Jump to content

Best Scanner for Medium/Large Format Film?


timlayton

Recommended Posts

<p>You're kidding yourself if you're judging the quality of a scanner based on 500X800 pixel images. That's 0.4 megapixels. No matter how good those 0.4 megapixels look (and they look "okay" in that they could be large format, could be a decades-old point and shoot at that resolution) they aren't proving anything. That's a 1x2 inch print at 400dpi! Then again, if 500x800 pixels is enough for your images then don't second-guess your scanner: it's certainly good enough.</p>

<p>One thing people forget is that most "pros" shooting medium format were printing optically, too. Even good cibachromes look way, way softer (though also more saturated) than digital prints do. So a digital print from a full frame dSLR is almost certainly as good as (and significantly more flexible than) a cibachrome print from roll film. If a medium format cibachrome is your definition of "good enough," then digital is way cheaper, way easier to control, and probably visually even better... A drum scan or Nikon scan of 6x7 might be slightly better in some ways than a full frame dSLR (it might not be) but it's more expensive and turnaround is slower. Full frame digital looks pretty great.</p>

<p>As for the Sony whitepaper, I haven't read it, but a good measure of sharpness is the Heybacher Integral, the area under the mtf curve. Digital is sharper until extinction, but film resolves finer details (not really: 25mp digital beats velvia in all respects, but for large format and medium format film has the edge due to sensor size), thus digital seems sharper than its extinction limits indicate.</p>

Link to comment
Share on other sites

  • Replies 111
  • Created
  • Last Reply

Top Posters In This Topic

<p>Daniel Lee Taylor - I do not think the first crop has a vibration issue (the same tripod placement and set up was used for both shots and the Fuji Gx680 is a camera that weighs over 10lbs and it's shot is sharp) - it is just an extreme enlargement from towards the edge of the 16-35 F2.8 II at 35mm and F2.8 where it performs at its worst. The issue is merely the size of the enlargement. On my 27 inch iMAC the crop is 10 inches wide by 5.7 inches high. Thus I am looking at an edge crop from a shot that would be 10 feet six inches wide by just over 7 feet high. This link from the Digital Picture test shots shows similar issues at a much lower enlargement. http://www.the-digital-picture.com/Reviews/ISO-12233-Sample-Crops.aspx?Lens=412&Camera=453&Sample=0&FLIComp=0&APIComp=0&LensComp=0&CameraComp=0&SampleComp=0&FLI=4&API=0<br>

In terms of the 7D sensor out resolving lenses this is documented by others including DXO labs. On the DXO mark website you will see that the 85 F1.8 lens (the highest resolution Canon lens they have tested to date) resolves 67 line pairs per mm on the 21MP 1DsIII body but only 47 lp/mm on the 18MP 7D - link provided http://www.dxomark.com/index.php/en/Lens-with-Camera/Compare-lenses/(lens1)/241/(lens2)/241/(onglet)/0/(brand)/Canon/(brand2)/Canon<br>

Interestingly the 85 f1.8 resolves 39 lp/mm on the EOS 40D with it's 10 MP sensor. If this does not show that the 7D sensor goes beyond the resolution of the 85F1.8 - the best performing Canon lens they tested) then I would be interested in your explanation.<br>

As to the 5DIi and 7D performing almost the same at low ISO it depends on what you are looking for - on small prints there is little difference but if you follow the EOS forum you will see a recent series of posts from others who see the same thing (Lots of samples). I own both the 7D and 5DII and have put lots of shots through both of them. If you own a 5DII that performs the same as you 7D at lower ISOs then perhaps you have a 5DII issue I cannot find a lens where the Digital picture has shots with the 5DII and the 7D but here is the 1DsIII vs the 50D and you can see a big difference (the 1DsIII is very close to the 5DII, the 50D is probably not that much worse than the 7D at lower ISO as it has similar MP - I have never owned a 50D however). THe DXO mark instrumented tests show the 5DII achieves significantly higher performance to the 7D with the same lens.</p>

Link to comment
Share on other sites

<p>"Notice how many of us would rather keep this ridiculous argument going, despite glaring evidence in favor of the quality the Epson scanner is capable of?"<br>

Exactly. I am not saying the more expensive scanners can't tweek out a bit more detail but as far as I'm concerned when you bump into the grain you bump into the grain. There is not a lot left after that.</p>

Link to comment
Share on other sites

<p>Michael - to me the advantage of the Nikon scanner is time saving for the same (or a better) result.<br>

Daniel - one final point on the 7D sensor and lens resolution. the 7D sensor has 232 pixels per mm in each direction. Thus at the Nyquist frequency (sensor theoretical limit) it would show 116 line pairs per mm. The best tested lenses can reach up to 130 line pairs per mm at good contrast levels and thus could resolve more than the sensor. In practical applications and lower contrast situations 90 lp/mm is a good lens and is out resolved ny the 7D sensor. The Luminous Landscape carried an interesting article about this (even though it predated the 7D by over a year). http://www.luminous-landscape.com/tutorials/resolution.shtml</p>

Link to comment
Share on other sites

<blockquote>

<p> The reality of film, at least for the hobbyist, is that 35mm is not worth the time, and 4x5 may not be worth the cost. </p>

</blockquote>

<p>No, let's be clear, that is the reality for you. You might just want to speak for yourself.</p>

Link to comment
Share on other sites

<blockquote>

<p>the 7D sensor has 232 pixels per mm in each direction. Thus at the Nyquist frequency (sensor theoretical limit) it would show 116 line pairs per mm.</p>

</blockquote>

<p>No, no. That would be 58 line pairs per mm.<br>

It's only a theoretical limit if you want to avoid aliasing. To help you do that, they put soft focus filters in front of sensors.</p>

<p> </p>

<blockquote>

<p>The best tested lenses can reach up to 130 line pairs per mm at good contrast levels and thus could resolve more than the sensor.</p>

</blockquote>

<p>The best tested lenses resolve over 200 line pairs per mm at normal contrast.<br />And even at half of that definitely resolve more than the sensor.</p>

Link to comment
Share on other sites

<p>Tim, you ever thought about sending your work out to bureau one that uses a drum scanner for your work. I might save you 20K. Otherwise I really don't see the point of scanning LF or even Medium format negatives unless it's a real necessity since all that does is fill up your hard-drive very quickly . If you are not going to be doing allot of special effects or stiching why not just purchase a 4X5 enlarger which is way cheaper than 20K and develop the film yourself ? </p>
Link to comment
Share on other sites

<p>QG are you talking about practical photographic lenses. The UK magazine Amateur Photographer (published for over 125 years) for about 20 years used to show high and low contrast lens charts with masks showing the best ever tested photographic lens and the worst ever tested. The best High contrast was a Minolta 85 F1.4 at 138 lp/mm and for low contrast it was a Leica 35 mm F1.4 at just under 110 lp/mm. These are both centre and F8 - F11 results. While with extreme high contrast trans illuminated targets lenses can be made to reach 200 lp/mm I personally have no interest in shooting them. Thus if the only way for a lens to match a sensors resolution is to shoot these types of test target then not would not call this a useful result.<br>

Indeed if you check the tests on the DXO Mark website you will find that a good combination 35mm lens and body (say the 1DsIII or 5DII with the 50mm F1.4 lens) achieves the 58 lp/mm figure at an MTF of around 10%. On the 7D this lens only achieves about 50lp/mm at an MTF of 10%.<br>

As to Nyquist Shannon I apologize for my mistake. Indeed with 232 pixels per mm you need 58 lp/mm of lens resolution. The theory states that you need to sample at twice the original signal frequency to completely resolve the signal. Thus a 232 pixel / mm sensor requires 58 lp/mm lens resolution to completely resolve. At lower resolution the higher frequencies would produce an alias. </p>

Link to comment
Share on other sites

<p>Yes, practical photographic lenses in normal circumstances.<br>

A good lens combined with a good film will be able to 'score' above 20% at frequencies in the 100 - 140 lp/mm range.<br>

A Rodenstock Imagon softfocus (!) lens drops below 30% at 30 cycles, but still manages more than 10% at 40 lp/mm.</p>

<p>So something is wrong. And it probably is the test procedure.<br>

MTF testing is notoriously 'fickle'. You can't really compare tests, not even of one and the same lens, done by two different testers.<br>

But the figures you quote are definitely too low.</p>

<p>Zeiss once published a test of films (not lenses, films. They did so, of course, to show that it pays to pay more for their good lenses), and they found that in real-life situations, a good film records well over 100 lp/mm. T-Max 100, for instance, has no problem recording up to 180 lp/m. Portra 160 NC, is capable of recording 140 lp/mm.<br>

That, using regular (but good) lenses, on regular cameras, focusing the way we all focus. No processing tricks. Just plain old photography.</p>

<p>Zeiss also published once how their cine lenses resolved well over 400 lp/mm. But that's pushing it a bit. ;-)</p>

Link to comment
Share on other sites

<p>The AP testing is very rigorous and I am sure you would have been impressed by the late Geoffrey Crawley (see this link for more info) http://en.wikipedia.org/wiki/Geoffrey_Crawley. The test charts they published for lenses were mainly done by him (indeed he pioneered scientific lens and shutter testing) and they did a low and high contrast test at the centre and the edge for over 20 years. A good lens would be one that reached about 130lpmm at high contrast and 90 lpmm at low contrast. GC was a big fan of Leica and Zeiss lenses and many were tested over the years. I still have an archive of over 1000 editions (It is a weekly publication) and I will see if I can determine their exact testing methodology. Later on they moved to MTF charts and still perform very rigorous tests.<br>

The only Rodenstock lenses I own are the three for my enlarger and they all perform very well. Perhaps the sharpest 35mm lens I own is the Contax G 90mm F2.8 while My Fuji GX680 lenses are very good performers - especially the 180 F3.2. My initial point was simply that scanned 6x8 images produce more resolution than I can with a Canon EOS 5DII and any Canon lens (I own most of there best lenses). I was merely reacting to the statement that an APS-C can produce similar or better results. Since I own all of the systems in question (5DII, 7D 645 and 6x8 plus a Nikon 9000 scanner) I was merely reacting to the posts by Daniel Lee Taylor and providing my own explanation as to why the resolution of the 6x8 is better. There are obviously a multitude of differences between film and digital and between the two workflows. The GX680 by necessity has a very slow workflow and I use a 5DII as a form of simple polaroid as I now the relationship between the Fuji shooting Velvia and the 5DII metering system after a lot of practice. Sometimes I still use a handheld meter but the 5DII does a great job.<br>

The reason why the 5DII crop looks so soft is that it is towards the edge, at full aperture and (at least on my screen) shown at a size of 12 x7 feet. The 5DII image is perfectly usable and indeed can be blown up to 20x30 with little difficulty. Unfortunately the DSLR lacks some of the magic of the large MF body and people looking at my prints (typical 17x11, 19x13 and at most 20x30) gravitate towards the scanned MF shots - even though they are unaware of the differences in body and process.</p>

Link to comment
Share on other sites

  • 6 years later...
  • 3 weeks later...
<p>Hi Everyone. I found some old threads about scanning 4x5 transparencies with an old Epson 3250. I purchased one on Ebay, but it came without the film holders. From these old threads some people mentioned that the film holders were crap and they just taped the transparencies to the glass. Is this possible? Will the scanner recognize the media without a film holder? If so, where does one tape the transparency? Any help would be appreciated. Luke</p>
Link to comment
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now



×
×
  • Create New...