Velvia vs. Sensors

<p>Come on Tim, get the digital images in focus. Look at the Nikon logo on the digital images. My granddaughter could do better with her Coolpix.</p>

<blockquote>

<p>At the least expensive end of Nikon DSLRs, I think that <a href="/photo/17506648" rel="nofollow"><em><strong>the sensor on the D3200</strong> </em></a>is really quite incredible.</p>

</blockquote>

<p><br /><br /><br />I am mainly a film photographer. I have 70+ cameras with only two of them digital.<br /><br />One of them is a D3200 and I agree with your comment. I use mine with the pre AI lenses I have for my Nikon F.</p>

The focus is just fine in Tim's images David.

<p ><a href="/photodb/user?user_id=528518">Dave Luttmann</a> , said:</p>

 

<blockquote>

<p>The focus is just fine in Tim's images David.</p>

 

</blockquote>

<p>How can you look at them and say that? Does anyone else agree with me? I see "NIKON" in focus on the film images a badly OOF on the digital images. Digital does NOT cause that much unsharpness. If digital were that bad, no one would use it.</p>

<p>IMO, it's either OOF or the lens is poor or something was done wrong in Raw conversion.</p>

 

<p>Without knowing how the film was scanned and by what, and how it was processed afterwards or how the digital image was processed too (sharpened etc.), comparing anything is a waste of time. Even if you knew how they did these things, then what does that mean? Maybe they processed it poorly. </p>

I think David that you haven't seen extreme enlargements then. I think Tim's credentials are just fine.

<p>Dave, I look at Canon 5D3 images at 100 and 200% routinely and they're sharper than that, by a long shot. Tim doesn't tell us what magnification we're looking at, but, given the subject, out have no reason to believe that it's more than 100%.</p>

<p>As Alan says, we don't know what's being compared. I'd expect the MF film to be superior to a full-frame sensor, but then, people that shoot MF are going to use MF sensors. Still, those full-frame digital images don't hold up as I'd expect at 100%, assuming good focus and a reasonable aperture.</p>

<p>IMOPO, I'd expect a Mamiya 7 with Adox CMS 20, or even Fuji Velvia 50, to out-resolve a Nikon D800E. However, the differences among the examples above is more than I'd expect. Why? I tend to suspect that very few lens-and-working-aperture combinations can really deliver all the resolution the D800E can capture (you'd need good MTF response at 102 lp/mm). What lensese were used for these examples, and at what apertures? Also, as soon as whatever you're shooting (even a landscape in a breeze) won't sit still enough for ISO 20 or even ISO 50, the D800E will start to pull away (I'd expect visibly so starting around, say, ISO 400).</p>

<p>Dave, I think we're agreeing.</p>

<p>You don't buy a D800 unless you've got a lens that's up to the challenge. If you're going to do a comparison to film, then you use a similar ISO as the film, you carefully focus and select an appropriate aperture and take a shot that's as good as each camera can achieve. I just don't think those D800 images are up to what that body can achieve and don't understand what's wrong. I get much sharper files with my 5D MkIII. (Everyone knows that Raw digital files require sharpening, don't they?)</p>

<p>David. I have some sympathy with your argument, but you're spoiling it by making unsupported assumptions.</p>

<ul>

<li>I doubt very much that most photographers migrating from MF film to digital chose MF sensors. There's simply a huge cost for doing that.</li>

<li>"have no reason to assume its over 100%". Surely the right approach is to ask Tim what magnification he's used? Which might not be the same for all shots. Until we know, there's no point to be made.</li>

<li>"You don't buy a D800 unless you've got a lens----" Again, there's no point here unless you ask Tim what he's used for this test.</li>

</ul>

<p>It's an interesting debate and I for one would like to understand the truth of it.</p>

<p>Let's call this the Dave/David thread...</p>

<p>David H.: Unsupported assumptions?</p>

 

<ul>

<li>So your telling me that all that stuff that Leaf and Phase One sends me about all the pros using their very expensive MF digital cameras is BS?</li>

<li>If Tim would like to tell us what in the world he did, he can step back in. Right now, it feels like he dumped some images on us with little or no explanation of what we're looking at. The point to be made at this point is we can't tell anything from those images except that the D800 ones look ghastly. (Which seems really hard for me to believe after looking at my own Canon images).</li>

<li>I only mentioned the lens because someone else mentioned that as a variable, given the D800's high resolution potential. Once again, it would be nice to know what we're looking at, including EXIF and at least a brief comment about Raw conversion used.</li>

</ul>

<p>I too am various curious to know what's wrong with those D800 images.</p>

<p>David Stephens, I suspect that the Nikon camera is so small in the composition that the D800 is simply not resolving the Nikon lettering properly. At some point the D800 resolution runs out. It also looks to me also like the D800 image could have been interpolated to match the size of the film scan.<br>

I can only speak for myself but I bought a D800 without having much in the way of lenses that could really deliver the max resolution. The D800 is still way ahead of my D700 though with most Nikkors out resolving the 12mp D700 sensor.</p>

<blockquote>

<p>Tim doesn't tell us what magnification we're looking at, but, given the subject, out have no reason to believe that it's more than 100%.</p>

</blockquote>

<p>I'm in agreement with David Stephens on this. I operated a graphics camera for about ten years back before digital and there's a bunch of information left out of Tim's comparison demo that goes beyond simple magnification issues.</p>

<p>First off I couldn't find anywhere in the thread where Tim states how far away those shots were taken from the subject which will greatly influence accutance as captured by the lens in relation to the focus plane, sensor or film platen and capture resolution.</p>

<p>On my 6MP Pentax K100D I can get objects around 50 or so feet away tack sharp captured handheld with my 18-55mm kit lens so I'm having a problem seeing how that comparison is useful considering we don't know a lot about how it was shot in the least whether a tripod was used and capture resolution to zoom view size.</p>

<p>OK <br>

Nikon D800E with Zeiss 25mm at f/5.6 - we focus bracketed and aperture bracketed. slight variations in center contrast but not really much perceptible difference in result. Basically we had the D800 running at sharp pixel levels (hence moire - if you know much about digital you'll know you don't get moire on fine detail unless you've got pixel detail). <br>

The test was done at the back end of this comparison. We came back and shot with a D800 and a D800E using a variety of lenses including the Nikon tilt shift and others. <br>

http://www.onlandscape.co.uk/2011/12/big-camera-comparison/<br>

The film was scanned on a Screen Cezanne Elite Pro (I own a Heidelberg Primescan, Fuji Lanovia and a Howtek 4500 and for straight up dpi the Screen Cezanne wins hands down). <br>

As a sample, here's some detail from a hand held shot on an OM2 taken with Fuji Superia 1600 film - the screen just eats grain where drum scanners seem to emphasise it. <br>

http://static.timparkin.co.uk/static/tmp/35mm-OM2-Fuji-Superia-1600-6000dpi.jpg<br>

Just for reference the scanned medium format film was approx 14,000 by 17,000 pixels so you can imagine that the D800 images had to be uprezzed so these are probably around 285%?<br>

I'm happy to supply the original raw files if anybody thinks that their D800E can resolve more pixel level detail :-)<br>

Oh and if you think that the IQ180 will beat medium format film here you go!!<br>

http://static.timparkin.co.uk/static/tmp/iq180-mamiya7-adoxcms20.jpg<br>

The hasselblad in the mamiya test was the same distance away but with a difference angle. The results are still interesting, if not as scientific... </p>

<p> </p>

<p>Tim, what was the distance of the Nikon camera logo which I'm assuming was the plane of focus for each individual camera?</p>

<p>IOW how far away were the cameras for each individual shot? Were they different from one another and did you use different focal lengths for each? </p>

 

<p>I think it could be useful for many to see the full frame D800 shot at so they understand just how small the Nikon FM was in the overall shot.</p>

<p>Ok, almost 300% crop. I understand. </p>

<p>It's neat to see the resolution power of MF and larger film, but I'm only printing up to 50" with mere full-frame resolution, I don't think that I need it. I need the speed of my smaller, faster camera to shoot my main subjects, birds and wildlife.</p>

<p>If I were mainly a landscape shooter, MF might be tempting.</p>

Didn't someone post an often referenced test about an 11mp DSLR beating 6x7 film? Maybe people can stop linking to it

all the time now.

<p>The cameras were approx 10m away from the target. All focal lengths were 24/25mm equivalents. <br>

Here's the full scene</p>

<p><img src="http://www.onlandscape.co.uk/wp-content/uploads/2011/12/studio-wide.jpg" alt="" width="1000" height="667" /><br>

As you can see the camera is fairly small<br>

As for printing sizes, we made multiple prints and showed them to a range of people and everything was pretty even up to about 16"x20". At that point the 24mp DSLR's were judged as looking worse than the other cameras. D800's topped out about 20"x24"<br>

IQ180 topped out above 30x40 and interestingly depending on the viewer was judged better or worse than 5x4 film. <br>

Medium Format film scanned on a drum scanner topped out at just about 25x30 but for typical 'off the shelf' dedicated film scanner I would say medium format film and D800 are pretty much on a par. If you use an Epson Vx00 then expect you MF scans to be just about 20mp camera equivalent. <br>

We tested some colour 35mm film and depending on the subject and viewer the quality was judged at between 10 and 20mp. On average in terms of sharpness it was about 12mp but for some colour subjects with lots of fine detail (i.e. many landscape photos) the quality was seen as a much better - probably around 16-18mp. <br>

Now of course you can print any format as big as you like and I would imagine about 50% bigger than the sizes mentioned would be fine for most purposes so for a D800 you could get an acceptable 30x40 print that wouldn't look too bad at an arts fair.<br>

re: david stephens. I wouldn't bother shooting wildlife with a film camera unless you were pure fine artist and were looking for a differentiating factor (Nick Brandt?) or after some stunning enlargements and can't afford an IQ180 and you really, really know when to press the shutter!</p>

 

<p>Tim: Thanks for the test and input. It cleared up a lot of questions. </p>

<p>Oh and I've updated my resolution test chart compilation with the scans on the Screen Cezanne.. <br>

http://static.timparkin.co.uk/static/tmp/cms20-vs.jpg<br>

As you can see the Mamiya 7 lens is absolutely stupidly sharp and the Adox CMS20 can resolve ridiculous amounts of detail (about 280 line pairs per mm)</p>

<p>p.s. Post processig of the D800E was done in Capture One (and I tried other raw converters that offered different trade offs of sharpness). Sharpening was done in Photoshop using deconvolution (smart sharpen). But this is all really just small amounts of trimming. Bascially I was getting about 90% contrast across two adjacent pixels on a high contrast edge (a black peice of plastic card against a lightbox). To give you an idea of what different raw converters did, the contrast changed by a few percentage points. </p>

 

<p>Of course the 8x10 drum scan is the best. The CMS20 is pretty amazing compared to Velvia. </p>

<p>What is microscope?</p>

<p>A microscope is what you look through to get very high magnification. It will show you what detail is actually on the film rather than just what the scanner can see. </p>

<p>e.g. here's the microscope that I used.. <br>

<img src="http://nimax-img.de/Produktbilder/zoom/20680_1/Optika-Stereo-microscope-SZM-4-Zoom-trinocular-7x-45x.jpg" alt="" width="800" height="800" /></p>

<p> </p>

<p>So the drum scan appears to do better than what the microsoft shows. Is that due to contrast or other adjustments built into the drum scanner?</p>