TS-E 17 mm for landscapes/nature

Discussion in 'Canon EOS' started by peter_e, Sep 26, 2010.

  1. Does anyone here have user experience with the TS-E 17 mm tilt-shift lens in landscape and nature photography? Given the large depth of focus at 17 mm, I would have to be very close to the foreground to take advantage of tilt. I calculate that, without tilt and at f=11, the hyperfocal distance on an EOS 5Dii is from about 1.5 ft to infinity. I can see some potential interesting shots, but how useful is it in practice? I do have the 14 mm and the 45 and 90 TS-E lenses and use all these lenses a fair bit. I don't do enough architecture to justify the expense of the TS-E 17 mm. Thanks for your comments.
  2. I rented one a few months ago (I mainly shoot landscape) and was blown away. I was torn between the 14 F2.8 and the 17 F4 TS. I found that I mainly used shift when shooting landscape as left shift for one shot plus right shift for the other gave me about a 14mm angle of view when the two images were merged (I use this technique with my Fuji GX680 where the wide angle lenses are not that wide). The lens is an outrageous price but it is very sharp - even at F4. Rent one and try it - mind you I have been saving AMEX points ever since (I am nearly there!)
  3. Peter,
    The 17mm TS-E is the best ultra wide Canon make, if 17mm is a core focal length for you the lens is worth the money regardless of the tilt and shift.
    Now your specific question is about tilt, but I'll get onto that shortly. The shift function, for landscape/nature work, is worth the money on its own! Try getting undistorted horizons when placed at any position of the frame with any other 17mm.
    Now the tilt, anybody that has read any of Harold M Merklingers quintessential work on focusing and dof will know the arguments against using hyperfocal focusing. Basically, using hyperfocal focusing guarantees your work will be to a low standard, infinity will be barely acceptable, that is the given. It is not the way to get the best out of your equipment. Tilt is. Tilt will give you pin sharp images from zero to infinity at f4, with judicious use of aperture you can actually have perfect focus throughout your entire frame. For large prints from small cameras tilt helps immensely, for large prints from larger cameras, it helps too :)
    Stitched horizontal panoramas from the 17 TS-E equate to an 11mm FOV, way wider than your 14mm. Even when stitched out to 14mm the 17mm TS-E makes the 14mm prime look like the $2,000 Lensbaby that it is.
    Get one, take the time to learn how to get the best out of it and don't look back. If you hate it (you won't) then sell it for a small loss, about the cost of a few weeks rental. Just don't tell every wide angle landscape shooter how good it is.
  4. For all of the above reasons I would love to have a 17 TS-E for landscapes. I used a Canon FD 17mm f4 lens for 15 years and switched to a Nikon 14mm f2.8 five years ago for it's wider and more dramatic capabilities. It would not be an easy decision to go back to 17mm but the TS-E would be extremely compelling if I could afford it. I would certainly take one before the 24, 45 or 90.
  5. I have been using the TS-E 24 II extensively this year. It's not as wide as the 17 mm lens, but I find tilt to be
    indispensable. When shooting the camera at eye level I rarely tilt more than one degree, but that renders the foreground
    and background very sharp. When working close to the ground I tilt q bit more.

    You can use the hyperfocal distance, but your background isn't really in focus as it is when tilt is used properly. Further,
    the HFD requires smaller apertures which add diffraction to the image.

    Tilt does not work in all cases as when the image contains near and far vertical objects (as in a grove of trees). But when
    it works the effect is amazing. And of course shift is always useful especially in the vertical (rise/fall) dimension. The TS-
    E 17 is high on my wish list.
  6. Thanks for all the comments. You all made a good case for it!
    In defense of the 14 mm, mine is very sharp corner to corner and I'm very happy with it. I've got my share of Canon-branded paper weights in my closet, but the 14 mm is not one of them.
  7. I am thinking of getting TS lens...
    should i start with 17 or 24?
    May i use 17 on 7d for 1.6 and on 5d for wider? In this case, i cover both 17 and 24 in a way?
  8. Not everyone who has read The INs and OUTs of Focus necessarily agrees with Merklinger that infinity focus is the best choice for general photography. Infinity focus obviously improves sharpness at infinity, but does so at the cost of foreground sharpness, as can be seen in the image at the end of Chapter 6. Moreover, the illustrations of the improved infinity sharpness use extremely large magnifications, more than would normally be used except perhaps for surveillance or pixel peeping.
    Ultimately, the choice of infinity vs. foreground sharpness is a personal choice. There are many, including Ansel Adams, who recommended sacrificing the former rather than the latter. So there's nothing necessarily wrong with “hyperfocal” focusing.
    It should also be noted that tilt suffices to get everything sharp only if the subject is planar; for anything not on the plane of focus, you rely on the DoF, just as without tilt. The difference when using tilt is that the DoF is wedge-shaped, and for a common scene that has more height in the background than in the foreground, tilt can allow you to get the same overall sharpness with an f-number a couple of stops smaller. My experience with the original TS-E 24 is that I seldom need the tilt, but if you do extreme near/far compositions it may be helpful.
    If your landscapes include trees, the shift can be very useful to avoid converging trunks (or diverging trunks, famously described here by Frank Sheeran as “acid-trip trees”). It's also helpful when you photograph a well-known landmark (like El Capitan) that people expect to have the correct shape. It's something you can correct in postprocessing, but it's extra work and you lose some quality. In architectural photography, preserving the shape of a building is often important, so correcting converging verticals requires two operations: a trapezoidal stretch to eliminate the convergence and a vertical stretch to restore the aspect ratio. For many landscapes, the aspect ratio isn't as important, so the second operation can often be skipped.
  9. Christopher,
    Yes you can, though the shift control is tight on cameras with built in flash. What focal length do you use most, a 24 or 17? That is your answer. However, you can use a 1.4x or 2x TC with either to further enhance your focal lengths.
    The 24 does have several advantages over the 17, you can use filters on it (both screw on and square), and a lens hood, both possibly important for some shooters and you can leave a filter on it for front element protection should you want to. The 17mm has a very exposed front element that is prone to flare, no hood and no integrated way of attaching filters. 17mm is very wide! The 24 is better at the popular "toy" reverse tilt type image too. The 24 is sharper as well but both are so good it is a small factor. I would think 24mm sales far out strip 17mm sales, but they are both very very good.
  10. Jeff,
    It doesn't matter if you subscribe to the infinity focus or hyperfocal focus technique, or even, as common sense dictates, a combination of the two, tilt circumnavigates the issue the vast majority of the time, and gives you a sharper image than any other method to boot. If you are not pixel peeping, or printing large there is no point to any of these superb lenses, if you are in the market for one the assumption must be that you are hitting a quality ceiling.
    One of the best things about the latest lenses is the ability to rotate the tilt and shift independently, this increases functionality immensely, you are still "limited" to a plane of focus, but you can put that in any orientation independent of any shift. Just look at the examples in Merkingers Focusing The View Camera to see how much flexibility that gives you, groves of trees can be shot, with thought! J is not a fixed distance to the ground from the lens, it is one of the keys to a world of possibilities.
    Is a 17mm TS-E lens the magical answer to landscape photography? No. But take some time to learn it and it does offer a level of flexibility, control and options that no other ultra wide does in the 135 format. It is a unique lens with unique abilities.
  11. which one does rotate?
  12. Christopher,

    Both the 17 & 24 MkII have the same functionality, you can orientate both the tilt and shift independently in any direction.
  13. Christopher, before you buy a tilt/shift lens (it's a big investment) I would suggest that you invest some time researching the subject of "camera movements". Once you understand this concept better, you'll have more of an idea what theses lenses can do for you. Even then, you might want to rent one for a day or two to see if it's the right tool for you. Don't drop two grand on something just because somebody said that it's a good thing to own. T/S lenses are definitely not for everybody.
    Keep in mind that these lenses have NO AUTOFOCUS capability. They aren't particularly useful for handheld shooting, so you'll need to use them on a steady tripod to get the most out of them. If you are willing to spend several minutes setting up a shot and making sure that everything is perfect before you expose it, then you might like a T/S lens. Otherwise, you might find it to be rather frustrating, and you'll end up leaving it at home.
  14. i understood what titl and shift requires of me...
  15. Scott, I must confess to using a combination of “hyperfocal focusing” and infinity focusing, at least in the days of MF lenses when one could give a slight “tweak” toward infinity focus (it can readily be shown that the approximations used in most lens DoF scales put the far limit of DoF slightly short of infinity). With current AF lenses, though, it's more an issue of feasibility than theory. The scales are so small that accurately setting the hyperfocal distance is nearly impossible, and with the unfortunate demise of Canon's DEP mode (the real one), traditional techniques for controlling DoF are difficult, despite the seemingly endless discussions on Internet photo forums. Infinity focus has one big advantage: it's usually fairly easy to accomplish. I nearly always use it when photographing a rising or setting Moon with a long lens. In that case, of course, the foreground is far away, so focusing on the nearest object would give a similar result.
    You're preaching to the converted on tilt; I've used all the original TS-E lenses for many years. But they're not always the answer: unless the scene is a good fit to the wedge-shaped DoF that obtains from tilt, a TS lens is of little use (except perhaps for the very usable DoF scales on the older lenses, which sadly have essentially been eliminated on the two new models).
    I think Focusing the View Camera is a far more significant work than The INs and OUTs of Focus. By far the most important observation is that with tilt, the plane of focus rotates about an axis below the lens, and that the distance to that axis is determined by the tilt. Scheimpflug obviously understood this (though his application was quite different), but prior to Merklinger, the only published description was by Urs Tillmanns in 1992, which described what Sinar had learned in developing the Sinar e in the late 1980s. And Tillmanns revealed rather little of what Sinar had rediscovered. The second most important observation was that the DoF is equal on each side of the PoF on a line parallel to the image plane; it's very helpful in determining the best position for the PoF in a scene with height as well as depth. Finally, and more implied than directly stated: better results (i.e., smaller f-number for the same region of sharpness) are often obtained with less tilt. In particular: except for an essentially planar subject, the best location for the PoF rotation axis is hardly ever in or above the ground plane. This is clear from several examples in Chapter 5 of the original version, and also from the “Craig Bailey” problem discussed in the Addendum. Unlike many of his readers, Merklinger understood this quite well.
    I'll take it a bit further and say that in most cases (at least for the photographs that I take), the optimal tilt is determined by passing the PoF through the two object points that determine the near angular limit of the DoF, and then setting the final position of the PoF by adjusting focus without touching the tilt. If you look at Merklinger's Figure 33 on p. 43 of the original version, you'll see that he arrived at essentially the same conclusion, perhaps without realizing it.
    Once the tilt is set, though, I know of no other systematic method of determining the final position of the PoF (by adjusting focus) than using the lens DoF scale in the same manner as without tilt. There are a couple of caveats, of course: while the DoF scales work fine with tilt, the distance scales do not, and that near and far refer to angular rather than linear distances. Stated otherwise: once the tilt is set
    1. Focus on the near; note the position of the focusing ring. If the tilt has been set using the near angular limit of DoF, this has already been done.
    2. Focus on the far; note the position of the focusing ring.
    3. Set the focusing ring so that the focus index is midway between the two positions just noted.
    4. Use the f-number whose markings align with the near and far positions noted.
    This, of course, relies on the same assumptions as in conventional DoF with equal CoCs for near and far, and essentially the same as Canon's DEP mode. So that approach is either valid or it is not. In my experience, it works quite well. If the primary objective is recognizability of objects, Merklinger's object field method (or simply infinity focus) is probably a better approach. To be honest, I've never figured out how to practically combine it with tilt.
    Is the TS-E 17 the answer? I'll return to my standard question: what lens would you otherwise use?
  16. Personally the best focussing method for me is to focus on the point in your composition where your eye will naturally rest - that should be pin sharp, everything else falls into place behind it. The problem is that sometimes you think it will be one place, but when you actually look at the resulting shot you find it is somewhere different to what you thought it would be. Hyperfocal focussing is generally a recipe for acceptable but not exceptional apparent sharpness. Certainly I do not agree that infinity focus is more important-only if the main compositional element is at infinity. I have to agree with Jeff too. I don't really see how a tilt-shift lens will always help you. It depends on the arrangement of the elements of the shot. The shift can indeed be useful to avoid diverging/converging verticals. In my case I always try to make the composition without diverging/verticals from the start. Certainly this is not always an allowable luxury.
    I guess what I am saying is that the 17mm is a very expensive and specialized lens. I suggest renting it first to see if it really worth it for you. It certainly is not for me.
  17. I'm salivating, since the announcement of the canon 17mm TS lens. If, If I'm not a long time dedicated and happy user of Nikon cameras and lenses, including the 14-24/2.8 17-35/2.8 and 14/2.8 lenses, (Used to have the 28mm TS )I would all ready had the Canon 17 TS. To bad, Nikon always to slow to produce such a useful lens. I know the Nikon 24 TS. But the 17 TS can be a beautiful tool for landscape and I wish I can have one made by Nikon. I have seen images with the canon 17TS and no comparison to anything els. It is possible to convert somehow this lens to Nikon mount? I don't think so. And one more point. Before one get a 17 TS, they has to learn how to use a very wide angle lens, before going to the TS lens. Many photographers don't even know how to use a 24mm wide angle lens, never mind a 17 or 14mm or a 17mm TS lens.
  18. Thanks for all the additional comments. I am obviously familiar with how tilt works. The concern I have especially with the 17 mm is the simple fact that vertical objects (e.g. trees, other plants) close to the camera maybe in focus close to the ground (if I tilt forward to get the ground in focus from close to infinity) but out of focus at the top. I anticipate that this is more of an issue with the 17 mm lens compared to longer TS lenses where I get more foreground into the picture. I probably have to find out for myself.
  19. Peter,
    I was in the middle of typing a different comment. In reply to your specific query, obviously the TS-E DOF is never any worse than any other 17mm, but with the vastly improved functionality over the earlier TS lenses of fully independent rotating tilt the issue you are concerned about is vastly mitigated. It does add to the thought process involved in getting the best image, where to best put the plane of focus, and how to get it there, as well as everything else, focus, dof, composition, shift etc. But it can be well worth it. There is no other lens this wide that enables you to get such focused focus. Having that wedge shape of focus, and the ability to put it almost anywhere, fits the image a surprisingly high percentage of the time. Obviously sometimes it won't work, but when that happens, as I said, you are no worse off.
    Below is an example of the flexibility you now have. It is not totally what you were thinking of but is close. So I decided that the quay was the important bit, so I tilted to get that in focus from near camera to the trees at the end of it, effectively infinity. You will see the trees to frame left are totally sharp from top to bottom, as per your example. The water to frame bottom right is not as sharp, but it doesn't need to be. The kicker? This is shot at f4! Had I wanted the water sharp as well then I could have optimised it with a differently angled plane of focus and/or a smaller aperture.
  20. Peter, the problem with vertical objects close to the camera with forward tilt is essentially the same as with any other TS lens (or with a view camera, for that matter). Unless your region of interest is wedge shaped, tilt (or swing) won't help much. For example, standing in the middle of two rows of bookshelves in a library and trying to get both sharp near to far, tilt is no help. Of course, you could easily use swing to get one of the bookshelves sharp. Your experience with the 90, and especially the 45, should give you an idea of what will work and what won't.
    The 17, of course, has an additional issue of obviously converging verticals unless the camera back is vertical, and that's where the independent tilt and shift rotation comes in to save the day. A common technique with a view camera is to set the camera with the back vertical, fine tune the framing with the shifts, and then set the tilt. The two new TS lenses let you do a pretty good emulation of this. With most TS lenses, the framing changes slightly as tilt is applied, and the independent rotation of the shift makes it easy to correct this if the framing is critical. Semi-permanently rotating the shift 90° on the older TS lenses of allows the same fine tuning.
    One other surprising detail with the 17 (and the 24 and any asymmetrical lens) is that when you use tilt, you get slight convergence of verticals even if the camera back is vertical. This effect is noticeable with the 45, very noticeable with the original 24, and I assume it is even more noticeable with the 17. Why? There's nothing wrong with the Canon TS lenses; all the lenses I've mentioned are retrofocus, so the entrance and exit pupils don't coincide with the nodal planes, and you get some shape distortion when you tilt or swing. The effect is easy to demonstrate: focus on a rectangular object such as the front of a bookcase and move the tilt through its adjustment range--the verticals converge and diverge. Of course as you apply tilt, the top and bottom of the bookcase go out of focus, so it's not a technique you'd normally employ, but there sometimes are more practical situations where you need to keep an eye on the verticals (or horizontals with swing). The only way to fix it is to tilt the camera back to fix the verticals, which throws off the tilt settings, so you need to repeat the process ... in practice it usually takes a only a couple of tries. At least with the shift rotated to be in line with the tilt, restoring the framing is easy.
    In practice, this seldom has come up for me, but as I had said, I seldom use the tilt on the 24, and I can count on one hand the number of times I've run into it with the 45.
  21. Jeff,
    Very astute theory with regards the converging verticals when applying tilt, I was playing last night and got this. The camera back was vertical. The short wall, top left, is vertical and the armchair is square!
  22. Here is an image I shot just for this thread a couple of hours ago. The two posts are less than three feet away, the entire image is very sharp, the aperture is f10, so no diffraction. 2º forward tilt with camera approximately 12" off the deck. The tops of the posts sharpened up just fine, way more than the background would be if I'd gone the hyperfocal route or if I had focused on the posts and gone for a smaller aperture to sharpen the background and the deck.
    Tilt is not the answer to every problem, but it sure made this image considerably sharper throughout the frame.
  23. And here's a case [the interior] where you probably did need the tilt even with a 17. The only way to avoid the problem is to get a view camera and use a symmetrical lens (and hope it covers) ...
    I'm glad I don't shoot a lot of stuff like this. Using a 17 for a landscape with trees would encounter the same problem, but I guess with practice one would learn to tilt the camera slightly up before setting the lens tilt.
  24. Jeff,
    There are no convergence issues in the second shot. The trees and posts are the angles they are, the walkway support posts confirm this.
  25. Sometimes the only way to see if tilt helps is to try it both ways and see which requires the smaller f-number. QT Luong's How to focus the view camera is arguably the best article I've seen on how to go about this. The main guideline, as always, is don't use too much tilt. Here the PoF rotation axis (the “hinge line”) is about 7 inches below the deck, which seems reasonable offhand. Putting the rotation axis at deck level, as so many folks seem think is the right approach, would not have worked nearly as well. But again, that's something Merklinger never suggested for a typical scene such as this.
    Convergence? I guess you learned pretty quickly. It's not that difficult as long as you remember to look for it. It's also less of an issue with less tilt, which is typical of what's actually needed in most practical photography. I seldom use even half of the 8° tilt on any of my TS lenses unless I'm really close to the ground. Again, though the effect is easy to demonstrate with contrived settings (i.e., maximum tilt), it's seldom been an issue for me in taking real pictures.
  26. I appreciate all your effort and expert opinions presented here! Very informative. Jeff, good observation on the retrofocus/tilt issue. I usually do not attempt to eliminate convergence of verticals completely, so it has not bothered me on the 45 mm. All this means, of course, that I need another excuse now for further procrastinating on ordering a $ 2200 lens... Well, I'm sure it is very heavy!
  27. No it's not! The convergence is very easy to eliminate too. No excuses! Don't tell anybody but I got my absolutely totally mint one, in box still wrapped in the original plastic bags, from a very well rated seller, off eBay for $1,725. A real bargain, but they are out there.
    Take care, Scott.
  28. Peter, when I first noticed the convergence when playing with the tilt on my 24, I thought I was losing it. I had just made a diagram proving that lens tilt doesn't change magnification. There was one small problem with the diagram ... I'd used the typical Gaussian thin-lens approximation, and my geometrical inferences relied on coincidence of the pupils and nodal planes; though the approximation works for most lens-tilt derivations, it doesn't work here. When I re-drew the diagram with the pupils separated from the nodal planes, my “proof” no longer worked. I never did get around to proving that magnification did change, so to be honest, I was partially guessing at the cause until I chanced upon this post by Emmanuel Bigler (a professor of optics) in the Large Format Forum, which confirmed what I had suspected.
  29. The short wall, top left, is vertical and the armchair is square.​
    The distortion of the armchair is a wide-angle lens effect. The near side of the chair appears to be larger than the far side. The tilting wall I cannot explain. I haven't seen a problem like this when the camera has been carefully leveled.
  30. Dan,
    The right hand side of the chair is "leaning" in, I meant the side of the chair is square. This is a known optical effect of retrofocus lenses, the shorter the focal length, the more pronounced. I'll do another shot latter if you like, but I can assure you the bubble level was perfectly lined up. It is an effect that is very easy to deal with though, as Jeff said, all you need to do is put a little "rear" tilt (the camera) to counteract it.
    As Jeff hypothesised, the extreme difference between the focal length of the 17mm, and the comparatively massive percentage of retrofocus, make the effect with 17 particularly pronounced.
  31. Here is a graphic example of short focal length retrofocus T/S lens convergence when tilt is used.
    Neither the camera or tripod were moved throughout the first three shots, only the tilt was adjusted. This is the run down of images
    1. Control shot, leveled and plumb with the two pillars.
    2. Full 6º down tilt.
    3. Full 6º up tilt.
    4. Full 6º down tilt with camera tilted back to zero convergence.
    5. Camera set up for shot 4 showing back tilt of camera (with pillar in background for reference).
    Hope this helps, Scott.
  32. You get similar effects with any asymmetrical lens. It's quite pronounced on the original TS-E 24, less pronounced but noticeable on the TS-E 45 (which is a retrofocus because the tilt/shift mechanism puts the glass so far away from the image plane), and barely noticeable on the TS-E 90 (which is slightly telephoto but close to symmetrical).
    It sometimes takes a couple of tries to fix the convergence if you absolutely must have vertical lines perfectly vertical in the image. But these are extreme examples: maximum tilt, and obviously vertical features near the vertical image edges. In more typical cases, a slight convergence often isn't even noticeable. If possible, try to anticipate the convergence before precisely setting the tilt; if you tilt the camera back after setting the tilt, you'll need to reset the tilt to get the PoF where you want it.
    Peter, the second image also addresses one of your questions. The beam at the top is clearly out of focus. If you shoot a lot of landscapes with a frame such as this, you might have a problem; if not, it's an unpersuasive excuse ...
  33. Scott, you have effectively illustrated the phenomenon. My question would be why you would ever need six degrees of tilt except in an extreme macro setting (or for special effects). I rarely use more than one degree of tilt with my TS-E24 II. That's plenty to get everything on a horizontal surface in focus unless the camera is very low to the ground.
    One degree of tilt at f/8. The camera was less than three feet above the wood surface. Vertical lines are unaffected.
  34. Dan,
    I completely agree, I would hate to cast aspersions on the 17mm TS-E and have people think it has a fault. I even posted a more extreme example than yours, the one where the camera is very close to the deck, it had 2º of tilt and no convergence. In real world use I can't foresee an issue, it was just an observation of an optical side effect.
    I am very very happy with my lens, as an added "advantage", with one shift and perfect auto stitch alignment, I have ended up with a 40mp medium format sized sensor.
  35. I'm glad to hear that you like the TS-E 17 so much, Scott. Maybe I'll ask the Easter Bunny to bring me one next spring. ;-)
  36. If your bunny can carry it, Dan, then do ask. It is the lens I have been looking for.
    I do have one bit of advice/comment I'd give to the Canon team for a possible MkII version in ten years or so. You are quite right, 6º of tilt seems a lot to be effectively usable on this focal length (so far), it has too much dof for counter tilt "toy" images, and closeups, where it might be really useful, can put the subject dangerously close to the front element. I do have one job where I might use it all though, a guitar, similar to this image, I'll know in a few weeks as I have a guitar shoot. What I would like to see though, is much finer gearing over fewer degrees, it only takes a short throw of the knob to tilt the full range, spread that out over a couple of rotations and it would be far nicer.
    Take care and I'll keep my fingers crossed for a good Easter for you :). Scott.
  37. Scott: The lack of finer control for tilt is also my only complaint with the 45 and 90 mm. Obviously, less ciritical there than for 17 mm.

Share This Page