures of my vacation with a P&S camera and share them with friends, I would answer differently: digital. </p>

[Dave Luttmann]

<p>To answer the OP, which I forgot to do....the answer is a partial yes. I still use an old Canon D30 and 10D as my point and shoots. On vacations, at the beach, camping with the kids, birthday parties and gatherings of friends are perfect for digital. </p>

<p>Doing street photography, landscape, or anything requiring high quality, I reach for film in various formats.</p>

<p>Besides, I love finding old gear on Ebay and giving it some use. I recently picked up a Minolta X700, with 28mm, 35mm, 50mm and 135mm....for $200. It's in perfect shape. I've run a number of rolls of Astia, Neopan and HP5 through it. Lovely seeing a good scan from it exceed what expensive digital gear can put out.</p>

<p>And besides....the camera and lens are so light, I can carry it around all day. Something I loathed doing with my 1Ds Mk2.</p>

[don_e]

<p>Considering the subjects I like to shoot and the way I like to shoot, the sizes I like to print to, and the distance I like to view from, I don't see much difference whether they are digital or film. And, if I were to examine them closely for differrences, those I find are not worth the time and energy taken to find them.</p>

<p>I prefer film, but it has little to do with technical "image quality".</p>

 

[mt4x4]

<p>Don,<br /> <br /> First of all. Ken Rockwell is a douche bag. He attempts to make a living through a blog, which has a sole purpose of trashing on digital. Personally, I think he has a major inferiority complex.<br /> <br /> Second, he forgot a few things. 1st, saying that someone who is shooting digital isn't concerned with composition is just lame. Second, he forgot a few things on the film side. Of course, this wouldn't have anything to do with KR being biased, now would it?<br /> <br /> 1) Buying and Storing Film<br /> 2) Loading film<br /> 3) Unloading film and storing it again (not to mention packing it around)<br /> 4) Taking film to get processed<br /> 5) Picking film up from getting processed ($$$$)<br /> <br /> In my opinion, these should replace at least 5 of the blocks on the left side of that table. <br /> <br /> Not that I have anything against film, just stating the other point of view here. If film is what you like, then by all means shoot film. But anyone who gets off trashing on people for choosing digital needs to get their head examined. Not to mention, I feel sorry for Ken's family, I'm sure he can't provide a very good lifestyle for them with the measly earnings from his crappy website.</p>

[don_e]

<p>"Of course, this wouldn't have anything to do with KR being biased, now would it?"</p>

<p>He's just messin' with ya, Keith. You might say his site is a huge troll of photographers. Sorta like a Luminous Landscape with a sense of humour.</p>

<p> </p>

[stwrtertbsratbs5]

<p>It should be no surprise that film is capable of producing excellent images. It's a mature technology that's served us well for a good 100 years.</p>

[mt4x4]

<p>Don, agreed.<br /> <br /> I guess I should know better than to feed the trolls at this point!<br /> <br /> And honestly, if it hadn't been a KR post... I probably would have laughed.... oh wells. I guess something about that guy just hits a nerve with me.</p>

[Mike Gammill]

<p>I tend to use both, but if I know that I want to post something quickly I use digital. If I need a wider angle than 28mm (film equivalent) then I use film since I can go as wide as 17mm. I can nowhere near afford a digital lens that will get me that perspective. Unless I need a lot of prints, I just take all my non-pro film to CVS for developing and a CD. No prints.</p><div></div>

[patrick j dempsey]

<h1>Have we all been duped?</h1>

<p>In a word... Yes. A 35mm negative measures 1.42" x 0.95". When scanned on a 2400dpi scanner... which is a modest scanner by all means, you get an image 3400 x 2280 pixels. That's 8 mega-pixels. With a nice slide film, theres quite a bit more resolution than that available... remember... slides where meant to be projected to WALL SIZE and still look good. There was a time... oh yes I remember it... when 3.5MP digital cameras that had awful color fringing and terrible artifacts were being tauted as being superior to film. Maybe better than supermarket-brand 800 speed film in a disposable camera... well... maybe not even then. With every pixel upgrade, we are told that digital rivals film. And yet... if digital REALLY was that much better, it would speak for itself and it wouldn't have to be repeated over and over and over like a brain washing mantra. For a few hundred bucks you can get yourself into a professional 645 kit that will produce 22MP scans on a cheap 2400dpi scanner, with better colors, better shadows and better highlights than a $6000 digital camera. This year, Kodak released updated versions of something like 12 of their films, with dramatic increases in resolution and grain minimization. Sure, you can use Noise Ninja and the latest greatest fractal interpolation software to get wall size prints off of a 12MP camera... but can't you do the same with film and start out with better shadows and highlights and color saturations?</p>

[patrick j dempsey]

<p>Oh yeah, and I'm still waiting for people to buy into the 22MP cameras with 24x36mm sensors to discover that they have severe softness due to diffraction and the fact that the sensor is resolving well above the line resolution of the lenses. </p>

[jack_welsh]

<p>I feel that a person should use whatever they want. Film or digital. About the comment that a digital medium format gives a better image than a film one. Hasselblad has a 50 meg digital. But, who has $39,000 to spend? I do just fine with my 120 and 4x5's. B&W film is not that expensive. Buying a several thousand dollar DSLR to beat 35 and maybe equal 120, for me at least, would be a huge expense. Even if it beat the 120. I bought a digital once, the plastic on-off switch broke in very little time. I have a friend whose digital did the same thing. So, I'm sticking to film. They last for decades. My 116a Kodak jr sure has. (I can still use 120 film in it.) The $300.00 digital didn't, it sat there for the first several years, because the software wouldn't fully install.</p>

[arjun_mehra]

<p>Oh, god, another "film v. digital"? thread?<br>

You know what: I like film, but it is <em>very</em> expensive. About five dollars for a roll; about another seven to have it processed. Making 8"X10" prints, let's figure about seventy cents a sheet, and three sheets a print (it takes some tweaking with test strips to get a "perfect print" — three sheets for one final print is not uncommon). Plus, the fee of using a laboratory: about $15.00/hour; we'll say it takes about an hour to get one print totally finished.<br>

So, the cost of one print (I'll divide film and processing by 36 to make some strange attempt at calculating the cost of "one" picture): 14 cents for the negative frame; 19 cents to have it processed; $2.10 in paper; $15.00 for the lab = $17.46 for one 8"X10" color print.<br>

Digital: About $4.00 at Walgreens. You buy the camera; you buy the memory card — that's about it. A respectable-capacity memory card might cost $10.00 (forget Circuit City: have you visited eBay?), and it should last you quite a long while. Digital cameras (D.S.L.R.s) are still more expensive than comparable (35mm) film, but no longer by very much. You can toss in the cost of a brand-new, top-of-the-line desktop computer and of Adobe Photoshop, but I find that to be a sneaky approach to performing the computation.<br>

In the long run, I find that going digital can be cheaper than going the analogue route. That said, as I stated, I do like film — just, let's not kid ourselves by pretending it's the affordable alternative to its modern counter-part.</p>

[jack_welsh]

<p>When I have an 8x10 printed from a slide at the camera store, $5.50, not over $17.00. ( I know they scan and print the photos digitally).A DSLR camera , I've seen them go for $600.00-700.00 and up.(Way up) I bought a 35 Rebel Ti for less than half. I do all my B&W processing at home. The very few 6x7 slides I do, are processed at the camera store. I no longer have an enlarger. But, since all I do is the old contact processes, don't need one. So, I can't comment on the price of paper.<br />That high end Hassy digital, I mentioned. BHphotovideo.com has it listed as a 39 MS for $43,995.00.<br />The only point I'm trying to make in these Digital vs Film debates, is that digital is really great, but it does it at a great cost. I made less than 30 digital negs up to 8.5x11". I need to buy another black ink cartridge, and, I have 2 more that are getting very close to needing to be replaced.</p>

[peter_ferling]

<p>I shoot both and scan my images to digital, and have thousands of them. I can't remember them all, but when browsing my images I do know without looking at the meta data which are of film origin. Only 10% of my digitals become selects over film choices.</p>

<p>Arguments abound over this subject of which is better, and everyone forgets the fact that you really cant compare because each one is a different format, and I find that those whom use both formats, tend to favor film for personal usage over digital. Which explains that it is more rewarding for the added effort of processing film as it's the end result we all want.</p>

 

[luca_stramare2]

<p><em> "As for Ken Rockwell....consider the source. His self-delusional fabrications are nearly legendary at this point...."</em><br>

I thnk they are not. I quote from the about page of Ken Rockwell website:<br>

 

<p align="left"><em>"This is my personal website. I do it all by myself. I'm just one guy with a computer who likes to take pictures. I have the playful, immature and creative, trouble-making mind of a seven-year-old, so read accordingly.</em></p>

<p align="left"><em><strong>This site is purely my personal speech and opinion</strong>, and a way for me to goof around.</em></p>

<p align="left"><em>While often inspired by actual products and events, just like any other good news organization, I like to make things up and stretch the truth if they make an article more fun. In the case of new products, rumors and just plain silly stuff, it's all pretend.<strong> If you lack a good BS detector, please treat this entire site as a work of fiction.</strong><strong> </strong></em></p>

<p align="left"><em>This site is provided only for the entertainment of my personal friends, dogs, family and myself. I've never promoted this site. If you're reading this, <strong>you got here on your own.</strong> "</em></p>

<p align="left">Sometimes I find myself in agreement with Ken and sometimes not. But I recognize the fact that he is selling his opinion and not the truth.</p>

</p>

[graham_thompson1]

<p>At school we used to see who could pee the highest up a wall.<br>

This was pretty childish but great fun at the time.<br>

As far as I know this trait is unique to boys.<br>

Now we discuss film v digital in fact anything v anything.<br>

I remember my father who was in the army listening to an endless discussion as to which was the best weponary.<br>

"Listen arsehole when that bullet hits you between the eyes it really matters not what kind of gun or who hand made it with loving care - you're dead."<br>

Same with an image. If it hits you in the eye it's a good shot.</p>

[mjferron]

<p>Folks as the original OP I'd just like to state that this was never intended to be a digital vs film debate. I'd be sad if I had to live without one or the other. My thoughts behind the post were about how easy it is to drop off a roll of print film and get a nice set of prints and a photo CD handed back. All done, very little fuss. An example I posted early on mentioned a backyard wedding I shot this past summer. I did it for free and shot raw with my D80. I edited 175 raw files and printed out 75 for the married couple. This service was free for a co-worker and I hardly received a thanks in return. If a were to do it again I would have used a 35mm film camera and handed them the processed prints and have been done with it. A $40 investment would have saved me a number of nights of messing around with the digital files. I'm not trying to tell folks what to shoot or how to process. I just think print film is becoming a forgotten tool in the bag.</p>

[marc_bergman1]

<p>Michael,</p>

<p>Have you ever taken your memory card from your D80 to wherever you get get your film prints made and see what they can do with the digital images?</p>

<p>Many places have print kiosks where you can do it yourself. I wonder how widespread this service is.</p>

[don_e]

<p>" At school we used to see who could pee the highest up a wall [...] As far as I know this trait is unique to boys."<br>

Well, yes, of course 8-)<br>

For those in need of a timeout after contemplating Rockwell's site:<br>

http://www.pbase.com/bret/image/71055999</p>

 

[mjferron]

<p>Marc I haven't but will consider giving it a try. Maybe I''l edit in photoshop and bring in a CD to see what they can do.</p>

[alan_rockwood]

<p>Commenting on the Rockwell link that Don E posted, http://kenrockwell.com/tech/film-resolution.htm, I have read the article.</p>

<p>Most of the technical points that Rockwell made are true. Summarizing a few of them:</p>

<p>As far as resolution calculations are concerned, a Bayer Sensor's megapixel rating has to be cut by about a factor of two. (Actually, a little worse than this, but let's not go there.) There is no way around this. It is a rigorous result that comes from the Nyquist theorem of digital sampling. (It would be even worse, more like a factor of three, except for the fact that the human eye perceives sharpness mainly in the green region of the spectrum, which is where a Bayer sensor is most heavily weighted in pixel-count.) Thus a 10 megapixel Bayer sensor is really a ~5 megapixel sensor as far as resolution is concerned.</p>

<p>Likewise, a foveon's megapixel rating (as given by Sigma) must be cut by a factor of three for the same reason. Thus, a 15 megapixel foveon sensor is really a ~5 megapixel sensor as far as resolution is concerned. Last monthin another place I gave my own version of this discussion: http://www.edn.com/blog/400000040/post/810011881.html?cache=FALSE&preview=TRUE&postComment=TRUE.</p>

<p>High resolution films are capable of well over 100 cycles per millimeter, far better than any current digital sensor on any 35mm sensor size digital camera. However, at these high resolutions the contrast is low.</p>

<p>Images from digital sensor cameras often look sharper than film because they hold contrast nearly up to their sampling limit, whereas contrast in film is already rolling off fast by then.</p>

<p>Rockwell's article is a bit weak in fully exploring the implications of the contrast roll off of film. It is not that he contradicts this. In fact, he does talk about it, including a discussion of the reason why digital images often look sharper than film, despite having objectively a lower resolution rating if one includes the low-contrast high-spatial-frequency image components of film. However, he does this in a way that seems to partially mask or ignore the implications of the contrast fall off of film at high spatial frequencies.</p>

<p>Consumer scanners (mostly 4000 dpi) cannot fully capture all the information in film. Note that if a scanner with more dpi is used it would be possible to digitally boost the high-spatial-frequency components of the image and produce a sharper image. However, this comes at the cost of increased noise. What this means in practical terms is that only modest degrees of sharpening are possible due to unacceptably high noise generated in the sharpening process. One could engage in endless debates about whether it is worth it to go beyond 4000 dpi, but theory holds that there is a potential benefit, albeit a small one.</p>

<p>This kind of resolution enhancement is not possible with digital sensors, which are already close to the theoretical limit imposed by the number of sample points. Actually, this statement includes a built-in assumption, which is that resolution of a given digital camera actully is already near the Nyquist limit. This is probably reasonably close to being correct in some cases. However, the anti-aliasing filter in front of some sensors may compromise this assumption somewhat.</p>

<p>That's enough technical discussion for a now.</p>

[danielleetaylor]

<p><em>As far as resolution calculations are concerned, a Bayer Sensor's megapixel rating has to be cut by about a factor of two. (Actually, a little worse than this, but let's not go there.) There is no way around this. It is a rigorous result that comes from the Nyquist theorem of digital sampling. (It would be even worse, more like a factor of three, except for the fact that the human eye perceives sharpness mainly in the green region of the spectrum, which is where a Bayer sensor is most heavily weighted in pixel-count.) Thus a 10 megapixel Bayer sensor is really a ~5 megapixel sensor as far as resolution is concerned.</em></p>

<p>You are confusing the terms megapixel and resolution (as in line pairs per millimeter), as well as confusing Bayer and Nyquist.</p>

<p>Megapixel simply means million pixels. If a device outputs a million pixels from a sensor with a million photo sites, then it is a megapixel device. No if's, and's, or but's. The colored Bayer grid does not reduce the number of pixels output. Megapixel is a way to describe the digital resolution of a sensor. Line pairs per millimeter is used to describe the analog resolution. Nyquist is how you convert between the two. (Another wrinkle to keep in mind is that megapixels is usually a description of the physical sensor, while lpmm is usually used to describe the results from a test which includes all aspects of a visual system.)</p>

<p>Bayer has nothing to do with Nyquist. A B&W sensor with no Bayer filter still obeys Nyquist theory. The Bayer filter is not the source of or reason for the application of Nyquist theory. Nyquist describes the sampling rate necessary to digitize a given maximum signal frequency. In a visual system the signal would be a change in tone. Traditionally we describe and measure this as line pairs per millimeter, which is really one black line (signal) against one white line (background). To reliably detect a given frequency of lpmm you need to have slightly more than twice as many pixels. Not 3x or 4x, but 2 and some fraction. 2.2x is sufficient and is the multiplier used in CD audio.</p>

<p>Nyquist does not mean that a 10 MP sensor is "really" a 5 MP sensor. This is a confusion of terms. A 10 MP sensor is always a 10 MP sensor. Nyquist tells you how to determine the analog signal resolution of your 10 MP digital sensor. Saying that Nyquist means a 10 MP sensor is really a 5 MP sensor would be like saying that a 30 centimeter ruler is really a 12 centimeter ruler because of the 0.39x conversion factor to inches. That's how you convert to inches, but the centimeter measure is the centimeter measure. It doesn't change.</p>

<p>You do not apply the Nyquist conversion again if there's a Bayer filter in front. Bayer alters the color value sampled at each pixel, but light is still sampled at each and every pixel, so your sampling rate does not change. For example, a Canon 10D has 135 pixels per mm. Nyquist theory says the maximum resolution of the sensor should be around 60 lpmm. If the Bayer filter in front of the sensor cuts this in half again, then we should only expect to get around 30 lpmm from a 10D. Norman Koren has published test results for the 10D and it resolves to 60 lpmm (http://www.normankoren.com/EOS-10D_3.html). The Bayer filter is not cutting resolution. (For that matter, the anti aliasing filter isn't doing any harm to resolution either, in this case.) If you removed the Bayer filter, you would still have a maximum resolution of about 60 lpmm.</p>

<p><em>Likewise, a foveon's megapixel rating (as given by Sigma) must be cut by a factor of three for the same reason. </em></p>

<p>Foveon's rating needs to be cut by a factor of three because that's the multiplier they use for misleading advertising purposes. Sampling RGB at each pixel site brings small improvements in tonality and per pixel sharpness, and makes the system immune to rare situations where a Bayer sensor might misinterpret color in details near the Nyquist limit. But it does not change the hard reality that a density of X pixels can only sample and reliably detect a certain maximum Y spatial frequency.</p>

<p><em>High resolution films are capable of well over 100 cycles per millimeter, far better than any current digital sensor on any 35mm sensor size digital camera.</em></p>

<p>This is only true if the signal contrast is 1000:1. I am hard pressed to think of any real life subject which would present high frequency detail at a contrast of 1000:1. Further, on most films that contrast would produce a monochromatic result, no color or tone.</p>

<p>We do sometimes find edge transitions in real life with that kind of contrast, but they occur at low frequency and are therefore easily resolved by either system. In the 35mm format, at 50 lpmm and higher you're basically talking about surface texture or small, repeating objects at a distance (foliage). These details do not occur at 1000:1 contrast. The details that do occur at 1000:1 contrast are relatively large, low frequency details, such as silhouettes against the sun, or edge transitions between objects with dramatically different reflectivity or lighting. There may be more than a 1000:1 contrast between regions of a scene with dramatically different lighting, but not between texture details which occur at 100 or even 50 lpmm.</p>

<p>1.6:1 contrast tests are a much closer approximation of the average contrast of the kind of details found at high frequencies. Most films score in the 50-60 lpmm range on these tests. A few exceptional films score higher.</p>

<p><em>Images from digital sensor cameras often look sharper than film because they hold contrast nearly up to their sampling limit, whereas contrast in film is already rolling off fast by then.</em></p>

<p>This is true.</p>

<p><em>Consumer scanners (mostly 4000 dpi) cannot fully capture all the information in film.</em></p>

<p>Unless you're shooting high contrast, high frequency tests, consumer scanners are sufficient to capture all the detail in the vast majority of photographs. There are some situations where high frequency detail occurs with a contrast higher than 1.6:1, and therefore the film resolves more of that detail and more than the scanner can pick up. Anyone can demonstrate this with a B&W line chart and access to both a 4000 dpi scanner and a higher resolution scanner. But test charts not withstanding, those are rare situations. One also has to ask just how much that small amount of additional detail adds to viewer perception of the final print. People respond much more strongly to contrast in lower frequency details, and lenses are designed with that in mind.</p>

<p><em>This kind of resolution enhancement is not possible with digital sensors, which are already close to the theoretical limit imposed by the number of sample points.</em></p>

<p>The upper bound imposed on resolution is the same one for film and digital: optics and diffraction. If a digital 35mm sensor is made which hits the limits of the format, then those same limits are an upper bound for film. In other words no film will be able to magically out resolve this theoretical sensor. If film can achieve some high resolution under 1000:1 contrast, then a digital sensor can be built to do the same. P&S sensors are close to the limits imposed by optics, but 35mm sensors have a way to go.</p>

<p>While I'm at it: Ken's article was a joke. If he intended it as anything other than a joke to draw traffic to his site, then someone needs to make 40x60 prints from 35mm Velvia and a 175 MP digital stitch and present them to him. Some how I doubt he would claim a 175 MP resolution for Velvia after that. Heck, you could change his tune with a print from a MF back or even a 5D mkII, assuming it's not purposely set to half resolution (see his review of the 5D mkII and his favorite settings).</p>

[wentbackward]

<p>I think people duped into buying digital camera's had it easy. I did not practice my photography and thought it had lost it's spark for me for years as I bought little digital cameras and they just didn't ignite any passion. Then a year or two back, finally some digital cameras with reasonable images came out, I got one, now I'm back to film!!</p>

<p>Let's face it we live in a mass produced world. People are led up the garden path by large corps and today not many people actually know how to cook to live, let alone people that have actually seen, for example, a whole fish on their plate with a head on it. In my home we eat whole food, we shoot film, we look back over our family snaps regularly, we go for walks outdoors in the forest. Most people sit at home, watch TV, play with their WII and take pictures of their socks and complain about how miserable their lives are and blame everyone but themselves.</p>

<p>I'd bet that most people in the film forums on PN are exceptions to the latter group in more ways than one.</p>

<p>Many of my favourite portraits were taken when film was not mass produced, people still managed to take pictures then, so film will never die as long as someone has a formula. The day that photography becomes harder, we'll all be better off as the passionate ones amongst us will no longer be lost in the chaff.</p>

<p> </p>

[alan_rockwood]

<p>Dennis,</p>

<p>The current practice among camera makers in listing megapixels is somewhat deceptive with respect to describing the sensors. When they say 10 megapixels they do not mean that there are 10 million green sensors, 10 millions red sensors, and 10 million blue sensors. They mean that (if it is a Bayers sensor) there are 5 million green sensors, 2.5 million red sensors, and 2.5 million blue sensors. The three types of sensor are interleaved on the chip. They produce three images, a green image sampled at 5 megapixel, a red image sampled at 2.5 megapixel, and a blue image sampled at 2.5 megapixel. There are 10 million sampling points in total, but not for each color channel. Foveon is an exception to being interleaved, but they have their own deceptive pixel specification scheme.</p>

<p>The Nyquist sampling theorem prevents the green image from having a resulotion any higher than that of a pure 5 megapixel green sensor. The Nyquist sampling theorem prevent the red image from having a resolution any higher than that of a 2.5 megapixel pure red senor. The Nyquist sampling theorem prevents the blue image from having a red image any higher than that of a 2.5 megapixel pure blue sensor. These are absolute hard and fast limits an no amount of image processing can ever restore information lost in the sampling process.</p>

<p>As a side comment, the spatial resolution of a sensor (lines per mm) is determined by the number of of pixels and the sensor area. Those two parameters determine the sensor spacing, which in turn sets the upper resolution limit. Note: for a give aspect ratio there are three parameters. Sensor spacing, number of pixels, and sensor area. You can specify any two and the third is determined by the algebra.</p>

<p>The camera processes the three separate rgb images an produces an output of an image of non-interleaved 10 megapixels in each color channel. It does this by interpolation, which basically means filling between the sampling points using a mathematical algorithm (i.e. a calculation with a built in guess as to what consitutes a valid value for the data in between the sampling points.) For example, the new image for green is composed of 10 million points, whereas the original green image was composed of 5 million points. In between those 5 million points there were some "missing" points, which were occupied by the red and blue sensors. There were no green sensors at the red and green points (foveon sensor being an exception), so there was no green information at those points, so they were "missing" in the original data array as far as green is concerned.</p>

<p>Now, back to interpolation. The interpolation scheme tries to fills in the information between the original sampling points using some kind of educated guess. The educated guess is determined by the algorithm (i.e. calculation method) used to fill in the missing points. This is a form of image processing, and as already mentioned there is no image processing algorithm that can increase the resolution of an image beyond that set by the fineness of the sampling grid. Therefore, the 5 megapixel green image in the original data array can never be interpolated to give a true resolution of a 10 megapixel array. Any apparent resolution beyond the Nyquist limit is at best an interesting fiction. The increased apparent resolution may or may not correspond to the real image. Generally speaking it will not. One can assume that any spatial features smaller than those resolvable by the physical array (5 megapixels in the case of the green image) is an artifact of the computation and should not be considered real. (In some cases artifacts due to sampling can be larger than the sampling grid due to aliasing.)</p>

<p>Now another word about the interpolation scheme: It may be that some schemes try to fill in the missing information in the green array by using information from the red and and blue arrays. The problem is that the red and blue arrays do not contain green information. Therefore, any attempt to fill in the missing information in the green array by using the red and blue arrays is doomed to failure because the information simply is not there to begin with. (I am ignoring a subtlety about overlapping spectral response between the sensor types. It turns out that this doesn't actually help.)</p>

<p>I discussed this so far mainly in terms of the green array, which is the most favorable case because the green array samples at the highest density. However, the same general considerations apply to the red and blue arrays as well, except that they sample on a less dense sampling rate and are therefore even lower resolution than the green array.</p>

<p>I need to go to work. I will discuss lens resolution and how that fits into this scheme later.</p>

<p> </p>

[roger_smith4]

<p>Michael- how did your self-scanned and processed film go? You mentioned you were going to try it above? I'd like to see the difference you find and the trade off of time vs quality.<br>

<br />The rest of this disucssion is pretty pointless.</p>