Medium format vs digital

[jason_hall5]

<blockquote>

<p>...I am still waiting for the response to the blond vs. brunette question <grin> above</p>

</blockquote>

<p>as for me....brunettes are more fun. :o)</p>

<p>Jason</p>

[leonard_malkin]

<p>To Daniel. Interesting, your use of the 7D for landscapes. I recently posted to dpreview and asked whether a 7D at ISO 100 and on a tripod would give landscape pics equivalent to a 5DII. As usual, I got answers all over the place. The conclusion was it could and couldn't and then the discussion drifted off to a discussion re DOF and other things. I guess I'll just try the 7D and see whether I like it. I might rent a 5DII to compare. I wonder if I could rent a MF somewhere?</p>

[leonard_malkin]

<p>To Eric. Your FF Nikon D700 has only 12 MP. With a Canon 5DII FF, you'd be working with 21 MP reducing the need to up-rez. Of course, there's also more noise to go along with the increased MP's.</p>

[stp]

<p>Leonard, for the last few weeks I've been preparing files for my new website. These photos came primarily from a Canon 1DsMk3 and Pentax 645, with a few from a Canon 5D Mk1, Pentax 67, and Mamiya 7II. The film has been scanned on my Nikon LS8000 scanner. As I prepared these files, it really drives home the point that the digital files have to be interpolated to reach a 16"x24" print, while I usually have more pixels than I need with scanned film. There are many other variables that will affect the final print quality, but the size of the file from the two different sources is a significant difference. I will be able to offer the largest prints only from the photos that reside on film. I sometimes envy those whose web offerings are based on 4x5 film; I tried that for a couple of years, but I just didn't enjoy the workflow. I think one has to find the right balance between shooting style, workflow, and print size, and that balance is a very personal decision. Yes, the question has been asked many times, but there is an answer for everyone who asks it. While you will hopefully get some helpful advice here, your own experiments and experiences will be most important.</p>

[andylynn]

<p>What's this x2.58 factor about? A 7D is 5184 pixels in the long dimension, and that's "pixels" in Bayer sensor speak where you get to count a pixel in one channel as a whole pixel - but taking that at face value, print at 40" in the long dimension and you're at 130 PPI. Take a 6x7 negative and scan it on the cheapest thing you can get, an Epson V500 for example, and use the low estimate for its true PPI without a special film holder - which is 1600 - and it comes out to 110 PPI on a 40" print. Not a lot of difference and this is the cheapest hardware you can get. Use a 4000 PPI Coolscan and you're at 275 PPI on a 40" print.</p>

<p>To review:<br>

Printing at 40":<br>

Canon 7D: 130 PPI<br>

6x7 with very cheap scanner: 110 PPI<br>

6x7 with good scanner: 275 PPI</p>

<p>If you look at <a href="../film-and-processing-forum/00WErk">this thread</a> you can see a comparison. It was done with a 40D and not a 7D, and the 7D is better but not so much better that it catches up to the medium format. Suppose a 5DII were twice as sharp - still not at the level of the scans. Mauro was using the same camera and lens combo I have and I can vouch for it also beating the snot out of Nikon DX cameras.</p>

[Ed_Ingold]

<p>Andrew,</p>

<p>Not all pixels are equal. In terms of acuity (e.g., at MTF = 50), one DSLR pixel is worth about 3 from scanned film.</p>

[danielleetaylor]

 

<p><em>What's this x2.58 factor about?</em></p>

<p>In the samples I uploaded, the 7D crop has been enlarged 2.58x to match the MF crop in size. If you want to judge how details will look at a given print size, you must follow the formula I laid out to get the picture of the two crops to print at the correct size.</p>

<p>For example, if you printed at 557 ppi, the two crops would appear as if they had been cut from a 16x24 print of the map. (5184 x 2.58) / 24 = 557.</p>

<p><em>If you look at <a rel="nofollow" href="../film-and-processing-forum/00WErk">this thread</a> you can see a comparison. It was done with a 40D and not a 7D, and the 7D is better but not so much better that it catches up to the medium format.</em></p>

<p>I made a number of posts to that thread. The 7D renders significantly more fine detail than a 40D, despite Mauro's insistence to the contrary. He has not tested a 7D. If you compare my 7D sample to the 40D sample at Les Sarile's site, you can see a very significant difference in sharpness and fine detail.</p>

<p>While a single 7D frame does not match a good MF scan, it is pretty darn close for some print sizes. That's the point, that you need to decide what your final goal is. Do you never print larger than 13x19 or 16x24? Then there's just not a real difference between a top tier DSLR and a MF scan.</p>

<p>When I print RVP Howtek Scan vs Canon 7D Upscaled, Version 2 at 557 ppi (16x24 equivalent), I cannot see a difference in fine detail or sharpness. The film scan has more noise and that's apparent even at this size, but otherwise they're pretty much dead even. Go ahead and try it.</p>

<p>There will be some variance based on subject matter, but that tells me that I can pretty reliably make 24" prints from my 7D that would be indistinguishable to the average viewer from 24" prints made from MF. In larger prints the differences become apparent, and when I know I will need or want to print larger than 24-30" I stitch 3 frames to get the same absolute quality as the best MF scans.</p>

<p><em>Mauro was using the same camera and lens combo I have and I can vouch for it also beating the snot out of Nikon DX cameras.</em></p>

<p>The OP asked about the 5D mkII. A 7D is pretty darn close to a 5D mkII. Amateur Photographer did a full frame vs APS-C article. Their conclusion was that at low ISO it was not possible to distinguish between large prints from the 5D mkII and 7D. On the other hand, they compared the D300s and the D3s and even though the MP resolution was identical, in the samples they presented the D3s recorded more fine detail and had better sharpness. So you can't judge how a 7D or 5D mkII will perform based on a Nikon DX, or a Canon 40D, or an old 5D, or anything else. You have to actually test the sensor in question.</p>

 

[danielleetaylor]

<p><em>To Daniel. Interesting, your use of the 7D for landscapes. I recently posted to dpreview and asked whether a 7D at ISO 100 and on a tripod would give landscape pics equivalent to a 5DII. As usual, I got answers all over the place.</em></p>

<p>Amateur Photographer March 6, 2010, has a review of full frame vs APS-C. In the case of the 5D mkII and 7D their conclusion was that, all other factors being equal, it was not possible to distinguish between large prints made with them. So the sensor is certainly up to the task.</p>

<p>I see the same thing in the Imaging Resource samples and in my own tests. (I do not own a 5D mkII, but have had access to one.) The 5D mkII does not start to pull away until higher ISOs, where it presents a significant noise and fine detail advantage. At low to mid ISO, it's a wash. Out of camera 5D mkII images will be a bit sharper, but that's easily compensated for in PS or using in camera settings. (Note that with the 7D RAW yields more fine detail than JPEG, and ACR yields more fine detail than DPP.)</p>

<p>Another thing in the 7D's favor is the optically excellent Tokina 11-16 f/2.8. I'm quite frankly shocked that the Canon 10-22 is often recommended on forums when the Tokina is, literally, one of the very best UWA lenses made, even when put against the 16-35L II or the Nikon 14-24. When I was deciding what lens to shoot the map test with I looked up the peak apertures/resolutions of all my glass at photozone.de. Imagine my surprise when I discovered that the Tokina UWA had the highest center resolution of anything, including my primes and L telephotos! I ended up using a 50 f/1.8 @ f/5.6, but the Tokina is razor sharp, corner to corner. It will edge out the 16-35L II on full frame, and will easily out perform a 17-40L on full frame.</p>

<p>I consider the 7D to be one of the most well rounded bodies on the market today. It's a top notch sports camera, a top notch landscape camera, has wireless flash control for the studio, and while it's not as good in low light as FF, it's no slouch either. I have gotten noise free 8x10 prints shooting at ISO 2000 and even 3200 with the 7D.</p>

<p>If you can, rent both and see which you like better. Your lenses might fit better on full frame, or you might prefer the "feel" of full frame (i.e. you might be use to thinking of focal lengths in those terms). Or you might just need the excellent low light capabilities. But give some consideration to the 7D. It's a brilliant piece of engineering.</p>

[andylynn]

<p>The part about also beating a Nikon was an aside. Of course I know that different cameras produce different results. Anyway, the 7D is a great camera but we've come to the same conclusion on this - a good scan from a well made large neg results in more sharpness and fine detail than even a great DSLR, which is what the OP was asking about.</p>

[eric friedemann]

<p>Leonard, yes, there are a number of variables when considering Canon/Nikon FF v. MF film. A 5DMII will yield somewhat more resolution at 21 Megapixels than the D700 at 12 Megapixels, though not as much more resolution as the 12 v. 21 Mp. would suggest.</p>

<p>I've made 12 x 18 prints from my D700 and, working in a camera store, have seen a number of similarly-sized prints made from images shot with the newer Canon high-Megapixel FF DSLRs. To me, images at that size from FF digital cameras look a little ... digital. The images are on- or past- the edge of what I would consider ideal resolution. And this impacts an image's apparent sharpness.</p>

<p>Opening up an uncompressed RAW image from a D700, I get, I think, a 50 Megabyte file. When I scan a 6x7 film, I get a file of 250 Megabytes or greater. Now, it could be pointed out that, when shooting film, you lose a generation of sharpness when scanning, which isn't the case with digital. Still, I find that for larger-than-8x10 prints, I get better-looking results from big files scanned from MF films.</p>

[stemked]

<p>Almost a decade ago I bought most of my used gear at one of the various National Camera shops in (mostly near) Minneapolis. They are a good shop.</p>

[leonard_malkin]

<p>Thanks to all for your responses.<br>

I plan first of all to get a 7D once I sell my 30D. I think then I'll rent a 5DII and do some comparison landscapes. It would be nice if I would be satisfied with the 7D - it'll save me about $2500.<br>

I still would like to try MF and I'm looking for someplace where I can rent. I plan to shoot some rolls and have a lab do the scanning and printing, just to see what's what.</p>

[andylynn]

<p>In Boston, and some other locations, Calumet will rent you a Hasselblad 6x6 or Mamiya 645 or 6x7 kit or a Fuji 617 - I keep trying to find and excuse to rent a 617 :) Anybody in the Boston area need somebody to take some huge panos? :)</p>

<p>I think the 7D is going to show you improvement in a lot of areas over the 30D. As far as print quality, depends how big you print...</p>

[andylynn]

<p>Ah, Les, but you didn't include a Canon 7D in your comparison, and as we know Canon 7D's are magic :) Also you can use them to create stitched images, which is a unique feature of the 7D and is entirely practical in all situations :)</p>

<p>Which cat is which? I have a guess but you didn't say.</p>

[philippartridge]

<p>Always an interesting issue, no less in 2010 than in 2002.<br>

For highest quality landscapes with optimal technique that are within scope for low ISO medium format cameras (focal length lens, dynamic range of scene), output from Astia 100F or E100 is hard for any DSLR to match. Of course that still leaves a lot of ground for top-end DSLRs. The Nyquist limit for today's DSLRs is around that of colour negative film, and very very far behind black and white film.<br>

To claim the 7D is as good as any large megapixel DSLR is ludicrous. Higher Mp FF DSLRs have higher IQ (MTF), simple as that, and of course they also have better DR and are much more forgiving of less than perfect lenses because final IQ is the product of lens MTF and sensor MTF.<br>

21Mp versus 12 Mp is around a 35% increase in resolution, Eric. Well worth having, I think.</p>

<p>As is pointed out above, however, even 12Mp DSLRs with excellent lenses and with careful sharpening and file preparation can rival medium format film in some cases for MTF, but in best case preparation for both, MF is still ahead...and you get the luscious look of film, no CA, depth of image and a lot of pixels - all handy advantages. </p>

<p>To inject some science into the issue from an impeccable source, consider reading the recent Zeiss papers:<br>

<a href="http://www.zeiss.de/C12567A8003B8B6F/EmbedTitelIntern/CLN_30_MTF_en/$File/CLN_MTF_Kurven_EN.pdf">http://www.zeiss.de/C12567A8003B8B6F/EmbedTitelIntern/CLN_30_MTF_en/$File/CLN_MTF_Kurven_EN.pdf</a><br>

<a href="http://www.zeiss.de/C12567A8003B8B6F/EmbedTitelIntern/CLN_31_MTF_en/$File/CLN_MTF_Kurven_2_en.pdf">http://www.zeiss.de/C12567A8003B8B6F/EmbedTitelIntern/CLN_31_MTF_en/$File/CLN_MTF_Kurven_2_en.pdf</a></p>

<p> </p>

[kelly_flanigan1]

<p>Here is a 35 megapixel LF direct capture from my 1996 4x5 scan back for fun:</p>

<p><img src="http://i4.photobucket.com/albums/y148/ektar/150mmRonar/Ronar150mmFull.jpg" alt="" width="600" height="417" /></p>

<p><img src="http://i4.photobucket.com/albums/y148/ektar/150mmRonar/Ronar150mmWWVCLOCK.jpg" alt="" /></p>

<p><img src="http://i4.photobucket.com/albums/y148/ektar/150mmRonar/Ronar150mmStones.jpg" alt="" /></p>

<p><img src="http://i4.photobucket.com/albums/y148/ektar/150mmRonar/Ronar150mmDOG.jpg" alt="" /></p>

[steve_t.1]

<p>National Camera Exchange- Minneapolis/St. Paul, MN. Yes, I can vouch for them. I have left plenty of money in their till over the past several years, from film to P&S to digital SLR to a few used lenses for my Pentax DSLR to camera bags and accessories...</p>

<p>Their medium/large format and other non-mainstream stuff is centralized from their main store in Golden Valley, MN. (Suburb next to Mpls to the west.) Their "Shutterbug" department (pro stuff) handles it. Even from a distance, phone conversations to their Shutterbug folks will net you a good experience, I'd bet. And yes, as others have said, inquire about the return policy. The few used 35mm camera lenses I've bought all had 30-day satisfaction guarantees with them, and the $100 I just spent there tonight on a new Lowepro slingshot bag came with the same 30-day test drive. Yeah, I can find new gear cheaper online many times, but I do spend a fair share in their stores, too. Their site for used gear- http://www.natcam.com/used-equipment.html <br /><br /></p>

[andylynn]

<p>It's one image file... can't use exif... but from the whiskers clue I see I was right :)</p>

[BernardMiller]

<p>There's all this talk about stitching images from DSLRs together, and yet I haven't heard one word about stitching scans from medium format film. And yet it certainly can be done, too! Apparently, Annie Leibovitz used the technique all the time with her RZ67, including this image, which doesn't seem a candidate for that sort of thing at all:<br>

http://espacoembranco.files.wordpress.com/2009/01/bush_crew_vf_leibovitz.jpg?w=493&h=344<br>

It would take a bit of work, no doubt, but can you imagine the landscape panos you could achieve with two or three 6x7 scans stitched together?</p>

[danielleetaylor]

<p><em>How about we go all the way to the current king of DSLR resolution the >24MP Sony A900 set at ISO 400 vs Fuji Sensia ISO 400 35mm film scanned with a desktop scanner with no pre or post processing applied except for the superimposition of both images into one...<strong>Since the top of the line DSLR does not outresolve 35mm film,</strong></em></p>

<p>I have to say Les, I am surprised whenever you post the cat comparison. It is not in any way a valid test and no conclusions can be drawn from it. Aside from the fact that neither the subject nor the shooting conditions are the same, we have no knowledge of the shooting technique. Camera support, lenses used, apertures, shutter speeds, focusing methods, file processing, all unknowns. What can be drawn from this? Nothing.</p>

<p>Given the performance of the Canon 7D at ISO 400 against Sensia 400 in your map test, I'm quite confident that the Sony A900, properly handled, would out perform Sensia 400 by a wide margin. Note that I underexposed the ISO 400 test and had to lift exposure in ACR by a half stop. While some noise is evident at this scale, it would not show in print.</p>

<p>For anyone who is interested, here is my Canon 7D ISO 100 RAW file along with an unprocessed and a processed crop of the area commonly compared at your site. There's also a Read Me covering some mistakes I made while shooting, and the processing performed. (The mistakes shouldn't affect center resolution, though the right side is blurred.)</p>

<p>https://docs.google.com/leaf?id=0Bz1sfHfXHVDCYTgzYmZhMGMtYmVkOS00MTI1LWI5ZWEtYzg5ZGIzNzEzZDNm&hl=en</p>

<div></div>

[danielleetaylor]

<p><em>Ah, Les, but you didn't include a Canon 7D in your comparison, and as we know Canon 7D's are magic :)</em></p>

<p>Andrew, I'm curious if you have printed the map crops at various resolutions to simulate different print sizes, and at which print size you feel the resolution advantage of MF becomes clear? Thanks!</p>

[danielleetaylor]

<p><em>To claim the 7D is as good as any large megapixel DSLR is ludicrous.</em></p>

<p>I believe I said it was fairly even with the 5D mkII at low to mid ISO, and representative of what top tier DSLRs can accomplish. Obviously the 5D mkII will show a small resolution advantage, and the 25 MP models will show some more.</p>

<p>If you can point me to tests which show my claims to be ludicrous, I will change my opinion and stop making them. Again I will note that Amateur Photographer came to the same conclusion in their well researched article. Can you reference articles which prove this conclusion wrong? If they exist I certainly would like to see them. Thanks!</p>

<p>Also, you imply in your post that CA is a function of digital sensors. It is not. It is a function of the lens in use. One digital advantage is the ease with which CA is eliminated completely when converting from RAW.</p>

[jussi_vakkala1]

<p>I have found a couple of shortcomings in many tests I read at net. First there is no detailed documentation of test conditions or post processing. Second a sole tester with an obvious bias (and sometimes also skills) to one technology, either digital or film, is invariably test after test finding his/her favourite approach better in almost every aspect.</p>

<p>To overcome these problems I arranged a test together with one photographer on opposite viewpoint of mine: he is a film aficionado while I have been using and favoring digital now for quite a many years. Together we set up a test plan for 35mm digital vs. film test (b&w, film is Kodak TMax 100, camera is 5DMKII at ISO100) with detailed rules on execution, post processing and documentation. We have gone now through test plan and actual shooting. Film will go into drum scanner and we will also print large, around 30"x20", to compare actual results. Test scenario is quite simple and easy to replicate, but it requires two photographers with known bias to his/her part of technology in test to maximize results on both sides + a few years experience in photography.</p>

<p>Test scenario consists of simultaneous shooting with 50mm prime lens on tripod at f8 in ca. 8-10m distances to minimize irrelevant technical flaws for this test, like lens aberrations, air movements, diffraction or misfocusing. The route to print for each part is free of choose, including analogic devices like enlargerers too. But only basic manipulations like usm and curves tweaking are allowed, while combining multiple exposure are certainly not. The purpose of test is to find out what an amateur behind each camera and capturing method can get out to print in _realistic_ context of everyday shooting, but not to max out any theoretical limits we can only talk about. Intermediate files of process (or in case of bandwith restrictions, part of them) are available for readers' own experiments too.</p>

<p>So far test plan is only in Finnish, but I'll try to translate the coming test report in English too. Stay tuned!<br>

<img src="http://www.jussivakkala.com/fd/testi.jpg" alt="" width="600" height="400" /><a href="http://www.jussivakkala.com/fd/testi.jpg"></a></p>

[danielleetaylor]

<p>Jussi - I look forward to the results of your test. My only complaint would be that it's not reproducible by other photographers, as that requires a commonly available test target. But I applaud your efforts to work with someone with a different opinion to come up with an agreeable set of rules.</p>

<p>The 5D mkII has a higher Nyquist limit than T-MAX's rated resolution at normal contrast. (78 lpmm vs. 63 lpmm.) But then contrast is a big wrinkle in any test as film's ability to resolve detail varies greatly with detail contrast. You could actually end up with a shot where film resolves more detail in one spot, and digital in the other, not because of lens variation but because of contrast. (I have seen this with real world comparisons.)</p>

<p>A couple things to consider...</p>

<p>* RAW resolves more than JPEG. ACR, to the best of my knowledge and testing, resolves more than DPP. When comparing to the drum scan the method of enlargement and subsequent application of USM will be critical. (Right now I'm experimenting to find the best combination of steps for scaling.)</p>

<p>* Developer and development time/technique will affect resolution, sharpness, and grain for the T-MAX, and should be documented.</p>

<p>Good luck, and let us know when the test is published and translated.</p>

[jussi_vakkala1]

<blockquote>

<p>My only complaint would be that it's not reproducible by other photographers, as that requires a commonly available test target.</p>

</blockquote>

<p>I cannot transfer light and temperature conditions either. Simultaneous shooting in natural light is in our opinion the best method for our test goal. 'Test target' in our plan is any subject with enough details and some variation with textures. In our case it subject consisted wooden structures, cardboard test targets, grass, branches etc.</p>

<blockquote>

 

</blockquote>

<p>About documentation: developers and such details will be documented by test plan. Neither participants of test cannot claim any 'if only' excuses, since both participants try to get the most out of their cameras. If some particular method is not available for a tester in our plan, it is then also practically irrelevant method for the purpose of test.</p>