This Medium Format forum has had a lot of discussion (and disagreement) on the value of higher bit-depth on Medium Format Cameras. To help clarify the issue, I asked a digital imaging expert: Tim Grey. He posted his answer in his Digital Darkroom Questions (DDQ) e-mail today.

Conclusion: There is a difference with higher bit-depth Medium Format Digital Backs, but it tends to be very subtle, and digital 35mm SLRs continue to improve and narrow the gap

Here is the full text of his answer from the July 1 DDQ:
Images from a medium format digital back do indeed have the potential for greater dynamic range in the print. Of course, that potential isn't always realized, but in terms of the information captured, in a general sense medium format digital will still exceed the capabilities of digital SLR cameras to some extent.

The bit-depth is certainly a contributing factor, but just because you have higher bit-depth doesn't mean you necessarily have greater dynamic range. For example, you could theoretically use the imaging sensor from a very old digital camera that wasn't able to capture any real dynamic range (think in terms of a really old mobile phone camera) and update the analog-to-digital (A/D) converter to create a 16-bit per channel file from that image. You now have high-bit data (sort of), but it is based on information captured from the sensor that is lacking significantly. The point here is that the bit-depth is really more about the size of the container for the image data, and isn't necessarily indicative of the data you are actually capturing. Dynamic range is the measure of the difference between the brightest and darkest values that can be recorded.

In order to understand the nature of the data you're capturing, you need to look at the imaging sensor itself, not just the way the data captured by the sensor is being converted. There are a variety of factors that affect dynamic range in an imaging sensor, but one of the specifications you can look at to get a sense of the relative dynamic range capabilities of an imaging sensor are is the size of the individual pixels on the imaging sensor (the pixel pitch), which is measured in microns. Larger photodetectors on the imaging sensor are capable of capturing more electrical charge (which is how the light it "sees" is recorded), and thus are able to capture a broader range between "empty" (minimum value) and "full" (maximum value). That translates into greater dynamic range (generally speaking), so in general larger photodiodes are a good thing (from the perspective of dynamic range and noise anyway), within reason of course. And of course for a given sensor size (physical dimensions) as you increase resolution you reduce the size of the individual photodetectors, so this is always something the camera manufacturers (and imaging sensor manufacturers) have to contend with.

To give you a sense of the potential of medium format imaging sensors, it can be helpful to consider the relative size of the photodetectors in real terms. One of the top professional digital SLR cameras is the Canon EOS 1Ds Mark III, which offers 21.1 megapixels and has a pixel pitch of 6.4 microns. The PhaseOne P25+ digital bak, which offers an effective resolution of 22 megapixels, has a pixel pitch of 9 microns. While these are both incredibly small sizes, as you can see the medium format digital back has much larger photodetectors in relative terms. At the lower end, a typical point-and-shoot digital camera with 7.1 megapixels has a pixel pitch of about 1.9 microns.

Of course, when it comes to the print there are other limitations. You can have all the bit-depth and dynamic range in the world, and if you're printer can't reproduce all those values the data isn't really worth much in the context of that print. So, while a medium format digital back is most certainly going to give you more information in your digital captures, today's printers aren't really capable of reproducing that information to the extent that it makes a significant difference in the final print. There is a difference, but it tends to be very subtle, and digital SLRs continue to improve and narrow the gap. There's no question there are inherent benefits to the imaging sensor used in a medium format digital back, but technology is developing to the point that the benefit is relatively modest, especially when viewed in context of the pace of change in digital photography over the last ten years.

-----
Note: there are other threads that touch on this topic, but they started to meander and were not as specific. This topic deals just with bit-depth. Here are links to the other topics:

RAW files: 1Ds3 and Phase P30+

Medium Format vs 35mm - Just More Pixels?

Can Stitching 1DsMk3 files = Med Format Quality

whatever.

You know, I think I'm going to go to Papaya King tomorrow for the Depression Special. I really like their hot dogs. I might also hit Century and buy a new belt.

15 minutes and I'm still ROTFLMAO!

My pleasure! I aim to please.

dood - bit depth doesn't matter at all. just buy a 1Ds and be done with it...

Yaaaaawn!

It misses the point entirely. The benefit of having a 16-Bit MFDB is that the file is more elastic / flexible.

Also far improved on long exposures.

I *think* (note sarcasm) it has been recommended before to go out and take some pictures. Work with them, then you will see.

If you don't, it doesn't matter, therefore you are comfortable with the 35mm choice.

Ok?

Jo S.x

Exactly. Plus I wouldn't even trust that testers know how to switch off default curves in their raw processing apps.

Exactly - the container is larger and when you know how to fill it up (as most owners of MFDB's would do) you have more to play with. No one uses a file without a curve or adjustment applied...

It's a fair assumption that those who have shelled out big bucks for a MFDB system will not be cutting corners with their printing processes either.

Getting that last bit of a quality, only available at the 'top' of any endeavour is always going to require more resources for less return - it's why fewer people go there. The differences may be subtle but significant. Often that difference only appears 'subtle' at first glance...

Wow, what are you on some sort of ridiculous crusade to justify your own unwillingness/inability to invest in a DB? The problem is that if you have ever worked a file hard in post from both 35mm and a DB the difference is blatantly obvious. So who are you trying to convince here, yourself perhaps? I use both on a daily basis and have no delusion that one can substitute the other, instead I embrace the pros and cons and use them for different things.

Enough of this BS already.

...and in the end it's always about the money... as a good man used to say; " if you have to ask for the price, it is to expensive for you" and one can always find a reason not to spent a certain amount of money. i think it is useless to keep on answering those questions: all what there is to say is more than a million times written....

I just love the cause and effect with web forums these days... a poster takes the time to serve up some banter on a "grey area" topic and most viewers either walk out of the room yelling BS or they instantly send back an insult. Reminds me of the senate on planet of the apes.

If you don't have something good to say... then just don't say anything.

I own both a 1ds3 and a p45+V and do notice a difference in final image quality as well.... but each "system" has it's merits... i.e I can't get the look (love) of the 85mm f1.2 wide open on the 1ds3 when I use the MFDB.... Likewise, I can't get the crispness (high detail) of the hyper focal abilities of the MFDB stopped down around f16 when I shoot with the 1ds3.

I also can't match the wide angle of view with wide lens' on the 1ds3 when I choose the "V" MFDB.

But..... I can say that when I use the 1ds3 as it should be used ( raw to 16 bit tiff & good glass... + proper technique ) it closes the gap considerably from the 1ds2 to MFDB.

As to higher bit depth mattering?.... If you are striving for the best image you can capture today? YES!

If your shooting technique right for the situation and nail your exposure without having to adjust much in post?.... NOT REALLY.

If you tend to do "looks" in post that require you to mangle your file around a bit ( like the infamous Dragon look )..... YES

If you shoot reproduction work of paintings.... YES

If you shoot hyper-real landscape.... YES

If you shoot fine art..... ( dirty look ) NO.... ( clean hyper-real look ) YES


To quote the Beatles.... "it's getting better all the time"

if you had only 1 bit - whould that make a difference?
If your answer is 'yes', you should understand why 16 is better than 8.

if your answer is 'no', well, use 1 bit then or ask yourself if 0 bits would make a difference instead.

There was a relatively technical article recently referenced in one of the ETTR threads and my recollection is that at some point, at around 12-14 bits, noise renders any "larger container" superfluous, at least with today's AD converters.

The responses thus far to thread have been largely pathetic and have provided no objective information. Tim Grey's essay is good, but he too provides no objective information.

Dynamic range and useful bit depth is limited by noise, and increased bit depth will not be helpful if the useful levels in the raw image are obscured by noise. A good analysis of this matter is presented by EJ Martin.

Discussion with presentation of additional data is presented in a DPReview thread listed below. The discussion is concerning the Nikon D3, whose pixel size is 8.4 microns, slightly smaller than the 9 microns of the referenced 22 megapixel medium format camera, but larger than the pixel size of the P45+ 39 MP back (6.8 microns). The same principles apply.

DP Review Thread

me, beeing back in gauting ( closed to munich/ germany ) just have eaten a great steak with
rice at the schlosscafe.

having my back resting in my bag, so i cant discuss about its 16bit qualities.

That sounds fantastic! My Depression Special was really good, and really cheap. I'm sure it wasn't as good as your steak, but man, there is something about a good hot dog that makes me really happy. Last time I was in Munich I had a great lunch of Weisswurst, spicy brown mustard, and a litre of Warsteiner. Man that was good!

Mmmmm...mmm. Just had a nice salad with a tad too much feta.

Beer and food pairing. Interesting article http://www.allaboutbeer.com/features/243beerandfood.html

"The flavor hook is the part of the beer’s flavor and aroma that matches, harmonizes or accentuates the flavors in your food. When the flavors meet on your tongue, they "recognize" each other and this creates a harmony.

Sometimes, rather than harmony, you’re setting up a pleasant contrast. Beer can have flavors of caramel, coffee, chocolate, bread, bananas, limes, herbs, smoke or raspberries–there’s a lot here to work with."

You can't simply compare DSLRs and MFDBs by bit depth, or say that anything beyond 12 to 14 bits is meaningless - as with most things in life, it's more complicated than that. People who want to boil this debate down to simple black and white answers are only deluding themselves.

A/Ds come in various quality levels, depending on cost and sampling rate, among other things. You can certainly get very high sampling rate 16-bit A/Ds, suitable for imaging applications, with 2 to 4 bits of noise, i.e. the A/D itself generates 2 to 4 bits of noise, leaving 12 to 14 bits of real, noise-free data (obviously any noise in the signal being sampled by the A/D just gets passed along, along with the additional noise contributed by the A/D itself). These are relatively expensive, however. Lower cost A/Ds typically generate more bits of noise, delivering fewer bits of noise-free data. You can get up to 24 bit A/Ds with more moderate sampling rates with similar noise capabilities, i.e. you get 20 to 22 bits of real, noise free data (by noise free, I mean noise contributed by the A/D itself - any image sensor output will almost certainly have plenty of noise in the lower bits if you are sampling it at 24 bits).

So if you are going to argue the merits of bit depth, the first thing you have to ask is what is the quality of the A/Ds you are using. All 14-bit A/Ds or 16-bit A/Ds are not created equal. I don't have any hard data to back this up, but I'm pretty sure all of the new 14-bit DSLRs are using actual 14-bit A/Ds, which means they are getting, at best, 10 to 12 bits of real, noise-free data, assuming they used expensive low noise A/Ds. Furthermore, I believe that at least some of the MFDBs are using more expensive 16-bit A/Ds (and maybe even higher bit depths), and downsampling to 14-bits (throwing away the noisiest bits), and then saving those 14-bits as 16-bit RGB values (a bit confusing, I know). If true, this means MFDBs are getting 12 to 14 bits of real, noise free data (or more, if they are using high quality higher bit depth A/Ds, and then truncating off the lower bits for storage as 16-bit RGB).

Given the price differential between DSLRs and MFDBs, I also suspect DSLRs are using lower priced, and hence noisier A/Ds, while the more expensive MFDBs can afford to use higher cost, and hence less noisy A/Ds. This is simply guesswork on my part, but it seems reasonable. So again, not all A/Ds are created equal. I find it very plausible that the higher cost MFDBs are using more expensive, lower noise A/Ds in their processing path than the DSLRs are.

Anyway, my point is that making a simplistic argument that 12-14 bits is as high as you can go is just that - simplistic. It certainly doesn't take into account the parts budget that the designer has to work with, and the quality of the parts he is forced to use for cost reasons, or the quality of the parts he can afford to use because he is designing them into a higher priced product. In other words, part of the reason MFDBs are so expensive is that they use higher quality, lower noise components (of course, there are economies of scale, etc, but parts cost is also a factor).

And there are a host of other noise sources that have to be accounted for, in addition to sensor noise and A/D noise. There's amplifier noise, and power circuit noise, and coupling noise. All of these are price sensitive - the higher your parts budget, the lower noise parts you can afford to use. And most of these get worse as temperatures rise, so your camera or back has to have good thermal management and heat dissipation to get the best out of the equipment.

So as I said at the beginning - You can't simply compare items by their bit-depth, or say that 12 to 14 bits is the effective limit of what you can have. It simply isn't so.

Thanks for th link. My friend owns a bar on McDougal, 124 Rabbit Club. Its a beer/wine bar. He was pawning off this RauschenBier, a German beer that complies with the Reinheitsgebot but adds a twist to the flavor. Tastes like a campfire. Kind of gross. I'll check your link to see if there is pairing for RauchenBier.

Abita Amber goes really, really well with spicy Louisiana food. Maybe I'll make Jambalyia tonight . . .

the problem with traditional bavarian food in our digital stoneage is that they just use 11 of their 20 available bits. the rest is beer.
and this makes me feel somehow crazy if i take more than 6 bits of it. man ....
and this are 9 !!!!!!!!!

maybe i should eat my 5d.
its not water- resistant so all the beer is running out of it even if you fill in 12bit. or to go with the 1ds3?
any advice highly appreciated.

Your suggestion makes since; I agree. As I have said on my previous posts, I actually have gone out and done side-by-side comparisons. Specifically, I have compared my 1DsMk3 to the Phase One P45+ and also the Hasselblad 39 MP camera.

The reason I have been persistent with my posts here is that I was surprised at how little difference there was. But my experience was limited to just a few test shots, so I was reluctant to base my decisions on a limited experience. I thought that those of you who use these cameras every day could shed some light (theoretical and practical) on the differences.

When I enlarged these images to ten-feet wide and then took a section from the center to print, the difference were much less than I expected. Clearly the MF backs shows more fine detail, but with a more vigorous application of Unsharp Masking the difference was impossible for me to appreciate at normal viewing distance (3-feet).

Regarding shadow detail, highlight detail and noise, I could see no difference whatsoever. Granted, these pictures were perfectly exposed and did not need radical adjustments; but in the real world, my exposures are almost always correct.

Before I go and spend large amounts of money on a new camera system and spend the hours it takes to learn how to use it, I want to reasonable assurance that the quality differences will be apparent to me and my clients.

Based on what I have seen with my own experience and have learned so far on this forum, I remain unconvinced that upgrading to Medium Format is worth it for my business. But I am trying to remain open minded. My business incidentally is to sell large nature photographs to commercial clients. The prints are make on an Epson Stylus Pro 11880 and sizes range to 5 x 8-feet and larger.

If you disagree with my conclusion, please try convince me otherwise.
Thanks!

I'd wait for the new Nikon. 14 bit, 24 megapixels, water proof, such that it will hold a liter of Pilsner or 750 cl of heavier Belgian Trappist or Bavarian dopple bock. I heard, and I can't confirm or tell you who told me (it was Annie Leibowitz) that the new Nikon will have a sausage maker built into the rear display. Nice. Real nice.

wow.
hicks.
urg ... ups.

Henry, if you've already made up your mind, why continue to beat a dead horse? Act on your conclusions.

I'm not dissing you here, I'm just confused about where you're coming from and what your purpose in posting this thread is. A lot of people on this forum think you are just trying to trash MFDBs, that you are carrying the torch for Canon. Hence the hostility you get on this forum.

Personally, I've compared images from my D3 and a friends 1DsIII. I'll take the D3's images anytime. Please try to convince me otherwise.

Personally, I've compared D3 and 1DsIII images with images from my P45+. I'd much rather have the P45+ images. Please try to convince me otherwise.

Personally, I find fall supperior to spring. Please try to convince me otherwise.

Personally, I find classic coke much better than pepsi. Please try to convince me otherwise.

"You have chosen to ignore hdomke. View this post · Un-ignore hdomke"

TMARK, I know they're supposed to be so good and all, but I've never been a fan of German beir in general. A friend of mine is German and she's always having us meet up at Zum Schneider and Loreley and the like. Fine places, but I'd rather a nice microbrew most often. I'll have to check out Abita Amber. Big fan of ambers, bitters, browns and pale ales.

i am with mort in this case:

"Personally, I find classic coke much better than pepsi. "

did you realize that he is saying CLASSIC ?
not sinar, not leaf, not phase, not hassy.
CLASSIC !!!

means analog.
thats out. isnt it? no bits. no 3d. no 5d. no canon.
classic.

just realized that it seems to me that the canon even offers 5D.
with my backs i only get 3D !!!

Well, I do complain if I find bits in my classic coke. So I guess you're right, that does mean it's analog.

My German friends like Coors. Seriously. They are mainly from the Luneberg Heath/North German Plane area, but still. "Coors, its crip and refreshing. From the Rockies!" I'm usually shocked.

I myself love the ambers and ales. Abita is like a smoother, more "hopsie" if you will, version of New Castle, which I also like. I have a soft spot for Louisiana products, and Abita is one of the best. Dixie, on the other hand, is one of the worst.

I do like Bavarian beer, but the wheats. I do not like lagers, and can obly tepidly support pilsners.

Sometimes I like King Kobra, mainly because of the Billy D commercials. I'm also known to drink Old English on occassion, but only because I really really like Ice Cube.

D'y' know, I haven't had a Trappist triple in a long time. Wonder are there any places here in Northern Ireland you can get that?

It's a bit of a poor show just having Magners and Pizza, when I read what all you guys are having.

Mind you, the food mixer has a 4 bit processor. Do you think the dough would be better if it had a 16 bit processor instead?

can anyone explain to me why i should work with a digital back if it only offers 3D,
meanwhile my much cheaper canon offers 5D. aside from its better iso performance?

It probably wouldn't be as lumpy - you know, with 16 bits, you'd get a smoother texture.

LOL!

I think you've had a little to much Bavarian doppelbock, Rainer!

Well, my Phase has a bottle opener. I heard Sinarbacks have an adapter system such that you can add a bottle opener to any MF camera. The 5D really only has 2D, anyway, because 3 of the D of the 5D are wasted, empty calories.

And Jon, a 16 bit food processor would make a much more layed, textured dough that would stand up to more stretching once processed.

no. the 5D has a certain 5D look. if you dont see it its not worth to discuss it with you.
sorry for that.

Imagine the look of the 40D......
8X the look of your 5D

I wish our Australian beer makers adopted some of those fine European brewing techniques...

We do have good wine though.

Some sobering facts

The "digital imaging expert" seems to understand of this subject about as much as most of the posters above, which is not much (if I am very generous).

Now perhaps it is time to consider some facts.

If and how far 16 bits are useful with MFDBs is a complicated matter, in most cases impossible to prove in either direction. I can prove on images of *certain* MFDBs, that there *is* a difference between 14bit and 16bit (slightly less noise in 16bit), but that proof shows, that the difference is *very, very tiny*, in fact it is useless.

The question needs to be adjusted a bit in order to become reasonable:

do MFDBs need greater bit depth than DSLRs?

When analysing for example Canon 1DsMkIII images, the same can be said: the difference between 12bit and 14bit is there, but very small (note: even so I am a proponent of 14bit recording).

When comparing 16bit, 14bit and 12bit images of a Sinar eM54, it can be shown, that reducing to 12bit does cause practical image degradation.

It is clear, that the larger dynamic range of some MFDBs *requires* more bits than a camera with 2 stops lower dynamic range.

However, the most sobering fact is, that the Phase One cameras' last two bits are not only useless, but they are *detrimental* to the image. Luckily, the raw processors don't make any use of those bits, except in extreme circumstances (that's what caused Edmund complain about striations with his P45+).

The following captures depict three segments of the image CF000841, shot by a Phase One P30+ by the fellow LL poster 203. The first one shows the original, with +3 1/2 EV. The second one shows the 15th bit only and the third one is created by the 16th bit only (i.e. the lowest order bit) . The next four captures show two other crops, with only the 15th respectively 16th bit.

The waves and horizontal strips are artifacts of the sensor, as is the vertical separation line, caused by the "stitching" of the sensor chips.

These captures demonstrate, that the two low order bits of Phase One raw files *must not* be used (this is not a speciality of the P30+, I saw teh same on the P25+ and P45+ as well). In other words: the 16 bit recording is worthless with these cameras.

I really like Sabretts.


Click to view attachmentClick to view attachmentClick to view attachmentClick to view attachmentClick to view attachment

Mort,
I appreciate your comment and will stop posting on this forum about this topic.

I'm sorry if people on this forum thought I was trying to trash MFDBs. I had no such intent. I was seriously considering buying a MF system, but wanted to look carefully before I jumped. My limited testing did not show a clear difference when comparing my 1DsMk3 and a couple of MF backs. That surprised me. I wanted to hear from those with more experience.

Nor am I wanting to promote Canon. I've happily owned and used Hasselblad, Nikon and Canon systems over the last three decades. All have served me well.

Thanks to all who provided feedback.

Henry,

I think this is a fair comment to make: By the looks of your work, you won't really see much difference in looking at files between a Canon 1DsIII and MFDBs. You aren't pushing your files past straight forward exposures and in no way pushing the abilities of any of these cameras through your chosen content of shooting. For people who are truly "bending" their files for look and style (very evident in fashion and many other genres of photography), then the advantages of MFDBs or just higher bit depth will show for sure.

For Panopeeper,

Here is a "sobering fact", you and I and many more here are graced with the participation of some world class photographers on this forum. They are world class because they are artists in the commercial realm. They, for the most part, don't give a rat's ass about the scientific bullshit. ART and CREATIVITY are, in my view and many talented photographers' views, the foremost keys to unlocking photography, not all this mumbo jumbo "anal-yzing".

I have a degree in Chemistry (it was easy for me), I was an organic chemist for several years. Science is soul-less without creativity of how to use it. Photography is soul-less without creativity to harness vision. In the end, the vision can last a lifetime or many lifetimes; gadgets, tools, programs, and anything else which in and of itself accomplishes nothing without human mastery will erode rapidly.

It's all about the image. I can look at the work from talented photographers (a hell of a lot more talented than I am) and see that they had that image in their head. The camera, hardware, software, film, scanners, lighting, etc. were just the means to the end, nothing else.

I'll also add, I stick around here because the forum is about problem solving so we can work more fluidly to create our art, not problem making. There are a lot of photographers with different shooting methods and in different photo genres, all that put together makes for a wealth of info. 1's, 0's, bits are just the least informational items possible.

I am very impressed, but I don't know, for what you mean it.

Is this an explanation for the relative low level of digital literacy, or an excuse for the schoolyard bully like behaviour?

Well said.

Mort54, you seem to be implying that all ADCs are adding 2 bits of noise as a rule of thumb but this is not true. In fact it is even conceivable that a poor quality 16 bit SDC will have a worse SNR than a 14 bit ADC.

Just look at this spec chart from Texas Instruments of 16bit ADCs. The SNR on these ranges from 79 to 95dB! That nearly 3 bits difference.
http://focus.ti.com/paramsearch/docs/param...R_2000084|EQ|16

So at least we can agree that not all ADCs are created equal. It is quite possible for a stunning 14bit circuit to outperform a poor 16bit competitor.

Did you test scenes with high dynamic range?
If the sensors from two cameras have different DR then how do you compare the exposures? Do you consider them equal when they are both equally weighted around 50%, or both exposed to just under the point of clipping? These are two different methods, and the former does not take full advantage of the higher DR sensor.
Actually you provide so little data about this comparison that I can't make any useful judgments at all, sorry.

Rainer, I hope you haven't wasted your money already. You can get 6D for only $70!
http://www.digitalgraphicsresources.com/in...PROD&ProdID=664

Interesting. You'd think that P1 would pad out the 2 least significant bits, even though this really doesn't have an impact in the real world unless you are stretching/pushing the files hard.