large sensor for large format

<p>Dear everyone,<br>

These questions have been circulating in my head for years:<br>

In the past several years, I have sold the medium format, bought the DSLRs but still keeping the large format camera with the hope that, one day, there will be affordable 4x5 inches sensor for large format camera. But I wonder will it happen in the coming 20 years? Also, seeing how close IQ is the medium format digital compared to the large format film, do we really need a sensor as large as 4x5 inches?<br>

Thank you in advance for clearing out my head for these questions.<br>

Paul</p>

<p> </p>

<p>Been around for years, look at Betterlight's scan backs, as well as several other manufacturers (Arigram, Phase One, as well as others)</p>

<p>From a quality standpoint, the latest 50+mp backs <em>theoretically</em> can acheive the same resolution, however there are numerous technical reasons that make it very, very difficult to achieve the level of resolution that high resolution MF backs are capable of. For a good analysis of this issue, see Joseph Holmes excellent article here: http://www.josephholmes.com/news-medformatprecision.html</p>

<p>With larger photosites a scanback has less issues with lens quality, virtually any recent lens (made in the last 50 years) will give you more sharpness than the back is capable of. Capture time and subject movement are the largest downside, as exposures in the minutes are the norm. From a bang for the buck, even a new Betterlight gives you far more than the 50+mp backs cost (both priced new), if you can live within the limitations.</p>

<blockquote>

<p>But I wonder will it happen in the coming 20 years?</p>

</blockquote>

<p>Real, honest to God 4x5 single shot sensors? That's never going to happen. But, as Eric mentioned, there are scanning backs from Better Light and Seitz (the other two he mentioned are out of the game).</p>

<blockquote>

<p>Also, seeing how close IQ is the medium format digital compared to the large format film, do we really need a sensor as large as 4x5 inches?</p>

</blockquote>

<p>That depends on what you shoot. There's still a demand for "get your face close to it" large prints. I often do "lobby art" at 360dpi on a 44x66 inch print, that's 376mp, the limit of the highest resolution scanning backs. I get there a different way, by stitching. It's nearly impossible, for a number of technical reasons, to get that sort of resolution out of a 4x5 scanning back (let alone MF) for the reasons that Joseph Holms laid out in the article that Eric linked.</p>

<p>Basically, the practical lens resolution limit for the best, most modern "digital lenses" for 4x5 is 60lp/mm, which is 180mp. ((60 lp/mm * 2 pixels/lp)^2 * 100mm * 125mm). You can crank the scanning back up to the advertised 340mp (or scan 4x5 film to a higher resolution, and get sharp images of the grain) but it's empty resolution, there's no added detail. And that 180mp involves absolutely perfect technique.</p>

<p>I don't think we'll ever get clean 180mp, let alone 376mp, out of a MF back. The current trend in MF is for downsized sensors, for faux affordable cameras. 33x44mm at $10-15,000, 36x48mm (or 37x49mm) the old "standard" at $20,000, and 42x53mm at $30,000. Assuming someone did manage to put 160mp into a MF digital back, you'd need...</p>

<ul>

<li>152 lp/mm for 33x44mm</li>

<li>137 lp/mm for 37x49mm</li>

<li>127 lp/mm for 42x53mm</li>

</ul>

<p>It's just not happening. What I'm seeing from the current crop of 50-60mp MF backs is that they've pretty much hit the wall, and we're drifting right into empty resolution. So, if you want to push past about 50mp, the only way is with a camera that stitches nicely.</p>

<p>I love how you guys are predicting what is impossible to happen in 20 years, think back to what the state of the art was in 1990 (twenty years ago) . . .</p>

<p>Steve, 20 years ago, lens resolution for the best primes was just about the same as it is today.</p>

<p>And 20 they were playing with APS size sensors, because those could be done in a single shot on IC steppers, and a couple of aerospace sensor manufacturers were using stitching techniques to produce larger sensors. And that's where we still are, today.</p>

<p>Here I have a 35 and 50 megapixel Phase One 4x5 scan backs. They both are cropped; like all 4x5 scan backs are. Mine scans 7x10cm area. I got one used in 1996; another used in 1998.</p>

<p>Thus here at least I have had folks ask me this question:</p>

<p>( WILL THERE BE )</p>

<p>"one day, there will be affordable 4x5 inches sensor for large format camera". ???</p>

<p>This at least for me using LF scan backs; the question is now 14 years old</p>

<p>One has theses LF backs now for about 16 to 17 years of the 20 years of digital cameras ; thus the question is real old</p>

<p>It is so old a question that my 35 megapixels scan backs recommended CPU is a Pentium; a 486 is usable but slow. My first scanning computer was our 3 grand Photoshop dream machine; one folks faught over; a 75 Pentium.</p>

<p>(1) You have to ask yourself if these LF backs have been around now for 16 to 17 years ; what magical thing will drop their price?</p>

<p>(2) There has never been a full frame 4x5 sensor/scan area yet thus what makes one do-able in the future?</p>

<p>(3) Folks who need these have mostly bought them already; thus the market is less than in prior times</p>

<p>*SURE*; most folks here want:</p>

<p><br /> (A) a low cost full frame 500 megapixel 4x5 back<br /> (B) Kodachrome to be made and processed forever<br /> © E6 labs to be right next door in every decent city<br /> (D) a giant paid for 300 foot yacht with babes and trust fund to support it <br /> (E) gold bars popping out of the street</p>

<p>A rational maker for a LF back sees 16 to 17 years worth of proven palty sales; and folks wishing they would be sold below cost.</p>

<p>It probably will not happen in another 20 years; there is already a 17 year track record.<br>

<br /> It is the same with farm tractors, metal lathes, bulldozers and big boats; we WANT them to be low in cost; but a functioning one is often still not affordable for a casual user.</p>

<p>This question pops up about 2 to 3 times a year on photo.net. If you ignore design; development; service and marketing costs you can say they should be lower in cost. Folks parachute in an say it is all due to the middleman; or it cost nothing to make a big sensor or big scan bar.</p>

<p>One has 17 years worth of palty sales over 3 or 4 players; it is a dying market; like the CRT market</p>

<p>Actual sales of LF backs is more like dialup modems or CRT's or APS cameras; or Bitney Spears music; it has past already.</p>

<p>Thus a rational person would not enter the LF back market</p>

<p>It is actually real super easy to get a 35 or 50 megapixel scan with my two LF 4x5 backs.</p>

<p>The pixel density is rather large; about 14 and 12 microns.</p>

<p>This makes the lens requirements just modest.</p>

<p>An 150mm F9 Apo Ronar or 135mm F5.6 Componon work real well over the entire 7x10cm area. Even a dumb Tessar is overkill in the central core; but is not enough for the far corners. The lowly 127mm F4.7 Kodak Ektar in the central core out resolves the 50 megapixel sensor. </p>

 

<p>The first ( not for newspaper relay usage) Nikon film scanner for 35mm was over 4000 dpi; it was marketed about 1988 to about 1992ish. Then Nikon came out with the LED coolscan series and the dpi dropped to 2700 about; later ones went up to 4000; now Nikon does not even make scanners any more</p>

<p>It really is older than any of us around that a given format only tends to capture so much info. It is OPTICS 101. You have the wavelength of light; and the quality of that plate in 1860; or film in 1910; or sensor in 1990 to today.</p>

<p>You are not going to get 100 horsepower out of a Cox 0.049 motor.</p>

<p>You not going to get 100 megapixels out of a 35mm slide</p>

<p>Every thread on photo.net where wildly high numbers are quoted; the cheerleaders ignore reality of the roll-off responses; ie transfer function of film; of lens; of the scanner; or the object one shoots.</p>

<p>part of the cheerleaders gambit is best case-ing a long string of things to "prove" the wildly high numbers</p>

<p>The entire concept of design margin is trampled with these "lets cook the numbers" folks</p>

<p>"Arigram"</p>

<p>I think you meant Anagramm. Which is more popular in the EU then the USA.</p>

<p>Kelly brings up some excellent points.</p>

<blockquote>

<p>You have to ask yourself if these LF backs have been around now for 16 to 17 years ; what magical thing will drop their price?</p>

</blockquote>

<p>There's only one or two things that can do that. One is the pieces becoming commodities because another market makes it so. There's no sign of that happening. The other is increased demand, and judging by the dearth of new product introductions, that isn't happening, either. In fact, we're seeing companies either</p>

<ul>

<li>pull out entirely, like Phase One</li>

<li>discontinue half their product line with no sign of updates or replacements, like Better Light, who has the words "sold out" on half their line.</li>

<li>pull back and target a specialty niche, like Anagramm, going after the art repro market and putting more emphasis on their copy stands than their scan backs.</li>

<li>gone smaller, like Seitz, the only company using 60mm sensors in a 6x17 back or the "roundshot" scanning cameras.</li>

</ul>

<blockquote>

<p>There has never been a full frame 4x5 sensor/scan area yet thus what makes one do-able in the future?</p>

</blockquote>

<p>Something that the industry just doesn't have: higher demand. Right now, scan backs "borrow" document scanning tech. That industry is quite happy with 72mm wide sensors. While it's possible to stagger a couple of 72mm sensors to make a wider effective one, one of the bars is going to be riding slightly behind the other, and that means there's going to be an outright "tear" in the image if something moves.</p>

<p>I wish it weren't so, but LF scanning backs seem to be a dead field. Better Light and Anagramm last introduced new products about 6 years ago. And, in Anagramm's case, "salvadore" appears to have been an attempt to increase parts commonality between their high and low end lines, dropping the chip used in "daguerre". Seitz launched D3 back in 2007. That's it: the industry only snores once every 3-4 years.</p>

<p>Oh, I should point out that Seitz, the only one of the current players who appears to be in anything resembling a "growth mode", charges $34,000 for their 6x17 D3 scan back.</p>

<p>By some standards, that's a bargain. The D3 is, essentially, a "consumer" version of a DALSA "Piranha" machine vision camera, mounted to a small scanning bed. It's the only scan back to use a TDI sensor or colorimetric filtering.</p>

<ul>

<li>Scan backs traditionally use off-the-shelf document scanning sensors. The filtration is set up with minimal overlap between the red, green, and blue channels. This allows the most information to be extracted from printed images, when you know the spectral characteristics of the image. That's why Phase One offers HP "ColorSage" spectral management software for the art reproduction customer. But for general purpose photography (mostly product and landscape for the scan back market) you get large problems with metamerism (failures in observer metamerism, for the pedantic). The Seitz has a "machine vision" heritage, and the sensor has more "vision like" overlaps in the red, green, and blue responses; it's more "colorimetric".</li>

<li>For those unfamiliar, TDI sensors use multiple, parallel lines on one chip. The sensor is more like a digital camera sensor, an array. It's just a very rectangular array, 7500x96 pixels. They can either accumulate in great detail, by using a scan rate that has each of those 80 lines pass over one line of the image, or scan at a higher rate and have each image line be every other, every third, etc. line of the TDI. You can then sum the overlapping lines to increase dynamic range, or ignore them to increase scan speed (think of the image as a "stitched panorama" made of 7500x96 pixel blocks. Theoretically, the Seitz can preview an image at full resolution (but worst-case noise) for the 617 in 1/5 second. I don't know if it can actually move that fast, though. </li>

</ul>

<p>And that's the cost of maintaining current technology.</p>

 

<blockquote>

<p>I don't think we'll ever get clean 180mp, let alone 376mp, out of a MF back. The current trend in MF is for downsized sensors, for faux affordable cameras. 33x44mm at $10-15,000, 36x48mm (or 37x49mm) the old "standard" at $20,000, and 42x53mm at $30,000. Assuming someone did manage to put 160mp into a MF digital back, you'd need...</p>

<ul>

<li>152 lp/mm for 33x44mm</li>

<li>137 lp/mm for 37x49mm</li>

<li>127 lp/mm for 42x53mm</li>

</ul>

<p>It's just not happening. What I'm seeing from the current crop of 50-60mp MF backs is that they've pretty much hit the wall, and we're drifting right into empty resolution.</p>

</blockquote>

<p>I'm not sure what "empty resolution" is or what "wall" MF sensors are hitting. Further, I doubt that future MF backs will require insanely high lppm numbers.</p>

<p>Every time Phase One increases their pixel count they expand the size of the sensor. The P65+ sensor is larger than the P45+. The beauty of this is that they haven't even reached the full size or a 6x4.5 negative yet. 645 format could conservatively reach pixel counts 120 MP or more. If they expand the size to 6x6 or 6x7 sizes, 200-300 MP could be possible from an MF back.</p>

<p>The idea of using a scanning back would be intriguing if the world stood still. I can't conceive of using a back that takes nine minutes to make an exposure. The beauty of the MF backs is that they capture everything all at once just like a piece of film. A very LARGE piece of film.</p>

 

<blockquote>

<p>Every time Phase One increases their pixel count they expand the size of the sensor.</p>

</blockquote>

<p>No, not every time. They actually launched a series of smaller sensor backs (33x44mm) where they upped the pixel counts past what they were using in the standard (36x48mm) backs.</p>

<ul>

<li>P45+ (33x44mm) is 40mp, P40+ (37x49mm) is 39mp</li>

<li>P30+ (33x44mm) is 31mp, P25+ (37x49mm) is 22mp</li>

</ul>

<p>Basically, they have three pixel sizes, 6, 6.8, and 9 micron, and each of those is used in 2 or 3 different sized sensors.</p>

<blockquote>

<p>The beauty of this is that they haven't even reached the full size or a 6x4.5 negative yet.</p>

</blockquote>

<p>They're pretty close. P65+ is 40.4x53.9, with 6 micron pixels. 645 is only 42x56mm, which is 65mp. You won't notice much difference going from 61mp to 65mp.</p>

<blockquote>

<p>645 format could conservatively reach pixel counts 120 MP or more.</p>

</blockquote>

<p>There's no physical reason. Canon just announced an experimental 120mp APS-H (29.2x20.2mm, or 1.3x crop relative to FF) sensor. <a href="http://www.dpreview.com/news/1008/10082410canon120mpsensor.asp">(link)</a></p>

<p>However, there is an economic reason. Phase One is not in the driver's seat, as far as sensor development. They don't have the R&D budget to go to Kodak or DALSA and say "we need a new, smaller pixel sensor". All they can afford is to say "we need a 40.4x53.9mm sensor with your stock 6 micron pixel. The stock pixel sizes are what Kodak's and DALSA's aerospace and document handling customers funded.</p>

<ul>

<li>12, 9, 6.8, 6, and 5.4 microns.</li>

</ul>

<p>The P65+ used the 6 micron cell right now. So, the highest resolution we'll see in a full 645 from P1 is a full 42x56mm with the 5.4 micron cell: 81mp. That's where medium format hits the wall.</p>

<blockquote>

<p>If they expand the size to 6x6 or 6x7 sizes, 200-300 MP could be possible from an MF back.</p>

</blockquote>

<p>6x6 is dead, gone, kaput. All the 6x6 makers went bankrupt in the mid 80s (yes, Blad too). All subsequent attempts to revive 6x6 have met with financial failure. 6x7 is possible, Phase One owns the remains of Mamiya, Shriro puts the Blad label on Fuji gear. A full 56x66mm 6x7 with those 5.4 micron pixels would have a resolution of 127mp. But consider this...</p>

<ul>

<li>33x44mm - $10-15,000</li>

<li>36x48mm - $15-20,000</li>

<li>40x54mm - $25-30,000</li>

</ul>

<p>Near as I can make it, that puts MF at about $11/mm2. So...</p>

<ul>

<li>56x66mm - $40,000</li>

</ul>

<p>And then, consider this...</p>

<p><strong>I have yet to meet a MF digital shooter who shows any sign of actually understanding the relationship between diffraction and resolution.</strong> With a 5.4 micron pixel sensor, you can't stop the lens down any farther than f11, or you start having diffraction limit your resolution to below that the sensor can resolve. How many MF shooters do you see stopping the lenses down to f22 or f32 pretty much every time they shoot in the studio? That's a leftover from 160 or 400 speed film, or those old 12 micron pixel sensors. So, all that $40,000 gets the MF shooter is the ability to say "you need me to shoot this, because my gear costs more than my competitor's gear". Empty resolution and bragging rights.</p>

<p>(now, to be fair, I don't know that many FF shooters who really have a good handle on diffraction, either).</p>

<p><em>The beauty of this is that they haven't even reached the full size or a 6x4.5 negative yet. 645 format could conservatively reach pixel counts 120 MP or more.</em></p>

<p>Nope. Or at least, it would be wasted. The problem is the lenses. They place some hard limits on things. Getting 120 MP out of a 41.5 x 56 mm sensor means the lens must deliver 114 lp/mm. That's about twice what the lenses can deliver in linear resolution, at most working apertures; in other words, as you get above about 30 MP, there are few lens-and-aperture combinations that will take advantage of the higher-resolution sensor. I realize that not all lp/mm are created equal, and not all of you put a ton of faith in <em>Popular Photography</em> magazine (US), but as an example, their test of the Hasselblad HC 80mm f/2.8 found that its best performance was at f/11, where the lens delivered 66 lp/mm in the center and 47 lp/mm in the corners. In other words, even with 'full frame 645' and the optimum aperture, the most this lens can capture is 40 MP. In fact, of the ten or so modern medium format lenses they've tested in the last decade or so, the very best performance of any of them at any aperture and any part of the lens' coverage was the Contax / Zeiss Planar T* 80mm f/2, which delivered 84 lp/mm at the center at f/5.6 (but only 28 lp/mm at the corners)--so even with 84 lp/mm, any 645-size sensor over 66 MP is wasted.</p>

<p>Of course, don't get me started on, to make an example, the Digital Rebel T2i / 550D. That 18 MP sensor would need 116 lp/mm for full-resolution results (neglecting the limits of the anti-alias filter)--how many of its kit lens' focal-length-and-aperture combinations will deliver that?! <em>Any?!</em></p>

 

<p>The only reason that Betterlight has been forced to discontinue backs is that Kodak withdrew from the linear sensor market, they bought all they could prior to the discontinuance, but there's a finite limit to how much a company can speculatively buy.</p>

<p>Medium format is still produced in Germany. See <br>

http://gadgets.softpedia.com/news/Rollei-Still-Available-For-Purchase-in-Three-Models-at-DHW-Fototechnik-8355-01.html</p>

Joseph,

 

Note that I said that every time Phase One increases resolution they increase sensor size. P45+ to P65+ is an increase in

resolution. P40+ is a miniaturization of a P45+ sized sensor.

 

120 MP on 645 format is possible at the P40+ pixel density.

 

Everything in photography is a trade off. Need a fast shutter speed? Lose some depth of field. The diffraction bogeyman

isn't going to eat anybody. F/16 won't be as sharp as f/11, but f/11 won't be as sharp as f/8. So what! A photograph can

never be optimized in every possible way. We always lose a little here to gain a little there. You can do a lot with f/11 and

movements or focus stacking techniques. At least Phase sensors don't have the AA filter issues that small-format digital

shooters deem acceptable every day. The Phase sensors are highly regarded. Somebody out there seems to have figured

out how to get a good picture out of them.

<p>Erie, thank you. Do you know if they're planning on a switch to DALSA sensors or if they're going to shut down?</p>

<p>Dan, thank you for that rather condescending analysis of basic tradeoffs. For the record, AA filters aren't just "deemed acceptable" by anyone, they're expensive and if the camera companies thought that they could get away with leaving them out, they sure would. I've had MF images ruined (as in so many hours of post needed that you might as well just redo the shoot) by the lack of an AA filter.</p>

<blockquote>

<p>Somebody out there seems to have figured out how to get a good picture out of them.</p>

</blockquote>

<p>That flies in the face of the reality of the MF industry having decreased sales every single year for the last 20.</p>

Joseph, there was nothing condescending in my comments. Do you assume that Newton's law about equal and opposite

reactions is condescending? Reciprocity law is an observable phenomenon and it creates trade-offs that we cannot

escape when exposing images.

 

MF sales are down because today's small format cameras produce high-quality results. You can now do with a Canon or a Nikon

what you used to need a Hasselblad or a Mamiya to achieve. That doesn't discount the ability of Phase One's best sensor

to deliver results that exceed the capabilities of anything less than 8x10 film at this point in history.

 

If something is bothering you, or if you just really dislike MF digital backs for some reason, I'm sorry. I didn't mean to

upset you by suggesting that this gear is worthy of consideration. Scanning backs are amazing, but I don't feel that they

are practical for any subject that moves. Personally, i'd rather have a Phase One in a good technical camera - if I could

only afford one!

<p>So —lets say a 4x5 one-shot sensor IS possible and can be sold for less than a Grand. What super computer wil we have to use to open this multi-Tera-byte file? Even if its possible, I just don't see it hapening.</p>

<p><em>So —lets say a 4x5 one-shot sensor IS possible and can be sold for less than a Grand. What super computer wil we have to use to open this multi-Tera-byte file? Even if its possible, I just don't see it hapening.</em></p>

<p>To say that anything approaching a 4x5-size sensor will never happen, or will never be even remotely affordable, is amazingly silly. Sure, we cannot foresee <em>how</em> or <em>when</em> it will happen because our perspectives are so tied to today's technology. And maybe when it's readily possible, nobody will have the urge to do it. But lots of relatively intelligent and informed people thought that heavier-than-air flying machines were impossible, too. Maybe in ten years my son will be in college and fabricating 4x5-size sensors as part of an undergraduate course lab session, using $100 worth of materials and a machine that the college lets undergrads use. By the time we're talking about my grandchildren, wanna make any firm bets?</p>

<p>As to processing it, put aside the point that computer speed has been increasing dramatically for so many years (I had a Sinclair ZX-81 with a 1 MHz, 8-bit Z80A CPU and 1 KB RAM, and am sitting here using an ancient Dell with a 3 GHz, 32-bit Pentium 4 with 2 GB of RAM; in 24 years, my computers' CPU clock speed times processing word size went up by a factor of 12,000 and RAM went up by a factor of 2,097,152). And look at todays 65 MP sensors. I don't want or need a phenominal resolution, just a sensor that works well with rise and tilt. Indeed, the diffraction limits of the lenses at any relatively normal working aperture mean that for most uses it's fairly pointless to use a back of over about 200 MP--which means 1.2 GB for a 16-bit TIFF.</p>

<p>One last thought: CNC machining systems have made relatively small runs of high-precision parts and even whole machines economically and technically viable. I'd hate to assume we'll never see their equivalent for fabricating IC's and sensors.</p>

 

<blockquote>

<p>So lets say a 4x5 one-shot sensor IS possible and can be sold for less than a Grand.</p>

</blockquote>

<p>It is, and it can be. I'm looking at a 30 inch monitor that is an active matrix LCD panel, with 12 mp (the way cameras count mp) and two transistors per pixel. That switching circuitry, transistors and diodes, is grown on a sheet of glass. The same sort of tech that makes LCD panel circuitry, solar cells, etc. can make 4x5 single shot sensors.</p>

<p>The only barrier is start-up capital. And the only barrier to getting that is demand. In an industry when product updates happen every 3 years, it's hard to convince anyone you've got enough potential business to justify investing.</p>

<blockquote>

<p>What super computer wil we have to use to open this multi-Tera-byte file? Even if its possible, I just don't see it hapening.</p>

</blockquote>

<p>First off, they're not "multi-terabyte" files, their only a few hundreds of megs. 4x5, at the limits of lens resolution, is about 160mp, or 960 megs for 16 bit, uncompressed RGB. You can buy 2 terrabyte USB flash drives, cheap, enough to store a couple thousand of the highest resolution 4x5 images. How many shots does the typical 4x5 shooter make in a day's work?</p>

<p>Second, those 160mp backs have been on the market for over 8 years. People process the images on laptops, old desktops, new desktops, etc. 8 years ago, I was stitching 25, 36, even 81 6mp images together into giant, high resolution images, on a desktop. Or scanning 4x5 film to 115mp (my scanner only does 2400dpi on 4x5. It has a 4800dpi setting, but honestly, the prints don't look any different, the limit of the machine is honestly 2400dpi). Yes, they didn't "snap" around in Photoshop the way a 6mp image from the D100 did, but they were a lot quicker to work with than burning and dodging in the darkroom.</p>

<p>Remember Moore's law? Processing speed and memory capacity doubles every 18 months. 8 years ago, the MF back makers had 22mp models out, and DSLRs were nibbling at 11mp. 8 years is 5 Moore "generations". 5 "doublings" is 32x more speed and capacity. You can buy a desktop today that can process a 160mp image quicker than a 2002 model could process an 11mp image.</p>

<p>It's 2010, and some people really are working with gigapixel images, typically by stitching DSLR images. There's even commercial automatic stitching hardware, like "Gigapan".</p>

<blockquote>

<p>Sure, we cannot foresee <em>how</em> or <em>when</em> it will happen because our perspectives are so tied to today's technology.</p>

</blockquote>

<p>Actually, we can. We don't have to see what the future technology will look like, we just have to be able to anticipate it being there. I learned this stuff from Ted Kiernan, who literally wrote the book on it. You're dealing with something called the "frequency of miracles". You go back and look at similar situations (I'd put the current large active transistor arrays, printed on glass, that make LCDs work as "similar"), what the obstacles were, and how many years an obstacle stood before a "miracle" overcame it. Then you apply that frequency of miracles to the new technology.</p>

<p>As far as I can tell, we're so far past the level of technology we need for large, cheap FF sensors that the fact that they don't exist serves to indicate that the demand level is incredibly low. You have a very valid point, someone developing the actual tech for reasons that have nothing, at all, to do with demand, could pop up with it at any time.</p>

<p>A one shot full frame 4x5 digital back could happen; so could practical low cost flying cars like I have seen in Popular Science for the last 50 years.</p>

<p>The "sell for less than a grand" is a problem; a "one shot full frame 4x5 digital back" would have to cost 50 to 100K if one looks historically at how oddball, obscure, low volume stuff has to be priced.</p>

<p>To make a "one shot full frame 4x5 digital back" would require a rich person who really wants to make a cool toy; than any sane businessman would avoid. It could happen; look at the super riches yachts.</p>

<p>Making a one shot sensor for a full frame 4x5 would expensive; maybe 10 to 100 million bucks. Lay folks cannot understand that doubling the sensors size causes the cost to more than double. Thus one has the Beavis and Butthead dream that the cost will drop. Doubling can cost 20 to 40 times more; but the Beavis brain equates it to pouring concrete; ie double.</p>

<p> </p>

<blockquote>

<p>To make a "one shot full frame 4x5 digital back" would require a rich person who really wants to make a cool toy; than any sane businessman would avoid. It could happen; look at the super riches yachts.</p>

</blockquote>

<p>Isn't that what happened with Leica, Andreas Kaufmann wanted it, and used his interest in ACM to acquire it? Leica collector, on a large scale...<br /></p>

<blockquote>

<p>Lay folks cannot understand that doubling the sensors size causes the cost to more than double.</p>

</blockquote>

<p>I'm not "lay folks", that's why I proposed an alternate technology. Single crystalline silicon is insane for a 4x5 sensor, one part per wafer. It's a technology that's useful for 35nm features. Even regular digital camera sensors, with their 1 micron features and 6 micron cells aren't pushing the tech. You can build a nice 160mp 4x5 sensor with 9 micron cells and 1.5 micron rules. My first instinct for a large sensor is polycrystalline silicon on glass. Use LCD technology: I'm not sure what the state of the art is for poly-Si, but EVF LCDs use 10 micron cells and 1-1.5 micron rules. It's also worth examining organic semiconductors.</p>

<p>Microcrystalline is probably the best choice, high electron mobility and good red sensitivity. State of the art is about 10nm crystals, which means that 100nm features should be a snap, and that a 9 micron cell would contain millions of nanocrystals and perform very consistently.</p>