4
\$\begingroup\$

A lens with a tilt adjustment is capable of producing a plane of sharpest focus that is not parallel to the image sensor. Could a lens with no tilt adjustment produce the same image, if many images with different focal distances were stitched together?

I've included a picture to help illustrate the question. I am asking from a technical standpoint, what effects or adjustments on the camera would cause differences, and how would they manifest.

and have ideas for developing more

EDIT

I thought more about the question and I think the question I am trying to ask is if the wedge shaped depth of field (Scheimpflug principle) produced by the tilt of a tilt shift lens is because the region of focus is getting closer to the camera (less depth of field) or because of another effect.

Improve this question
\$\endgroup\$
5
  • \$\begingroup\$ A tilted lens also affects the perspective (usually correcting diverging lines). Your technique wouldn't do that, so while you might achieve focus throughout your subject, it's not realy matching a tilt-shift lens. Yes you'll get the subject "plane" in focus, but you will get a very different image I would think \$\endgroup\$
    – MikeW
    Commented Mar 2, 2013 at 20:46
  • 1
    \$\begingroup\$ Yes, I've excluded shift since that is easy to produce electronically. I thought that tilt only changed the regions of focus, and not the perspective? \$\endgroup\$
    – Phil
    Commented Mar 2, 2013 at 21:13
  • \$\begingroup\$ There are lenses that tilt, and ones that shift, and those that do both. I think it's the tilting that accomplishes the change in perspective - en.wikipedia.org/wiki/Perspective_control_lens - I edited your title, you may want to correct if I've misinterpreted \$\endgroup\$
    – MikeW
    Commented Mar 2, 2013 at 21:22
  • 2
    \$\begingroup\$ @MikeW - As far as movements go, it's the shift that changes the geometry. (Oy—complicated bit here.) It's actually the orientation of the sensor in relation to the subject that determines the shape of the projection. When shooting a tall building from ground level, you'd ordinarily have to point the camera upwards, which would keystone the building. The shift allows you to shoot with the sensor parallel to the building, keeping things square, but move the sensor lower in the lens's image circle. (Character limit coming up; continuing...) \$\endgroup\$
    – user2719
    Commented Mar 2, 2013 at 21:47
  • 2
    \$\begingroup\$ (con'd) The effect is similar to taking a very wide-angle shot that includes a lot of foreground, then cropping down to just the building. The relative tilts of the sensor plane and the lens plane determine the plane of focus. In a perspective-correction (only) shot, the sensor, the lens plane and the subject are all parallel. (Do note, though, that if the lens does not deliberately distort to produce unnatural recitiliearity—and most normal lenses do—you would have to use some small amount of sensor tilt to artificially widen the top of the building, even with everything parallel.) \$\endgroup\$
    – user2719
    Commented Mar 2, 2013 at 21:52

1 Answer 1

Reset to default
2
\$\begingroup\$

You could, but it wouldn't work exactly as your illustration suggests except in very simple scenes.

If the aim is to produce an image that has as much in focus as possible (the traditional purpose of tilt movements), then focus stacking without regard for an imaginary plane of focus is a simple, effective and highly automatable solution. Despite the fact that (OMG!!!) you'll lose some of the original sensor data, the mere fact that things aren't poking inconveniently in and out of a real plane of focus that is limited by real-world physics makes the image better than the real thing. (You only get so much depth of field, even with a view camera's movements and f/64.)

But when you're trying to place a distinct plane of focus (along with out-of-focus areas) into a complex scene, there would be a lot of manual work and you'd need to have a pretty good mental model of what an actual tilt image of the scene would look like. It would almost be easier, I think, to do a regular "get everything sharp" focus stack, then take the resulting composite image into an application (or module, plugin, or whatever) that allows you to apply a depth map and add lens blur. The process would be similar with a stack of images; you'd need to paint in the things that should be in focus, as well as the bits from other images that are appropriately out of focus for a given part of the image (and would probably have to do some gradient blending as well to get the right bokeh transitions).

Improve this answer
\$\endgroup\$
3
  • \$\begingroup\$ Thank you for the answer. I've added some more detail to the question. The part I am still having trouble understanding is if the variance in depth of field across the image is appropriately modeled by a composite 'panorama' of thin strips of identically framed images taken at different focal distances. \$\endgroup\$
    – Phil
    Commented Mar 2, 2013 at 23:00
  • 2
    \$\begingroup\$ @Phil - It would be -- it's just the transitions that'll kill you if you try this in practice unless you take a lot of images and make your strips very, very narrow—at least at the close end of the "wedge", where DoF is very narrow. You'd need to do a lot of blending to avoid the appearance of focus jumps. \$\endgroup\$
    – user2719
    Commented Mar 2, 2013 at 23:20
  • \$\begingroup\$ Don't some of the focus stacking applications available automate a lot of that blending? \$\endgroup\$
    – Michael C
    Commented Feb 23, 2016 at 3:41

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.