It all comes down to the print

<p>And then tell me what should I use for (lines per picture height):<br>

10MP DSLR high contrast: ___<br /> 20MP DSLR high contrast: ___<br>

<br /> 10MP DSLR low contrast: ___<br /> 20MP DSLR low contrast: ___</p>

<p>All the numbers in my previous post are in lp/mm. None of them is lines per mm</p>

<p>To get a realistic estimate of the LP/PH, it is better to use the MTF10 lp/mm number. For Velvia that is about 120 lp/mm from the MTF curve. Paired with a 120 lp/mm lens, the resolution is 60 lp/mm.<br /> 35mm Velvia lp/ph = 60*2*24 = 2880.</p>

<p>You might note that I used 80 lp/mm in my earlier post to calculate the lp/ph for 35mm Velvia and that was because you had specified the 1.6:1 resolving power number. This number is not used much these days and the MTF based estimate is considered more reliable.</p>

<p>The MTF10 for TMAX 100 would be around 200 lp/mm from the MTF curve and the lp/ph would be 75*2*24=3600, assuming a 120 lp/mm lens.</p>

<p>Ok. If the high contrast resolution of Velvia 35mm is 2880 lines per picture height (and this is before the scanner you imply I assume);</p>

<p>Then the explanation for Velvia's resolving 3650 lines with just a Coolscan is?</p>

<p>Or a Velvia 35mm scan (above) outresolving an 18MP DSLR is?</p>

<p> </p>

<blockquote>

<p>Ok. So the explanation for Velvia's resolving 3650 lines with just a Coolscan is?</p>

 

</blockquote>

<p>You tell me. In any case, 3650 is still less than the 3800 lp/ph for 7D as tested by DPR</p>

<p>I think your numbers are confused: (see DPR test below)</p>

 

<table width="507" align="center">

<tbody>

<tr valign="top">

<th width="150"><strong>Camera</strong></th> <th width="178"><strong>Measurement</strong></th> <th>

<strong>Absolute<br /> resolution</strong>

</th> <th>

<strong>Extinction<br /> resolution</strong>

</th>

</tr>

<tr>

<th rowspan="2"><strong>Canon EOS 7D </strong></th>

<td><br /></td>

<td align="right"></td>

<td align="right"><br /></td>

</tr>

<tr>

<td>Vertical LPH</td>

<td align="right">2450 </td>

<td align="right">* 3050</td>

</tr>

</tbody>

</table>

<blockquote>

<p>I think your numbers are confused: (see DPR test below)</p>

 

</blockquote>

<p>Look at the second page of the resolution tests for the raw image. It is 3800.</p>

<p>LOL The 7D has only 3456 pixels.<br /></p>

<p>I believe it is due to the sharpening applied.</p>

<p>5184 x 3456</p>

<p>I am puzzled as to how sharpening could have anything to do with that. If there are only 3456 vertical pixels, then there is no possible way to resolve higher than that.</p>

<p>"Sharpness is not resolution": the first axiom of any kind of analysis of this sort. It may look sharp as a razor, but that does not mean that it is resolving anything real. It is possible to print digital artifacts, but that would be more than a little strange.</p>

<p>--Lannie</p>

<blockquote>

<p>Sharpness is not resolution</p>

</blockquote>

<p>No argument with that. But sharpening can enhance response attenuated by the sensor (or film) and lens, if that response is present. If that is not intuitive then should the extinction resolution be limited to 3456? Even if it is, it is still more than the 2880 I get from the MTF curve.</p>

<p>Les mentioned rotation above, which I presume implies the possibility of resolving along the other axis--but you will have to ask those guys about that.</p>

<p>I do notice that the comparison above between the 7D and the 35mm Velvia scan (the map) shows rivers below Port Gentil, but the 7D gives darker lines while losing virtually all the color. </p>

<p>In any case, I am not really literate on all this--and still trying to make sense of it. I am pretty sure that more detail can be pulled out of film, but the question for me is how much of it is usable. That question brings us unfortunately back to digital as well: how many of the rated megapixels are usable?</p>

<p>With MF, there is no contest, of course--at least not to me, although I certainly do not know a thousandth of what Ellis Vener knows, and he told me once that 35mm digital can now out-resolve MF film. I frankly cannot see how, but all of these guys on both sides of the fence know more than I do.</p>

<p>--Lannie</p>

<p>Whoa? What's all this talk about pixels? I thought it's all about the print, right? </p>

<p>I want to look at the print, and make my decisions from comparing the prints. Then, I go pixel-peeping and reading about MTF and resolution to help understand why I'm seeing what I see. </p>

<p>Stuart wrote:</p>

<blockquote>

<p>Many other also found that they prefered the grainless digital prints which is what I believe caused many to say the prints were rather like medium format prints. It was not about the resolution but the smoothness of the prints.</p>

</blockquote>

<p>My experience agrees with Stuart's... I think prints from digital capture are excellent. </p>

<p>A little story... Visiting old friends in Japan, a nice couple, they take us out for dinner. She pulls out her 4x6 prints of pictures from recent vacation. They are really good prints! Strong, punchy, detailed. Terrific. What camera? P&S digital. Then, he pulls out his prints. They are OK, but just not as good. What camera? A very good 35mm P&S film camera. That's opened my eyes... Ordinary people, good consumer cameras, same place, same time, printed at same shop, the evidence was clear in the prints.</p>

<p>I'm sure his film camera out-resolved her digital, but the prints told a different story.</p>

<p>Please, no one tell me that some resolution test result disproves what I saw!<br>

<br /></p>

<p>I've done my own tests. In my opinion, looking at 8x12 prints, I see about equal quality from 12 MPx DSLR as from Ektar 100 in a 35mm SLR scanned on a Coolscan V. Yes, that's with noise reduction and sharpening.</p>

<p>I've got a pretty deep background in science, measurement, and data, so I've looked at it that way too. My numbers agree quite precisely with Les's, Mauro's, and DPReview. 35mm film, scanned on a Coolscan V or 5000, out-resolves a 12MPx DSLR, by the usual measures. But, I judge the prints as equal in quality. Huh?</p>

<p>How to square these results? Stuart suggested smoothness coming from low noise in digital capture. Might be. I rather like the theory in the <a href="http://bssc.sel.sony.com/Professional/production/productsite/files/24PTechnicalSeminar2.pdf">Sony Tech Briefing</a> I linked earlier... That the human visual system responds more favorably to high MTF than to ultimate resolution, and digital has higher MTF at moderate detail.</p>

<p>One more comment... I printed Mauro's test file that included the two women. Moving from one print size to larger and larger, it was the image of the women that fell apart first, before the test chart fell apart. That was surprising to me; I thought any differences would be easier to see in the test chart than in the continuous tone image.</p>

<p><br /></p>

<p>I don't know why Richard but I am very happy with my 8x10/12 inch prints from my DSLRs. I spent from 2000 to 2006 working as a wide format printer so I have made a lot of large prints but the encad printers we used back them were coarser being only 600dpi devices. Prints looked good though but things have moved on since then. I've printed some pretty huge prints from DSLRs, 35mm film scans as well as medium format scans on those printers, most of them would be large prints in hotels, resturants, bars, theaters and shopping centres. Personaly I prefer much smaller prints for myself and usually prefer B&W prints I rarlely go above 8x10/12 for myself. I got the dslr around 2005 and have been very pleased with the DSLR at those smaller print sizes.</p>

<p>Mauro states,</p>

<blockquote>

<p>In contrast, I do not print from my 10MP 40D larger than 11x14 ...</p>

</blockquote>

<p>Richard Karash also asks,</p>

<blockquote>

<p>out-resolves a 12MPx DSLR, by the usual measures. But, I judge the prints as equal in quality. Huh? How to square these results?</p>

</blockquote>

<p>Neither would I. 10MP fills out an 8x10 print nicely with a bit of margin to crop. The point is that resolution limited as a DSLR may be, what pixels are available tend to be much higher quality than those originated from 135 film. </p>

<p>Look at the numbers C. Sharon pulled out from the data sheet. Velvia at 70% MTF is 35 cycles/mm, or roughly 5MP for a frame of 135. Source image features finer than this starts to be recorded at noticeably reduced contrast.</p>

<p>Okay, so say the frame is scanned on a 4000dpi Nikon 5000. This yields a 20MP file. Yes, all 20MP has real image information, but much of it is a recorded at lower fidelity and not as easily usable.</p>

<p>Pitch the file at a 6x9 300ppi (~5MP) print and everything looks sharp. This is because the printer is effectively just rendering the 70% and higher MTF portion of the film's recording regime. However, do a 12x18 print from the same 20MP file and visible fine details are now decidedly a bit fuzzy: the printer is now reproducing the 30% and lower MTF response part of the film.</p>

<p>The other aspect of film is that it's much noisier than the same sized digital imager. Keep in mind where effectively the printer low pass filtering kicks in. Printing a film scan at moderate sizes, e.g., less than 5X enlargement, places the film grain above the limit of reproducibility of the printer.</p>

<p>Taken in aggregate, this is why it can take a 6x7 to equal a 8MP DSLR on a 8x10 print. </p>

<p>One more thing though, the noise reduction and sharpening tools available is of course agnostic as to where the digitized file came from. These are perfectly suited for improving the film scan quality - if you're bent that way.</p>

<p>Les - <em>To be even more precise, because the <strong>scanned film</strong> results actually has more resolution then your processed 7D setup it is therefore able to distinguish even finer detail and is able to render those details more precisely that I pointed out.</em></p>

<p>No Les, that is purposely <strong>less precise.</strong> Which has more resolution depends entirely on the contrast of the detail being resolved. It is the 7D which is able to distinguish even finer detail, and is able to render those details more precisely, where the details are low in contrast. Velvia 50, the best color film by a wide margin, on an Imacon, one of the best scanners ever made, could not resolve the relief texture in the map that the 7D picked up with ease.</p>

<p>Looking at a sample print at the same ppi as a 30x45" full map print, I can observe that the 7D resolves more low contrast detail and gives a superior rendering of relief texture. I wouldn't consider it an important or huge difference, but it's observable. I cannot observe any difference in the high contrast detail (i.e. the text). I have to view at 100% to see that.</p>

<p>In the real world fine details occur overwhelmingly at low contrast. The texture of a model's skin, fabric threads in clothing, dust on a petal, foliage at a distance...all are defined by small variations in color and tone. They're all low contrast. I'm hard pressed to think of any fine details which occur at high contrast in real scenes, with the exception of human signs (i.e. black text on bright backgrounds). That means for most real scenes, most of the time, the 7D will yield a higher resolution image and print than even Velvia 50 on an Imacon.</p>

<p>"Taken in aggregate, this is why it can take a 6x7 to equal a 8MP DSLR on a 8x10 print."<br>

Agree at 8 x 10.</p>

<p>If you print the test sheet I offered (the one of the two ladies) you will see that you are not far from reality. At 8x10, the 6x7 film and 10MP DSLR produce somewhat similar results. At 11x14 the differences become more visible, and pass that point, (in my opinion) the 10MP DSLR crosses the line where the print quality is materially impaired.</p>

<p>A 20MP DSLR gives you about a size up from a 10MP DSLR (i.e a 20MP DSLR can produce a similar print at 16x20 than what a 10MP DSLR can produce at 11x14).</p>

<p>35mm film (Velvia, TMAX, Ektar) scanned with a Coolscan, give a significantly higher quality print than a 10MP DSLR at 16x20 prints or larger. This in particular I tested rather extensively. (Some comparisons you can find in this thread to print yourself).</p>

<p>I'm not particularly interested in getting buried in a discussion about Mauro's charts. I will say, however, that I have no idea where he got the numbers or thresholds, and that they directly contradict what we just observed with the 7D vs film map samples.</p>

<p>I will also point out, again, that LPH cannot be converted to MP the way Mauro and Les are converting it. The 40D is not a 5 MP camera, it is a 10 MP camera.</p>

<p>And I would like to reiterate that in real world photography the finer the detail, the lower the contrast. It is rare to find a scene which contains detail that is both very fine yet also high in contrast. Most fine detail is texture, and most surface textures are defined by very subtle changes in color and tone. Foliage and Earth (i.e. soil, rock) at a distance exhibit the same characteristic.</p>

<p>I have some trips coming up at the beginning of summer when I can test the following assertion with side by side landscape shots. But knowing that the Canon 7D exhibits higher resolution than 35mm Velvia 50 when dealing with low contrast detail, I predict that it will give me higher resolution images and superior prints, regardless of size, than 35mm Velvia 50 will when dealing with real world landscapes.</p>

<p>Now if I were to go into a lab and shoot some B&W test charts, I'm sure Velvia would give me higher resolution images and superior prints, assuming an Imacon or drum scanner. But prints of lab shots of B&W test charts don't interest me very much.</p>

<p>If this exercise has taught me anything, other than how to maximize processing for extreme enlargements, it has taught me that the resolution argument is dead.</p>

<p>Daniel,<br /> If you wish, I can scan your Velvia shots when you return from your trip. That will be a very relevant comparison.</p>

<p>I hope you use Velvia 50 (not 100). </p>

<blockquote>

<p>A 20MP DSLR gives you about a size up from a 10MP DSLR ...</p>

</blockquote>

<p>Right, pretty much. It's got just enough pixels to nicely fill out a 12x18 print, and as Daniel mentioned high quality pixels, i.e., almost perfect fidelity and almost no noise. In actual use, this means being able to record the skin textures on even a model with perfect skin.</p>

<p>Where the 20MP DSLR starts to fall teeter and fall apart is on enlargements above 12x18. Unlike film, there is <em>no</em> image information recorded above this threshold.</p><div></div>

<p>Very nice!</p>

<p>Cool snowman....</p>

<blockquote>

<p>A 20MP DSLR gives you about a size up from a 10MP DSLR ...<br />Right, pretty much. It's got just enough pixels to nicely fill out a 12x18 print</p>

</blockquote>

<p>And that is at 300 dpi. (300x12 = 3600 on one side and 300x18 = 5400 on the other: 3600x5400 = 19,400,000) If one is content to print at lower resolution, it can go much larger.</p>

<p>I know that this is obvious to the people who post here, but it is a good reminder to all of us who do once a while like to print.</p>

<p>I wonder how many here routinely print at 300 dpi.</p>

<p>--Lannie</p>

<p>I print at 360 dpi (native for my 24" printer).</p>

<p>A 20MP dslr can generate quality prints up to 16x20, in my opinion, with all conditions being ideal.</p>

<p>Mauro says"A 20MP dslr can generate quality prints up to 16x20, in my opinion, with all conditions being ideal."<br /> Mauro I've noticed something that doesn't get mentioned much. I often print small usually 5x7 or maybe 6x9. I've been shooting a lot of 35mm Tri-x with my M6 and a V/C Nokton 40 1.4. This somewhat grainy film doesn't look all that great on screen at 100% when compared to a decent DSLR. But!! that same grittiness that doesn't look so great when magnified creates a print that is so sharp and detailed that it's blowing my mind. The tonalities are superb and I can't get my 40D to begin to match detail when it comes to things like vegetation, grasses and the like. Wish I could show folks the print of the attached photo.</p><div></div>

<p>Michael, that is very true - tri-x has very interesting tones. Also grain may add to the sharpness of the print. </p>

<p>That is a fantastic picture. Please let me know if you have a link where I can download it for print. (you may want to put a watermark on it).</p>