phase versus hassleblad

[siba]

Mark Dubovoy has done an outstanding, pretty much unbiased test, which is pretty interesting.

And then he writes:

"The dynamic range of a typical high-end professional 35 mm DSLR is around 7 F/stops. Medium Format cameras are closer to 13 F/stops of dynamic range. You can see the difference from 30 feet away in a small print"

You can see the difference from 30 feet! well.....

I'm a phase one user. A professional photographer, who uses the phase one (P 45) for 99% of my work. I also use a canon 5D when light is bad and I'm shooting people, and I want to use the ambient light. I will bet everything I own, and everything my wife owns, that no person on this earth could tell the difference between a small print taken with a high end dslr and a high end medium format digital back from 30 feet (10metres) away.

Any takers?

Stefan Siba

[Schewe]

I will bet everything I own, and everything my wife owns, that no person on this earth could tell the difference between a small print taken with a high end dslr and a high end medium format digital back from 30 feet (10metres) away.

You sure about that bud? I used to do brackets of 1/6 of a stop on 8x10 film (and yes I COULD put the in the correct order on a 6' light box).

If you want to hold him to the term 30" "literally" then you might squeak out your bet hedge...but on the other hand, it's not hard to tell simply by subject matter and composition. at a first glance reaction, what was shot of a "serious" camera and a DSLR...

Obviously you aren't gonna be shooting a P 65+ in a seriously low light condition...so if I were a betting man, I would put my money on Mark being able to be more than 50% correct guessing on whether a given image was done on a real camera back vs a DSLR on a smallish print from 30" away...and I seriously wouldn't be betting MY wife on that deal! (ok that's not entirely what you said you would wager, but if you want to make it interesting).

:~)

Really, just from the point of shooting style and visual appearance, it ain't hard to tell what is shot handheld vs tripod...

[EricWHiss]

I found this article interesting partly because I've been testing my own ixpress 528c  multishot back against my own phase p20.   I've come to the opposite conclusion as Mark has - but I eliminated the camera and lens differences in his testing by using the same camera and lens - a Rollei 6008AF with schneider 150mm apo macro (and some testing with the 90 apo macro).   I had a job to shoot paintings in a gallery for catalog and possible reproduction and carefully shot both backs using the same lighting, camera and lens.  It appears that he the ixpress has better color fidelity, dynamic range, and actually a cleaner iso 400 shot too. The multishot and microstep functions of the 528c just leave the phase P20 for dead when it comes to color, tonality, detail, noise, etc.     These are older backs, but the current software for processing.     YMMV.

[barryfitzgerald]

Am I the only one who finds it curious about 13 stops of DR, and then we have the outside shot that's blown out on both?
The cyan shift is not uncommon when the blue channel blows.
Not sure where folks get their DR numbers from. I'm working with colour negative for some scenic stuff, and you really have to go some to get a blow-out, it's a rare event indeed. Now we can debate about the DR of neg film, but somebodies numbers are not adding up in the real world here.

Still interesting read, I liked the rant about pixel peeping, and even more so when I saw the 100% crop down the page ;-)

[hsmeets]

.....I would put my money on Mark being able to be more than 50% correct guessing on whether a given image was done on a real camera back vs a DSLR.....

So you imply you think he is a lucky guy in guessing.....not that he really knows or can recognize, see the difference   :-) :-)

[Graeme Nattress]

First - what definition of DR is being used? Second, you need to shoot proper backlit wide dynamic range charts to properly measure and test DR. If you just shoot a scene - even with as many parameters controlled as you can, how you can you measure it correctly when one camera will be more or less sensitive than the other. That will mean there might be highlight clipping on one camera, but much better shadow performance. Without a calibrated chart how can you see and measure that properly?

Graeme

[BernardLanguillier]

7 stops against 13...

For the sake of LL's credibility, wouldn't this better be amended?

Cheers,
Bernard

[tokengirl]

Photos of newspapers and circuit boards are so helpful.  Because I really dream of making the most excellent photos of newspapers and circuit boards.

[bjanes]

First - what definition of DR is being used? Second, you need to shoot proper backlit wide dynamic range charts to properly measure and test DR. If you just shoot a scene - even with as many parameters controlled as you can, how you can you measure it correctly when one camera will be more or less sensitive than the other. That will mean there might be highlight clipping on one camera, but much better shadow performance. Without a calibrated chart how can you see and measure that properly?

Graeme

If you have a full range shot, you can get a pretty good idea about the DR of a camera from a raw file as Emil Martinec shows in this post on the Open Photography forum. Since digital sensors are linear, the exposure is proportional to the raw value. Unfortunately, this type of analysis is rare on this forum.

Regards,

Bill

[bjanes]

7 stops against 13...

For the sake of LL's credibility, wouldn't this better be amended?

Cheers,
Bernard

Yes, indeed. You could post a reference to the DXO site or post your own data, but it would be a waste of time since it would be ignored be the elitists on this forum who substitute "expert consensus" for data and scientific analysis.

[michael]

Slow down everyone.

Mark Dubovoy is one of the most respected people in the digital camera technology community. He also is a Phd nuclear physicist, which doesn't necessarily give him expertise in photography, but does speak to his understanding of the scientific method.

Mark regularly consults for some of the major companies in this industry, so of they listen to and respect his opinion don't be so quick to shoot him down, even though you may not agree with his assessment in this case.

I imagine that he'll respond here in due course, so lighten up on the name calling.

Michael

[Mark D Segal]

...................Mark Dubovoy is one of the most respected people in the digital camera technology community. He also is a Phd nuclear physicist, which doesn't necessarily give him expertise in photography, but does speak to his understanding of the scientific method.

..............................
Michael

What Mark Dubovoy set out to do in this article is a bit courageous because making these kind of comparisons is really and truly not easy. One needs to "normalize" for many variables, do lots of testing both on display and in print, then think hard about what the results say. Hence, I was impressed with this article. I think Mark went to great length with painstaking methodology and lots of time to address all the potential pitfalls - at least that I could think of - and more, so cudos; and well-written. This was done by a scientist with the relevant experience to know what to do and how to do it.

That said, there were only two points of detail which I picked-up on: (i) the DR question; according to DxO the latest crop of high-end DSLRs are in the 12 stop range, while the Phase P40/P65 are 13.7 stops, so it wasn't clear to me where 7 stops came from; (not to say that DxO is an icon of perfection - they claim Phase P40/65 backs are 15 bit, but they are 16 according to Phase-1, and they have different low light ISO ratings for the two sensors which are identical except for the number of pixels); (ii) I was wondering about the use of f/11 for the lenses - perhaps a tad smaller aperture than ideal in terms of diffraction - another complex issue - or are the guidelines different for MF than for 35mm DSLR (i.e. about 2 stops above maximum aperture for minimizing diffraction and maximizing resolution - admittedly *guidelines* - not rules of law for all lenses). I have no doubt Mark selected f/11 for a reason, so I'll be interested to hear why.

[Alan Goldhammer]

If you want to hold him to the term 30" "literally" then you might squeak out your bet hedge...but on the other hand, it's not hard to tell simply by subject matter and composition. at a first glance reaction, what was shot of a "serious" camera and a DSLR...

Jeff, the article says 30 FEET, not inches.  Unless someone has superhuman vision, telling any difference between like prints from different cameras at this distance would be amazing.  I don't have any horse in this race other than this statement.  I found the comparison quite interesting to read given these cameras are way beyond both my interest and price range.  What I will do is my own experiment to compare a small print from my Canon S90 and Nikon D300 and see at what point I can tell the difference.  Will report back to all but I don't hold much faith that I can do this at 30 feet.

[luong]

Just a general question for those who conduct such comparisons. If measurements rather than subjective evaluations are desired, why not use Imatest ? Norman Koren was featured in the LLVJ.

[fredjeang]

Well, I certainly beleive Mark's comment about the differences that can be seen from a distance. 30 feets? 20 or 40 does it matter?
Just look in LU-LA Michael's photos. I've always found a noticiable difference when he uses MFD, and all the images are drasticaly downsampled for the web, despite, it is already visible at that size-resolution.

I found the article well done and much more precise than the Rockwell's Mamiya review for example.
It gives tendencies and, I think, brings some light about 2 star products.
In that sense Mark has succeded very well.

Regards,

Fred.

[siba]

I enjoyed the article, and I always find everything Mark Dubovoy writes interesting.
When I started this thread I certainly didn't want anyone criticizing the article per say.
It would be great if more people took the time and energy to give us such objective insights and comparisons.

It was purely the 30 feet away ability to tell the difference between a DSLR image and medium format back image which stood out as a surprising thing to claim. I took it as slightly toungue in cheek, and thus was willing to bet even my wife's possessions. 30 feet is a long way away to be able to tell anything much at all about a small print.

regards

Stefan

[PierreVandevenne]

Mark regularly consults for some of the major companies in this industry, so of they listen to and respect his opinion don't be so quick to shoot him down, even though you may not agree with his assessment in this case.

Hamamatsu, Nikon, etc... all agree on this definition

"The dynamic range value, however, corresponds to the limiting situation in which the full well capacity of a sensor element is reached, and is defined as follows:
Dynamic Range = Full Well Capacity (electrons) / Read Noise (electrons)"

A difference of 6 stops is.... hmmmm.... quite significant, and you don't have many factors to play with.

[ErikKaffehr]

Hi,

Measurement is one thing and perception another. Whom to believe? Perception is capricious. If measurements don't agree with perception we need to refine the measurement, or what we measure. I have this far not seen anything that could explain a five stop advantage with MFDB over DSLR in a scientifically feasible way.

It's a bit similar to audio. Simple measurements don't necessarily reflect perception. On the other hand, some people can hear difference in gold plated TOS cables and plain ones, which is obviously nonsense.

My last guess is that lenses play a major role. Better lenses with better coating, carefully built,  with blackened edges on lenses and baffling to keep internal reflections down will be better than similar lenses built with less care.

Best regards
Erik

What Mark Dubovoy set out to do in this article is a bit courageous because making these kind of comparisons is really and truly not easy. One needs to "normalize" for many variables, do lots of testing both on display and in print, then think hard about what the results say. Hence, I was impressed with this article. I think Mark went to great length with painstaking methodology and lots of time to address all the potential pitfalls - at least that I could think of - and more, so cudos; and well-written. This was done by a scientist with the relevant experience to know what to do and how to do it.

That said, there were only two points of detail which I picked-up on: (i) the DR question; according to DxO the latest crop of high-end DSLRs are in the 12 stop range, while the Phase P40/P65 are 13.7 stops, so it wasn't clear to me where 7 stops came from; (not to say that DxO is an icon of perfection - they claim Phase P40/65 backs are 15 bit, but they are 16 according to Phase-1, and they have different low light ISO ratings for the two sensors which are identical except for the number of pixels); (ii) I was wondering about the use of f/11 for the lenses - perhaps a tad smaller aperture than ideal in terms of diffraction - another complex issue - or are the guidelines different for MF than for 35mm DSLR (i.e. about 2 stops above maximum aperture for minimizing diffraction and maximizing resolution - admittedly *guidelines* - not rules of law for all lenses). I have no doubt Mark selected f/11 for a reason, so I'll be interested to hear why.

[dubomac]

I can see that my remarks about Dynamic Range have sparked quite a bit of discussion.  Thank you all.  This is very healthy and quite interesting.  

So let me explain where I am coming from:  

About a year ago I was very curious about how the dynamic range of DSLR's compared to film. Part of my curiosity came because I had heard a few professional photographers complain that DSLR's could not approach film in this regard. I made a number of calls to Canon, Nikon, Kodak and Fuji.  There was a universal consensus among the technical experts at these companies that top DSLRs delivered roughly 7 F/stops of dynamic range while most transparency film (there are obviously variations) delivered about 8 F/stops of dynamic range.  The point was that they were relatively close. Negative film had somewhat higher dynamic range, according to Fuji about an extra F/stop.

It is important to note that during those discussions dynamic range meant how many F/stops you could capture and still deliver texture and detail.

No, I did not perform any tests.  Since there was universal agreement, I took them at their word.

Now, turning to Medium Format, the factory spec for PhaseOne backs is approximately 13 F/stops. I know from experience shooting with these backs that they do deliver this kind of dynamic range.

This is how I came up with the numbers.

My experience with a Canon 1DsMKIII and my P65+ back certainly bears out a huge difference in dynamic range capabilities. I have never done a side-by-side measurement, but there is definitely a very large difference; 6 F/stops would not surprise me at all.

The reason for my comment of 30 feet is  based on recent anecdotal evidence. About one week before I started writing the article, I went to the camera store and I had with me an 8x10 inch print from an image that I shot using an ALPA camera with the P65+ back.  A couple of photography enthusiasts entered the store, and from about 30 feet away I heard one of them say "WOW, look at the dynamic range in that picture".  Then, they approached me and kept asking how many shots it took, and how I did the HDR processing of the multiple shots, etc.  They were stunned that this came from a single image capture. They also commented that "you could never get that kind of dynamic range out of a single image with a small format DSLR".  There are two salient points here:  First: One of the striking things about high end Medium Format backs is their dynamic range capabilities.  Second: You can detect it from quite far away.  If you do not like 30 feet, then make it 10, or whatever makes you feel good.

I would never state as fact something based on a single instance of anecdotal evidence, but this short piece of anecdotal evidence is typical and constantly repeats itself.  I find that (if the original subject has it), the first thing that strikes the viewer about a Medium Format digital image is the dynamic range.  The second thing is the level of detail retrieval and natural looking sharpness.

Having said all that, I am always interested in new evidence and new data, and if there is proof that some newer small DSLR (I have never used or tested a D3x for example) can match the Medium Format backs in dynamic range, I will be the first one to be ecstatic that the technology is getting better, and I will also stand corrected.  If anyone has the right measurements and/or the proper visual experience, please be so kind as to share it.

To summarize: My visual experience to date, combined with what I have heard from the manufacturers themselves points to a very significant difference in Dynamic Range capabilities between Medium Format and smaller DSLR's.

I hope this is helpful, and again thanks to all of you for the discussion.

Mark Dubovoy

[feppe]

No, I did not perform any tests. Since there was universal agreement, I took them at their word.
...
A couple of photography enthusiasts entered the store, and from about 30 feet away I heard one of them say "WOW, look at the dynamic range in that picture".
...
To summarize: My visual experience to date, combined with what I have heard from the manufacturers themselves points to a very significant difference in Dynamic Range capabilities between Medium Format and smaller DSLR's.

I'm speechless - good thing LL is free as otherwise I'd be demanding my money back.

My experience with a Canon 1DsMKIII and my P65+ back certainly bears out a huge difference in dynamic range capabilities. I have never done a side-by-side measurement, but there is definitely a very large difference; 6 F/stops would not surprise me at all.

DXOmark lists Canon 1Ds MkIII you have tested to have quarter of a stop less dynamic range as Phase P65 Plus. While there are certainly different ways to measure and interpret the results, going from 1/4 to 6 EV difference in DR with any interpretation would be shocking.

Several people have also shown empirically on these forums that MFDBs have little, if any, edge on DR. In fact, 1D MkIV has one stop more DR than both P65 and Hassy H3DII 50.