Does Nikon Need Some Good Canon Lenses?

[jamie_robertson2]

<blockquote>

<p>Lest I be misunderstood, yes, Canon, we will abandon you in an instant if you do not stop the price gouging and if you keep coming in second to Nikon where bodies are concerned! The D3X over the 1Ds III was one thing, but enough is enough! Come out with the best, or at least have the common courtesy to drop your prices.</p>

 

</blockquote>

<p>It's obvious from that statement that you're of the opinion that more pixels = superior camera. It's a ludicrous opinion. Why don't you ditch all your camera gear and buy that silly 42MP Nokia camera phone? After all, it has more pixels than the D800 so must surely be better.</p>

[Landrum Kelly]

<blockquote>

<p>It's obvious from that statement that you're of the opinion that more pixels = superior camera.</p>

</blockquote>

<p>Trying to hijack the thread, are you? See Scott Ferris' thread to discuss that and related issues:</p>

<p>http://www.photo.net/canon-eos-digital-camera-forum/00a5IN</p>

<p>Canon gear in general is increasingly over-priced. The Canon prices will have to fall. That is my point. There is no single right camera for everyone, and I have never made such a ridiculous claim as you suggest. You sound like a brand loyalist who cannot bear to see his team/brand criticized.</p>

<p>The 5D III does sound impressive, and it will be very useful for many photographers. I have the 5D II, however, and I do not see in the 5D III a compelling reason to upgrade at this time.</p>

<p>The more Nikon samples that I view, however, the more obvious it is that Nikon has truly hit a home run, both in terms of quality and in terms of price. I hope that Canon will follow up with a camera that is friendlier to landscape photographers and to bank accounts. Not all of us are shooting action pictures, and not all of us have a lot of money.</p>

<p>I would, for the record, like to see someone develop a lens that can nail the corners. This thread is primarily about recent product announcements in both camps as they relate to optical demands made by newer sensors. Except for the corners at very short focal lengths, I see very little in the D800 to criticize. I am not as happy with the 5D III.</p>

<p>--Lannie</p>

<p> </p>

[sravan]

<p>Lannie i hear that using the Canon 17 TSE you can take two photos with moving the camera - keeping the lens same place and get the angle of view of a 14mm. I believe this will give you a much better picture than any 14 mm including the much praised Nikon 14-24mm lens.<br>

I dont have this lens and so only going by my reading of forum posts from other owners.</p>

[i_liang_siu1]

<blockquote>

<p>Sad, isn't it? A wide zoom is the best wide angle lens that Nikon has?<br /> I rather doubt that.<br /> --Lannie</p>

</blockquote>

<p>Just curious, which Canon wide zoom would you consider to have less distortion* and better corner sharpness than the example you referenced?</p>

<p>*My understanding (which may or may not be correct) is that the distortion you refer to is actually an effect of capturing a very wide angle of view, and not due to any intrinsic lens deficiency. In other words, a theoretical lens with perfect optical design of the same angle of view would produce the same amount of distortion.<br>

** I just realized that Sarah had already posted a good explanation regarding distortion.</p>

[richard_chalfan]

<p>Several answers, like Sarah's, addressed the important points about distortion. But perhaps it would help to address the issue explicitly as there are two types of distortion you can see in a photographic image, and I think they are getting confused in some posts. There is rectilinear distortion and geometric distortion.<br>

Rectilinear distortion is one of the 5 "Seidel monochromatic aberrations" that can be present in your lens (there are also chromatic aberrations but they aren't relevent to this discussion). If the aberration "distortion" is present, straight lines in the object are not straight in the image. The commonly observed "barrel" and "pincushion" distortions are samples of the optical aberration, distortion. These aberrations are a result of the image scale, or magnification, not staying constant across the image plane. If the image scale increases away from the center of the frame you get positive, or barrel distortion. If the image scale decreases away from the center of the frame you get negative, or pincushion distortion.<br>

Geometric distortion is NOT an optical aberration, it is not caused by the lens but is caused by the geometry of the scene. Look at an example. <br>

Say you draw a precise circle a foot in diameter at your head height on a wall. Now if you back away (say 10 feet) from the wall but stay exactly opposite the circle it will still look circular (no surprise). Now, move 10 feet to the right and still stay 10 feet from the wall. Now look back to your left and look at the circle. It no longer looks like a circle, but it appears elliptical from your viewing angle and it is smaller because you are further away. That is geometric distortion. There is nothing "wrong", that is just how things really look off to the side, and that is how they will look in your image too.<br>

All lenses corrected for the aberration "distortion" will display geometric distortion. Just a fact of life.</p>

[Dieter Schaefer]

<blockquote>

<p>Nikon has truly hit a home run, both in terms of quality and in terms of price.</p>

</blockquote>

<p>As a Nikon user I don't doubt the quality part ;-) but in terms of price - Canon just goes were Nikon has gone before. Every Canon lens in recent years I can think of had been cheaper than the Nikon counterpart - so much so for the superteles that you could buy a 7D for the difference. With every introduction of a new lens that I can recall, Canon has aimed at equalizing that difference - in many instances "overcompensating".</p>

[Landrum Kelly]

<blockquote>

<p>My understanding (which may or may not be correct) is that the distortion you refer to is actually an effect of capturing a very wide angle of view, and not due to any intrinsic lens deficiency.</p>

</blockquote>

<p>I-Liang, have such limits been established purely deductively or empirically? I am always a bit skeptical when someone claims that something cannot be done. Next thing you know, someone has done it.</p>

<p>As for Canon v. Nikon zooms, I really have no idea which are better. I would think that a prime lens of either brand would be more likely to be able to nail the corners than a zoom, but it goes without saying that I could be wrong.</p>

<p>I do want to point out that I recognize that most shots which will be made with the D800 are not going to test the limits the same way that that library shot did. Most people are not even going to notice such problems in actual prints unless the prints are very big and persons are standing very close.</p>

<p>--Lannie</p>

[Landrum Kelly]

<blockquote>

<p>Lannie i hear that using the Canon 17 TSE you can take two photos with[out] moving the camera - keeping the lens same place and get the angle of view of a 14mm. I believe this will give you a much better picture than any 14 mm including the much praised Nikon 14-24mm lens.</p>

 

</blockquote>

<p>Sravan, that bears looking into. Thanks for that suggestion. If you are correct, I can buy one after my next convenience store heist.</p>

<p>--Lannie</p>

[Landrum Kelly]

<blockquote>

<p>Say you draw a precise circle a foot in diameter at your head height on a wall. Now if you back away (say 10 feet) from the wall but stay exactly opposite the circle it will still look circular (no surprise). Now, move 10 feet to the right and still stay 10 feet from the wall. Now look back to your left and look at the circle. It no longer looks like a circle, but it appears elliptical from your viewing angle and it is smaller because you are further away. That is geometric distortion. There is nothing "wrong", that is just how things really look off to the side, and that is how they will look in your image too.<br /> All lenses corrected for the aberration "distortion" will display geometric distortion. Just a fact of life.</p>

 

</blockquote>

<p>Richard, you might well be right, and I have no idea whether such problems can ever be overcome. Optics is one of the most difficult branches of physics that I have ever delved into, and I cannot see what might be possible <em>a priori</em>.</p>

<p>--Lannie</p>

[Landrum Kelly]

<p>Dieter, I guess that you are right, and essentially, the "problem" is that the demand is apparently there. I won't ever be able to afford an AF 500mm f/4 or AF 600mm f/4 of either brand. That is why I hang onto my old Nikon Ai-S manual focus 600 f/4--it is the only 600 that I will ever be able to afford, if indeed I can continue to afford to hang onto it. (I got it on eBay in 2006.) </p>

<p>This for me is an expensive hobby--and getting more expensive every day at just that time in my life when I have to start cutting back. I cannot cry too much, though. I still have my first digital camera, the Olympus E-20, as well as a Kodak PRO SLR/n, not to mention my Sony NEX-3, etc., etc. There will always be something that I can afford to shoot unless I become totally indigent.</p>

<p>--Lannie</p>

[sarah_fox]

<p>To bring Richard's explanation home, if you were to stand to the side of the circle on the wall (which appears as an elipse from where you're standing) and photograph it with your excellent Sigma 12-24 rectilinear lens pointed perfectly perpendicular to the wall, that circle would be projected as a perfect circle in your photograph, just as though you were viewing the wall from a much greater distance. If you were to photograph it with a fisheye, it would look like an elipse, just as you view it from where you're standing.</p>

<p>Now consider yourself standing in a room full of spherical Chinese lanterns hanging from the ceiling. From where you're standing, they all look "circular." Photograph them with a fisheye, and they will all look circular. Photograph them with your Sigma 12-24, and they will appear stretched in the corners and edges (quite a lot at 12mm!).</p>

<p>Generally speaking, a rectilinear lens is good at projecting from flat surface to flat surface and not as good with 3D objects, while the reverse is true of a fisheye.</p>

<p>FAIW, these are not so different from the issues involved when mapping a spherical surface (the Earth) to a piece of flat paper. There are several ways to do it, each of which introduce their own particular distortions. There is no perfect, natural way to map a sphere onto a plane.</p>

[Dieter Schaefer]

<p>Landrum, you are probably right in that one should never say<em> it can't be done</em> - but SLR wide-angle lenses are burdened by the fact that they are all retro-focus designs by necessity - there's got to be space for the mirror movement. This complicates the optical design - and certainly contributes to distortion and corner softness. Primes in general will show less distortions than a zoom at its widest setting (not too surprising considering the more complex lens design of the latter). Rangefinder lenses should do a bit better in that regard - but some don't play nice on a digital sensor because their rear is too close to the sensor and the incident light angle too shallow. AFAIK, the Voigtlander 15mm lens has very low distortion, but produces color shifts in the corner when used on a Leica M9.<br>

I am sure that a wide-angle lens could be designed with optimized corner performance and negligible distortion - what I don't know is what the price for such a lens would be and what other side effects (like maximum aperture of f/5.6, for example) it would entail. Some Zeiss and Leica wide angle lenses seem to do quite well in terms of corner sharpness; I have never seen one that is totally distortion free though. On the other hand, distortion in the range of +/-1% are usually negligible in everyday shooting - and many can be completely eliminated in post processing.</p>

[Landrum Kelly]

<p>Sarah, you sound as if you are saying that these problems can be logically seen to be insurmountable. The amazing thing, however, is that lens technology has advanced to the stage that it has. I am not sure what the theoretical limits really are. At present we seem to be seeing problems at perhaps 15mm or less. (I really do not know where the corners start to show obvious distortion.) </p>

<p>I do understand your analogies, but I remain unconvinced as to what constitutes a proof, especially when it comes to specific focal lengths. Could newer designs push the realm of what is considered acceptable into presently uncharted (or unattained) territory?</p>

<p>Dieter, my response to Sarah applies to you as well, I suppose. I am not saying that either of you is wrong. I simply do not know.</p>

<p>--Lannie</p>

[Landrum Kelly]

<p>Dieter, your remarks about post-processing do raise some interesting theoretical and practical questions--can post-processing do what lens technology cannot?</p>

<p>--Lannie</p>

[i_liang_siu1]

<blockquote>

<p>I-Liang, have such limits been established purely deductively or empirically? I am always a bit skeptical when someone claims that something cannot be done. Next thing you know, someone has done it.</p>

</blockquote>

<p>As Richard and Sarah explained, the distortion that you are describing is a product of the rectilinear geomerty design. It doesn't matter if you replace the Nikon with a Canon/Leica/Zeiss rectilinear lens of identical field of view - you'll see the same geometrical distortion. It's a product of projecting a curved surface onto a flat plane. <br /><br />Sravan suggested stitching two images from a shift lens as a way of achieving a simliar field of view with higher image quality. However, depending on how the stitched image is processed, it will either end up with the exact same rectilinear distortion, or it can eliminate the rectilinear distortion but then will introduce a different sort of distortion. In this link:<a href="http://www.cambridgeincolour.com/tutorials/image-projections.htm"> http://www.cambridgeincolour.com/tutorials/image-projections.htm</a>, compare the rectilinear vs cylindrical projections under the section Examples: wide horizontal field of view.<br /><br />The previously mentioned arguments are qualitative, emprical ones. If a rigorous, mathematical proof is necessary to convince you, I can't provide one.</p>

 

<blockquote>

<p>As for Canon v. Nikon zooms, I really have no idea which are better. I would think that a prime lens of either brand would be more likely to be able to nail the corners than a zoom, but it goes without saying that I could be wrong.</p>

</blockquote>

<p>That's a very valid point if you're arguing that a prime lens would out perform a zoom, but the title of your original post basically asks if a Canon lens (not specifically a prime) would do a better job than a Nikon lens (specifically the 14-24 zoom in the example you highlighted). My feeling is that there are some lenses in both the Canon and Nikon catalog that are fully capable of taking advantage of the increased resolution; they can be identified by reviewing various metrics (MTF, CA, etc) that have been posted on well known lens review sites.<br /><br />Or you could wait until respected photographers have had a chance to evaluate both cameras on multiple lenses and give their anecdotal impressions.</p>

[hjoseph7]

<p>The only major question I have is that I had time to jog around the block with my bad knee and all before this picture downloaded. Is it going to take this long for every picture that comes out of the D800 ? </p>

[michael_young3]

<p>Harry, I was impressed in the other direction, that the image was indeed large enough to challenge my networking infrastructure. It's a serious case of the sour-grapes and pixel-envy if one can look at this as anything but an improvement, if your interest is detail. 35 MP is a few pixels shy of double my 7D's 18 MP, and appreciably more than half again as much as each frame from a 5D2. Each image is almost as tall (the short direction) as mine are wide (the long direction). I would have to stitch two portrait oriented shots to equal one landscape shot from the D800, and likely still come with lower IQ (due to losses in the processing). Quite literally, a shot I can print at 8x12 can have the same detail printed at 19x13. Isn't this an improvement?</p>

<p>My 7D can be set to over-sample RAW images at 10 MP, rather than its native 18 MP. I have never ever seriously considered doing so. There are a few reasons why I *might* consider doing so, but I have never come across a reason compelling enough that I *would* do so. Does anyone really routinely capture images smaller than the camera is capable of?</p>

<p> </p>

[ilkka_nissila]

<p>Interesting choice for you to pick on the 14-24mm Nikkor. The 14-24mm easily has better performance than Nikon fixed focal length superwide angle lenses. Nikon developed some new technology to make it:</p>

<p><a href="http://imaging.nikon.com/history/scenes/25/index.htm">http://imaging.nikon.com/history/scenes/25/index.htm</a></p>

<p>Its performance was compared to the 14mm L II here:</p>

<p><a href="http://www.16-9.net/lens_tests/canon14l2_nikon1424/nikon1424_canon14l2_a.html">http://www.16-9.net/lens_tests/canon14l2_nikon1424/nikon1424_canon14l2_a.html</a></p>

<p><br />That said, if you're specifically looking for a sharp, distortion free image, the Biogon type wide angles (Hasselblad, Contax G, Zeiss Ikon and Mamiya 7 43mm and 50mm lenses) satisfy those criteria (0.03% distortion), but they do not perform well with digital sensors and they produce prolific vignetting at wide apertures even on film. The 14-24 has some barrel distortion at 14mm but the distortion is very low from about 17mm to 24mm. Vignetting is low and color uniformity excellent. The 14mm end is sharpest and 24mm is a bit weaker but still gives fixed focal length lenses a run for their money.</p>

[JDMvW]

<p>Do these sorts of wild assertions--Nikon bodies are best, Canon lenses are best, etc. -- really do any good at all? They are not even novel or interesting trolling. Remember the old law:</p>

<blockquote>

<p>At any given time T, camera marque A will have the lead in some respect over marque B.<br>

At time T+1, B will have the lead over A.</p>

</blockquote>

<p>There was a long time in which the top Nikon cameras had a max of 6MP, now Canon has only 23MP on the D5 mk iii, and Nikon has 30MP. Horrors, "I must change my whole camera system right now!"<br>

Those who change cameras every time one or the other pulls ahead, are doomed to change cameras often.</p>

[Dieter Schaefer]

<blockquote>

<p>your remarks about post-processing do raise some interesting theoretical and practical questions--can post-processing do what lens technology cannot?</p>

</blockquote>

<p>Nikon and likely also Canon (I just don't know) already have in-camera correction for lens "shortcomings" - like CA correction. Software often allows further corrections - like eliminating lens distortions (DxO is probably the most advanced one here, but given decent lens profiles, even photoshop does OK).</p>

<p>This thread is starting to confuse me a little - it had been my understanding that we were discussion lens properties - like corner sharpness and lens distortion - but not the geometric distortion that is inherent when projection a 3D object onto a 2D surface.</p>

<p>My understanding of the "theoretical limits in lens design" is that the lens designer has a complicated system to deal with that requires careful balancing of different abberations and other "lens errors" in order to arrive at a design that suitable for the job and fits within the design parameters. Very often, minimizing one or two lens parameters goes at the expense of making others appear more pronounced. The fact that with digital cameras another optical system is within the optical path (the microlens array on the sensor) adds further complications to the lens design process.</p>

<p>I tend to look at it a different way - if distortion-free and "sharp-in-the-corner" lenses were easy to design and manufacturer at a reasonable cost - then it would be done (or would have been done already). I believe that whatever high-end lens appears on the market today is the best possible under a set of given circumstances. A different set of optimization parameters and a different set of priorities may result in a different lens - the manufacturer needs to make the decision which route to take.</p>

<p>I owned a Novoflex 400mm and 600mm rapid-focus system - both lenses did fairly well on film but were a disaster on my D200 - CA was horrible. Those lenses were simple designs, a triplet for the 400 and a single lens (or a duplet, I forgot) for the 600 - out of necessity they were as long as their focal length (not tele designs). Much more complicated designs with speciality glasses are used in today's high-end teles; AF and IS/VR further add complications to the lens design. Rather than complaining that a photographic lens system isn't perfect, we should be happy with the fact that they have become as good as they are now.</p>

[Landrum Kelly]

<blockquote>

<p>Interesting choice for you to pick on the 14-24mm Nikkor.</p>

 

</blockquote>

<p>Well, Ilkka, there is a silver lining to my criticism: the library picture was so perfect that I was almost stunned to see the distortion in the corners. Truth be told, I did not know which lens it had been shot with. I guess that I got a copy of the file from some other source and did not know what the EXIF data was, and I didn't run it through PS to find out for myself what focal length was used. Yes, it looked wide, but I had no idea how wide it really was!</p>

<p>Knowing now just how wide the lens actually was certainly tempers my questioning, benign though it was. Now that the question has been raised as to how (or whether) a lens can be made much better, I am all the more interested in <em>that</em> question. The library photo was truly an impressive photo. I was really quite in awe of it, and still am.</p>

<p>Sometimes a thread takes one in a direction that one could not possibly have anticipated when one started it. When Sarah posted the fisheye shot with her explanation, I knew that I was into some very interesting territory, territory that is much more interesting than my original question. Sarah has quite a mind, and she has still left me thinking about the problems of optics.</p>

<p>There is a reason that we hang onto some of our Nikon glass when we switch to Canon or some other brand, as I did when the 5D came out and Nikon was not indicating any strong commitment to ever going full-frame. The reason? Well, if you shoot Nikon, I don't have to tell you. . . .</p>

<p>That said, I don't expect to switch back again, although I do have a D90 which I shoot just for fun. I feel sure that further developments from Canon will lead into the same territory that Nikon has now covered.</p>

<p>--Lannie</p>

[Landrum Kelly]

<blockquote>

<p>Do these sorts of wild assertions--Nikon bodies are best, Canon lenses are best, etc. -- really do any good at all? They are not even novel or interesting trolling.</p>

</blockquote>

<p>JDM, if you are talking about my original question, that was not the point at all. I thought that I had bent over backward to express my awe at the D800. As for the lens, as I just told Ilkka, I did not know just how wide it really was when I wrote the original post. Silly me.</p>

<p>--Lannie</p>

[sarah_fox]

<blockquote>

<p>Sarah, you sound as if you are saying that these problems can be logically seen to be insurmountable.</p>

</blockquote>

<p>Well, I wouldn't quite put it that way, because I don't consider the particulars of these geometries as "problems." They are what they are. Of course you can always transform a photo from one geometry to another through an app such as PhotoShop, usually with somewhat mediocre results. However, just understand that a geometry is what it is. </p>

<p>Think of it this way. Mount your Sigma onto a FF camera and dial it out to 12mm. Stand in front of a large painting, making sure to keep the camera square with its orientation. Frame it up so that you fill the frame with the entire painting. Take your picture, pull it up on your computer, and give it a good look. Absolutely nothing will look distorted. The lens is doing its job correctly. </p>

<p>Now line up your family for a group photo at 12mm. Be certain your wife is standing either on the left or the right. Print and frame the picture, hang it on the wall, and prepare to sleep on the couch. She even won't have to ask, "Does this photo make me look fat?" ;-)</p>

<p>It's only when you photograph something 3-dimensional (or something that lies in a plane that isn't perpendicular to the axis of the lens) that you will see distortions. In that case a different geometry of lens (fisheye) might (or might not) be preferable, depending on what you are photographing. Or perhaps the distortions are kinda' cool. A lot of people love UWA rectilinear landscapes, with swoopy clouds in the margins! </p>

<p>In any event, UWA photography of any geometry is problematic (and always will be, despite technical advancements), because you're taking a hemispherical scene and re-mapping it to a flat surface. Something has to stretch and distort somehow to make that happen. (And a skilled photographer will make absolutely certain that it's not his/her significant other!)</p>

[Landrum Kelly]

<blockquote>

<p>Rather than complaining that a photographic lens system isn't perfect, we should be happy with the fact that they have become as good as they are now.</p>

 

</blockquote>

<p>Dieter, I'm not whining. Every day presents me with new delights, having been a resolutoin freak since the age of thirteen when I first started splitting double stars. I had to deal with my share of chromatic aberration in those days as well, in spite of the fact that the objective on that first telescope was said to be "achromatic." My present small refractor has an apochromatic objective, and it is awe-inspiring. The lens designers keep doing things that I never thought possible. I am still in awe of all of it.</p>

<p><strong>I am simply glad to have lived long enough to have seen all these developments in both lens technology and digital imaging--and that is from the heart. </strong>Brand names be damned. We live in interesting times, and my only regret is that I will not live long enough to see everything that is coming where imaging and communications technology is concerned--and who knows what else. Gosh, I feel left out already. . . .</p>

<p>As for [several] changes in the direction of the thread, well, I have been delighted there as well. Sarah's fisheye shot brought me out of a chair with a benign (but not mild) expletive which I shall not repeat here. Her lasting intellectual challenge in her explanation might keep me up nights for a while. </p>

<p>Let's face it, though. Given the weakness of my original post, the thread had no way to go but up, and I take no credit for the fact that it has improved vastly as it has gone along.</p>

<p>--Lannie</p>

[Landrum Kelly]

<blockquote>

<p>It's only when you photograph something 3-dimensional (or something that lies in a plane that isn't perpendicular to the axis of the lens) that you will see distortions.</p>

 

</blockquote>

<p>Thanks, Sarah. I think that I finally, finally, finally get your point. Could I hire you as a personal tutor?</p>

<p>--Lannie</p>