medium format vs. dslr lens sensor resolution

[SecondFocus]

Did I read somewhat recently that Leica is selling S systems as fast as they can build them?

[Rainforestman]

I own both a D800E and an IQ260.

Diffraction and lenses are the limiting factor on DSLR's.  According to Lloyd Chambers testing with the D800E, diffraction starts at f8 and is well and truly underway at f11.  Then.add the fact that the glass hasn't kept up with the demands of the high resolution sensors.  Even some of the best Nikon glass is quite frankly disappointing in the corners compared to the Schneider Kreuznach lenses.

I can happily shoot the IQ260 at f11.  I don't need to crop because my corners are comparatively sharp.

That is why I have gone medium format.  The D800E is a very good camera but it is not in the same league as the IQ260.  Even if a 54 Megapixel sensor were to be released in DSLR, the system will struggle to match medium format in terms of diffraction.  Not to mention all the other advantages of medium format.

[ErikKaffehr]

Hi,

I would say that there is a lot of potential for development on the 24x36 front. What I see is that we are having a new generation of high performing lenses from Sigma and Zeiss, like the Sigma Art-series and the new Zeiss Otus and 135/2 APO Sonnar. These lenses are sharp corner to corner, mostly. Many of today's lenses are very good over a large part of the field, but have weak corners. The Zeiss 25/2 is an example of that (according to the data I have seen).

The D4X is a rumor, we have not yet seen. It may be or it may be not. Sony seems to have a new generation full frame chip in development, it has shutter for each pixel, non bayer design and at 54 MP, but it is 1-2 years from market introduction.

On the other hand, Sony is also said to cooperate with Hasselblad on an MFD chip. Hasselblad seems to have a lot of good technology, but perhaps not the best sensors, so working with Sony may be a good option. I would also say that small pixels (like sub 4 micron) would be nice for MFD. Better detail reproduction and reduced aliasing.

Best regards
Erik

with the rumors of a 54 mp nikon d4x and the death of medium format (ha ha), at what point do you reach in either dslr or medium format, the ability of the lens not being able to resolve fully for the large mp sensor. the new zeiss lens for dslr is said to resolve about 31 mp and these lenses are very expensive. has the race to the top in the mp race become a marketing game instead of really being a benefit to the photographer.

[torger]

I'm afraid that small pixels will murder tech cam lens designs, for each new MFDB sensor generation the color cast has become worse and the manufacturers don't seem to care as they rather want users to use their SLR systems instead. The IQ260 was fortunately a break of this trend though.

I'm quite pessimistic. What probably will happen is that we'll see new sensors with smaller pixels and worse color cast, and therefore tech cam lenses is forced to become more and more similar to ordinary retrofocus lenses as required by 24x36mm CMOS sensors, tech cams will become more and more similar to an ordinary mirrorless (like Alpa FPS) and we come to a point when there's no longer any meaningful difference.

I think tech cam makers, Arca-Swiss, Alpa, Cambo, Sinar, Linhof should form a joint company and make their own backs and steer the development in a direction better for them. Maybe Leica can revitalize Sinar's line of backs and make it more than a tethered studio product.

[torger]

Diffraction is a zero-sum game. With smaller formats you get diffraction earlier, but you don't need to stop down as much either as you can use shorter focal length and thus get more depth of field with larger apertures for the equivalent field of view. To get the same depth of field as with f/8 on your D800E you need to stop down to f/11-f/13 on your IQ260.

What I think can prove to be a problem with smaller formats in terms of resolution is the increased requirement on precision in optics. I think it's too early to say if the 135 is too small to be able to reach resolving power currently seen in medium format due to precision issues in manufacturing, time will tell.

Another issue is that wide angles need to be strong retrofocus due to the strong color casts of the current CMOS sensors, and those retrofocus lenses need to be extremely complicated heavy and costly designs (see Rodenstock 32mm tech cam lens, it would be even worse than that) to resolve well over the whole frame, and while it may be technically possible to make, the market for a 2 kg $6000 wide angle lens would be too small for the manufacturers to care making one, when you can make something almost but not quite as good for a third of the price. However, the new Zeiss Otus lens is not exactly cheap or light, so maybe I'm wrong and we'll actually see wides that have these extreme designs.

Medium format did not before require strong retrofocus and wide aperture lenses, that's why we can find exceptional performance in tech cam lenses with relatively simple and inexpensive designs. This segment is about to die though, as MFDB makers don't seem to prioritize low color cast, and retrofocus lenses are more user-friendly in terms of live view (less vignetting and larger focusing aperture).

I own both a D800E and an IQ260.

Diffraction and lenses are the limiting factor on DSLR's.  According to Lloyd Chambers testing with the D800E, diffraction starts at f8 and is well and truly underway at f11.  Then.add the fact that the glass hasn't kept up with the demands of the high resolution sensors.  Even some of the best Nikon glass is quite frankly disappointing in the corners compared to the Schneider Kreuznach lenses.

I can happily shoot the IQ260 at f11.  I don't need to crop because my corners are comparatively sharp.

That is why I have gone medium format.  The D800E is a very good camera but it is not in the same league as the IQ260.  Even if a 54 Megapixel sensor were to be released in DSLR, the system will struggle to match medium format in terms of diffraction.  Not to mention all the other advantages of medium format.

[ErikKaffehr]

Hi,

I was not really considering technical cameras, my mistake. On the other hand I really feel that small pixels have benefits. My main interest is using Hasselblad lenses or perhaps Pentax 67 lenses and all those wide angles are retrofokus designs, so I have a bias.

Best regards
Erik

I'm afraid that small pixels will murder tech cam lens designs, for each new MFDB sensor generation the color cast has become worse and the manufacturers don't seem to care as they rather want users to use their SLR systems instead. The IQ260 was fortunately a break of this trend though.

I'm quite pessimistic. What probably will happen is that we'll see new sensors with smaller pixels and worse color cast, and therefore tech cam lenses is forced to become more and more similar to ordinary retrofocus lenses as required by 24x36mm CMOS sensors, tech cams will become more and more similar to an ordinary mirrorless (like Alpa FPS) and we come to a point when there's no longer any meaningful difference.

I think tech cam makers, Arca-Swiss, Alpa, Cambo, Sinar, Linhof should form a joint company and make their own backs and steer the development in a direction better for them. Maybe Leica can revitalize Sinar's line of backs and make it more than a tethered studio product.

[ErikKaffehr]

Hi,

I don't really think manufacturing precision is the problem. The lens for the Nokia 41MP cell phone is a quite impressive and seems to work very well on those very small pixels. The Zeiss Otus does impress, but so does the Sigma 35/1.4 A lens, albeit the latter one seems to have problems with sample variation.

Another point may be that larger formats make less demands on the lenses. A simple lens like a double gauss design can do really well at say 40 lp/mm but if you need to got to 60, 80 or even 120 lp/mm more complex designs are needed. There is probably some kind of optimum regarding sensor size, but cost will always be a factor finding that optimum.

Best regards
Erik

Diffraction is a zero-sum game. With smaller formats you get diffraction earlier, but you don't need to stop down as much either as you can use shorter focal length and thus get more depth of field with larger apertures for the equivalent field of view. To get the same depth of field as with f/8 on your D800E you need to stop down to f/11-f/13 on your IQ260.

What I think can prove to be a problem with smaller formats in terms of resolution is the increased requirement on precision in optics. I think it's too early to say if the 135 is too small to be able to reach resolving power currently seen in medium format due to precision issues in manufacturing, time will tell.

Another issue is that wide angles need to be strong retrofocus due to the strong color casts of the current CMOS sensors, and those retrofocus lenses need to be extremely complicated heavy and costly designs (see Rodenstock 32mm tech cam lens, it would be even worse than that) to resolve well over the whole frame, and while it may be technically possible to make, the market for a 2 kg $6000 wide angle lens would be too small for the manufacturers to care making one, when you can make something almost but not quite as good for a third of the price. However, the new Zeiss Otus lens is not exactly cheap or light, so maybe I'm wrong and we'll actually see wides that have these extreme designs.

Medium format did not before require strong retrofocus and wide aperture lenses, that's why we can find exceptional performance in tech cam lenses with relatively simple and inexpensive designs. This segment is about to die though, as MFDB makers don't seem to prioritize low color cast, and retrofocus lenses are more user-friendly in terms of live view (less vignetting and larger focusing aperture).

[torger]

Large sensor size is indeed unpractical in terms of cost, but the major part of the extreme high cost in the MFDB market is not due to the sensor, but due to the business model. There's no interest/ability to sell on volume, and there's still enough business users happy to pay. Also for full-frame DSLRs the sensor cost is a problem, but cameras are sold in a way where the sensor is allowed to make up more than half of the manufacturing cost, rather than a fraction like for some of the MFDBs.

The development in the coming 4-5 years will be most interesting to see.

[spotmeter]

I just purchased a CFV-50 back for my 'old' 501CM and the files blow away what I get with my Nikon D800E with top grade Zeiss glass and very careful technique.

Plus, I am rediscovering the joy of looking down the chimney finder and seeing my image on a huge, glorious 6cm square focus screen.

In addition, my newer Hasselblad lenses can resolve 8 point type at 20' from center to corner. None of my 35mm lenses are this good.

Because you don't use the corners of the Hasselblad lenses with the CFV-50, you get the best part of each lens.

Highly recommend this route for medium format digital.

[ErikKaffehr]

Hi,

I have gone a similar route, except I bought a P45+ plus a Hasselblad, the main reason was that Hasselblad V lenses are available at a very low price. I am quite satisfied with my lenses, even if I see some weaknesses. I feel that most lenses work best at f/8 - f/11, so that is what I use. What I also see is that I use my 120/4 a lot. It didn't do well in my initial testing, but does fine in real world. The Sonnar 150/4 is my best lens, but I consider replacing it with a 180/4, as I have 80, 120 and 150 right now.

I have not tested the 250/5.6, yet. Regarding chromatic aberration, LR 5 takes well care of lateral chromatic aberration on all lenses except the 40/4 Distagon. The 150/4 has some longitudional chroma, I would say that my Sony Alpha 70-400/4-5,6 is better in that sense, no color fringing at all.

I am quite happy with the Hasselblad 555ELD / P45+ combo, but the main advantage over my Sony cameras I see is sharpness/resolution.

Best regards
Erik

I just purchased a CFV-50 back for my 'old' 501CM and the files blow away what I get with my Nikon D800E with top grade Zeiss glass and very careful technique.

Plus, I am rediscovering the joy of looking down the chimney finder and seeing my image on a huge, glorious 6cm square focus screen.

In addition, my newer Hasselblad lenses can resolve 8 point type at 20' from center to corner. None of my 35mm lenses are this good.

Because you don't use the corners of the Hasselblad lenses with the CFV-50, you get the best part of each lens.

Highly recommend this route for medium format digital.

[david distefano]

if nikon comes out with a d4x and 54 mp sensor and the photographer uses the new zeiss otus lenses, with an 85mm f1.4 lens coming in 2014 to go with the 55mm f1.4 otus, is the image quality of the d800/e now stepping on the toes of medium format digital up to about a 30 to 40mp. even though the new zeiss lenses are expensive, a refurbished phase one p45+ bundle is around $16,000 which gets you back, body and one lens. for that price you could get the new hypothetical d4x at 54mp (questamated price going around the web $8-9,000)  plus 2 of these new lenses for about the same price, and the price of new mfdb, plus equipment would set you back a few years mortgage. i think hasselblad, before they made the stupid decision to discontinue the v line, had almost the right idea on pricing for the cfv-39 and cfv-50, for the serious amateur.

[Chris Livsey]

Hasselblad have discontinued the camera V range but I do not recall any statement that they would not in the future produce digital backs for them. They would be foolish not to tap into that market but then again that is not beyond the realm of possibility!

[synn]

if nikon comes out with a d4x and 54 mp sensor and the photographer uses the new zeiss otus lenses, with an 85mm f1.4 lens coming in 2014 to go with the 55mm f1.4 otus, is the image quality of the d800/e now stepping on the toes of medium format digital up to about a 30 to 40mp. even though the new zeiss lenses are expensive, a refurbished phase one p45+ bundle is around $16,000 which gets you back, body and one lens. for that price you could get the new hypothetical d4x at 54mp (questamated price going around the web $8-9,000)  plus 2 of these new lenses for about the same price, and the price of new mfdb, plus equipment would set you back a few years mortgage. i think hasselblad, before they made the stupid decision to discontinue the v line, had almost the right idea on pricing for the cfv-39 and cfv-50, for the serious amateur.

Would be nice to see all the analysis paralysis people keeping the said 54MP camera at f/4 regardless of content to avoid diffraction.  

To me, the only way that 35mm can match up and exceed MFDB is if foveon-like full color sensors become prevalent. A year and a half ago, when I got my D800, I was one of those people who thought "Oh here's some real competition to MFDB". All I can say is, people who make these comparisons just haven't experienced MFDB.


While the Otus might be a pretty nice piece of kit, it is useless for a lot of photographers. A manual focus lens limits its use for me as a location based portrait photographer. The D800 viewfinder is nowhere as bright as the old film Nikons for MF and the live view is pretty terrible to obtain critical focus too. Also, the 55mm focal length is rather silly. The 85 is more interesting, yes. But unless and until Zeiss hugs and kisses Canon and Nikon to get access to their AF algorithms, color me disinterested.

I stirred up the hornet's nest in the other thread, so here's some more fuel.

On the D800 is the 85 f/1.8 G, while not an Otus, is a very sharp lens in its own right. 80mm LS on the Credo.


Fair warning. Images were shot months apart, in both cases, casually out of my window. Take it for what it is, if you will.

[AreBee]

Synn,

Fair warning. Images were shot months apart...

I really, really hope that the building in the background was painted in the intervening time...or something is very, very wrong with one of those files.

[synn]

I'll bite. What is this "Very, very wrongness" that you speak of?

[ErikKaffehr]

Hi,

One is with a long lens with focus on the foreground leaves the other with a shorter lens with pretty much everything in focus. You could as well process the Nikon image in C1, as C1 one has a more aggressive sharpening. Or you could try amount=45, radius=0.7 detail=100 in LR, that works fine for me.

Best regards
Erik

I'll bite. What is this "Very, very wrongness" that you speak of?

[synn]

Hi,

One is with a long lens with focus on the foreground leaves the other with a shorter lens with pretty much everything in focus. You could as well process the Nikon image in C1, as C1 one has a more aggressive sharpening. Or you could try amount=45, radius=0.7 detail=100 in LR, that works fine for me.

Best regards
Erik

Dear Erik,

That is a very small crop from a very big scene. Both cameras were set to f/8 from what I can remember with pretty much everything in focus. Both lenses are within 5mm of each other, if you read my post carefully again.
I can sharpen the file even more in LR or process in C1 for sure, but it's not going to magically bring back detail that wasn't captured in the first place.


The simple point is, can a DSLR file be made to look like or come close to an MFDB file? Sure you can. I've been doing that for a while now. But it takes time and a lot of jumping through the hoops.
An MFDB file on the other hand looks "Right" pretty much 10 seconds after import.

[david distefano]

i use to own the cfv-16, shot many times with the cfv-39 and large format film up to 8x20. imho an 8x20 contact print or any other contact print from an ultra large film format system is the tops. now there is a special look.  the reason for the questions was, for many people like myself who enjoy photography for itself and not a business, the prices for mfdb's are way too expensive new and the value loss after purchase is worse than cars. i like to purchase new. there is something that i find special when i open new photographic equipment. also i am not talking about 50 or 60 or 80mp backs for the professional or the wealthy. af is not important to me because i use the hasselblad digital backs as well as roll film holders and sheet film on an arca swiss camera. with live view on dslr's for static objects it makes focusing that much easier for these old eyes. So yes i "know" the medium format look.

[ErikKaffehr]

Hi,

I have never used larger format on film than 6x7 (cm), but it is my understanding that film can give excellent results if properly scanned.

Regarding the MF look I cannot really say. My normal print size is A2 (16x23), I seldom print smaller and printing larger means sending to a lab. Recently I made 70x100 cm print from 24MP full frame and a 60x80 cm print from 39 MP MF. The MF was sharper, no question. The 24 MP print was about what I got from a 67 Velvia scanned on CCD film scanner at 3200PPI (I happen to have a similar print of the same subject in same light taken 10 years ago on Velvia).

I bought my P45+ for about 10k$ (as I recall), you can buy a Pentax 645D for the same price. Hasselblad also has some interesting cameras at affordable prices now and than.

Best regards
Erik

i use to own the cfv-16, shot many times with the cfv-39 and large format film up to 8x20. imho an 8x20 contact print or any other contact print from an ultra large film format system is the tops. now there is a special look.  the reason for the questions was, for many people like myself who enjoy photography for itself and not a business, the prices for mfdb's are way too expensive new and the value loss after purchase is worse than cars. i like to purchase new. there is something that i find special when i open new photographic equipment. also i am not talking about 50 or 60 or 80mp backs for the professional or the wealthy. af is not important to me because i use the hasselblad digital backs as well as roll film holders and sheet film on an arca swiss camera. with live view on dslr's for static objects it makes focusing that much easier for these old eyes. So yes i "know" the medium format look.

[AreBee]

Synn,

I'll bite.

For the avoidance of doubt my previous post in no way was an attempt by me to 'set you up'.

What is this "Very, very wrongness" that you speak of?

Only that while photographers can and do argue until the cows come home about the difference in colour between MF and 35mm camera files, there is no question that a significant difference in colour of the background building wall exists. Matching, say, the colour of foliage in the two files will not alter that fact.