Ultimate digital chip for LF

[chris jordan]

Eric, that's right where I am too-- watching with interest, while still shooting 8x10 Astia 100F and negative films. They are still better than anything I've seen with digital, though I've heard that the Betterlight scanning backs are amazing. My dream piece of equipment is an 8x10 digital capture device that's about the same size and shape as an 8x10 film holder. You slide it in the back of the camera, pull out the dark slide, make an exposure just like normal, and within a minute or two it records a 2GB file right into its battery-powered 100GB internal flash memory. Now THAT would rock, eh?

[bglick]

Eric, as for the digital lenses, they can cover 6x9, but its probable the loss of imaging area will compensate for the added resolution, getting you back to where you started. The point I was making about this was.... as you also mentioned, the advancements in technology such as digital lenses are being geared towards digital, another strike against film in the years to come. Not only will there be no new LF lenses, I would be quite surprised if the current line up continues to be manufactured.... speically by Fuji...and I beleive Nikor already quit.....Fortunately, I bet Schneider will stay the course... Rodenstock ?? (I am sure Bob will tell us)

Agreed on your outlook of color film, we are safe for a couple of years, the scary part is, if and when they decide to cut film, it could be abrubt.... its seems lots of us are on the fence waiting for the digital product that will push us over the edge... sad...but...it's reality...

Chris... BetterLight backs do produce wonderful images, and for still photography, they are tough to beat. But for landscapes, I can NOT swallow 15 minute exposure times...sometimes more for sunsets. I also hate the odd effects they produce when the light changes or wind slightly moves tree branches on different lines of the scan. BetterLight reps coin this as artistic touch.... I see it as un predictable results which you often do NOT have the opportunity to re shoot - as quite often with landsacpes, you only have a short window of opportunity to capture a scene. As fast as the single shot backs are progressing, I question how long scanning backs will be around. It seems more prudent to buy one back that can perform both tasks, vs. buying two backs.... so I think the digital progression will diminish the market of scanning backs to the point of extinction, IMO. Just like film scanners, a very short product lifespan.

For color, IMO, 8x10 neg film is really the cats MEOW, as it offers more than sufficient resolution (although not as much as chromes, but surely enough for most print sizes up to 60") and has the huge advantage of much larger exposure lattitude. The only compromise is, you have to be careful with the near / far distances as those long fl lenses are not DOF friendly. I have learned to pick n choose my scenes carefuly based on the gear / film I have with me.

Carrying a 50 MP back on a small view camera set up would open up tremendous possiblities all in one camera system. But like you, I still would like to see a bit more improvements in exposure lattitude and the use of wide angle lenses.... which IMO is probably digitals biggest shortcoming. Till then, I use film!

[MJSfoto1956]

Please tell us where the LL DVD is "pseudo-science." How is the scanned film vs a RAW digital shot flawed?

Dear David, I've already stated it as clearly as the English language will allow: scanning film attenuates any existing grain, making it "grainier" than it really is. In other words, scanning film ALWAYS puts film at a disadvantage. The only fair apples-to-apples comparison is to NOT scan the film, but examine the printed output.

Secondly, you are fooling yourself if you think that RAW import even at "zeroed" settings doesn't still clean up the data significantly. ALL popular RAW importers agressively eliminate Bayer artifacts (and then do a small amount of sharpening to boot even if you say "none"). This is why you've been led to believe the data is noise-free even when hidden deep inside the RAW file it is a mess. Your problem is simply that you do not own a tool that extracts the pure data without cleaning it up. If you did, you'd know that what I am stating is true. Sadly, RAW ain't RAW.

I think you have a little bit of reading to do on this topic as there appears to be a bit of apparent bias in your statement.

Hah! Ok I'm biased: against marketing hype and the people who continue to blindly promote it. As for reading, perhaps you should read my book on digital scanning written back in 1994 (revised in 1996). In the 1990s I lectured extensively at Seybold Seminars and MacWorld on the topic of digital scanning. FWIW, I've been scanning film since 1988 and have used every version of Photoshop since v1.0 (actually beta v1). I also was an early adopter of digital photography for advertising doing my first studio photoshoots back in 1993 with a three-shot 2k x 2k Leaf Back (man was that fun!). I've never stopped shooting/printing with digital ever since.

I'm afraid you simply have it wrong: I'm a BIG fan of digital -- it's just that all of my comparisons to date with Bayer arrays have been against a BetterLight scan back which delivers about as pure digital data as you can get**. And I find the comparison wanting. I simply distrust marketing hype and know it to be false when it patently is so (and I've got nearly 25yrs in advertising and I know what I'm talking about). And lastly, I'm not particularly fond of "groupies" who embrace whatever new fad comes around.

J Michael Sullivan
Editor/Publsiher, MAGNAchrom
www.magnachrom.com

** A BetterLight 16-bit digital file -- without any post-processing whatsoever -- demonstrates subtle discernable noise even at low ISO. It is the nature of *ALL* electronics I'm afraid. In spite of it being clearly visible, the BL noise is consistent and scales linearly with higher ISO. So it is predictable. This is what I like about the BetterLight software -- it does not lie, it does not fabricate. Adobe RAW and other RAW converters on the other hand, have to lie since the data they start with is significantly WORSE than the much purer BetterLight data. I will state for the record: a BetterLight pixel has twice as much data and half as much noise as any CMOS/CCD Bayer sensor on the market today. So if you are seeing "noise free" out of your Bayer digital back via RAW conversion, then you know that someone has been messing with your data behind the scenes. Personally, I don't like that. But then again, I don't have to rationalize a $50,000 investment in equipment that will be obsolete in 24 months...

[Eric Leppanen]

The feedback I received from various folks here in L.A. earlier this year also indicated that, with 39MP MF digital backs available today, a 20+MP DSLR available from Canon early next year, and 50-60MP MF backs available within maybe two years, the scanning back market is withering away.

An interesting question is going to be how higher digital back megapixel counts are going to square with lens resolution limits. As I understand it, to keep lens manufacturing costs within reason, increasing lens resolution ultimately means reducing image circle (as Rodenstock recently did with their HR digital lens series, which supposedly is already capable of supporting a 50-60MP back). However, increasing resolution by increasing pixel density in turn requires larger lens apertures to avoid becoming diffraction limited, so a higher resolution digital back incurs a reduction in DOF, similar to moving to a larger format. So what digital sensor size will ultimately prove the "sweet spot" in reaching an optimal combination of lens resolution, sensor noise, DOF and cost?

We live in interesting times!

[MJSfoto1956]

An interesting question is going to be how higher digital back megapixel counts are going to square with lens resolution limits.

Great points Eric,

For those of us with significant investments in "big cameras" it is a shame that 6x7 and 6x8 sensors with LOW pixel/mm count aren't available. (you *want* low pixel/mm count so that the sensors act like big photon buckets, resulting in low noise and high sensitivity without all the fake RAW witchcraft) If such a back were available, then your 20-50 year old 6x7 and 6x9 cameras and backs would continue to provide value.

The only manufacturers likely to pull that off would be the Chinese (particularly those on Taiwan) since they don't have an alterior motive to sell you new lenses and/or bodies!! One can only hope. I pray that there are some people out there paying attention.

J Michael Sullivan
Editor/Publisher, MAGNAcrhom
www.magnachrom.com

[bglick]

Michael... although those old bodies may get use, I still question the use of older lenses with any digital sensor, even a less dense sensor.

I don't mean to start a war, but I am not following the benefits of the larger sensor. The larger the sensor, the longer the fl, the higher the f stop. At the sizes we are talking about, this all equates to higher amounts of lens diffraction where even small amount of DOF are required and slower shutter speeds, both are undesireable.

I follow you on the noise issue, but to make bigger pixels, you are trading a little noise for a lot of diffraction. Not sure that is a good trade off in the final print. In the end, it's all a question of degree, and from what I have seen so far, I would clearly desire the smaller, more dense sensors.... I'll take the little extra noise and gain much sharper digital lenses, less diffraction and faster shutter speeds. In the end, I will come out ahead.

A good example of this is the tiny digicams. An 8 MP digicam can have 350 pixels per mm... and of course they are very noisy. However, when you don't push these pixles beyond their capability, such as making a 16x20 print with 8 MP, the noise erases nicely with noise reduction programs. To me, it's all part of the technology of digital..... you take the gains where you can get em, and in this case the gains are smaller cameras, smaller lenses, faster ss's, less diffraction.... in return, you have a small bit of post processing work, which can often be batched processed. And the noise in the sensors we are referring to in this thread, will not be anywhere near as noisy as the digicams.


Before I would attack noise as a huge culprit, I would attack digital sensors undesireable means of recording tonal range. 75% of all the tonal ranges are in the top two highlight stops.... (half in the top stop, half in the next stop, etc. etc.) this is not desireable... I know this is true in Bayer sensors, but not sure if the same holds true in CCD sensors. Assuming it does, this might be the area where film still holds an advantage, specially neg film.

[MJSfoto1956]

Michael... although those old bodies may get use, I still question the use of older lenses with any digital sensor, even a less dense sensor.

I don't mean to start a war, but I am not following the benefits of the larger sensor. The larger the sensor, the longer the fl, the higher the f stop. At the sizes we are talking about, this all equates to higher amounts of lens diffraction where even small amount of DOF are required and slower shutter speeds, both are undesireable.

Good points B,

Funny you should mention this. I absolutely AGREE with you that the smaller the sensor the more everything will be in focus. However, for me this is the MAIN reason to LOVE big cameras! There is nothing like a 4x5 or 5x7 @ f8. I find it surreal. And I know others here do too.

As for "big pixels" the benefits are two fold:
1.) better light-capturing ability (higher dynamic range + lower noise)
2.) lower lens resolution requirements per millimeter

The second fact listed above is the cool reason it would make so much sense for a true 6x7 or 6x8 sensor. If we take 56mm x 69mm as the image area and use a sensor with 5040 x 6210 pixels (31.2 MP) you get a line resolution of only 90 pixels per millimeter -- most 30 year old large format and medium format lenses can readily produce that! And at 90 pixels per millimeter, you would have pixels that are nearly 2.66x as light sensitive as the P45 which has a pixel density of 147 pixels per mm. This would be a GREAT back!! I'd buy one in a hearbeat.

Question: how many of you would buy such a back? (assuming it were priced "reasonably" and would extend the life of your current equipment)

J Michael Sullivan
Editor/Publisher, MAGNAchrom
www.magnachrom.com

[Daniel Geiger]

The big vs. small pixel issue is interesting, particulary in context of the overlooked third capture type, the Foveon Chip. It is a three layer chip (no hated Bayer pattern), so much more like film, which does not have any issues with interpolation/aliasing.

It still is a chip, but at least some of the problematic issues with digital devices are removed.

But apart from a small chip in a Sigma digiSLR, it is non-existant. Big pixel on large chip (4x5 or 8x10) with three layers, that is certainly more appealing. Does anyone have any intel on the future (or lack thereof) of Foveon?

[bglick]

> However, for me this is the MAIN reason to LOVE big cameras! There is nothing like a 4x5 or 5x7 @ f8. I find it surreal.

Well, I would prefer the Canon 200mm f2 for this effect :-) With large format lenses, i find the problem with f8 is the lenses perform so poorly, that even the in-focus area is not that sharp. Maybe in a non enlarged print, but not so in most enlargement. The only real reason most LF lenses have low f stops is to see a brighter image on the gg. Most large image circle lenses are so abberated at f8....I never saw much need to shoot at them.


> you would have pixels that are nearly 2.66x as light sensitive as the P45

Explain "light sensitive" ? Do you feel such pixels would be of faster speed? ISO 25?


Daniel, I too often wondered about that Foveon.... there must be some reason that technology grounded itself. If it had merit, I would think one of the big chip makers would license it from Sigma. The Bayer method of capturing one color per pixel is much inferior, however, since the Bayer technology overwhelmed the chip with pixels, it more than compensated for this shortcoming... and in the end, Bayer sensors combined with interpolation actually do an incredible job of getting the color right, and it makes sense, its a great mix of hardware and mathematics / software. The one huge benefit of Bayer is the fact you are recording only 1/3 of the pixles in the camera that the finished raw file will be after raw processing. This represents storage savings of 66% in the field. Hence why I think Bayer lives on... and we may have seen the last of Foveon....

[MJSfoto1956]

Explain "light sensitive" ? Do you feel such pixels would be of faster speed? ISO 25?

The math is easy: take the square of one pixel dimension (e.g. 1/90 ^2) and divide it by the square of another pixel dimension (e.g. 1/147 ^2) and you'll get a very close approximation of the capabilites of any two sensors, all things being equal.

In short, the light capturing capability is increased significantly as you REDUCE the number of pixels per mm. I.E. the sensor is bigger, thus it is more sensitive to light.

So what does this mean? The best analogy is to consider a sensor a form of a photon "bucket" -- and to recognize that a small bucket can only hold so many photons. A larger bucket can hold exponentially more photons due to its greater volume. In the case of comparing 90 pixels per mm vs 147 pixels per mm the math reveals a 2.66 increase in the light capturing capability. Typically with digital sensors, this plays out in less noise (higher signal to noise ratio since there is more "signal"). Meaning, you can increase the ISO and get significantly less "noise" out of it.

Thus, to answer your question, yes, a "big bucket" sensor @ 400 ISO might produce roughly the same noise as a "small bucket" 100 ISO sensor. Or conversely, it would deliver significantly purer data at 100 ISO. And purer data is the key to getting expanded dynamic range. In fact, it is the only way.