jeff_conrad

<p>The dpreview link that Philip posted is particularly good because the example at the end is one of the few that illustrate both the aberration/diffraction tradeoff in the plane of focus <em>and</em> the defocus/diffraction tradeoff at the DoF limit. The topic actually was covered in great detail years ago by H. Lou Gibson and Lester Lefkowitz, but most recent analyses have seemed to examine only the situation in the plane of focus, which is only one consideration in landscape work.</p>

<p>The tradeoff between DoF and sharpness in plane of focus increases with magnification, so the tradeoff for most landscape work will be much less than that for the dpreview example. For many situations, simply set the aperture for the required DoF, and diffraction won't be an issue.</p>

<p>Although Live View is a great feature, especially for T/S lenses, it's interesting that large format and small format have gone in opposite directions here.</p>

<p>The DoF calculator on Sinar LF cameras was a feature adapted from small-format lens DoF scales; it was added to replace stopping the lens down and viewing through a loupe on the groundglass. It wasn't really any better, but it was a lot faster, immediately giving the optimal focus and the required f-number (which depends on having optimal focus).</p>

<p>Small-format cameras with AF lenses have done just the opposite, going from quickly determined settings using the lens DoF scale to visual sharpness assessment. If done carefully, it should work just as well, though the f-number won't be optimal unless the optimal focus has also been found, which requires additional fiddling. The small-format camera has the great advantage of using a larger aperture for the same DoF, so the viewfinder and LCD are much brighter. Prior to Live View, the screen was so small that it was tough to see whether something was really sharp. With the ability to use 10x magnification anywhere in the image area, it's a lot more manageable.</p>

<p>So indeed Live View is a great feature, but it's now the only tool available. It eliminates most of the guesswork, but it's much slower than using DEP, which took only a few seconds.</p>

<p>Diffraction on crop-format cameras is an overrated problem (extreme closeups excepted). Diffraction and defocus blur both essentially scale with format, so the tradeoff between the two is the same for 8x10 as it is for APS-C. As Daniel indicated, the key is to scale the aperture (and preferably also the focal length) as well.</p>

<p>Diffraction is a real phenomenon, and it's easy to see if you look for it on a monitor at high magnification. But under normal conditions of enlargement and viewing, it's often imperceptible.</p>

<p>In the plane of focus, there is a tradeoff between aberrations and diffraction; for most small-format lenses, diffraction is the greater problem for apertures smaller than about f/8 or f/11. At the DoF limits, there is a different tradeoff, between defocus and diffraction. The optimal aperture depends on the specifics of a scene: the greater the required DoF, the smaller the optimal aperture for the DoF limits. In most cases where DoF is of interest, the optimal aperture for the DoF limits will be smaller than that for the plane of focus. But if both are imperceptible, it doesn't really matter.</p>

<p>As Daniel mentioned, the best way to avoid diffraction is to use the largest aperture that will give the desired DoF. On a view camera, this can be determined directly by measuring the difference between the near and far image distances. On a manual-focus small-format lens, it can readily be determined using the lens DoF scales. With an AF lens, it's not quite as simple; Canon's DEP mode (the real one, not ADEP) did this quite well, which is why it's unfortunate that it isn't provided on current models. The only practical method then is to visually assess DoF sharpness; it's more of a chore than it should be, but Live View is a big help.</p>

<p>In any event, there's no reason not to use a 7D or any other APS-C camera for landscapes. It's just a slightly smaller format than FF 35 mm, and won't capture quite as much detail. Much as FF 35 mm doesn't capture the same detail as 4x5 ...</p>

<p>Tim, you're not the only one confused. Background blur is definitely trickier than it seems ... I misspoke on getting the same background blur with both the APS-C/24 mm lens and the FF/35 mm lens. When the subject <em>distance</em> (rather than the subject <em>magnification</em> ) remains constant, the final-image background blur is proportional to format size. So with the 24 mm lens on APS-C, you'd get 0.625 the background blur that you'd get for the “same picture” (i.e., same angle of view) with the 35 mm lens on full-frame 35 mm.</p>

<p>As several others have mentioned, you'd get about the same DoF as you would with a 35 mm f/2.2 on full-frame 35 mm.</p>

<p>The background blur is trickier than it may seem. Although the background blur spot is indeed larger (it's roughly proportional to focal length), the magnification of the background object is increased by the same factor as the blur spot, so it may not appear more “blurred.” This is demonstrated pretty well by the two photos in Bob's article. The image on the left has greater blur, but the background is arguably as recognizable as in the image on the right. The longer lens still has an advantage, though: the angle of view is narrower, so it's easier to crop out distractions such as the white blob in the photo on the right.</p>

<p>The situation changes a bit when format is changed. Assuming everything is the same except for format (i.e., the focal length is scaled to maintain the same angle of view), the blur spot is smaller with the smaller format. But the smaller format needs greater enlargement, so the background blur in the final image is the same for both formats.</p>

<p>Though given scant coverage in most treatments of DoF, this issue actually has been discussed. Merklinger touches on it in <em>The INs and OUTs of Focus,</em> showing that separately controlling DoF and recognizability of background objects isn't possible without changing the subject-to-background distance. The topic is also covered in the Wikpedia article on <a href="http://en.wikipedia.org/wiki/Depth_of_field">Depth of Field</a> , and illustrated in Paul van Walree's page on <a href="http://toothwalker.org/optics/dof.html">Depth of Field</a> . Nonetheless, DoF and background blur remain among the most misunderstood areas of photography, especially when comparing different formats.</p>

<p>So, to repeat, you'd get 1.6x the DoF, but about the same distant background blur. If the blur is the main objective, this should be fine.</p>

<p>I agree with Peter that changing the orientation of the tilt and shift functions is no big deal, but it's not something you want to be doing very often, especially in the field. For landscape work, changing the orientation once will probably suffice if you even need to do it. For one who does different kinds of work, the independent rotation is a nice feature. I know of someone who carried two of the original TS-E 24 mm lenses, one with the orientation of the features each way. He's obviously delighted with the new version that allows him to carry only one lens. I can count the times I've used tilt on the 24 on one hand, so it's not such a big deal for me.</p>

<p>Though the independent rotation would be a nice feature, it's hardly essential I've done fine without it, and quite honestly don't need to use both tilt and shift all that often. YMMV, of course.</p>

<p>Yet another alternative would be a <a href="http://www.hartblei.com/">Hartblei</a> Super-Rotator, which has had independent rotation for years. I've never used one, so I can't comment on the IQ. I also don't know how easy it is to find a used one.</p>

<p>Again, if you need the tilt and shift functions, the TS-E 45 is a great lens.</p>

<p>I fully expect Canon to eventually release new versions of the TS-E 45 and the TS-E 90 (or thereabouts). In particular, the ability to independently rotate the tilt and shift functions would be a great improvement—for what I do, it would be far more useful than on the 17 and 24. But the new versions may come at commensurate prices, and we don't know when (or even if) this will happen. So I agree with Mark: if you need the movements, just get the lens and don't worry that it may not be perfect. Nothing is.</p>

<p>The lens was first released in 1991, so it certainly wasn't with a digital sensor in mind. It's nonetheless a good lens, as most of the linked reviews indicate. There definitely was some vignetting with the original 24, especially wide open at full shift, but it's not nearly as bad as some might have you believe (I've had mine for 15 years; it's a good, if perhaps not great, lens). Vignetting isn't an issue with the 45 because the angle ov view is much narrower.</p>

<p>It boils down to this: if you need either tilt or shift on a lens of that focal length (and I use both), you don't really have much of a choice. The essential question is what you plan to use it for.</p>

<p>You might find quite a line ahead of you ...</p>

<p>Bruce Miur wrote:</p>

<blockquote>It sounds like you are saying you cant control depth of field without a DEP mode...</blockquote>

<p>Well, you can, after a fashion. Obviously, you can focus on something and adjust the aperture to change the DoF. And for some types of photography, this is just fine. For other types, such as landscape or architectural, it's often desirable to have the DoF extend between a near and far distance (with the latter often at infinity) at the minimum possible f-number.</p>

<p>With current models, you can certainly play around focusing at various distances and looking at the DoF scales (to the extent that they can be read) until the DoF extends between the desired distances, but this certainly seems a clunky procedure compared with DEP mode.</p>

<p>The K1000 obviously didn't have DEP mode. But with most manual-focus primes, setting the DoF as above was easily done using the lens DoF scales. The technique was sometimes known as “zone focusing,” used often as a means of “pre-setting” focus and f-number when there wasn't time to do so when the picture was actually taken. The settings were often made just by glancing at the DoF scales, but the distances could also be determined visually by focusing on the near and far points. The great advantage of this technique is that the process is deterministic; if the desired near and far limits are known, the focus and f-number directly follow.</p>

<p>With most AF lenses, the DoF scales are so small as to be almost useless; moreover, AF lenses need to be focused manually to set the focus deterministically for given DoF limits. The DEP mode works much like visually finding the near and far distances with a manual-focus lens DoF scale, except that it's much faster and easier. Results seem about the same as those obtained from MF lens DoF scales. Despite the alleged 7/17 split of the DoF (which never really was completely explained), some fairly extensive testing seemed to indicate that DEP mode actually split the image distances equally, just as one would want. So DEP mode worked better than Canon led us to believe.</p>

<p>DEP mode hasn't been included in any of the EOS digital bodies since the original EOS-1Ds. Some of the other digital bodies include ADEP mode, which is similar to DEP except the camera chooses the near and far points. This is probably fine for shots such as group portraits where everyone is within the AF-point area, but not so good for landscapes because the desired near and far points are often outside the AF-point area. And in any event, photographer has no control over the points chosen.</p>

<p>My initial statement was probably unnecessarily dismissive. Properly, the current models do not provide a means of deterministically setting the focus and f-number so that the DoF extends between desired near and far limits at the minimum possible f-number. But if one does this on a substantial portion of his shots, both statements work to about the same end.</p>

<p>Allegedly, DEP mode was dropped because the firmware required too much memory. I dunno ... In my (and probably Canon's) experience, only a small percentage of people used or even understood DEP mode. And many photographers today have probably never even used a manual-focus lens, let alone done “zone focusing” using the lens DoF scales. So Canon may have simply eliminated a feature that they thought no one used.</p>

<p>Perhaps Canon also think no one uses “zone focusing” even on MF lenses. The new 17 mm and 24 mm TS lenses have great optical performance, but the DoF scales are very small and hard to read, very much in contrast to the very good DoF scale on the old TS-E 24. I hope the new 45 mm and 90 mm TS lenses (if they're ever introduced) include more usable scales.</p>

<p>What this country needs is a good EOS-3D ...</p>

<blockquote>I think I will like the DEP setting, but I haven't used it yet.</blockquote>

<p>One of the best features the Canon EOS line ever had. Try it and learn to use it well. Dropping DEP was one of the worst decisions Canon ever made.</p>

<blockquote>The last film SLR I owned was a Pentax K1000.</blockquote>

<p>One of the all time classics. The EOS 3's DEP mode will let you control depth of field just as you were able to do with the K1000. Something you can't do even with the current gigabuck EOS-1<em>x</em> series.</p>

<p>Correct. For the most part, the effect of diffraction in the final image is independent of format. It becomes the limiting factor in any format when you need huge DoF, especially at high magnification. But diffraction shouldn't be an issue at all under the conditions you've mentioned.</p>

<p>Expanding a bit on Bob's comment: the difference between APS-C and FF 35 mm is much less than between FF 35 mm and larger formats. With a 5D Mk II, you will get 0.625 the DoF that you get with APS-C. But the 5D Mk II's improved noise at higher ISO may by itself make up for the difference (and perhaps then some), as Sheldon mentioned. And if you ever need a wider angle of view or a shallower DoF, you have it, in addition to many other benefits. I wouldn't let the slightly less DoF with FF deter you if that's really where you want to go. As Hal said, people shot FF for quite some time and managed to survive. You're likely to do so as well. And probably then some.</p>

<p>For a given lens, the angle of view changes with format, but the focal length doesn't. But as I said, it all depends on what is kept constant. If you keep the same lens, subject distance, and final-image size, you <em>do</em> get less DoF with the smaller format. Why? The magnification doesn't change, but the image from the smaller format needs greater enlargement, so the circle of confusion is necessarily smaller. Of course, when you keep the same subject distance, the angle of view is less, so you don't get quite the same picture. If you reduce the focal length to get the same angle of view, you get more DoF with the smaller format. Why? The effect of reducing magnification is greater than the effect of the greater englargement.</p>

<p>Again, article on DoF that I linked above shows the interaction of the various factors quite well (though it doesn't explicitly cover the case of keeping the same subject distance.</p>

<p>It's questionable whether you get a greater usable aperture range with FF. Again, it depends on what you assume. The larger format can use a smaller aperture before diffraction becomes a problem; again, the smaller format needs greater enlargement. But the larger format requires a smaller aperture to get the same DoF, so there's greater diffraction. So you end up with the same diffraction blur in the final image.</p>

<p>At the DoF limits, the tradeoff is between defocus and diffraction. In the plane of focus, the tradeoff is between aberrations and diffraction. With the same lens, the optimal aperture in the PoF doesn't change with format; if you consider the optimal aperture for the PoF as the minimum, you might actually get a slightly greater usable range of apertures.</p>

<p>But the optimal aperture for the PoF is usually in the middle of the range (e.g., f/8 to f/11), which doesn't apply if you're usually working close to wide open. For much practical photography, the loss of sharpness in the PoF when you use a smaller aperture for more DoF isn't even noticeable under normal viewing conditions. If you blow the image up on a monitor and go looking for trouble, it's usually easy to find. But is that really how images are usually viewed?</p>

<p>The tradeoff between PoF sharpness and DoF can be significant at magnifications greater than one. But it's usually not a big issue in general photography. So I don't think it's a big factor here in making the choice.</p>

<p>It's really pretty simple. HTML has no math facilities, so it would be much easier to look at the the Wikipedia article on DoF linked above, which covers it pretty well. That article also derives DoF equations from scratch if you want to get into that much detail. The Circle of confusion article covers choice of CoC if you want more detail on that than the DoF article provides.</p>

<p>Bear in mind that the relationship is approximate, valid perhaps between about 10 times the focal length of the larger format and about 1/3 the hyperfocal distance of the smaller format (so the range of validity varies with aperture). But it's still useful for a general comparison.</p>

<p>All of the Canon TS lenses, including the oft-unloved 45, can be very useful in certain situations. But a TS lens should solve a problem before you fork over the considerable cash to get one.</p>

<p>What are you now having trouble with that you think a tilt/shift lens will fix? Moreover, what focal length(s) are you using when you're having this trouble?</p>

<p>It seems to me that we're discussing the DoF that obtains from different format sizes.</p>

<p>It's more than just the focal length; were this the only factor considered, the DoF would be in inverse proportion to the <em>square</em> of the format size. When the enlargement (and hence the CoC) is also considered, the relationship is only inverse linear. So yes, size <em>does</em> matter.</p>

<blockquote>These wider angle optics inheritantly give greater depth of focus at any given aperture.</blockquote>

<p>It's not quite that simple. If the subject distance is held constant (so that magnification with the wider-angle optic is reduced), the DoF at the same f-number increases. But if the final-image size is the same for both formats, the greater enlargement of the image from the smaller sensor also comes into play.</p>

<p>Whether short-focus lenses give inherently greater DoF depends on what is assumed. If the subject distance doesn't change, the short-focus lens does give greater DoF, as previously stated. But at the same subject magnification, all lenses give approximately the same DoF except at distances approaching hyperfocal (where a short-focus lens does give slighter greater DoF).</p>

<p>DoF depends on the subject distance, lens focal length, lens f-number, and the acceptable circle of confusion. The circle of confusion depends on the viewer's visual acuity, the viewing conditions, and the enlargement of the final image. How these interact is described in this Wikipedia article on <a href="http://en.wikipedia.org/wiki/Depth_of_field#DOF_vs._format_size_2">DOF</a> ; Factors affecting the choice of CoC are discussed in the article on <a href="http://en.wikipedia.org/wiki/Circle_of_confusion#Circle_of_confusion_diameter_limit_in_photography">Circle of confusion</a> .</p>

<p>To summarize: when essentially the same picture (i.e., same subject magnification) is taken in two different formats using the same f-number, the DoF is approximately in inverse proportion to the format size, as has been said. So the APS-C format has approximately 1.6× the DoF of FF 35 mm. Subject magnification can be held constant either by adjusting focal length to keep the same angle of view or by adjusting subject distance to maintain the same field of view (though the perspective will obviously be different).</p>

<p>Yakim,<br>

It's still not clear why you want a TS lens. As several others have said, the most common application for a 17 mm length, even with APS-C, is the prevention of converging parallel lines, and you've indicated that this wouldn't be of much use to you. It's possible to use tilt with a lens that wide, but you need to be doing some pretty extreme near/composition for the tilt to be of much benefit. This isn't to say that such situations don't arise—see, for example, <a href="../nature-photography-forum/00UXL1">this thread</a> , which involved a PC-E 24 on a D700, and would be comparable to what you'd get with a 17 on APS-C. But you've also indicated that you seldom do landscapes, so I'm not sure I see the benefit there, either.</p>

<p>Although you didn't mention it, perhaps you're interested in selective focus for portraits. Some great work has been done with this, and has been discussed in this forum. But there also seems to be a fascination with “tilt-shift” (whatever it may be) for its own sake, and I think that novelty will fade before too long. In any event, if you want selective focus for portraits, you'd probably be much better off (and much money ahead) with either the 45 mm or 90 mm TS lens.</p>

<p>There's nothing wrong with TS lenses—I've had three of them for years, and find them invaluable in many situations. But I shoot mostly landscapes and architecture, using tilt and shift in the conventional manner. I also almost always use a tripod; I wouldn't say it's impossible to set tilt handheld, especially if you set it to the max for selective focus, but trying to maximize sharpness handheld would seem to be adding an unnecessary obstacle.</p>

<p>As I have said here many times: I can't see getting a TS lens just because it may be cool. Get one because it solves a specific problem you're having with what you currently have. Otherwise, I'd probably avoid scratching the itch unless you have a lot of extra scratch.</p>

<p>One needs to be very careful when using “stop.” If <em>stop</em> really does mean detention under circumstances of <em>Terry v. Ohio</em> , there needs to be an articulable suspicion of criminal activity. For the folks in the Lexus, speeding alone justified the stop, and attempted flight essentially demanded it. The cause to stop in the second two cases is far less obvious; from the description, though, it doesn't sound like either was actually a stop.</p>

<p>It's not uncommon to use <em>stop</em> loosely. But the distinction <em>is</em> important; careless use of <em>stop</em> last fall cost Palo Alto, California Police Chief Lynn Johnson her job. The distinction often gets lost in some of the endless discussions in this forum, as in <a href="../street-documentary-photography-forum/00UVVh">this thread</a> , where the OP wasn't stopped at all.</p>

<p>Again, “suspicious” isn't at all the same as “casing a job, a stickup.” People may differ on whether photographing a government building or an industrial complex is “suspicious,” but absent something else, it's not a basis for detention. Not even close.</p>

<p>This behavior isn't new to NYC. In 1999, the New York Attorney General's office released a <a href="http://www.oag.state.ny.us/bureaus/civil_rights/pdfs/stp_frsk.pdf"> report</a> (PDF) on the NYPD's “stop-and-frisk” practice. Unsuprisingly, the report noted that there sometimes are disparities between what's legal and what actually happens. Police everywhere have long stretched the law when they felt it impeded doing their jobs, and in many cases, it probably is motivated by a genuine concern for officer safety.</p>

<p>What I find disturbing is that the article states, “The practice is perfectly legal” without mentioning the conditions to which it is subject. A stop is lawful if the police have specfic and articulable facts that suggest that the person stopped is involved in a crime; a frisk is lawful if, in addition, police have specific and articlulable facts that suggest the person stopped may be armed and dangerous. A good starting point for an example is <em>Terry v. Ohio</em> : Detective McFadden observed three men who appeared to be “casing a job, a stickup.” The situations described in the article don't seem to come even close; in fact, photography might seem suspicious by comparison.</p>

<p>Fortunately, the police in the SF Bay Area don't seem quite so aggressive, and in particular, don't generally seem to harass photographers, as Eric mentioned. In 30 years, I haven't even been approached by the police. So far, anyway.</p>

<blockquote>

<p>The centre of the iamge is now on the edge of the lens image circle</p>

</blockquote>

<p>I don't think that's the problem—the diameter of the image circle is given as 58.6 mm; its radius is 29.3 mm. The maximum shift is 11 mm, so the center of the image isn't even close to the edge of the circle.</p>

<p>Like Robin, I've hardly ever found the need for exposure compensation using the TS-E 24. Full disclosure: I usually use spot metering, and with the 24 I usually use an external meter to get a small enough spot size to measure what I want. But I meter as I would for any other lens, so again, I don't think it's an issue with the lens. As Robin also mentioned, it's important to meter essentially the same image framing that you will use in the picture, and I'd say this is true with any metering pattern. The only way to do this is to meter with the camera pointed up to approximate the image framing you get with the shift applied.</p>

<p>You will get noticeable vignetting at the corners, just as you would with any ultra-wide-angle lens. At full shift in the long direction, corners are as far from the lens axis as with a 17 mm lens, leading to a 2.9 EV light loss from cos⁴ alone. With the original TS-E 24, you're also at the edge of the image circle, so there may be some additional losses from mechanical vignetting. This should be less of a problem at f/11 that it would be with the lens wide open.</p>

<p>From what you described, it's tough to tell whether what I've just discussed applies. Again, I'd try metering the same scene with another lens and see how that compares with the TS-E 24 just to ensure that there isn't some problem with the information transmitted from the lens to your 5D.</p>

<p>To repeat: though use of full shift, especially in the long direction, results in noticeable darkening in the corners, I've not found that shift per se requires any exposure compensation.</p>

<p>From what you've described, your technique seems fine—you need to meter before applying tilt or shift, which you've done. What metering pattern are you using?</p>

<p>If you have another lens (the closer to 24 mm, the better), you could take a reading of a typical outdoor scene using the TS-E 24 and another with the other lens; If you get the same readings, especially if both are close to the “sunny 16” rule, the lens is probably working properly.</p>