Scheimpflug with TS-E lens: where is the lens plane?

[ed_avis2]

<p>I've recently bought Canon's 90mm tilt-shift lens and I am learning about how to use tilt. I have read explanations of the Scheimpflug rule which I think I have just about grasped: take a plane one focal length in front of the lens plane and intersect it with the focal plane to give a line; the subject plane also intersects that line and can be 'hinged' around that line by focusing. But a couple of questions:</p>

<p>1. There is a discrepancy between different explanations of the rule: some talk about the lens plane and have diagrams showing this plane passing through the centre of the lens. But others refer to a plane parallel to the lens plane and one focal length <em>in front of</em> the lens. Which is the case here?</p>

<p>2. Where is the lens plane with this particular lens? Is it a plane parallel to the front element and passing through the middle of the front element? The front element extends when focusing - does that change the position of the lens plane?</p>

<p>My camera body has a line marking the position of the sensor and so the focal plane. But I do not see any marker on the lens showing the lens plane.</p>

[John Crowe]

<p>Without going and reading about it myself I would suggest that they are referring to the plane of focus. So, you have the film/sensor plane and the plane of focus. The plane of focus occurs at the distance where you have the lens set to focus on. Then you have the depth of field in front of, and behind that plane, which is determined by the set aperture. When you tilt the lens, the plane of focus, and it's accompanying depth of field, tilts with the same angle as the front element. So, yes, the plane of focus could be said to be parallel to the plane passing through the centre of the front element. However some front elements are curved so on what part of the front element you chose to place this plane has no bearing. It is simpler to consider that the focal plane remains perpindicular to the centre axis of the front element. So, as you tilt the lens, you tilt the axis, and you tilt the focal plane.</p>

<p>I have the original Canon EOS lens book and they have a great photo with a train on a platform. The camera and 24 TS-E is placed beside the train, looking straight down the platform so the film/sensor plane is perpindicular to the side of the train. With the train on one side, and with a wide aperture set, only a small section of the train is in sharp focus. With the camera in an identical position but with the lens tilted to the side so that the front axis is pointing more towards the side of the train, the entire length of the train falls within the depth of field of the focal plane, and is in sharp focus.</p>

<p>I have used my 17 for shift only so far, but I will play with the tilt in the spring. It is useful being able to change the angle of the focal plane but while you are including some things in the photo into the field of focus you will still be removing others. The trick is finding appropriate subjects, like the train, to make the effect work well.</p>

<p>Have fun with it. From what I have seen you will observe the most significant changes with the tilt effects of the 90mm over the other TS-E lenses.</p>

[joe_fikes]

<p>While the diagrams you see in books and articles are correct, I don't think they're of any use in actually setting the focus and tilt of the lens. Here's what I do with that lens, realizing that I'm trying to bring a tilted subject plane into sharp focus:</p>

<p>1. Focus on the closest part of the subject plane by twisting the focus ring as you normally would.<br>

2. Then focus on a distant part of the subject plane using tilt alone. Don't move the focus ring.<br>

3. Go back to the closest part of the subject and repeat 1 and 2 as many times as needed to get the entire subject plane in focus.</p>

<p>In my experience, it seldom takes more than 2 or 3 iterations of this procedure to get everything sharply focused. If you can't get the procedure to work it's probably because the lens won't tilt far enough to achieve the result. But the 90 mm has enough tilt for most subjects.</p>

<p>The effect is most noticeable with the 90 mm. I seldom see an improvement using tilt with my 24.</p>

<p>I hope this helps.</p>

[scott_ferris]

<p>The seminal work on the subject, the thing that everybody tries to explain is <a href="http://www.cs.cmu.edu/~ILIM/courses/vision-sensors/readings/FVC16.pdf">here</a>. Scheimpflug is only half the theory, the hinge line is the real key, <a href="http://www.northlight-images.co.uk/article_pages/using_tilt.html">this link</a> has two gif's that really demonstrate what is going on.</p>

<p>The lens plane, for all intents and purpose, is through the front element. Technically I forget, but the definitions don't work for retrofocus lenses anyway, I believe it is one focal length in front of the sensor, but that is only for simple lenses, like those found on view cameras.</p>

<p>The focal length of your lens changes when you adjust focus, so again, technically your lens plane moves, but these are theoretical factors, they are too small to make any difference in actual application.</p>

[scott_ferris]

<p>However, focusing is what makes the hinge line function. Focusing is key to getting your plane of focus where you want it. The difference between your lens plane and the sensor are critical, the smallest movement of focus will change the angle of plane of focus.</p>

<p>The j1 point, the Scheimpflug line, is what is moved with tilt, the Hinge line, is what is moved with focus.</p>

<p>Think of the plane of focus as a door, Scheimpflug tells you where the edge of the door is, the Hinge line tells you how open that door.</p>

[akocurek]

<p>Thank you Joe Fikes! I finally know how to use the tilt on my 24 TS-E. </p>

[nathangardner]

It is a wedge in the space between the lens plane and film plane. So if you tilt the lens down, the bottom of the frame

will have shallow dof because the LP and FP come together and intersect. As you move up the frame the two planes

get farther apart and the dof increases.

[John Crowe]

<p>I stand corrected. Thank-you Scott for the link. I will refer to this in the spring when I start to play with tilt. </p>

[zml]

<p>The biggest issue with pretty much all lenses sporting any kind of movements designed for tiny format cameras (24x36 mm) is that the engraved scale is way too tiny for any precise control of said movements. For instance, if the camera (24x36 mm sensor) fitted with a 24 lens is around 3 feet - circa 1 meter - from the ground and the focus point is roughly 2/3 distance to the far object in the frame, the drop (tilt down) just shy of 1 degree is required to get everything in focus, from the tripod legs to the far object in the frame. Try this with any T/S lens and you'll notice that the movement is not fluid and controllable enough for repeatable results (is it .95 or .85 of a degree...a huge difference at f/4...) </p>

[scott_ferris]

<p>Excellent point Michael, it is the one aspect of my 17 I'd swap to a MkII for. It can be mitigated slightly, if clunkily, by using an iPad or smartphone etc with an inclinometer on it. I have a great free app called Clinometer, and another, Measures HD, that can give you very accurate measures. Nothing like as easy as an accurate, well marked, finely adjustable angle movement and scale on the lens though!</p>

[michael_young3]

<p>Just so, Michael. Without quibbling with the larger thought, the tilt you give for 3' height is for 17mm, not 24mm.</p>

<p>It's the relatively short focal length, compared to focal lengths common for larger formats, that does this. The 24mm we use on our teeny tiny sensors compare to 90mm on 4x5. The offsetting advantage for our small lenses is the very short J distance their small tilt ranges allows. For the same TS-E 17mm, its maximum tilt of 6.5 deg. allows it to sit just 6" above its focus plane.</p>

<p>Even so, getting back to your original thought of ultra sensitivity, the problem isn't quite so profound in practice, once we step away from our desks, spreadsheets, and sliderules. The tilt adjustment is rather coarse, but this is a mechanical issue, chosen by Canon. The Hartblei 45mm Super-rotator, as a counter-example, has a very smooth worm drive on its tilt setting. It takes about 5 3/4 turns to tilt through its 8 degree range. The same 1/10 degree you mourn is 25 degrees of turn on that worm. The problem is actually on the other end of the scale, at near full tilt, where the tilt is much less sensitive to distance.</p>

 

[ed_avis2]

<p>Thanks for the links Scott. It looks like I mixed up two lines: the Scheimpflug line and the hinge line. As you suggest, I will assume that the lens plane is through the front element (and that the lens is 'thin' in the terminology, meaning there are not two separate lens planes to consider). The 90mm Canon lens is a Gaussian design, not retrofocus, which simplifies things a little.</p>

<p>The guide makes it clear that I do need to allow for movement of the front element when focusing. In the view camera diagrams, focusing is typically shown as a movement of the film plane, keeping the lens fixed - but I will assume that keeping the camera body fixed and moving the lens is roughly the equivalent operation, except perhaps for very close subjects.</p>

 

[ed_avis2]

<p>Joe Fikes - thank you for the advice. I am trying out this procedure and general trial-and-error experiments. My camera body does not have Live View so manual focusing can be difficult to get right. (It is a 1Ds. I have the EC-C IV focusing screen; I've thought about the EC-S matte screen but since the lens is only f/2.8 it would not make much difference I believe?)</p>

[joe_fikes]

<p>Ed, Live View really makes all the difference in the world. When I shoot with my TS lenses Live View makes the experience almost like using a view camera. You frame the shot and get all of that handled and then you pop in the 10X magnification. It feels very much like using a good Schneider loupe (well, not THAT good, actually, but pretty good). I've tried it a few times without Live View and it's pretty difficult. Some of the young eagles might see well enough to make that work well, but my eyes aren't up to the job. </p>

<p>There is a technique I've used that does not depend on Live View that sometimes works. Make a pure guess at a tilt angle based on the tilt of the plane you're trying to match. Focus on the middle of the scene and rock the focus back and forth. If the desired focus plane is "distinct" enough, then you'll see the region of sharp focus move toward and away from you as you rotate the ring. If the region of sharp focus approaches you as you turn the ring for closer focus, then you aren't tilted enough. Conversely, if you focus closer but the region of sharp focus moves away from you, you're tilted too much. When the tilt angle is right, then the whole plane will come into focus at the same time. I think if you practice on something flat with a good bit of texture (like carpet on your floor), you'll get the hang of it. Just remember, you focus closer by turning clockwise as viewed from in back of the camera and vise-versa.</p>

 

[jeff_conrad]

<p>As Michael Young suggests, most people get far too hung up on precise calculations. About the only time you want the PoF rotation axis in the ground plane is when you’re photographing something completely flat—like your mother-in-law’s dining-room rug. For most 3-D scenes (assuming you tilt down rather than up, such as when photographing a ceiling), you want the rotation axis below the ground plane. Sitting at a computer, it’s easy enough to calculate the settings, but in the field, it’s almost impossible.</p>

<p>The technique Joe Fikes describes is one variation of Howard Bond’s “Focus/Check” method. QT Luong’s <a href="http://www.largeformatphotography.info/how-to-focus.html">article</a> on the Large Format site describes that method and some others, and includes a link to a copy of Bond’s article. Much the same result obtains from setting tilt and adjusting focus until near and far points are sharp; I prefer that approach because I have much finer control with the focusing ring that I do with the tilt knob. But I’ve found either approach a lot easier than playing with tilt and focus simultaneously. This approach works especially well with Live View (which, alas, I do not have) because you can set a magnified view of the near or far point, get it sharp, set a magnified view of the other point, and check its sharpness without constantly moving the view between the near and far points.</p>

<p>I agree that the smallish angle scales make setting precise tilt difficult, but you have the same problem with most view-camera scales, and most field cameras don’t even have scales. With a purely visual method such as Bond’s, though, this isn’t an issue.</p>

<p>By far the greater challenge with most 3-D scenes is finding the optimal tilt, especially if the scene includes any significant height. The only way to be sure is to use the approach QT describes: set a tilt, set the PoF and check the required f-number; set a different tilt, check the required f-number, and compare, repeating until you get the smallest f- number (or at least one that is acceptable).</p>

<p>How do you find the required f-number? It’s possible to play with the aperture and estimate when everything is sharp, but I have a hard time doing this with a small, dark viewfinder. At least for me, the easiest way is to use the lens DoF scale, much as with any manual-focus lens. One caveat: when the lens is tilted, the marked distances don’t mean anything, so you note the near and far positions of the focusing ring rather than the marked distances. Unfortunately, the DoF scales on the new 17 mm and 24 mm lenses are so small that they’re mighty tough to use for this purpose (I have the older 24 mm). Another caveat: the “near” and “far” points are angular distances from the image plane rather than distances from the camera, so learning to recognize them takes a bit of practice. Once you get used to it, though, it’s fairly easy.</p>

<p>In any event, though I think it’s a good idea to read Merklinger to understand what’s going on, a method like Bond’s is usually a much easier way to set tilt in practical photography. As always, of course, YMMV.</p>

[zml]

<p>The lack of scale for swings on view cameras (even though many do have scales) is irrelevant to focusing because of three things: a large focusing screen; a focusing loupe, and a piece of black cloth. These three things used in unison allow a very accurate settings using whatever method one chooses. Also, most photographers, especially in a studio, simply memorize their settings and/or keep good notes; degree and direction of a swing (most studio cameras swing in all directions), focus setting, f/stop, rear standard setting (again, studio cams often have rear movements as well.) For those who photograph landscapes and want everything in focus from the flowers at their feet to the furthest mountains, it is really easyto tilt/shift (geometric distortions are largely irrelevant) but if you do product photography and need tilts/shifts it gets a bit more complex to avoid turning a soccer ball (round) into a football (American) ball, which is not round.</p>

[peter_e]

<p>Two things I might add:<br>

1. With all Canon TS lenses, it is possible to change focus with one hand, and tilt/swing with the other, while watching the result in the finder or in live view, and applying proper focusing technique by focusing beyond focus and back through multiple iterations.<br>

2. I take multiple images with different focus/fstop combinations, in particular of 3D scenes where the the effect is difficult to predict and perhaps also difficult to observe in life view.</p>

[dan_south]

The lens plane is separated from the film/sensor plane by the focal length at the camera's level. If you are standing and

the camera is approiximately five feet off the ground, and if your focal length is 24 mm (one inch), the theorem yields a

very tall and narrow triangle. I find that approximately one degree of tilt works well in this scenario.

 

If your are working close to the ground (subject plane) more tilt till be required. I can't recall using more than three

degrees of tilt on my TS-E 24 II, but regardless that small amount of tilt can make a huge difference.

 

Here is my focusing method.

 

IMPORTANT: First, know clearly how your focusing ring rotates in order to focus on distant versus close objects. I focus

my TS-E 24 II with my right hand reaching around from behind the camera. To focus FARTHER away I rotate the lens to

the LEFT, i.e. in the direction of my fingers. To focus on a NEAR object I rotate the lens to the RIGHT, i.e. in the direction

of my thumb. (The direction of the focus ring may vary by the model or brand of lens that you are using, so please verify

this before attempting to use tilts/swings.)

 

1. Compose the scene. The camera may be set level or tilted toward the subject. It doesn't matter.

 

2. Begin with no tilt at all. Optionally (not required, but saves time) begin with an initial tilt based upon the theorem for this

focal length and the height of the camera above the subject plane (for 24 mm, 1 degree at standing height, 2-3 degrees

when close to the ground). Either approach will work. An initial tilt is not required.

 

3. Use live view to focus on a near object or point on the subject plane.

 

4. Noting which way you need to rotate the focus ring to do this, focus on a distant object/point on the subject plane.

 

5a. If in Step 4 you rotated in the FAR focus direction (left for Canon) to bring the distant object into focus, you need

MORE tilt.

 

5b. If you rotated in the NEAR focus direction (right for Canon) to bring the distant object into focus, you need LESS tilt.

 

5c. If you didn't need to turn the focus ring at all (the distant point is already in sharp focus), the tilt is correct and focusing

is complete. Give yourself a high five and make your exposure.

 

6. Adjust lens tilt as described in Steps 5a and b, and repeat from Step 3 until focusing is complete.