Largest print size from medium format ?

<p>I had a 6x4.5 neg printed at 1 meter X 1.5 meters (I guess about 40 x 60 inches) and it looks great. It is soft and grainy close up, but from normal viewing distance it's really nice. IMO you can go much larger than many people say you should. Wall size prints don't need 300 dpi - text on a page needs 300 dpi. You can test yourself at A4 to get a feeling for it.</p>

<p>Jens, I've just come in on this and have no "beef" with anyone here, but honestly what you're claiming is just ridiculous. And that sample you've posted just looks blurry as anything to me.<br>

If we're making silly claims; here's mine. I'm attaching a shot from an 8MP bridge camera (honestly) with about the same area cropped at 100% from the frame. IMHO my crop looks far sharper than yours. Notice the Amnesty International symbol on the young lady's tabard? Let's stop the war right now!</p><div></div>

<p>jens,</p>

<p>I have a question on your resolution article.</p>

 

<blockquote>Fuji Velvia 50 resolves 160 lppm [ line pairs per millimeter ], Fuji Provia 100F Professional as well as Kodak EliteChrome resolve around 140 lppm. This is the finest level of detail it can resolve, at which point its MTF [ Modulation Transfer Function ] almost hits zero.</blockquote>

<p>How are these tests performed? Do they use a high contrast target like 1000:1?</p>

<p> </p>

Let's not go down another track, particularly not one that has a big sign reading "Dead end" in big letters and in plain view at its beginning: high contrast or low contrast does not matter.<br><br>Yes, you lose sight of things sooner when a low contrast is reduced even more than when starting out with a high contrast.<br>But resolution is resolution: the ability to keep two points separate. A matter of point spread size and energy distribution therein.<br><br>And another 'yes': films like the ones mentioned have no problem (in everyday, 'real life' situations) resolving those numbers of lp/mm. There is absolutely no point in trying to contest those. (If you want to have something to niggle about, jump on that "lppm", which has one "p" too many and is one "m" short ;-) ).

<blockquote>And another 'yes': films like the ones mentioned have no problem (in everyday, 'real life' situations) resolving those numbers of lp/mm. There is absolutely no point in trying to contest those.</blockquote>

<p>So a 1000:1 contrast ratio is found in everyday, 'real life' situations?</p>

 

<blockquote>But resolution is resolution: the ability to keep two points separate.</blockquote>

<p>Then why do film manufacturers give lower resolution numbers for lower contrast targets? Isn't resolution resolution?</p>

<p>Films are resolving those numbers through what lenses?</p>

<p>What about the lenses in scanners?</p>

Going down that Dead End anyway? Use the net and inform yourself. Those figures are nothing but extremely realistic.<br>And another Dead End? Yes, lenses resolve even more. Even mediocre thingies have no problem matching or surpassing film's capabilities.<br>And a third Dead End: scanners have no problem resolving detail even finer than both film and taking lenses can (re)produce.<br><br>If you (or anyone else) want to fight a battle, pick one that actually is one. The Big Contentious Issue still is (see this thread) whether sensor technology has advanced to the point it can match the lenses that are put in front of sensors (the answer to that still is "not yet") and the film it is replacing (undecided yet, mainly because it depends a lot on what you are looking at, and for).

<p>Why wont you allow Jens to answer? I directed my original question to him. Why is this a dead end?</p>

<p>He has written several articles addressing this issue so I wanted to ask some questions. Why isn't that allowed? We have a discussion concerning large prints. Why aren't resolution figures part of that discussion?</p>

<p>Can you point to any film resolution tests that show these film resolution figures to be extremely realistic?</p>

<p>Why is asking a question tantamount to fighting a battle? It seems to me you are trying to stifle the discussion. Why is that?</p>

 

<i>"Can you point to any film resolution tests that show these film resolution figures to be extremely realistic?"</i><br><br>Yes. One that proves to be very useful here on PN: "The resolution of photographic film", Zeiss Camera Lens News, March 23, issue 19, page 6.<br>(It used to be available on Zeiss' website, but the publication has been changed into a blog, and the archive no longer contains a working link to that issue. It's too large to post, but i will extract the page and post that later.)<br><br>Asking a question is, of course, not tantamount to fighting a battle. But asking about things that are well known, well established, serves no purpose (hence "Dead End"), except, perhaps, distracting from the real issues (Tactics? Battle?).<br>How long do you suppose have we been using lenses and films that we need to establish what they are capable of, now? ;-)<br>But when it comes to whether or not digital has caught up, the issue, without fail always turns to what film can do. As if we don't know that... and as if saying that one thing isn't that hot would make another thing, that isn't better, look good.<br>But we know what films and lenses can do. What can digital do?<br><br>And before we get entrenched in another Dead End battle: no, i have nothing against digital, am not saying anything about the relative merits of the two media. Just commenting on The Usual Way the issue gets obscured and bogged down by the Seemingly Inevitable Dead End Questions.<br>;-)<br><br>Now, who is not allowing Jens to answer anything???

And as promised: <a href="http://home.tiscali.nl/qnu/PN/CLN19_en%206.pdf" target="_blank">Zeiss Camera Lens News, 2003, 19, p. 6</a>

<p>Marc, the film fanatics can't produce any detailed photos to back up their claims, so they keep referring to those meaningless numbers. Be wary of anyone defending Jens' article or maths - they clearly live in a land where their film is sprinkled with pixie dust to give it a magical 175 megapixel resolution and they can't be reasoned with :P<br /> Kodak themselves say on their website that "resolution level (2048 x 3072 pixels) captures all the image data 35 mm film has to offer". Ouch, that's 6.3MP. http://www.kodak.com/digitalImages/samples/fiveResolutions.shtml<br /> Here is a list of tests including real world image tests of film v. digital. They seem to agree with my original assessment that 35mm maxes out in the vicinity of 10MP. By the way, I used to work at one of the largest repro houses in Melbourne in the film heyday and we would never scan a 35mm frame at more than 4000 pixels long (10.6 megapixels), because there was no more useful image information after that point. That was on a top Heidelberg drum scanner.<br /> Ok, now to some links with real world results:<br /> Here the author concludes that 35mm film is beaten by an 11 megapixel digital camera: http://www.grafikogfoto.dk/photographical/canon_1ds_prints.html<br /> Here the tests show that 35mm Provia gives very similar results to a 6MP digital camera: http://www.sphoto.com/techinfo/ocesideharbor.htm<br /> Same conclusion here: http://www.sphoto.com/techinfo/santaysablefarm.htm<br /> Here the author finds that an 11 megapixel camera beats 67 film: http://www.luminous-landscape.com/reviews/shootout.shtml<br /> Here the author finds that 39 megapixels almost catches up with 4x5 film: http://www.luminous-landscape.com/essays/Cramer.shtml<br /> Shutterbug magazine finds that 6MP digital is approximately equal to 645 film: http://www.shutterbug.com/content/digital-files-vs-scanned-medium-format-filmbrwe-put-mediums-test<br /> Here are some test results of a variety of film and digital cameras: http://www.luminous-landscape.com/essays/back-testing.shtml<br /> There are hundreds of tests like this out there, all with approximately the same results.</p>

<i>"[...] the issue, without fail always turns to what film can do [...] as if saying that one thing isn't that hot would make another thing, that isn't better, look good."</i><br><br><i>"Marc, the film fanatics can't produce any detailed photos to back up their claims, so they keep referring to those meaningless numbers. Be wary of anyone defending Jens' article or maths - they clearly live in a land where their film is sprinkled with pixie dust to give it a magical [...]" </i><br><br>See?<br><br>How very boring. Every time again, the same inane thing...<br>Luckily, most people know what they get from whatever medium they use, and use them precisely because they do know what they get from them, be it film, digital or both.

<blockquote>

<p>Marc, the film fanatics can't produce any detailed photos to back up their claims, so they keep referring to those meaningless numbers. Be wary of anyone defending Jens' article</p>

</blockquote>

<p>What about me, a 100% film user, defending your 80MP image and reversing my opinion of it having been able to look at an actual file?</p>

<p>There are film fanatics who use huge pieces of film - up to 20" x 24". These have more detail than anyone will ever need or ever get from a digital sensor but it's not a competition is it?</p>

<p> </p>

<p>Steve, you're right - it's not a competition. I am just trying to correct some false information posted earlier in the thread which several people are still trying to defend. I actually bought 4 rolls of 120 last week and am looking forward to shooting B&W for the first time in years!</p>

<p>I'm not sure where these silly figures for film resolution are coming from, but they certainly don't tie in with my 45+ years experience of using film.<br /> Someone asked about the methodology used to get these figures, and I can partly answer that: The measurement of resolution figures for film doesn't involve use of a lens. They're either contact printed, directly projected onto the film with a moving slit and lamp arrangement or more likely these days scanned with a laser. However they're produced, they're totally divorced from any means commonly used in real life to capture an image by using a camera.<br>

<br /> I'll attach an extract from a PDF by Zeiss - the company, not some Fanzine using Zeiss in its title.<br /> As a little background; the MTF curve of film or a digital sensor needs to be multiplied together with that of the lens used to produce a combined system MTF. So, for example, if the two MTF curves hit 30% contrast at the same resolution figure, then the combined contrast would be 30% of 30%, giving 9%. This would take the combined MTF below the accepted perceptual threshold of 10% contrast, and would therefore be seen as the cutoff point of resolution.</p>

<p>The MTF curves below show the combination of a very good lens working close to the diffraction limit at f/5.6 and what Zeiss considers to be a typical colour film. You can see that the 10% limit is reached at around 80 line pairs per millimetre, which relates closely to my own experience. In pixel terms this would be equivalent to just under 22Mp on a 24 x 36mm frame - a long way short of 175 Mp!</p>

<p>Remember this data originated with Zeiss, so if anyone wants to argue with it, argue with the big Z.</p>

<p> </p><div></div>

<p>Thanks for posting that Joe. I would argue that 10% MTF is too low. 30% seems more reasonable, which happens at 50 lp/mm, or an equivalency of 8.6 megapixels. Of course you can argue about where to draw the line and there is no correct answer, and the result will often be lower anyway due to the lens or film not being up to the high standard shown here.<br>

By the way, your maths wasn't correct. The interaction of MFT works like this - the combined MFT is 1/((1/a)+(1/b)) where a and b are the two individual MFTs. Example: at 60 Lp/mm, the above film has an MFT of 32 and the lens, 70. The combined MFT is 1/((1/32)+(1/70)) = 22%, which is the same result shown by the graph.<br>

Your example of two MTFs of 30% combining would give an MTF of 15%, not 9%.</p>

<p>In reality, the figures for resolution of sensors or scanners or quality of lenses and film and pictures of test charts mean anything.</p>

<p>To me, all that is important is the quality of a print. If it looks good, it is good. </p>

<p>I meant: '<strong>don't</strong> mean anything'.</p>

R.J.<br><br>If that publication you were calling a fanzine is Camera Lens News: you should know that it is the publication produced by the very same people from the very same Zeiss Camera Lens Division you give thanks for providing the graph. ;-)<br><br>A more serious point though: yes, you can measure film resolution using film only, i.e. without lens. But doing it the way the Camera Lens Division did obviated the theoretical attempts to answer the question what that would mean when we put that film in a camera, put a ens on that camera, and expose the film as we normally would.<br>Theory can be fun, but you do have to know when it is helping us to get closer to knowing 'what's up', and when it does the exact opposite.<br><br>So though you say you can <i>"partly answer"</i> what <i>"the methodology used to get these figures"</i> was, you demonstrate that can't. Notice how the people from Zeiss' Camera Lens Division themselves explain how they arrived at those <i>"silly figures"</i>, while you put your trust in some theoretical considerations instead! ;-)<br><br>And finally: the length of my experience with film easily match yours, and i can confidently say there's nothing silly about those figures.<br>So much for that sort of "evidence". ;-)

<p>Since we're all having so much fun, and nobody's posted an image from a D3x or M9 yet, I just thought I'd chime in with a quick look at the images Q.G. posted.<br />

<br />

The larger (presumably full-frame) image is 450x294.<br />

The inset rectangle image is 450x300.<br />

The inside of the red rectangle in the larger frame (and I'm going with the <i>inside</i> to give the maximum estimated resolution) is 17x12 pixels. Obviously the rounding isn't quite right for the aspect ratio, but we'll go with it.<br />

Pixel peeping at the inset image, I can't see an edge that transitions in fewer than four pixels. That's being extremely generous in my opinion, but I'm sure it can be debated.<br />

<br />

Blown up to the resolution of the inset image, the whole image would be something like 450x450/17 by 294x300/12, or 11912x7350, or 87.6MP. That would tally with a 2x enlarged 4000ppi scan of 37.8mm by 23.3mm, which is a reasonable approximation to full frame.<br />

<br />

Now, I claim that the linear resolution of the inset image, given that each transtion seems to take four pixels, is no more than a quarter of this. In other words, actual detail is something like 5.47 megapixels.<br />

<br />

Now, you can argue that a digital camera with a low pass filter and a bayer sensor can't resolve its nominal pixel count, and that to beat this definitively we need either a camera with four times the resolution (such as a D3x, or - more or less - a 5D2), no filter (such as an M9), or no bayer sensor (SD1). And you can argue whether the lenses are up to it anyway, although it helps that the thickness of the film isn't an issue for a digital sensor. The film grain is also much more visible than any digital noise is likely to be at similar ISO.<br />

<br />

I love film. I shoot film - and digital. But, while I'm still waiting for someone with a high-end digital camera to make the point, I thought it was worth pointing out that the apparent resolution of the example film image isn't all that great. All I can show off is a D700 image (a 161x116 crop scaled up to 450x300, which is the same proportion of the total frame shown in the red box) - this is unsharpened, JPEG from the camera, and simply an example I could find easily, remembering that the D700 is known for having a strong low-pass filter and only has 12MP. Still, I think it's comparable to the film image.</p><div></div>

<p>For reference, here's the source image.</p><div></div>

<i>"[...] a quick look at the images Q.G. posted."</i><br><br>Just for the record: i haven't posted images.

<p>Oops - terribly sorry, Q.G. I got confused while multi-tasking. The images I referred to were, of course, Jens's.<br />

<br />

Of course, I should also say that the film image is a much better example, and also a much nicer photo (especially since the latter was a documentary image that I just pulled out as the first thing I could find that was ISO 200 and shot with a macro lens). But that's not the camera's fault. :-) I'd say the differences are small (looking at the scratches on the silver on mine, and the detail in the painting on the scan); the film is possibly slightly sharper, but the grain is pretty distracting at this scale.<br />

<br />

Of course, there's a lot more resolution and (relatively) finer grain to be had from medium format film - or 5x4 - and I'm in no position to go shopping for a medium format digital back. So I won't be stopping shooting film any time soon, but nor do I expect my 35mm slides to give a significant improvement in resolution over my DSLR.</p>

<p>If you take the trouble to go to Jens' 3rd link, he shows and gives more details on the same 35mm image and same 200% crop. The important point is that he states that his full 35mm scan is "21 MegaPixel". Now that's way more reasonable than the 87 megapixels scanning that he claims is necessary for 35mm in his 2nd link. And it agrees with the "just under 22Mp" that Rodeo Joe related to us from his extract of Zeiss "best case 35mm" work.</p>

<p>But the ironic thing is that Jens' image doesn't even come up to "best case 35mm" performance. When I copied the 200% crop from his 21 megapixels scan into my image viewer and examined it up close, I saw that it is clearly oversampled. The sharpest linear features and edges (such as the arch supports in the castle "picture within a picture") have transitions which are 4 pixels wide (the arch supports are double-edged so they are 8-9 pixels wide). I resized the image by 50%, flipped back and forth between the original and 50% version, and all of the same detail was still present, albeit more pixellated of course. I then resized the 50% version by 2x, back up to 100%, the same sampling as the original. This resizing necessitated interpolation of course. The original and resized-from-half-sampling version are stunningly similar - which verifies my original impression that it was oversampled. I will acknowledge that the original was a tiny bit sharper in some of the arch supports. So it's not fully 2x oversampled; I'll be generous and estimate 1.5x oversampled (where 1x means correctly sampled). 1.5x oversampled in both axes means 2.25x too many pixels in area. And 21 megapixels divided by 2.25 is...damn close to the 10 megapixels which Jens was dismissing out of hand!</p>

<p>I will say one more thing - Jens' treatment of Graham in this thread was absolutely appalling. To call him a "Dishonest liar" (he did so in bold font just to make sure that no-one missed it!) shocked me to the core. In my country, that would mean a very large libel payout!<br>

http://indigo.ie/~kwood/defamation.htm<br>

"An actionable defamatory statement has three ingredients:</p>

<ul>

<li>it must be published, </li>

<li>it must refer to the complainant and </li>

<li>it must be false."</li>

</ul>

<p>Jens' outburst meets all three criteria.</p>

<p>And it is certainly false. I knew that Graham was being truthful and accurate all along - I am familiar with his professional work and his tools. Troll cartoon aside, he showed more patience than his antagonists deserved, and did the right thing - he came back with more and more evidence to back up his position.</p>

<p>This being the lawless internet, I have a feeling that Graham won't be looking for trans-national libel payouts: but a public apology from Jens would be in order.</p>

<p> </p>

<p>Ray, thanks but I'm not bothered by his accusations when everyone can judge his behaviour and competency for themselves. His 'scanning back' response was an instant fail classic.<br>

By the way, Jens' blog article which he linked into this thread gets worse - if you keep reading, he asserts that "we would need a digital camera of 87 x 2 = 175 MP to see every last detail compared to film". He and Q.G. are still sticking to this story too <img src="http://forums.rennlist.com/rennforums/graemlins/roflmao.gif" alt="" width="39" height="15" /></p>