Choosing a tilt-shift lens

<p>Hi all,<br>

I am weighing which tilt-shift lens to purchase. My main application is architecture photography. I have a Nikon D800 and would consider the Nikon 24 TS lens given the great reviews. However, it lacks a fundamental feature that is found on the Canon equivalent as well as some third-party lenses. That feature is the ability to shift and tilt simultaneously. For those who use TS lenses for the built environment, how significant do you see this "shortcoming" of the Nikon TS lens? Is this capability in the "nice to have" category or "essential?"<br>

Thank you,<br>

Andrew</p>

 

<p>You sure you are not taking about the ability to change the angle between the axis of shift and tilt?</p>

<p>As Chuck points out, on Nikon's 24mm/f3.5 TC-E lens, you can tilt and shift simultaneously, but the tilt and shift movements are either both in the same axis or on different axes. That is a change that one needs to take the lens apart to do. (You might be able to do that at home too, but there is risk to breaking some ribbon cable, resulting in some expensive repair.)</p>

<p>What Canon's version can do is to change axes in the field. That makes the lens very complex mechanically but it is an engineering accomplishment.</p>

<p>Personally, I don't use tilt and shift simultaneously so that it doesn't matter to me. But I don't use that lens very often to begin with. For other people, the capability the Canon version has could be significant.</p>

<p>As I understand it with the Nikon, the relationship between tilts and shifts is fixed. So If you tilt up or down, you can't swing left or right. You can tilt in only one axis at a time, and it can shift in only one axis at a time. That's my understanding anyway.</p>

<p>Andrew, the exact restriction is not that simple to describe in words; anyway exceeding my capability. Perhaps someone else can provide a better description. However, I think it is best that you visit a large camera store to check that lens out in person. If that is not convenient, maybe a search on YouTube and see whether someone has done a video on how that lens operates. There is, of course, the rental option.</p>

<p>Andrew:</p>

<p>The lens can do two types of movements. One is shift, where you translate the entire optical element train perpendicularly to the long axis of the lens. The other is tilt, where you pitch or yaw the optical elements so the central axis of the optical element train is no longer perpendicular to the plan of the sensor or film.<br>

Yes, on the Nikon lens you can shift and tilt at the same time. <br>

You can also rotate the lens while it is on the camera so the plane of the shift lines up with whatever object you want to photograph. However, the plane in which you can tilt is fixed with respect the plane in which you can shift. You can't hold the plane of shift fixed while changing the plane of tilt, at least not on the fly. AS Shun indicated, You have to disassemble the lens to change the orientation of the plane of shift with respect to the plane of tilt. This is probably most safely done by Nikon, who will charge you $400 for the privilege, or so I was told by several people who bitched extensively about it. Furthermore, you have to send the lens back for another $400 hit each time when you want to adjust the angle.<br>

Canon doesn't charge you anything for doing this because it is a built-in feature of the lens. There are also third party tilt-shift lenses which cost 1/2 as much as Nikon's TC-E lenses, and which lets you adjust the relative orientations of the two planes on the fly.<br>

I actually don't think it need to be mechanically very complicated to preserve all functions of the lens while letting the angle between tilt and shift be adjustable on the fly over a 90 degree range, which is all you need. But Nikon is determine to have its pound of flesh, I meant $400.</p>

<p> </p>

<p>Perhaps the problem is terminology. "Tilt" and "swing" are the same thing, just that one is defined vertically and the other horizontally but in any case this can only be done on one axis, which can be rotated to the desired orientation. As Shun says, you can definitely tilt/swing and shift at the same time; however the second movement is constrained to a parallel or orthogonal axis that cannot be readily changed in the field. Thus a landscape photographer will typically have it set up one way, and an architectural photographer perhaps the other. For instance, I will usually tilt for depth of field and then shift sideways for a limited panorama. This sideways shift makes for very easy stitching to make the final image.</p>

<p>The 24mm PC-E is my favorite lens on my D800E. I mostly shoot small town architecture with it. I've never needed both movements on one axis. After a year of owning, I either use shift, or I use tilt. If I ever need more complex movements (very rare), I own three 4x5 field cameras. BTW, the shift makes a HUGE difference in my shots. Don't know why I waited so long to get that lens.</p>

<p>Kent in SD</p><div></div>

When shooting architecture I've always found the tilt movement of a tilt/shoot lens to be of limited utility. Still life is

different case entirely

<p>To echo what Ellis says, for architecture, I mainly use shift to correct the converging verticals. For landscape, I mainly use tilt to alter the plane of focus to gain near/far sharpness. That is why I don't use both at the same time very often.</p>

<p>I've long had one of the old 35/2.8 shift lenses from film days, and it is very nice for correcting perspective, and also for eliminating one's shadow from directly lit scenes, and removing oneself from windows and mirrors. In fact, I like it so much that I use it as the normal lens on my DX digital camera, despite the hassle of preset aperture. </p>

<p>I also have the older D version of the 85/2.8 tilt shift lens, which I got mainly for macro work (that's an excuse of course. I got it because I lusted after it). It's a pretty long lens on DX. On that, I rarely use the shift (though it makes a pretty nice panorama), but find the tilt useful for perspective control. </p>

<p>If I were not eagerly anticipating busting my lens budget soon on a 200-500 zoom, I'd be lusting after that 24, which would sit very nicely on a DX body! I confess to a weakness for shifting lenses. </p>

<p>Though I think Nikon could have done a little better and made their current lenses more flexible, I don't find the limitation particularly annoying except in principle. I figure if Canon could do it, Nikon ought to have been able to also. That said, I'd buy the 24 if I could come even halfway to justifying the extravagance. </p>

<p>Matthew--</p>

<p>I think the only DX bodies the 24mm PC-E will work with is the D300/D2 series. Since it's a short lens, there are clearance issues with the lens hitting the pop-up flash housing above it. In fact, I wanted the 24mm PC-E so bad it's the main reason I bought the D800E. I.E., I bought the camera just so I could use the lens! (Was using D7100 at the time.)</p>

<p>Kent in SD</p>

<p>Being able to control the vertical and horizontal plane of the lens is really useful in architecture when you are in tight places and you have to correct perspective in camera. That's just one use, but the one that you would expect to encounter shooting some interiors as well as exteriors. If that lens does that, you should be in good shape. </p>

<p>Kent, thanks for the update. I guess I'll have to stick with the 35 for now until I convince myself that I need an FX body too! The 85 would be easier to use with a better viewfinder for sure! </p>

<p>I've had the PC-E 24mm f/3.5 for about a year. I did the 90 degree rotation almost straight away, as that suited my needs. There are on-line instructions on how to do it. Just be careful and take your time. However, it is not something that you'd change very often, as there is still risk. From what I have read, you cannot do the same with the 45mm and 85mm PC-E lenses as the cable is too short. They must be converted by Nikon.</p>

<p>For the record, Nikon's inability (shared by the older Canon lenses) to make an in-the-field choice about the relationship between tilt and shift annoys me as well, and almost pushed me to the 24mm Samyang. I have an old 65mm Hartblei super-rotator (originally in Canon mount, switched to Nikon with much argument in which they claimed that I couldn't possible fine-tune the mount alignment and I asserted that it couldn't possibly matter for a tilt-shift lens - they were kind enough to relent) which has independent axes, although annoyingly the detents don't line up with shift and tilt in my preferred direction. I also have a Kiev 35mm which mechanically slides the shift along the tilt hinge, so it can't be adjusted at all.<br />

<br />

Like Tony, I most generally use the tilt and shift aligned so that I can use the tilt for the ground plane and the shift to correct vertical perspective. While I can see uses for the other configuration, I remain very surprised that everyone seems to think of the alternative as the "default" (both Nikon and Canon's previous lenses had to be rotated to work like this).<br />

<br />

Regarding the viewfinder, I had been asking Nikon for a while to offer a live view mode in which different sections of the LCD could be positioned and scaled separately, making it possible to determine when the focal plane is in the desired configuration. The D810's split live view is a first step towards this, although I'd have preferred the four-way split that I initially suggested even if it caused a reduction in refresh rate. The D810 split view only allows two views with the same alignment and size along the short axis of the sensor (it is allegedly for checking level horizons), but can still be useful for checking the ground plane if you don't mind the camera being in portrait configuration. My experience has been that scrolling around a live view image to check points for focus is painfully slow and prone to moving the camera unless you have a very secure support. Focus peaking would help a lot, too, if Nikon ever get around to catching up with the mirrorless cameras and offer it. I have to assume that tilt-shift lenses are not high on Nikon's priority list - although they might sell a few more if they did pay more attention to usability! (I'd certainly use the ones I had more if it were less painful. I've proposed a DSLR with rear movements to Nikon in the remote hope that they'd consider it - that would make me very happy, but I'm not holding my breath.)</p>

<p>To only allow cross-shift with vertical tilt and vice-versa by default is absolute madness. Shift is nearly always needed in the same direction as tilt/swing in order to bring the "sweet spot" of the lens back into centre-frame and reduce vignetting.<br>

That was Scheimpflug 101, as an introduction to LF camera movements at photo-school.</p>

<p>Further proof that these days it seems Nikon's designers haven't a clue about photography as practised in real life. So I won't be throwing the old DeVere monorail away quite yet.</p>

<p><br />BTW Andrew, any rigid bodied camera has back movements, it's just that the lens is always stuck pointing in the same direction as the camera back. What's needed is a gimbal lens-mount to keep the lens axis horizontal at all times. :-) Maybe a modified VR mechanism could double-up as an electronic swing/shift?</p>

<blockquote>Maybe a modified VR mechanism could double-up as an electronic swing/shift?</blockquote>

 

<p>Hmm. Interesting. With enough coverage, possibly. I'd been thinking in terms of five-axis sensor-based IS as a route to supporting this. I'm way more interested in tilt than I am in shift - fixing distortion in post is easy, for certain levels of image quality (and you can always crop from a wider lens), but adjusting the focal plane is non-trivial unless you're capturing a light field (you're Lytro or you're focus stacking). (Rear) tilt would likely have less of a problem with moving the sensor outside the image circle than shift would. You're a bit hosed when it comes to the mirror mechanism behaving sensibly and it complicates the shutter mount, of course. It's much easier just to put an adaptor with bellows on a mirrorless camera and allow the flexibility there (especially since several of them have focus peaking) - but I'd really love some motorized worm gears to move the shutter around to achieve autofocus while tilted. Not that I've ever used a mirrorless camera with a tilt-shift - though now I think about it, I could try my micro 4/3 mount to F mount adaptor on my GF2 and see how it behaves - but there's no focus peaking on the GF2, so it doesn't really get me anywhere.</p>

 

<blockquote>To only allow cross-shift with vertical tilt and vice-versa by default is absolute madness.</blockquote>

 

<p>Quite. And yet I have a lens which physically can only work this way, and most tilt-shift lenses in which the axes can be adjusted in relative configuration using a screwdriver are shipped in this configuration. I've always been told that it's for product photography and architecture, where you may want a vertical surface (wall, packaging) in focus but you want to shift the other way to correct perspective. Which is fine, but for landscape shots, I'm with RJ.</p>

<p>I have the 45mm PC-E for my D800, its a focal length that suits my needs very nicely, it's also a very fine lens.</p>

<p>But someone does have to mention, so it might as well be me.... if you haven't played around with "lens correction" in Lightroom or seen what you can do with "distort" in Photoshop before you invest a lot of money in a PC-E you should, I think.</p>

<p>Distortion correction can compensate for shift in a lens - it applies the same effect that keystone correction does on a projector. Obviously you're stretching some pixels, so you can expect a loss in quality, but not necessarily a large one, especially if you started with a high-quality image. This may be no worse than handling the peripheral image quality of some tilt-shift lenses, so I agree that there are reasons not to bother spending money on one of these. Those who use them a lot may prefer to correct things in camera, although it's pretty easy to do this in post and tends to be more accurate.<br />

<br />

Tilt changes the orientation of the focal plane so that it's no longer aligned with the camera. Without recording a light field or focus staking, or faking it by blurring areas of the image (if you wanted to tilt the focal plane for blur rather than additional sharpness), that's hard to handle in software because the image doesn't capture 3D information. It's sometimes possible to apply selective sharpening to different portions of the image to compensate partly, but it's painful and likely to require human intervention.</p>