Is the difference of DR on MFDB vs 35mm dslr discernible on print?

[archivue]

for me, the main reason to go with a MfDB was the distortion with TS-E lens... so i went with a 35 xl digitar, a 55 apo sironar digital, a 90 apo sironar digital... and i've just bought a 45 apo sironar digital... my 5DII is sleeping... night and day for my purpose, and i only use an aptus 22 on my beloved arca 69 for now !
so it mainly depends on your type of shooting than pure pixel comparison !

i will go for a DX3 if i was in need of hand held shooting, af, and hight iso... just different tools as were M6, hasselblad, rolleiflex tlr, 8x10...

[TMARK]

Amen brother.  

I love photography as my profession but at the same time could drop it like a ton of bricks and go do something else creative of which I am passionate, and I would have no regrets and probably never look back. My ability to leave it has nothing to do with actual photography or creativity but the down spiral of the industry. But I stay in it, am an ASMP chapter president looking to help the industry and others however possible regardless of times I think it doesn't matter.

I see it like this, a client calls photographer A and says what kind of images they need and photog A says great, I have this 60MP digital back I sleep with every night that makes 1's and 0's and I can give you the biggest cleanest file you've ever layed your eyes on, it's dynamic range is xyz, it's bit rate is 16 and that's the real bit rate not fake like other brands, cough-cough DLSRs are only 14bits.

Then a clients calls Terry Richardson and says what kind of images they want and he says, sure I'll bring my point and shoot and some Kodak Gold 200 that we can process at Walgreens.

Find a happy medium. Learn the craft, learn the art, give clients value, no attitude. Make them want to hire you again. I'd garner to say 80% of my clients are repeat. That's the biggest confirmation to myself of my work and it never had a thing to do with what camera, file size, sharpness of this lens or AA filter of that camera.

If people want to constantly debate 1's and 0's then go work for Phase, Leaf, Sinar, Adobe. Add to the discussion by making some new wiz-bang thing.

And my advice to medium format companies is to promote their product by what it will actually do to help, and that's all that is needed. How will a 60MP camera help a photographer. Sometimes it will and sometimes it won't regardless of price and affordability for the photographer. Competing on numbers, bits, sensor will lose in the end because users do see the hassle, they do see less frustration with a 5DII or D3x. All medium format companies should have 5-10 years ago grouped together and pushed Dalsa and Kodak to make 6x6, 6x7 sensors (and even larger) a reality in 2009, so cameras that have worked fine can continue working fine. Cropped sensors on 6x6, 6x7 cameras are lame. Most 645 cameras feel lame. Once frame size is fulfilled, we may not care by then, I don't know. The intent of all the formats in their intrinsic qualities other than detail on film may be lost to the newer generation.

We went from Minox cameras to 11x14 film and larger in zee ol' days (a broad format base which has use), now we're only arguing on which is better, 35mm sensors or 645 sensors (we've squashed it in the middle to nothing). The purpose of photography has been lost at this point.

[gwhitf]

Amen brother.

Several of the Lab Coat Experts here actually drive out to the countryside, find their scene, measure the contrast and dynamic range of the scene that they may someday want to actually photograph, and then they drive back to town to choose the camera on their shelves that matches the dynamic range of the scene. The camera that they'd choose has nothing to do with what they enjoy photographing with; it's solely driven by the capturing capability numbers.

But by the time that they drive back to town to fetch their 12-Stop Capturing Device (12-SCD), the light has changed to eleven stops, thus, they consider it a failed trip and drive back to town, without ever having shot a frame.

Day after day they do this.

[bcooter]

Day after day they do this.

Day after day, try year after year, because this conversation will go on forever.  Usually somebody gets really crazy rude and Michael shuts it down, then 4 months later somebody makes the mistake of saying, "uh should I buy a 400megapixel back or keep shooting trees with my 5d2? . . . then the floodgates open, the dealers line up, we see 110 not so pretty jpegs of clouds, charts, alleys, brick walls, more charts and then of course histograms.

The weird thing is somebody that knows their stuff and actually designs this cameras will explain it, like that Grame Nattress guy and everybody just passes that by and keeps arguing about their medium format back of their dslr stitched into 400 frames, or their medium format back stitched into 24 frames.

It's a shame that this section ever was named medium format, or now large sensor photography, rather than the Professional Photography Section, or something that would actually provoke talk and thought about what it takes to make a beautiful photograph, or make a living at photography (past giving seminars), or anything other than a clipped highlight, an oversharpened image, or histogram.

I hate to spring this news to everyone, but there is no format anymore in digital photography, other than the lcd screen, whether it's 3inches on the camera , or 30 inches on a monitor.  

Shoot a big gig, switch cameras, from canons to nikons to any medium format back and sit down to edit with a client and they will absolutley never look at what "format" camera was used.

They will look for that one photograph that captures the essence of what they are paying for.  Period.

The photography industry is going through the most abrupt and overwhelming changes since Matthew Brady and you'd think that would bring up one or two conversations, but instead we get this.

This conversation goes nowhere and I'll admit by writing this I'm just as nuts as everyone else because I'm adding fuel to the fire, so please forgive me.

Now back to work.

BC

[asf]

While I agree with Cooter, almost no one save perhaps 2 of 3 of the people contributing and reading this will "shoot a big gig, switch cameras, from canons to nikons to any medium format back and sit down to edit with a client", and that's why this goes on. No one listens to the couple actual pros here, and maybe they shouldn't. It's way too depressing. Pros haven't gotten it into their heads they no longer matter in the whole equation.

[Dick Roadnight]

Several of the Lab Coat Experts here actually drive out to the countryside, find their scene, measure the contrast and dynamic range of the scene that they may someday want to actually photograph, and then they drive back to town to choose the camera on their shelves that matches the dynamic range of the scene.

I am glad that there are people who actually choose cameras based on their requirements... and do not just use the same DSLR (that they have used for half a decade) for everything.

How often do how many on this forum use mirror free cameras?

[Ray]

How often do how many on this forum use mirror free cameras?

Probably quite frequently. All modern Canon DSLRs have a Live View mode which is mirror free. You can even bracket exposure at a rate of 6 fps on some models, up to 8 fps on the new 7D, without the mirror flipping up and down.

[Dustbak]

Probably quite frequently. All modern Canon DSLRs have a Live View mode which is mirror free. You can even bracket exposure at a rate of 6 fps on some models, up to 8 fps on the new 7D, without the mirror flipping up and down.

Try using one of the lenses that is supposed to go on a body that is not hindered by a mirror box (eg. Schneider Digitars or Rodenstock HR). I was stunned by how much better my Digitars were compared to my HC lenses. Huge difference, in my opinion. Using mirror up doesn't make up for a lens design taking into calculation it has a mirror box in between the lens and the capture surface.  

I am pretty sure this is what Dick meant, somehow I believe you know this too but are simply trying to argue for argue's sake.

[bjanes]

Hi,

That may depend a little bit on how DR is defined. The normal (technical) definition of DR is SNR = 1, and normally the read noise is considered. Read noise seems to be lower on CMOS and the technology Canon is using, so by this definition Canon get high DR-rating because of their low read noise.

Now, noise in normal photography is not dominated by read noise but noise resulting from the poisson distribution of photons. That essentially means that you need to collect about perhaps 100 photons on average to acceptable noise levels. In the first case:

DR = maximum electrons / read noise

and in the other case

DR = maximum electrons / (read noise + 100)

This will be something like 3-5 stops less than the first value. It's quite clear that having larger sensels have two advantages:

1) They collect more photons so they need less exposure to achieve good statistics
2) Bigger sensel can hold more electrons which increases dynamic range

This is a very short explanation and not a scientific one.

For a good explanation check:

http://www.clarkvision.com/imagedetail//do...el.size.matter/

http://theory.uchicago.edu/~ejm/pix/20d/te...oise/index.html

Best regards
Erik

Eric, you are misquoting Emil Martinec in reference 2. See his section Big pixels vs Small Pixels. Big pixels do collect more light, so that the signal:noise improves. However, if you keep the overall sensor size constant (say full frame 35 mm), the total amount of light falling on the sensor stays the same so the total poison noise (shot noise) will be the same if other factors are held constant. For example, consider full frame 35 mm sensors with 6 MP and 24 MP respectively. The ratio of the pixel spacing is 2:1. The large pixel sensor will collect 4 times as much light per pixel, but there are only 1/4 of them. Overall, the amount of light will be the same. If you downsize the small pixel camera to the same resolution as the large pixel camera, 4 pixels will be combined into one and total shot noise will be the same. There is a trade off between dynamic range and resolution. Since noise adds in quadrature, the 4:1 downsampling will improve the N:R by a factor of 2.

Contrary to what you say, read noise is usually relevant in the deepest shadows where the noise floor used to determine DR is set. In the above down sampling where 4 pixels are combined into 1 outside of the sensor, the superpixel formed by combining 4 pixels into 1 still has 4 read noise components, so down sampling in the shadows will be less effective than in the highlights. However, as Emil explains, large pixels have a relatively high read noise at low ISO, so the actual difference in total read noise will be less than 1:4. At high ISO, the large pixel will have a relative advantage to the small pixel. These considerations help to explain why the DR of the Nikon D3x is better than that of the D3 at base ISO, while the D3 is still better at high ISO.

In Roger Clark's example comparing a dSLR with a P&S, the pixel size is much larger in the dSLR but so is the overall sensor area. Emil has criticized that study, since it did not control all the variables. Roger's conclusion is based on a per pixel analysis.

[ErikKaffehr]

Hi,

I wasn't exactly quoting Emil Martinec. Regarding the noise issue I stand corrected.

Best regards
Erik

Eric, you are misquoting Emil Martinec in reference 2. See his section Big pixels vs Small Pixels. Big pixels do collect more light, so that the signal:noise improves. However, if you keep the overall sensor size constant (say full frame 35 mm), the total amount of light falling on the sensor stays the same so the total poison noise (shot noise) will be the same if other factors are held constant. For example, consider full frame 35 mm sensors with 6 MP and 24 MP respectively. The ratio of the pixel spacing is 2:1. The large pixel sensor will collect 4 times as much light per pixel, but there are only 1/4 of them. Overall, the amount of light will be the same. If you downsize the small pixel camera to the same resolution as the large pixel camera, 4 pixels will be combined into one and total shot noise will be the same. There is a trade off between dynamic range and resolution. Since noise adds in quadrature, the 4:1 downsampling will improve the N:R by a factor of 2.

Contrary to what you say, read noise is usually relevant in the deepest shadows where the noise floor used to determine DR is set. In the above down sampling where 4 pixels are combined into 1 outside of the sensor, the superpixel formed by combining 4 pixels into 1 still has 4 read noise components, so down sampling in the shadows will be less effective than in the highlights. However, as Emil explains, large pixels have a relatively high read noise at low ISO, so the actual difference in total read noise will be less than 1:4. At high ISO, the large pixel will have a relative advantage to the small pixel. These considerations help to explain why the DR of the Nikon D3x is better than that of the D3 at base ISO, while the D3 is still better at high ISO.

In Roger Clark's example comparing a dSLR with a P&S, the pixel size is much larger in the dSLR but so is the overall sensor area. Emil has criticized that study, since it did not control all the variables. Roger's conclusion is based on a per pixel analysis.

[bjanes]

Hi,

I wasn't exactly quoting Emil Martinec. Regarding the noise issue I stand corrected.

Best regards
Erik

In any case, the reference to Emil's article is most valuable. I see that he has visited this thread a number of times, but has not yet commented. Perhaps he lacks data on MFDBs, but his theoretical insights would be welcome. Aside from the DXO measurements and a brief post by Guillermo with a comment by Gabor, no one has presented actual data from MFDBs, but some photographers with these backs make claims which seem doubtful and it would be nice to have some data, not merely pontifications. As Guillermo has stated, anything visible on a print should also be visible on screen.

[cjmonty]

If you look at the finest detail in your landscape shots (leaves, distant architecture, distant figures, cement gravel...), the MFDBs resolve it noticeably finer than the Canon (and I suppose Nikon & Sony) sensors do.  This is constant over all types of lenses I have tested in front of both types of sensors.

To me, it is reminiscent of the difference between a fine grained 4x5 film and a fast 120 film.  Both have lots of detail, but one is a wee bit clumpy on closer inspection.

The catch is that you only notice this noticeable difference in huge prints or 100% pixel peeping.  

So, the MFDB pays if you are going for Gursky-esque detail, or are showing off on your laptop.

For most people, its not necessary.  But the difference is there.  Just like in the old days of film- there are always trade-offs for stepping up your format.

[ErikKaffehr]

Hi,

Michael Reichmann, Bill Atkinson and Charlie Cramer did a shootout with Canons, MFDBs, Film and Better Light Scanning Back in 2006. The results of this shootout was available on DVD which I happen to have. I'm looking at it right now.

Bill Atkinson made a "scientific" comparison, but that was i now way easy partly because tonality adjustments and sharpening are much depending on personal taste. I try to reproduce the setup with what I have, it contains a $1 bill and a "Mini Color Checker". In Bill Atkinson's comparison the difference in sharpness between the Canon 1DsII and the Phase 1 backs is obvious. Regarding DR and so I don't think that differences are obvious. My main interest right now is comparing with scanned 67 film, but that is another issue.

Best regards
Erik

In any case, the reference to Emil's article is most valuable. I see that he has visited this thread a number of times, but has not yet commented. Perhaps he lacks data on MFDBs, but his theoretical insights would be welcome. Aside from the DXO measurements and a brief post by Guillermo with a comment by Gabor, no one has presented actual data from MFDBs, but some photographers with these backs make claims which seem doubtful and it would be nice to have some data, not merely pontifications. As Guillermo has stated, anything visible on a print should also be visible on screen.

[Ray]

Try using one of the lenses that is supposed to go on a body that is not hindered by a mirror box (eg. Schneider Digitars or Rodenstock HR). I was stunned by how much better my Digitars were compared to my HC lenses. Huge difference, in my opinion. Using mirror up doesn't make up for a lens design taking into calculation it has a mirror box in between the lens and the capture surface.  

I am pretty sure this is what Dick meant, somehow I believe you know this too but are simply trying to argue for argue's sake.

I may appear to be arguing for argument's sake, but I'm really just trying to get some reliable information. The original question from Jim2 was about differences between a 1Ds3 and MFDBs on print.

The only reliable information in this long thread, specific to the original question, is from DXOMark. It can be seen there that the 1Ds3 has lower performance than any of the 3 Phase DBs reviewed so far. The DR and tonal range of the 1Ds3 is noticeably less than the P40+, p45+ or P65+ at the smaller print size, and by extrapolation, presumably at any print size.

The noticeable exception is the D3X which at the smaller print size remarkably still has more DR than any of the 3 DBs featuring on the DXOMark website. In other areas such as color sensitivity, total range and SNR, the DBs still marginally outperform even the wonderful D3X, but only at the DB's base ISO which is almost a stop slower than the D3X base ISO. Increase the DBs' ISO by just one stop so it's closer to that of the D3X, (ISO 96 v ISO 78) then remarkably in all parameters tested the D3X is either equal to or better than all of the DBs tested at DXOMark..

However, when discussing lens performance, it's a different ball game. I'm very receptive to the idea of choosing a camera system because of the exceptionally fine lenses that may be available. The main reason I now own a Nikon D700 is because of the very fine Nikkor 14-24 zoom lens. I bought the lens first, tested it on my Canon 5D using an adapter, compared it with my existing Canon-mount wide-angle lenses, and found it to be so clearly better I decided to buy a camera body designed to fit the lens and allow full functionality.

For the benefit of everyone else, I showed my comparisons on this forum. I'm naturally skeptical of generalised claims people make without a comparison. Photography is a visual medium. Seeing is believing.

It may well be the case that some of the Schneider Digitar lenses and the new HR Digaron_S 23mm ultra-wideangle are far superior to any Canon equivalent lens. If that's the case, then let's see how much better they are. I promise I won't be upset if any of them knock the socks off the new TS-E lenses from Canon.

[Christopher]

Well ray, from what I have seen so far. The 17TSE is quite good, I would probably set it equal, perhaps a little better than the 14-24. Now the zeiss 21mm is already a lot better than both. I can say from personal experience that nearly any modern large format lens is as good or better than the Zeiss 21mm. I don't need to make a 1:1 test for that. It is just how it is. (Note, I'm talking about the orginal zeiss 21mm, I have no clue how the new one is)

[Ray]

Well ray, from what I have seen so far. The 17TSE is quite good, I would probably set it equal, perhaps a little better than the 14-24. Now the zeiss 21mm is already a lot better than both. I can say from personal experience that nearly any modern large format lens is as good or better than the Zeiss 21mm. I don't need to make a 1:1 test for that. It is just how it is. (Note, I'm talking about the orginal zeiss 21mm, I have no clue how the new one is)

Christopher,
It's not possible to determine that's 'just how it is' without at least someone doing a comparison. One of the advantages of 1:1 comparisons is to provide an example of how significant a performance advantage is, and at what apertures.

No-one expects a zoom, even the Nikkor 14-24, to match the performance of a first rate prime. However, if the performance advantage of the prime is relatively small, the adjustable focal length of the zoom may be of greater benefit in practice.

A useful comparison would be between the the HR Digaron-S 23mm with the P45+, and the new Canon TS-E 24mm with 5D2, using full shift with 5D2 vertical to create an effective 5D2 sensor area of 48x36mm, which would match the both the sensor size of the P45+ and its pixel count fairly closely.

[rethmeier]

The original Zeiss was useless for straight lines,because of the extremely hard to correct waveform(moustache) distortion.
The new ZF 18 has similar problems,so I expect the new ZF 21 to be the same.
The Nikkor 14-24 @ 21 is pretty much perfect.

Regarding the Rodenstock 23HR that cost $11.000 in Australia, I doubt that is 4x better than the 14-24 @$2400

A P65+ will set you back $55.000 here in Sydney.

Is that P65+  8x times better than a D3x?

[Dustbak]

I may appear to be arguing for argument's sake, but I'm really just trying to get some reliable information. The original question from Jim2 was about differences between a 1Ds3 and MFDBs on print.

The only reliable information in this long thread, specific to the original question, is from DXOMark. It can be seen there that the 1Ds3 has lower performance than any of the 3 Phase DBs reviewed so far. The DR and tonal range of the 1Ds3 is noticeably less than the P40+, p45+ or P65+ at the smaller print size, and by extrapolation, presumably at any print size.

The noticeable exception is the D3X which at the smaller print size remarkably still has more DR than any of the 3 DBs featuring on the DXOMark website. In other areas such as color sensitivity, total range and SNR, the DBs still marginally outperform even the wonderful D3X, but only at the DB's base ISO which is almost a stop slower than the D3X base ISO. Increase the DBs' ISO by just one stop so it's closer to that of the D3X, (ISO 96 v ISO 78) then remarkably in all parameters tested the D3X is either equal to or better than all of the DBs tested at DXOMark..

However, when discussing lens performance, it's a different ball game. I'm very receptive to the idea of choosing a camera system because of the exceptionally fine lenses that may be available. The main reason I now own a Nikon D700 is because of the very fine Nikkor 14-24 zoom lens. I bought the lens first, tested it on my Canon 5D using an adapter, compared it with my existing Canon-mount wide-angle lenses, and found it to be so clearly better I decided to buy a camera body designed to fit the lens and allow full functionality.

For the benefit of everyone else, I showed my comparisons on this forum. I'm naturally skeptical of generalised claims people make without a comparison. Photography is a visual medium. Seeing is believing.

It may well be the case that some of the Schneider Digitar lenses and the new HR Digaron_S 23mm ultra-wideangle are far superior to any Canon equivalent lens. If that's the case, then let's see how much better they are. I promise I won't be upset if any of them knock the socks off the new TS-E lenses from Canon.

I would not really mind showing you images that have been captured with either one these lenses, 120 & 90 mm Schneider or any of the HC lenses since I own all of these. However since most of my 'testing' is done on work I have I tend to use only one system and the lens I need. I rarely repeat the same setup with another set because for me that would mean a waste of time. This means I can show you images of all lenses but I don't do tests. I don't have 2 of the same images made with different lenses.

I totally understand that you are sceptical about claims people make about performance. I have the same which means that I try the things I am interested in myself. I don't want to rely upon anyone's claims this includes DxO. My testing parameters are my eyes, my time and the perception of the people I make my images for. Yes, these things are no scientific parameters (mostly) but they do count for me

It is not that I don't want to do the tests people like you ask about but I don't see the point in doing so. They don't mean anything to me. I don't feel the need to proof anything. Whatever I say is the way I think I have experienced something whether someone chooses to believe me or not is up to them. I couldn't care less.

[Dustbak]

The original Zeiss was useless for straight lines,because of the extremely hard to correct waveform(moustache) distortion.
The new ZF 18 has similar problems,so I expect the new ZF 21 to be the same.
The Nikkor 14-24 @ 21 is pretty much perfect.

Regarding the Rodenstock 23HR that cost $11.000 in Australia, I doubt that is 4x better than the 14-24 @$2400

A P65+ will set you back $55.000 here in Sydney.

Is that P65+  8x times better than a D3x?

Grappenmaker

No naturally it is not. It probably will be a bit better depending on many other things besides the equipment. Handled improperly it can even be worse. Anyway I don't think that needs to be expained to you.

My god is that 23HR that expensive?!!?? No wonder I instinctively never dared to look at its pricing

[Christopher]

Grappenmaker

No naturally it is not. It probably will be a bit better depending on many other things besides the equipment. Handled improperly it can even be worse. Anyway I don't think that needs to be expained to you.

My god is that 23HR that expensive?!!?? No wonder I instinctively never dared to look at its pricing

Well that is Australia, which is a freaking expensive country. (only saying M9... around 8700 before tax, where i can get it in Germany for under 7k before tax) In Germany or the UK you can get the 23HR for a more normal price.

Is the P65 8 times better ? Well no, if you don't need to print large or have time to stitch. (Side note I would say a P65 is only around 5 times more than a d3x, but that once again depends.)