Pixel density and lens quality

[kelly_flanigan1]

<p>Here with my old 4x5 digital scan backs; the lens requirements are only modest. The 35 and 50 megapixel backs have a 7x10cm scan area. The small guy shoots a 5000x7000 pixel image; the big guy a 6000x8400 pixel image. 10cm is 100,000 microns; thus the smalll guy has a pitch of 14 microns; the big guy 12 microns. If the lens is *good enough* for the sensors pitch; one can get a sharp image.<br>

The poor assumption you all have led your self into a mental rut is assuming that a 100 buck P&S digital has a bad lens; since your assumption is gravely in error; you conclusions are faulty and in error. Optically it is easier to make a good lens for a smaller format camera that a larger one. P&S digitals are made by the millions too; thus the tooling costs per unit sold are way less than a low volume lens.</p>

[kelly_flanigan1]

<p>Here with my old 4x5 digital scan backs; the lens requirements are only modest. The 35 and 50 megapixel backs have a 7x10cm scan area. The small guy shoots a 5000x7000 pixel image; the big guy a 6000x8400 pixel image. 10cm is 100,000 microns; thus the smalll guy has a pitch of 14 microns; the big guy 12 microns. If the lens is *good enough* for the sensors pitch; one can get a sharp image.<br>

The poor assumption you all have led your self into a mental rut is assuming that a 100 buck P&S digital has a bad lens; since your assumption is gravely in error; you conclusions are faulty and in error. Optically it is easier to make a good lens for a smaller format camera that a larger one. P&S digitals are made by the millions too; thus the tooling costs per unit sold are way less than a low volume lens.</p>

[kelly_flanigan1]

<p>Here with my old 4x5 digital scan backs; the lens requirements are only modest. The 35 and 50 megapixel backs have a 7x10cm scan area. The small guy shoots a 5000x7000 pixel image; the big guy a 6000x8400 pixel image. 10cm is 100,000 microns; thus the smalll guy has a pitch of 14 microns; the big guy 12 microns. If the lens is *good enough* for the sensors pitch; one can get a sharp image.<br>

The poor assumption you all have led your self into a mental rut is assuming that a 100 buck P&S digital has a bad lens; since your assumption is gravely in error; you conclusions are faulty and in error. Optically it is easier to make a good lens for a smaller format camera that a larger one. P&S digitals are made by the millions too; thus the tooling costs per unit sold are way less than a low volume lens.</p>

[g dan mitchell]

<p>Here is a simple thing that every photographer who wants to understand sharpness, aperture, and how the two interact in their gear.</p>

<p>On a slow day, take your camera and tripod and lenses out somewhere and shoot a series of test photographs. With each lens carefully focus on a subject that provides a decent target - don't use a test target, just a scene of the sort you usually photograph, but pick one that doesn't provide any real DOF challenges. With each lens:</p>

<ul>

<li>Camera on tripod.</li>

<li>Focus by whatever means you prefer.</li>

<li>Turn off AF.</li>

<li>Shoot RAW.</li>

<li>Starting at your largest aperture make an exposure at each whole f stop in aperture priority mode. Use mirror lockup (or live view) and a remote release.</li>

<li>If you use a zoom repeat the process at several representative focal lengths across the range of your lens.</li>

<li>Try with each of your lenses.</li>

<li>Take the resulting images through your normal basic workflow including sharpening.</li>

</ul>

<p>Take a close look at what you end up with. Compare center sharpness and corner sharpness at a variety of apertures. Note the amount of vignetting. Note the effects of barrel/pincushion distortion.</p>

<p>The object is not to see if you have a "good copy." It is not to determine what aperture you should always shoot at. The goal is to quickly understand the capabilities of your lenses so that you can intelligently apply choices about lens, aperture, etc to your photography. You'd learn much of this eventually by trial and error, but you can significantly jump-start the process by this means.</p>

<p>This I discovered about my lenses include stuff like:</p>

<ul>

<li>One lens has significant barrel distortion and vignetting at the wide end at f/4 but is still very sharp.</li>

<li>Another provides surprisingly good bokeh for a f/4 lens.</li>

<li>Another turns out to produce quite decent IQ at its longest FL but is even better at a slightly shorter FL. </li>

<li>Resolution of one zoom is so good that it usually isn't worthwhile using a prime in its FL range.</li>

<li>On a full-frame camera, diffraction blur is inconsequential at f/16 even in quite large prints.</li>

</ul>

<p>All the theorizing in the world becomes quite irrelevant in the light of what you find out about your own gear by using it.</p>

<p>Dan</p>

[yoshio]

<p>Dan</p>

<p>noone wants to know for themselves, they may not believe their eyes. It requires them to interpret the evidence and have faith in themselves. Most will prefer to be told by experts. It reminds me of a "monty python" episode ...</p>

<blockquote>

<p>"Is this the right room for an argument"</p>

</blockquote>

<p>but your test is a good idea</p>

<p> </p>

[mike_hitchen]

<p>Michael Young: Have you taken the same picture with a 300D and 7D and printed them <em>to the same size</em>? Surely that is the only test that matters? I would tend to agree with Dan onthis one, viewing each image at 100% is irrelevant if your end point is to print so surely that should be the ultimate test.</p>

 

[michael_young3]

<p>Mike, I first looked at slides and negatives through a microscope in 1972. Why don't we pick up the validity or value of understanding the image micro-structure some other time? The fact of the matter is, photographers will continue to examine their images up close, whether they plan to print that size or not. Well, maybe not photographers per se. Real Photographers such as yourself care only about the message conveyed by the image, the art of the image, rather than the image itself or the technology behind it.</p>

<p>Alas, photographers are also camera owners. Male camera owners in particular, being male and owners of newly acquired high tech equipment, will examine in excruciating detail the performance of their newly acquired, expensive equipment. They will have questions about what they perceive to be softness and inferior performance. Will you have an answer?</p>

<p>The newest, very fine pitch sensors are now able to resolve diffraction effects at surprisingly wide apertures. Ultimately, your answer of "Don't do that. It's stupid and meaningless." is itself stupid and meaningless. If instead you can put a number on it that they can understand -- f/6.9 on the 7D -- they'll move on and start taking some Real Pictures and stop shooting lens tests, sooner rather than later.</p>

[mike_hitchen]

<p>Michael - my point was not whether your comaprisons are stupid and meaningless (your words not mine) but was to pick up on Dan's comment about the validity of comparing 100% views from sensors of different resolution. By printing at the same size you are forced to compare apples with apples. Do you think a picture with f8 on a 7D (past its diffraction limit) will be sharper or less sharp than a picture taken with the same lens on a 300D (within its diffraction limit)</p>

<blockquote>

<p>They will have questions about what they perceive to be softness and inferior performance</p>

</blockquote>

<p>Yep. Sure they will, and to do this it is not the data you collect, but (a) the question you are asking and (b) how you analyse the data and I feel you are not analysing the data in the correct way.<br>

If you had the 7D and the 300D, and you wanted to shoot a landscape using f10, you would put aside the 300D and pick up the 300D because the 7D is limited at f6.9?</p>

<blockquote>

<blockquote>

<blockquote>

 

</blockquote>

</blockquote>

<p>Male camera owners in particular, <em>being male and owners of newly acquired high tech equipment</em>, will examine in excruciating detail the performance of their newly acquired, expensive equipment.</p>

</blockquote>

<p>I agree completely :). Been there, done that with hifi as well as cameras. And I still do sometimes :)</p>

[michael_young3]

<blockquote>

<p>would put aside the 300D and pick up the 300D because the 7D is limited at f6.9?</p>

</blockquote>

<p>Of course not, Mike. Is that how this conversation shapes up?</p>

<p>I hear the 550D will have similar MP as the 7D. Get ready for the deluge of even less sophisticated but completely expectable Y-chromosome driven questions about its crappy resolution.</p>

[g dan mitchell]

<p>Mike observed:</p>

<blockquote>

<p><em>By printing at the same size you are forced to compare apples with apples.</em></p>

</blockquote>

<p>Absolutely true! I believe that comparing at 100% on the screen actually tells us very liitle when it comes to comparing cameras with different pixel dimensions and sensor sizes. However, making photographs than then taking them through a typical workflow to a typical final medium at a given size - be that a print (my preferred form) or a uniform size on-screen display - is a much more meaningful real world test.</p>

<p>The classic misunderstandings are represented by a couple of types of posts that we frequently see on forum boards:</p>

<ul>

<li>"My 5D2 isn't as sharp as my 5D when I compare 100% crops from the two cameras." (Of course it isn't - you are looking closer at a smaller portion of the image!)</li>

<li>"My 5D2 is diffraction limited at an earlier point as I stop down, so a 5D is better for shooting at smaller apertures." (The two cameras will produce precisely the same amount of diffraction blur in a print made from respective photographs at the same apertures.)</li>

</ul>

<p>These and similar misconceptions arise all the time when people obsess over 100% magnification crop comparisons on the screen... but they disappear in actual output.</p>

<p>Dan</p>

[michael_young3]

<p>Dan, knowing it won't stop them from clicking just once on the image in LR and getting the 1:1 view. Wouldn't it be nice to have the magnification default to a preset for "8x10 at 240 dpi, corrected for Epson 3800 dithering" instead? We don't have that today. It just snaps to 1:1. On what else would they base their questions?</p>

<p>The irony is we're all saying the exact same thing. Me personally, I would much rather have someone say to me, "Nah. Put on your sharpest lens and shoot it again at f/(whatever). It'll be tack sharp or something else is wrong." That curmudgeon stick of "Well, don't look at it if it bothers you" is so completely meaningless I haven't the words to begin to rebut.</p>

[g dan mitchell]

<p>Michael, it still wouldn't resolve (pardon the pun) the issue, though there is a "as printed" sizing option in Photoshop and I presume in Lightroom. Here you would still dealing with different and less than optimum interpolation.</p>

<p>Really, the only meaningful result is one that is based on the final photograph in whatever form you intend to present it. If you always (or mostly) go to jpg at size X by Y, then it is very important to understand how different lens/body systems will or will not differ. If you print, seeing the result in a print is the most significant thing.</p>

<p>Dan</p>

[scott_ferris]

<p>Michael,</p>

<p>Your assumptions are that less educated owners are as naive as you are obtuse. You are trying to make a point about others not understanding something that you can see. They can see it too, but don't think they don't have the common sense to realise they can't do 30mpg at 180mph. Whilst your point is valid, anybody with an ounce of common sense can see through your "issue".</p>

<p>Take care and time for another bottle of wine, Scott.</p>

[michael_young3]

<p>Let me see if I understand you right. Downsampling the larger image to the smaller image's downsampled dimensions gives an equivalent image that's valid for comparison.</p>

<p>Thanks a bunch. Why not point out that water is wet? Just how much new understanding did that inject into the conversation?</p>

 

[michael_young3]

<p>Not at all, Scott. I shot something at f/7.1 and it looks sharp. I stopped down further with the expectation that it would be sharper still, and it wasn't. Same body, same conditions, same lens, same viewing software, moments apart. Shouldn't I want to understand why? Now that I can put a number on it, I do wonder why water being wet is such an insight while this is so shruggable.</p>

[scott_ferris]

<p>That is now a second issue, are you trying to stir up a "my 7D is not as good as my 300D" comment or are you ignoring the answers to "my 7D is not as good at f8 as it is at f7.1"?</p>

<p>You know the answers to both. It takes no real understanding.</p>

<p>Again, your 7D is magnifying everything, CA/diffraction/lens limitations everything, more than your 300D, enlarge them the same and the 7D will look better. Your 7D images look sharper at f7.1 than at f8 because the engineers have done such a good job you can now resolve diffraction blur accurately at f8, your 300D wasn't good enough to do that. So again your 7D is better than your 300D.</p>

<p>Your new 7D does not break the laws of physics, diffraction is more at f8 than f7.1. Zooming in to 100% on the 7D allows you to see that, on the 300D it did not. You have bought a better magnifier, why are you complaining because you can see smaller things?</p>

<p>Like I said, you raise a point, a small and inconsequential one, but a point. Accept that anybody that noticed this, seeming anomaly, and did not have your understanding of it would have been lead to the answer long before this.</p>

<p>Back to your bottle of wine, Scott.</p>

<p> </p>

[michael_young3]

<p>I can hardly claim credit for discovering that images suffer from diffraction. I merely learned to determine with precision where that limit is. You dense brick. I'll be by tomorrow to install the breathalyzer on your keyboard.</p>

[scott_ferris]

<p><a href="00Vhah">"(PS: I have no problems with sharpness or detail with any lens. The 7D is a peeper's delight.)"</a></p>

<p>or</p>

<p><a href="00Vhah">"I'm a little surprised to find so many folks have problems with it."</a></p>

<p>or<br /> <br /> <a href="00Vbwn">"depleted sherry glass in hand and the week's troubles beginning to drain into a hazy past." </a></p>

<p>Michael, thanks for the thoughts but I don't drink, you on the other hand obviously do...............</p>

<p> </p>

[michael_young3]

<p>Yup. All of that. Followups to usenet, please.</p>

 

[mike_hitchen]

<p>Micheal. I ams till trying to udnerstand what pont you haev been getting at in this thread.<br>

OK, the 7D is diffration-limited at 6.9, the 300D is diffraction-limited at 11.6. Now that is fair enough and the only conclusion you can draw is that the 7D is able to demonstrate the diffractino limitations before the 300D is.<br>

But after that you seemed (to me at least) be then using these figures to try and define one characteristic of 'the better camera'. Given that diffraction is a characteristic of the lens not the camera, if you used the same lens on the 7D and 300D, I would hope to God the engineers at Canon have done a good enough job that despite its diffraction limit being 6.9, the 7D would still be better than the 300D at 11.6 when using the same lens on both cameras.<br>

And this being the case, who cares how the diffraction limits compare - the diffraction limit is only relevant when using a specific camera under specific conditions and you want to know its limitations. Inter-camera comparison is almost meaningless.</p>

<p> </p>

[g dan mitchell]

<blockquote>

<p><em>OK, the 7D is diffration-limited at 6.9, the 300D is diffraction-limited at 11.6. Now that is fair enough and the only conclusion you can draw is that the 7D is able to demonstrate the diffractino limitations before the 300D is.<br /> But after that you seemed (to me at least) be then using these figures to try and define one characteristic of 'the better camera'. Given that diffraction is a characteristic of the lens not the camera, if you used the same lens on the 7D and 300D, I would hope to God the engineers at Canon have done a good enough job that despite its diffraction limit being 6.9, the 7D would still be better than the 300D at 11.6 when using the same lens on both cameras.<br /> And this being the case, who cares how the diffraction limits compare - the diffraction limit is only relevant when using a specific camera under specific conditions and you want to know its limitations. Inter-camera comparison is almost meaningless.</em></p>

</blockquote>

<p>At the risk of getting sucked into something that I may regret... an observation that might help people understand what "diffraction limited" <em>doesn</em> 't mean.</p>

<p>I don't know the figures, but let me grant for the sake of the example that a 7D might be "diffraction limited" at f/6.9 and the 300D might be "diffraction limited" at f/11.6. If you aren't careful, you might misinterpret this to mean that you should shoot the 7D at f/6.9 and the 300D at f/11.6, or you might imagine that you would get a worse photograph from the 7D at f/11.6 than you would from the 300D. Both assumptions are wrong and betray a serious misunderstanding of what so-called diffraction limiting means and how it affects photographs.</p>

<p>Imagine the following experiment. You have a 300D and a 7D. You have a really excellent lens that will work on both cameras. You first set up the 300D and carefully focus and make a photograph at some aperture, let's say f/11. Then you remove the 300D from the tripod, mount the 7D in its place and attach the same lens you used for the first image made with the 300D. You carefully focus and using the same aperture you make the same photograph.</p>

<p>Take both photographs through optimized workflow and make prints at the same size. The amount of diffraction blur in each will be <em>identical</em> . Repeat the experiment at any aperture and the result will be the same - the amount of diffraction blur in your print will be the same.</p>

<p>Dan</p>

 

[scott_ferris]

<p>Mike and G Dan,</p>

<p><a href="http://photography-on-the.net/forum/showthread.php?t=747761">This is a very good link</a> , it explains, and demonstrates the difference between diffraction limited aperture and the diffraction cutoff frequency, via another thread in which I also posted that link I seem to have placated Michael somewhat.</p>

<p>But shhhhh, we don't want to wake him :-)</p>

[michael_young3]

<p>Sheesh, Guys! No angst here, or at least not any more than usual. Why don't we start over?<br /> <br /> I first stumbled across DLA mentioned in passing in Merklinger's article on the Luminous Landscape site. I later read Scott's first words about it written in Oct, 2008. TDP reviews first reported DLA figures in their equipment reviews with their review of the 50D. Me personally, I had been living under a rock until buying the 7D in November. Up till then, I had been happily shooting the 300D, in all its glorious 6 MP, since it was first released in, when was that, September, 2002? 2003? If there's any contradictions in the thoughts I expressed earlier, it's because I've been trying to make sense of the concept. Stay tuned for even for more refinements, revisions, and reversals. DCF used to mean discounted cash flow. Now, it'll be a mantra until I work out a better figure of merit for choosing an aperture when shooting smaller than DLA. After that, I'll likely fall back asleep under that rock until the 7D's replacement obsoletes physics.<br /> <br /> Merklinger gave the simple relationship of: DLA = 1600/pixelPitch in millimeters. Equivalently, that's DLA=1.6 * pixel spacing in um. How convenient. I multiply by 1.6 easily in my head, having apparently hardwired some neurons to do so for crop sensor related thoughts. The long and short is, I soon had two numbers in hand: f/6.9 for the 7D, and whatever it was for the 300D. They match my experience exactly. Where f/11 had been a perfectly good best sharpness aperture for film and the 300D, it produced mush on the 7D. Mush only at 1:1 to be sure, but mush nonetheless.<br /> <br /> Mush relative to what?<br /> <br /> Sure, the print is the final measure of image quality. No question about it. I don't exhibit in galleries, and I don't hawk them at craft shows. The ones that are good enough to print, I've tended to print as large as I can. I am therefore looking for the best image quality for the circumstance, and that includes sharpness of detail. I care very much about mush and its converse. With all due respect to anyone else's measure of goodness, I buy L series lenses for reasons other than their color scheme. Surely Canon doesn't make them just for me, in case I might want to buy one. I'm not alone in caring about sharpness.<br /> <br /> I recognize that peeper-baiting is common sport here and perhaps elsewhere. In an equipment forum, though, it can't be anything but noise and flaming. I'm not here to discuss pinhole cameras and alternate processes, or even fine art and exhibitions. I'm here talking with you because the quality of the image means something to me. <br /> <br /> And so that brings us back to DLA, and DCF. Knowing where it is on your gear is a good first step, but not enough. I want to know what I'm trading off when I stop down further, as is often the case, for other considerations. I believe the trade off can be expressed meaningfully as "no larger than X by Y at full-frame at this aperture." <br /> <br /> As to why focus only on aperture, ... We can solve global warming and Bayer noise some other day. Aperture isn't the only choice or decision to be made, but it's the one that changed the most for me when I bought the 7D.</p>

[g dan mitchell]

<blockquote>

<p><em>Sure, the print is the final measure of image quality. No question about it. I don't exhibit in galleries, and I don't hawk them at craft shows. The ones that are good enough to print, I've tended to print as large as I can. I am therefore looking for the best image quality for the circumstance, and that includes sharpness of detail. I care very much about mush and its converse. With all due respect to anyone else's measure of goodness, I buy L series lenses for reasons other than their color scheme. Surely Canon doesn't make them just for me, in case I might want to buy one. I'm not alone in caring about sharpness.</em></p>

</blockquote>

<p>You won't get any less detail at f/11.6 on the 7D than on the 300D in your "large as I can print" images.</p>

<p>If you prefer to test with something other than a print, consider an onscreen jpg at some realistic size and do the same comparison - even though you won't be able to see the whole image on a laptop try 1000 x 1000 pixels.</p>

<p>The "diffraction limited" business never favors either camera over the other. It is absolutely neutral and is a lens property that is independent of whether you use a $50 used kit lens or a $2,000 prime - you get the same diffraction either way.</p>

<p>The camera that becomes diffraction limited at a larger aperture actually has an advantage if you intend to make big prints. If you compare the two cameras at a small aperture equal to the diffraction-limited aperture of the camera with "more megapixels" the effect on sharpness will be equal and might be the limiting factor to some extent assuming you use good lenses. If you use very good lenses and continue to open the aperture the "more megapixel" camera can continue to improve the resolution of the image - but the "fewer megapixel" camera will not get any better.</p>

<p>Dan</p>

 

[michael_young3]

<p>Dan, the only comparison I'm making is between what I'm now holding in my hand and what I'm now holding in my hand. I'm interested in capturing the most detail I can, if that's what I'm doing, consistent with the shot in front of me that moment. If it mattered all that much, I would have driven the pickup truck and unlimbered the view camera trunk instead. But I didn't and likely won't ever again. What matters that moment, when considering which aperture to shoot, if I decided that's what I care about, is whether f/7.1 or f/whatever else is more optimal. It's much simpler this way. I don't even carry the 300D.</p>