Can you see a difference in small prints between an MFDB and a DSLR?

[Bart_van_der_Wolf]

I don't understand why Bart's "original raw conversion" is so soft. I never seen anything so soft coming out of a camera without low pass filter, even with no sharpening applied.

Hi Jerome,

That's what was puzzling me as well, although it also allowed to show that there is still a lot that can be done to salvage the situation. However, I usually get better base material to work from, even with an OLP-Filtered sensor array, so there may still be an issue with the lens. To ferret out that aspect, which is beyond the topic of the thread, some more systematic testing is required (as I already said, a star target shot may help to determine the actual limiting resolution, but focusing on a flat subject must be close to perfect).

Cheers,
Bart

[ErikKaffehr]

Hi Bart,

I made a series of shots with the Planar 120 at 1 cm intervals it is 50x70 kappa board with five of your resolution targets, shot at 1 cm intervals.

http://echophoto.dnsalias.net/ekr/Articles/ForBart/Planar120_f8_series/

Here are some selected shots from the 2006 MFD shootout by Michael Reichmann, Bill Atkinson and Charlie Cramer using different combos. Please don't redistribute. The copyright owners permitted pesonal use: http://echophoto.dnsalias.net/ekr/Articles/ForBart/2006ShootOut/

There is a file called CaptureInfo.xml describing all combos. I just posted a selection of files. They have the extension TIF but they all raw.

Best regards
Erik

Hi Jerome,

That's what was puzzling me as well, although it also allowed to show that there is still a lot that can be done to salvage the situation. However, I usually get better base material to work from, even with an OLP-Filtered sensor array, so there may still be an issue with the lens. To ferret out that aspect, which is beyond the topic of the thread, some more systematic testing is required (as I already said, a star target shot may help to determine the actual limiting resolution, but focusing on a flat subject must be close to perfect).

Cheers,
Bart

[Bart_van_der_Wolf]

I made a series of shots with the Planar 120 at 1 cm intervals it is 50x70 kappa board with five of your resolution targets, shot at 1 cm intervals.

http://echophoto.dnsalias.net/ekr/Articles/ForBart/Planar120_f8_series/

Hi Erik,

Of this series, CF044614.IIQ looks to have the best center focus, CF044613.IIQ is marginally less it seems. The series shows how much the blur diameter grows with just a few centimetres difference, resolution is affected easily by defocus. However, I have a question because the target(s) seem to be shot from too close-up to really stress the lens.  I know that indoors with longer focal lengths is can be difficult to get enough distance.

Maybe I'm mistaken in my interpretation, but the target (star diameter) measures approx. 1224 pixels which, on a 6.8 micron pitch sensor, means that the maximum star diameter is 1224 x 0.0068mm = 8.3232 mm on sensor. The original target's star diameter, when printed at the indicated PPI, measures 100 mm. That means that the effective magnification factor was 0.083232, or approx. 1:12 . A magnification factor  of 1:12 with a 120 mm lens means that the shooting distance was approx. 1.56 metres, or 13x the focal length. My recommendation is to use 25x to 50x focal length. That will reduce the influence of the print quality and stress the performance of any lens, and also produce a smaller target which is easier to post full size crops of.

Is my reconstruction of 1.56 metres shooting distance approximately correct? I'd prefer doing an analysis on something more critical, say 3 metres for a 120mm lens, just to minimize the influence of the print quality.

And could you tell me which aperture was used, that info is not recorded in the EXIF.

Cheers,
Bart

[ErikKaffehr]

Hi Bart,

I will redo it, no problem.

There was a couple of  reasons I did it this way. I wanted to have your test charts in the corners, and this was the largest board I have found at home. The other was that the lens is optimized for short distances, at least according to those who make it.

Just to note, there is a newer version called CFi wh ich is much improved according to Zeiss. MTF curves are basically same, but Zeiss says the new lens has much higher contrast. I pressume they really say that flare levels are reduced.

Best regards
Erik

Hi Erik,

Of this series, CF044614.IIQ looks to have the best center focus, CF044614.IIQ is marginally less it seems. The series shows how much the blur diameter grows with just a few centimetres difference, resolution is affected easily by defocus. However, I have a question because the target(s) seem to be shot from too close-up to really stress the lens.  I know that indoors with longer focal lengths is can be difficult to get enough distance.

Maybe I'm mistaken in my interpretation, but the target (star diameter) measures approx. 1224 pixels which, on a 6.8 micron pitch sensor, means that the maximum star diameter is 1224 x 0.0068mm = 8.3232 mm on sensor. The original target's star diameter, when printed at the indicated PPI, measures 100 mm. That means that the effective magnification factor was 0.083232, or approx. 1:12 . A magnification factor  of 1:12 with a 120 mm lens means that the shooting distance was approx. 1.56 metres, or 13x the focal length. My recommendation is to use 25x to 50x focal length. That will reduce the influence of the print quality and stress the performance of any lens, and also produce a smaller target which is easier to post full size crops of.

Is my reconstruction of 1.56 metres shooting distance approximately correct? I'd prefer doing an analysis on something more critical, say 3 metres for a 120mm lens, just to minimize the influence of the print quality.

And could you tell me which aperture was used, that info is not recorded in the EXIF.

Cheers,
Bart

[Bart_van_der_Wolf]

Hi Bart,

I will redo it, no problem.

There was a couple of  reasons I did it this way. I wanted to have your test charts in the corners, and this was the largest board I have found at home. The other was that the lens is optimized for short distances, at least according to those who make it.

Yes, I understand. What I usually do to determine the maximum performance of the corners, is shoot a single target positioned in a corner and individually focused one at a time. The reason is that 5 targets on a flat plane tell more about field curvature than optimum corner performance. Also, when the 4 corners are individually focused to the best possible resolution, then a well centered lens will give the same resolution in all 4 corners. When one or two corners on one side perform worse, then decentering is in play.

The difficulty with lenses optimized for the close up or macro range, is that the target needs to have even higher quality to out resolve the capabilities of the best lenses. An inkjet print at 600 or 720 PPI can 'resolve'/dither detail up to 1/23.6 or 1/28.3 of a millimeter (which will diffuse a bit depending on the paper coating). To make sure that a very good lens cannot possibly resolve that level of detail, we would need at most a magnification factor of say 1:10, or 1.32 metres with a 120mm lens, still not very close-up. Printed targets (not bi-tonal, but grayscale, such as sinusoidal) are a limiting factor for close-up testing.

That's also a bit too close to what you used if the lens is very good (which we want to establish as true or false). So if your lens is very good, it may be the (dithered) print quality that sets part of the apparent system resolution limit. Hence my recommendation for at least 25x focal length. When your lens is not perfect, which we suspect, then the current resolution limit of an approx. 104 pixel blur diameter, or 88% of Nyquist (64.8 cycles/mm out of 73.5 cycles/mm maximum for a 6.8 micron sensor), will not change very much with a more demanding target distance because it's mostly the lens that limits system performance. This would also suggest that you'd not be able to print as large as with a better system resolution for an identical visual resolution. Not a real drama, but still something like 12% can be considered a significant size difference.

Of course, diffraction continues to have it's influence, and a lens often has it's peak performance 1-2 stops narrower than wide open, which may be limiting for single shot DOF. Compromises, compromises, ...

Cheers,
Bart

[EricWHiss]

[quote author=ErikKaffehr link=topic=87650.msg717378#msg717378 date=1394482648

The essence was:
- If you use same processing the SQF values will be pretty close (within 5%) for low end MF and high end DSLRs
- If you use use different processing you can get almost any SQF you want

Anyway, I really learned a lot.

[/quote]

Yes, I wrote that as well in the very beginning - that the outcome of the imatest results can vary depending on how the image is processed.

More importantly I also wrote that detail is but only one facet of a number of image qualities that can distinguish a Medium format image from a 35mm image.

There is also: Tonality, Color separation, Image compression effects due to different magnification levels, roll off from in focus to OOF areas… etc.   All of these will affect the appearance of depth and lend a more palpable quality to even printed images which is what your original topic was about - if a person could see the differences.

Like so many of the these threads, most of the qualities of an image are thrown aside because they are hard to measure and compare and the discussion congeals around sharpness.  But this leads to may false comparisons and discounts so many important aspects of an image - including the concept, idea, and composition which are paramount.   


Much of what Bart wrote about processing images is very useful information and I thank him for that, though I admit that I haven't been able to get results from focus magic or topaz as easily as he does, nor do I have the time to handle batches of images in this way.  Instead I use a two step process - initial sharpening in C1 with modest amounts of sharping and then a 2nd step in Lightroom with the detail slider set high and also modest amounts of sharpening.  This seems to work well for me and fits into a workflow that can be easily handled in batches. 

Erik, I am happy that you "learned a lot" and am thankful that Bart, Synn and others have been able to show you that both your images may be processed better and also that some are not focused well.  I have been sending you personal messages about potential issues with your camera or handling for many months but you never seemed to notice.  I hope that now you are aware of this you will revisit many of the 'report's and findings you have put forth to the world and see if you have not come to any wrong conclusions now that you know you can get better results with your MFDB. 

[ErikKaffehr]

Well,

Much of this discussion would not be needed if MF dealers like yourself posted carefully made comparison images available as raw files. Unfortunately this is seldom the case, and some of the cases I have seen the DSLR files from MF vendors they have been quite crappy.

Imaging Resource has posted good raw images shot with the Pentax 645D, and I used those files for my SQF calculations. Similarly I got some images from Diglloyd so I could compare Leica S2 with Nikon D3X, my findings on those images were published here: http://echophoto.dnsalias.net/ekr/index.php/photoarticles/38-observations-on-leica-s2-raw-images, with kind permission of Mr. Lloyd Chambers.

So, my findings are not limited to my own shooting with my equipment.

Regarding sharpening, I feel it is perfectly OK to compare images with similar sharpening, using the same toolkit. It may be that Capture One is a better choice than my favorite tool Lightroom, but I would say that for a proper comparison an identical toolset is needed, with either identical or optimal settings. Establishing optimal settings can be difficult, so I feel identical settings are a better choice.

What I also stated, quite clearly, is that there is a very obvious difference between my prints using a 5.5X loupe, so much of the difference in image quality is carried over to print, just not or possibly barely visible in the print.

Quite a few posters have found that there was no significant difference in prints up to say 20" wide, which is pretty consistent with my observations. Some posters have other observations.

My recommendation is that anyone planning to buy an MFDB should either arrange a rental, or loan if possible. Or download high quality reference raw images that he/she can develop on his/her own. For a proper comparison proper comparison images are needed from existing equipment.

I have seen and used images from the following sources in forming my opinions:

- Marc Calmont: Phase One IQ 180 and Nikon D800E (and used his images with his kind permission)
- Tim Ashely: Phase one IQ 180 and Nikon D800E (and used his images with his kind permission)
- Tim Parkin: Published several comparisons between different cameras, supplied me with images from Phase One IQ180 and Nikon D800/D800E. Tim Also suggested that I look into MFD colour rendition.
- Lloyd Chambers: Nikon D3X and Leica S2.
- Chris Barret was nice enough to publish raw images from IQ 260 and Sony A7r

Some of the images were used as a basis for this article: http://echophoto.dnsalias.net/ekr/index.php/photoarticles/71-mf-digital-myths-or-facts which was reviewed by Tim Ashley, Tim Parkin and Marc McCalmont. All three reviewers felt the article was very good. I asked all contributors to read the article and suggest improvements.

So, I have quite a few sources who are experienced users of high end MF equipment.

Best regards
Erik Kaffehr

[quote author=ErikKaffehr link=topic=87650.msg717378#msg717378 date=1394482648

The essence was:
- If you use same processing the SQF values will be pretty close (within 5%) for low end MF and high end DSLRs
- If you use use different processing you can get almost any SQF you want

Anyway, I really learned a lot.



Yes, I wrote that as well in the very beginning - that the outcome of the imatest results can vary depending on how the image is processed.

More importantly I also wrote that detail is but only one facet of a number of image qualities that can distinguish a Medium format image from a 35mm image.

There is also: Tonality, Color separation, Image compression effects due to different magnification levels, roll off from in focus to OOF areas… etc.   All of these will affect the appearance of depth and lend a more palpable quality to even printed images which is what your original topic was about - if a person could see the differences.

Like so many of the these threads, most of the qualities of an image are thrown aside because they are hard to measure and compare and the discussion congeals around sharpness.  But this leads to may false comparisons and discounts so many important aspects of an image - including the concept, idea, and composition which are paramount.  


Much of what Bart wrote about processing images is very useful information and I thank him for that, though I admit that I haven't been able to get results from focus magic or topaz as easily as he does, nor do I have the time to handle batches of images in this way.  Instead I use a two step process - initial sharpening in C1 with modest amounts of sharping and then a 2nd step in Lightroom with the detail slider set high and also modest amounts of sharpening.  This seems to work well for me and fits into a workflow that can be easily handled in batches.  

Erik, I am happy that you "learned a lot" and am thankful that Bart, Synn and others have been able to show you that both your images may be processed better and also that some are not focused well.  I have been sending you personal messages about potential issues with your camera or handling for many months but you never seemed to notice.  I hope that now you are aware of this you will revisit many of the 'report's and findings you have put forth to the world and see if you have not come to any wrong conclusions now that you know you can get better results with your MFDB.  

[ErikKaffehr]

Hi Bart,

I posted two series, shot at 3.9 m (measured from approximate sensor plane). The images were shot with 2-3 cm intervals. In the first series I used the Planar 120 at both full aperture using incandescent light and at f/8 using electronic flash. I found an asymmetry in the off axis targets and readjusted them to be within 1 cm measured from the lens and reshot a single image a 0 offset, marked unfortunately as 150/8.

I also added a series using my Sonnar 150/4 shot at f/8.

I used the split image wedge with the PM5 viewfinder and there was no visual difference on the wedge with the camera movement I had.

Regarding aprture, I shoot mostly f/11 if I can. I see some loss of resolution at f/11, but focusing in the field is not easy, so I rather add a tiny bit of DoF.  On the Distagon 40/4 I feel there is a need to stop down to at least f/11 because it is my belief, based on Zeiss MTF data, that the lens has significant field curvature, with edges bending toward the camera. This can often yield sharp foreground. My interpretation of the Zeiss MTF data may of course be faulty, there may be other reason for the low MTF off axis.


Best regards
Erik

Yes, I understand. What I usually do to determine the maximum performance of the corners, is shoot a single target positioned in a corner and individually focused one at a time. The reason is that 5 targets on a flat plane tell more about field curvature than optimum corner performance. Also, when the 4 corners are individually focused to the best possible resolution, then a well centered lens will give the same resolution in all 4 corners. When one or two corners on one side perform worse, then decentering is in play.

The difficulty with lenses optimized for the close up or macro range, is that the target needs to have even higher quality to out resolve the capabilities of the best lenses. An inkjet print at 600 or 720 PPI can 'resolve'/dither detail up to 1/23.6 or 1/28.3 of a millimeter (which will diffuse a bit depending on the paper coating). To make sure that a very good lens cannot possibly resolve that level of detail, we would need at most a magnification factor of say 1:10, or 1.32 metres with a 120mm lens, still not very close-up. Printed targets (not bi-tonal, but grayscale, such as sinusoidal) are a limiting factor for close-up testing.

That's also a bit too close to what you used if the lens is very good (which we want to establish as true or false). So if your lens is very good, it may be the (dithered) print quality that sets part of the apparent system resolution limit. Hence my recommendation for at least 25x focal length. When your lens is not perfect, which we suspect, then the current resolution limit of an approx. 104 pixel blur diameter, or 88% of Nyquist (64.8 cycles/mm out of 73.5 cycles/mm maximum for a 6.8 micron sensor), will not change very much with a more demanding target distance because it's mostly the lens that limits system performance. This would also suggest that you'd not be able to print as large as with a better system resolution for an identical visual resolution. Not a real drama, but still something like 12% can be considered a significant size difference.

Of course, diffraction continues to have it's influence, and a lens often has it's peak performance 1-2 stops narrower than wide open, which may be limiting for single shot DOF. Compromises, compromises, ...

Cheers,
Bart

[EricWHiss]

Erik,
All of the images I won the 1st place (fine art) category in the Prix de la Photography Paris were taken with medium format, same with all the IPA merit awards.  All of my images published in Black and White magazine were taken with medium format.  In fact all of them were taken with one Rollei or another and a MFDB. It's because I loved the Rolleiflex system so much I became a dealer, but I remain very interested with my own art and image-making and still do some product, scientific, and art reproduction work - most of which I can't show or share freely.

I have posted very many images in various places, some in full res, and frequently correspond with friends and clients where we share images.  Occasionally, I'll post some images here to LuLa as well with 100% crops when its necessary to explain something.  I don't understand your constant requests for RAW images when you have your own MFDB system, and see them as purely self indulgent.  Additionally, from what I have observed of your posts and how you have recently used a list of other people to put yourself higher,  the probability for you to misunderstand or mis-represent me and my images is very high.  I disdain the idea of you mushing around with one of my files when the only purpose is for you to "measurebate".

Much of this discussion would not be needed if MF dealers like yourself posted carefully made comparison images available as raw files. Unfortunately this is seldom the case, and some of the cases I have seen the DSLR files from MF vendors they have been quite crapp

So, I have quite a few sources who are experienced users of high end MF equipment.

[ErikKaffehr]

Hi,

The reason I suggest raw files is simply disclosure. I am in no way suggesting that you present any of your award winning images as raws.

Regarding my sources, I just explain that my opinion is based on a wider selection of images than my own.

Best regards
Erik

I have posted very many images in various places, some in full res, and frequently correspond with friends and clients where we share images.  Occasionally, I'll post some images here to LuLa as well with 100% crops when its necessary to explain something.  I don't understand your constant requests for RAW images when you have your own MFDB system, and see them as purely self indulgent.  Additionally, from what I have observed of your posts and how you have recently used a list of other people to put yourself higher,  the probability for you to misunderstand or mis-represent me and my images is very high.  I disdain the idea of you mushing around with one of my files when the only purpose is for you to "measurebate".

[synn]

Hi,

The reason I suggest raw files is simply disclosure. I am in no way suggesting that you present any of your award winning images as raws.

Regarding my sources, I just explain that my opinion is based on a wider selection of images than my own.

Best regards
Erik

The dealers here have always been more than helpful to prospective customers and have provided them with raw files fitting their requirements. I am one such example.

There are several raw files available for download on the mamiyaleaf website and a few on phaseone's site (no test charts though, so you might be disappointed). How many raw files can you download from the canon, Nikon or Sony sites? Or from b&h or adorama, the biggest dealers  of these brands?

Nobody owes you anything Erik, let alone raw files. The resources are always there for people with genuine needs. If you can't find what you want to pixel peep at 400% and measurebate, that's your problem, no one else's.

[ErikKaffehr]

Hi,

I feel that if some post statements on a forum claiming the superiority of equipment he sells, it would be nice if he also would support his statements with images, and processed images say little about the starting point. Bart's processing of my image is a good example of that. I do post raw images mostly. I am not asking for raw images, I am sharing mines.

I don't see those images as piece of art, just as the most basic and correct information I can deliver.

But, yes I agree none has obligation to share images, it just seems to me that it is quite honest to do that.

Best regards
Erik

The dealers here have always been more than helpful to prospective customers and have provided them with raw files fitting their requirements. I am one such example.

There are several raw files available for download on the mamiyaleaf website and a few on phaseone's site (no test charts though, so you might be disappointed). How many raw files can you download from the canon, Nikon or Sony sites? Or from b&h or adorama, the biggest dealers  of these brands?

Nobody owes you anything Erik, let alone raw files. The resources are always there for people with genuine needs. If you can't find what you want to pixel peep at 400% and measurebate, that's your problem, no one else's.

[synn]

The superiority of equipment shines through when someone with superior taste, skill and technique uses it. Sharing raw files will do nothing for those who do not possess those qualities (files that are btw, available. Read my post again).

in simpler words, full access to Fernando Alonso's Ferrari won't make you drive like him. You actually would drive it slower than you do your Volvo. Incidentally,Alonso would drive your Volvo faster than you do too.

the question is, why are you not settling with the Volvo you clearly are comfortable with and are obsessing with the Ferrari you can't make head and tail of?

[ErikKaffehr]

Hi Synn,

I actually drive a Toyota, and I actually can see the difference between a Volvo and Ferrari. But this thread is about visible differences between relatively small prints. I don't think so, presuming similar processing, on the samples I have seen. Larger print sizes, where resolution exceeds the resolution of the human eye, there is absolutely a difference.

You made a pretty neat test yourself, published on this thread on LuLa: http://www.luminous-landscape.com/forum/index.php?topic=87158.0

If you print those images in size A2 (or smaller) and show to a few uninformed friends, can they tell the difference?

Best regards
Erik

The superiority of equipment shines through when someone with superior taste, skill and technique uses it. Sharing raw files will do nothing for those who do not possess those qualities (files that are btw, available. Read my post again).

in simpler words, full access to Fernando Alonso's Ferrari won't make you drive like him. You actually would drive it slower than you do your Volvo. Incidentally,Alonso would drive your Volvo faster than you do too.

the question is, why are you not settling with the Volvo you clearly are comfortable with and are obsessing with the Ferrari you can't make head and tail of?

[EricWHiss]

The superiority of equipment shines through when someone with superior taste, skill and technique uses it. Sharing raw files will do nothing for those who do not possess those qualities

+1

Nice one Synn …. A truer statement can not be made. 

And this may explain why so many people don't see the difference.  I'm not sure I'd say MF takes more skill to use properly, because that probably isn't true, but it is different. 

[hjulenissen]

I asked you to look for good light as a photographer to make pleasing images, which also happen to be good test images.
...
It just so happens that no matter what camera you use, you get better results with better light.

I usually find it more pleasing to watch portraits of interesting/pretty people than mundane people.

I don't think that an evaluation of camera capabilities needs to include pretty people, though. I take it that people with an interest in photography are able to make that connection themselves.

-h

That does not negate my point, nor is it contrary.

(I have pasted quotes of the relevant parts of posts in good faith to represent the discussion as I see it.)

A skilled photographer ought to be able to see the qualities of a good camera even if the model is ugly or the light is boring. A less skilled photographer, on the other hand, might need "impressive" shots in order to appreciate the qualities of a camera.

-h

[Chris Livsey]

Surely a good camera, even if the model is ugly and the light boring, will assist the photographer to make the best of what is presented whilst a bad camera just adds to the problems faced.
This links to the "tool for the job" argument.

A similar question to the original has been posed on another forum :- "What does a DMF give in exchange for it's astronomical costs?"
These are mainly amateurs but with disposable income, up to now it is the old chestnuts, bit depth and lenses no one has yet said the camera/viewfinder.

[MrSmith]

What if you don't look through the viewfinder to make images?
I rarely use the viewfinder shooting still-life.

In answer to the OP. No I don't see a difference but then I don't shoot to print.

[bjanes]

The superiority of equipment shines through when someone with superior taste, skill and technique uses it. Sharing raw files will do nothing for those who do not possess those qualities (files that are btw, available. Read my post again).

in simpler words, full access to Fernando Alonso's Ferrari won't make you drive like him. You actually would drive it slower than you do your Volvo. Incidentally,Alonso would drive your Volvo faster than you do too.

the question is, why are you not settling with the Volvo you clearly are comfortable with and are obsessing with the Ferrari you can't make head and tail of?

I find this smug and condescending comment directed to Erik Kaffer inappropriate for collegiality and good taste on the forum. Erik is a well respected forum member who has an engineering background and is an inveterate experimenter, and I have learned much from his posts and website and the discussion they engender. He brings scientific rigor to his analyses. One reason for requesting raw files is to minimize the variables that are inherent in rendered images (demosaicing, sharpening, noise reduction, etc.) Synn mocks Eric for wanting raw images, yet his own analysis of some of Erik's raw images demonstrate why raw images are necessary for proper evaluation.

Synn apparently considers him self to be a great artist who can get the most from his MFDB equipment. I've looked at his images on his web site and he does good work, but he is no Ansel Adams or Yousuf Karsh. I have seen similar quality work shown by talented amateurs at my local camera club. (I can't wait for him to challenge me to show my own images, but that is not the point. Does an art critic need to have painted on the level of Monet?)

The automobile comparison is in poor taste and implies that Erik doesn't know what he is doing. However, the comparison is appropriate in one way: a Ferrari is not a very practical car. It is temperamental, uncomfortable, and high maintenance as well as expensive. It does make a statement by rich car aficionados. The same qualities might also apply to MFDB setups. Artistic merit can be judged at relatively low resolution and one does not have to have MFDB setups to achieve artistic merit. Many of Ansel's iconic images are not all that sharp by modern standards.

Bill

[Manoli]

I find this smug and condescending comment directed to Erik Kaffer inappropriate for collegiality and good taste on the forum.

+1

Not just this comment but quite a few others - and not just Synn. As a previous poster said 'civility has always been a feature of this forum' – shame that a few felt the need to disrespect that.