MF vs 1Ds3

[Ray]

Ray,

if you really want to compare FF 35mm with a DMFB then do compare it by taking such a camera and such a back. Period. You won't compare the same type of sensors in play with your test.

Thierry
[a href=\"index.php?act=findpost&pid=189337\"][{POST_SNAPBACK}][/a]

No can do, Thiery. It's difficult enough and expensive to even compare a Nikon D3 with my Canon 5D, here in Brisbane. Last time I checked with the hire company, their priority was to fill the back orders for sales of the D3 before considering hiring out a camera. And the particular Nikkor lens I was interested, the 14-28/2.8 was not on the list for hire because of it's inability to take a protective filter (lens too bulbous).

There'd be no point in my comparing a DB with my 5D. I don't need to be convinced that a 21, 30, 45mp large format sensor produces technically higher quality images than a 12mp miniature camera.  

[Ray]

Thiery,
It's true that such a comparison (between the 40D and 5D) could not reveal the differences between the CMOS and CCD sensors. Are these significant? Do we have any examples in recent camera models, of two sensors of the same size and pixel count, where such differences can be seen in their own right, isolated from other distracting effects?

One (perhaps more significant) difference between the two sensor types is the fact that DBs generally have no AA filter. Can we compare the significance of such different qualities; on the one hand the subtle effects of the nature of the CCD sensor compared to the CMOS sensor (at base ISO), as opposed to the subtle effects of the presence of (or lack of) an AA filter? Which consequence is greater?

As it happens, it is possible to have the AA filter removed from the 5D. However, I'm not going to go to that trouble and expense just for the purpose of getting a more valid comparison, but it would certainly remove one more qualitative difference between the two sensor types, leaving, as far as I can see, just the CCD versus CMOS issue, which might be a non-issue, at least at base ISO.

[Chris Livsey]

Thiery,
One (perhaps more significant) difference between the two sensor types is the fact that DBs generally have no AA filter. .
[a href=\"index.php?act=findpost&pid=189560\"][{POST_SNAPBACK}][/a]

AFAIK the only non DB camera with no AA filter as stock is the Leica M8 where depth behind the lens was paramount, as was image quality, YMMV as to which consideration came first. Certainly in my experience the sharpening required in post, for prints within each cameras native output, is very similar between the P20 and M8 and a lot different to the Canon 1Ds3 which is different to the 2 variant.
Your problems are compounded in comparing what may be subtle differences in the variations present between the types of sensor and the variations seen between different serial number examples of he same model/back where POV and lenses etc can be matched. Edmund for example has clearly shown major differences in performance in backs of the same model that have passed QC. That of course depends on the tolerances QC are applying to a pass. Not all backs are created equally a fact also seen, and frequently commented on, in lens QC discussions. Most commentators are happy when they find a lens which is on spec. it would be nice if one out of spec. on the good side of the distribution curve, could be found, the same for backs.

[Graham Mitchell]

AFAIK the only non DB camera with no AA filter as stock is the Leica M8
[a href=\"index.php?act=findpost&pid=189610\"][{POST_SNAPBACK}][/a]

There is also the Leica DMR, which uses a CCD sensor with no AA filter. You could then use the same Leica lens with the DMR and a Canon body.

In fact there was a large thread on FM with this very test. Guy Mancuso tested the DMR against a Canon 1DsII. He sold his Canon gear after that test.

[eronald]

There is also the Leica DMR, which uses a CCD sensor with no AA filter. You could then use the same Leica lens with the DMR and a Canon body.

In fact there was a large thread on FM with this very test. Guy Mancuso tested the DMR against a Canon 1DsII. He sold his Canon gear after that test.
[a href=\"index.php?act=findpost&pid=189629\"][{POST_SNAPBACK}][/a]

Guy also then moved his whole practice to the M8.

Edmund

[Chris Livsey]

Guy also then moved his whole practice to the M8.

Edmund
[a href=\"index.php?act=findpost&pid=189638\"][{POST_SNAPBACK}][/a]

Now, ie very recently, he has bought a Nikon D3.

And yes I had forgotten about the other Leica although perhaps I can get away with it by adding "in current production" to my assertion. It does however mean as you say the lens can be taken out of the equation.

[httivals]

Actually, Guy ended up not buying a Nikon D3.  He bought a Mamiya ZD back.  I'm not making this up -- it's in a couple of threads at the www.getdpi.com medium format forum.

Now, ie very recently, he has bought a Nikon D3.

And yes I had forgotten about the other Leica although perhaps I can get away with it by adding "in current production" to my assertion. It does however mean as you say the lens can be taken out of the equation.
[a href=\"index.php?act=findpost&pid=189656\"][{POST_SNAPBACK}][/a]

[Ray]

Edmund for example has clearly shown major differences in performance in backs of the same model that have passed QC. That of course depends on the tolerances QC are applying to a pass. Not all backs are created equally a fact also seen, and frequently commented on, in lens QC discussions. Most commentators are happy when they find a lens which is on spec. it would be nice if one out of spec. on the good side of the distribution curve, could be found, the same for backs.
[a href=\"index.php?act=findpost&pid=189610\"][{POST_SNAPBACK}][/a]

I also returned my first Canon 5D within days of receiving it because I saw disturbing banding in the deep shadows in the first test shots I took, even at base ISO. The second copy was noticeably better.

As regards lenses, in the absence of real MTF charts applicable to the individual lenses, I find that all I can do is compare one lens with another. Today I picked up my Canon 50/1.4 which I'd sent in for calibration a few weeks ago. I sent the lens for calibration because one day I found during testing that the lens was marginally less sharp than my cheapest lens, the Canon 50/1.8 II which is less than 1/4th the price of the 50/1.4. It's difficult for me to know if the 50/1.4 is average and the 50/1.8 exceptionally good, or, if the 50/1.8 is average and the 50/1.4 is below par. Both lenses are sharper than my 24-105 zoom at 50mm.

Anyway, I'll be soon testing the 50/1.4 with my 40D using LiveView to get perfect focussing. I might as well shoot the same target with my 5D and TS-E 90/2.8 to get an idea of the magnitude of the differences between a 1Ds3 and ZD. The Mamiya ZD has an AA filter which is user removable, hasn't it?

[Jack Flesher]

Guy and I work together.  He does in fact now shoot with the M8, D300 and ZD back on a 645 AFDII.  He needed some longer-reach lenses for podium work, and file quality was not as important to him as focus speed/accuracy, hence the D300 and a few longer lenses.  He still shoots the M8 for some product and location work, as well as personal artistic projects.  MF digital, even entry level or earlier generation MF digital,  blows away any DSLR for image quality, so he made that move for his product/corporate work.  

The rest of the story:  Both of us have been waiting for Leica to release the R10, assuming slightly larger than full frame sensor and AF, all with quality Leica glass and non-AA's sensor -- IOW mini medium format digital.  However, I started doing the acquisition math and realized I could get a ZD outfit now for less than a similar R10 outfit would probably cost, so I made the move about 2 weeks ago. Took one look at the files and sold almost all of my Canon gear and bought up a bunch of Mamiya glass and a few specialized pieces of Hassy glass.  Guy saw my ZD files while we were teaching our printing workshop in Carmel last week.  Camera West hosted a private reception for our group he second evening, and happened to have a demo ZD outfit.  Guy bought that demo unit instead of the D3 he was planning on picking up, sold some of the overlap M and Nikon glass and is in the process of building up his ZD outfit.  

More of the story with images in the above referenced thread.

Cheers,

[Ignatz_Mouse]

I have a Leica R9/DMR combo. The DMR is a very "special" digital device with its, let's say, own character. The lack of the AA filter is a big advantage in terms of sharpness per pixel over the usual bunch of high end DSLRs. The R9/DMR is a good studio and still subjects camera but not a very versatile  DSLR compared to the Nikons and Canons. In this sense, I think it's closer to the philosophy of a MFDB. The quality of the files at 100, 200 ISO is really gorgeous but it decreases substantially at 400 and 800 (the inclusion of a 1600 Hi ISO option is just testimonial). The DR is also very good as far as I can tell. The Leica glass, needless to say, is incredible. Sometimes you get some strange green or magenta casts and the shadows noise could be an issue when processing the files with ACR:  C1 and Flexcolor offer the best results (althogh I hate Flexcolor usabilty).

I can`t tell how the DMR compares with a 1DsMkII or 1DsMkIII. I've been thinking for some time about getting the 1DsMkIII (I have a lot of Canon L glass) instead of taking the MFDB route (too much money for me) but its "soft" file output is keeping me away from it. You can`t also use the Leica R WA lenses with the 1Ds due to mirror clearance issues.

[Ignatz_Mouse]

As regards lenses, in the absence of real MTF charts applicable to the individual lenses, I find that all I can do is compare one lens with another. Today I picked up my Canon 50/1.4 which I'd sent in for calibration a few weeks ago. I sent the lens for calibration because one day I found during testing that the lens was marginally less sharp than my cheapest lens, the Canon 50/1.8 II which is less than 1/4th the price of the 50/1.4. It's difficult for me to know if the 50/1.4 is average and the 50/1.8 exceptionally good, or, if the 50/1.8 is average and the 50/1.4 is below par. Both lenses are sharper than my 24-105 zoom at 50mm.

Anyway, I'll be soon testing the 50/1.4 with my 40D using LiveView to get perfect focussing. I might as well shoot the same target with my 5D and TS-E 90/2.8 to get an idea of the magnitude of the differences between a 1Ds3 and ZD. The Mamiya ZD has an AA filter which is user removable, hasn't it?
[a href=\"index.php?act=findpost&pid=189701\"][{POST_SNAPBACK}][/a]

I think the Canon 50/1.4 is a very good preformer for its price. I use the 50/1.4 and the 24-70/2.8 L and I still find it better than de latter for a short margin.

The Mamiya ZD back has an AA filter that is user removable. I'm very interested about how it compares against the 1DsMkIII althogh I have mixed feelings about this back and this camera. The actual price of the ZD is really tempting but I prefer to wait and see for the release of the AFDIII and the new digital back. I'm still considering the Rollei Hy6 with a film back (thinking about a future migration to a MFDB).

The rest of the story:  Both of us have been waiting for Leica to release the R10, assuming slightly larger than full frame sensor and AF, all with quality Leica glass and non-AA's sensor -- IOW mini medium format digital.  However, I started doing the acquisition math and realized I could get a ZD outfit now for less than a similar R10 outfit would probably cost, so I made the move about 2 weeks ago. Took one look at the files and sold almost all of my Canon gear and bought up a bunch of Mamiya glass and a few specialized pieces of Hassy glass.  Guy saw my ZD files while we were teaching our printing workshop in Carmel last week.  Camera West hosted a private reception for our group he second evening, and happened to have a demo ZD outfit.  Guy bought that demo unit instead of the D3 he was planning on picking up, sold some of the overlap M and Nikon glass and is in the process of building up his ZD outfit.   


[a href=\"index.php?act=findpost&pid=189705\"][{POST_SNAPBACK}][/a]

This one is the next option. But, how much is going to cost the R10? How will it work with the actual manual focus lenses (in one of his last interviews Andreas Kaufman didn't totally clear out this point)?

[Conner999]

Man this argument just won't go away. It's like the old Nikon (APS-C) vs. Canon (FF) debate that filled the ether prior to the advent of the D3,

SLRs and MF only 'compete' across a relatively narrow spectrum where portability, fps and ISO don't matter. Horses for courses.

No matter what systems you test; with equivalent lenses, the camera with the larger sensor and LARGER (not more, but larger) photocells will outperform the latter in terms of IQ (res, color, DR, smoothness, etc etc).

That said, a 24mm x 36mm 1Ds2 sensor could likely match a 36mm x 36mm CFV sensor, center image, with the same glass  as both have 9um2 cells, but the SLR has an AA filter to overcome, not to mention the different manner in which CMOS and CCD 'draw'.

Then there is the issue that most sensors can deliver more than the vast majority of the same manufacturer's lenses can deliver, so unless you're sticking a premium (no, not Canon's BS "L" as in cough, 'premium', cough, but 'premium' as in exceptional tested performance) lens on the front (which in most cases would be from a third party anyway), of your uber-MP sensor DSLR, you're wasting money on the sensor anyway.

Nikon with the 14-24G is starting to get the picture, Canon (to name one), not so much. Leica and in most cases Zeiss have been there already - in designing (and pricing) for premium rendition from design day 1.

In MF and LF, Rodenstock and Schneider have gotten the message that people are moving from film on their bodies to MFDBs and it's time to cull the lens portfolio accordingly

FF DSLRs are now running into the same issue their APS-C brethern had; there are only so may photosites you can stick on a given mm2 of substrate before you  have to hit the 'reset' button and start looking elsewhere for increased performance.

One can also see the same thing coming in MF - how many cells/mm2 can you fit on a CCD substrate before you hit the wall and have to either increase substrate size ($$$$)or, alter the substrate tech, or do like Hasselblad and build around tighter lens/sensor integration so that end-to-end manufacturing tolerances and/or firmware corrections/compensation can tighten the IQ chain?

Aaargh, time to get another coffee.  




As an aside, the Jan-Feb '08 copy of Photo Techniques (available for download) has an interesting test of a 1Ds2 vs H2+P45 vs H3-39 vs Linhof+P45 that some might find interesting.

[Jack Flesher]

The Mamiya ZD back has an AA filter that is user removable.

To clarify: The ZD back comes with an IR-cut filter that is removable.  You can purchase an OPTIONAL  combo IR-cut/AA filter IF you need AA for your shooting application.  This filter is spring loaded and snaps in and out without tools, so ostensibly you could shoot direct IR capture with this back as well.  I am hoping to test that this week.

This one is the next option. But, how much is going to cost the R10? How will it work with the actual manual focus lenses (in one of his last interviews Andreas Kaufman didn't totally clear out this point)?

Those are the big questions and both unknown at this time. It's why I decided to go with what I did, since I can get it now...

[Chris Livsey]

Guy and I work together.  He does in fact now shoot with the M8, D300 and ZD back on a 645 AFDII.
[a href=\"index.php?act=findpost&pid=189705\"][{POST_SNAPBACK}][/a]

My error for which I offer apologies I had D3 not D300 in my head. My thoughts are running to the D300 for reach reasons.
My elderly (relatively) P20 still has a massive quality edge but the frame rate on the V is more in the frames per min range.

[ErikKaffehr]

Hi,

One of the reasons that the 1.4 is more expensive then the 1.8 is the extra half stop of aperture. There is no reason that an 1:1.4 would be better than an 1:1.8 at medium apertures.

Larger aperture needs bigger lenses which actually also are a constraint on design. Higher price may make it possible to use more expensive glass which may be good for performance.

Everything in life is a compromise...

Best regards
Erik

I also returned my first Canon 5D within days of receiving it because I saw disturbing banding in the deep shadows in the first test shots I took, even at base ISO. The second copy was noticeably better.

As regards lenses, in the absence of real MTF charts applicable to the individual lenses, I find that all I can do is compare one lens with another. Today I picked up my Canon 50/1.4 which I'd sent in for calibration a few weeks ago. I sent the lens for calibration because one day I found during testing that the lens was marginally less sharp than my cheapest lens, the Canon 50/1.8 II which is less than 1/4th the price of the 50/1.4. It's difficult for me to know if the 50/1.4 is average and the 50/1.8 exceptionally good, or, if the 50/1.8 is average and the 50/1.4 is below par. Both lenses are sharper than my 24-105 zoom at 50mm.

Anyway, I'll be soon testing the 50/1.4 with my 40D using LiveView to get perfect focussing. I might as well shoot the same target with my 5D and TS-E 90/2.8 to get an idea of the magnitude of the differences between a 1Ds3 and ZD. The Mamiya ZD has an AA filter which is user removable, hasn't it?
[a href=\"index.php?act=findpost&pid=189701\"][{POST_SNAPBACK}][/a]

[ErikKaffehr]

Hi,

Another question is how much of what is lost in the AA-filter that can be regained with optimal sharpening. AA-filters don't come cheap, so there is certainly some reason that they are used, and that reason is certainly not to save costs.

If you compare formats you need to do that with appropriate post processing which will not be the same for different sensors.

High resolution backs probably outperform the lenses at small (< f:11) apertures.

Best regards
Erik

To clarify: The ZD back comes with an IR-cut filter that is removable.  You can purchase an OPTIONAL  combo IR-cut/AA filter IF you need AA for your shooting application.  This filter is spring loaded and snaps in and out without tools, so ostensibly you could shoot direct IR capture with this back as well.  I am hoping to test that this week.
Those are the big questions and both unknown at this time. It's why I decided to go with what I did, since I can get it now...
[a href=\"index.php?act=findpost&pid=189720\"][{POST_SNAPBACK}][/a]

[Ray]

Hi,

One of the reasons that the 1.4 is more expensive then the 1.8 is the extra half stop of aperture. There is no reason that an 1:1.4 would be better than an 1:1.8 at medium apertures.

Larger aperture needs bigger lenses which actually also are a constraint on design. Higher price may make it possible to use more expensive glass which may be good for performance.

Everything in life is a compromise...

Best regards
Erik
[a href=\"index.php?act=findpost&pid=189763\"][{POST_SNAPBACK}][/a]

It's true that all good lenses tend have similar performance at F8 and I would not expect to see any significant differences between the Canon 50/1.8 and 50/1.4 at apertures from F5.6 to F11.

However, when shallow DoF is required, the smaller format needs to be used at wider apertures. If the lenses that one happens to use for that smaller format all tend to have rather poor performance wide open, then one might tend to draw the false conclusion that the larger format, of similar pixel count, (the DB) is directly responsible for that significantly sharper result including this elusive sense of 3-dimensionality, simply because it's larger.

This principle can be demonstrated very graphically by the following MTF 50 charts I've copied from Photozone. They use an APS-C cropped format camera for testing purposes (the 8mp 350D I believe), so centre resolution is very relevant to the results one would get using a 1Ds3 which has the same pixel pitch as a 350D. However, the border resolution, as shown in magenta on the charts, would be much worse on the 1Ds3.

The really interesting points one can glean from these charts, are:

(1) The 50/1.4 at F1.8 and F2.8 is significantly sharper than the 50/1.8 at F1.8 and F2.8.

(2) The TS-E 90/2.8 (at F2. is significantly sharper than the 50/1.4 at F1.4.

Now, the significant point here is, if I were in a studio using my 5D to shoot glamour pics, using the TS-E 90 at F2.8, I would probably be fairly pleased with the results.

However, if I then switched to the 40D and placed the camera on the same tripod, shooting the same model from approximately the same distance, I would need to use the 50/1.4 at full aperture to get the same DoF.

I predict that the results would be awful with regard to sharpness of eyelashes (although the model might be pleased that her complexion imperfections were suitably disguised).

We have a situation here where, not only is the lens on the 5D significantly sharper at the aperture used, than the equivalent lens on the 40D at the equivalent aperture, but such a lens (the TS-E 90mm) would still give better results on the 5D even if it were not sharper, because the 5D has wider pixel spacing.

I can predict the results of such a comparison. However, I need to actually make the comparison in practice to get a handle on the significance of such differences on prints.

[attachment=6145:attachment]

[Dansk]

How soon things change. 10 years ago medium format held a massive advantage over 35mm and there was not much to compare between them. The fact that we can even COMPARE a DSLR to a MFDB now in my opinion  is simply amazing. The homogenizing effect we are seeing will continue until we no longer call it MFDB or DSLR it will be back to

Hasslebald

Mamiya

Canon

Nikon

etc. systems and thats where it will stop. The format issue being argued here is soon to be moot.

FWIW I think the Canon does a wonderful job vs MFDB's and it really comes down to style and familiarity rather than quality for 99% of the shooters out there that need this much resolve.

We're splitting some very fine hairs here gentleman

[witz]

I agree...

also... I'm a little bothered when the discussion of dslr vs. mfd is left at that.... The 1ds3 has massive improvements in image quality over the 1ds2... so when the argument is not specific to the camera in question it has little significance. Also... many people compare dslr in camera jpegs to mfdb raw processed... again not fair.

if the user treats the 1ds3 like they do a mfdb, and with well chosen glass.... it's a kit hard to beat.

[Jack Flesher]

How soon things change. 10 years ago medium format held a massive advantage over 35mm and there was not much to compare between them. The fact that we can even COMPARE a DSLR to a MFDB now in my opinion  is simply amazing.

No offense, but what everybody has been telling you here is that you CANNOT compare them, there is no contest, the MF backs blow the DSLR's out of the water on resolution, tonality and color.  A 3 or 4 generation old Kodak 16MP DCS back mounted to a Contax 645 that was discontinued three years ago, will outperform a 22MP Canon 1Ds3 at comparable image ratios and ISO 100/200.  I know you don't want to hear that, but everybody that's shot both is telling you it's the case.  Where the DSLR's shine by comparison is where they've always shined, portability and focal length selection and with digital, you can add high ISO availability. End of story.  

Re the AA filter --- YES IT MATTERS!  IMO it is precisely what allows the 16MP Kodak back in my above example to outperform the 22MP Canon 1Ds3.  What you notice when you shoot a camera without one next to a camera withone, is how much more "smeared" the color looks on the AA file. THere is also a relative lack of acutance, though adding some local contrast (clarity) can help, but not replace.  I am equally sure this AA smearing is partly what allows for the higher ISO performance of the DSLR, the rest obviously is built in NR...  FInally, most MF backs allow for turning on NR.  However, if you do, you immediately notice a similar smearing to your color -- and associated loss of clarity -- in the MF file.

Cheers,