Rise/shift the same as raising the camera or relying in lens geometry distortion?

[Vaughn]

...
Of course it wouldn't be like just moving the camera a few inches higher. That much is empirically evident and logically deducible. But is it equivalent to moving the camera up a meter or three (or lowering it)?
Any helpfull visualisations or analogies?

My best guess is that the amount of object size reduction onto the film plays a part in that. If the size of an object is reduced from 10 foot tall to an image 1 foot tall (thinking of an 11x14 camera), that is a 10 to 1 reduction. So in this case, would one inch of front rise be the equivilent of raising the camera 10 inches (but without greatly changing the point of view)?

 

[MattKing]

What is the ratio from camera rise to moving the whole camera higher?

I, for one, am only 6 feet tall.
To get the same effect from moving the camera higher, you would have to move it several/many feet - at least with a building.
By the way, the angle of view of a lens is related to both the focal length and the size of the film/sensor being used with it.
An ideal lens (which doesn't exist) would have a magnification that is set by the focal length, and a field of view/angle of view determined by the film size. Given a single film/sensor to lens distance (the focal length), a big sheet of film would give you a wide angle field of view/angle of view. A little tiny sensor would give you a narrow field of view/angle of view, and a a middle size of film or sensor would be in between.
Real world lenses have physical and optical limitations that mean the big sheet of film would only have an image on the centre part, and only have a sharp image on the centre of that.

 

[greg zinselmeier]

We are distinguishing between lens aberrations and moving the center of the lens projection right, left, up, down on the GG. Correct?

IMO, I would not consider rise/fall or shift to be a “distortion “ . ???? But who am to blow against the wind.

I consider lens aberrations to cause distortion.

Geometry of an image or correcting geometry I do not consider “distortion.

For me, I realize this is a semantic issue, which most people will not agree with me, which is ok

 

[Mr Bill]

If the camera is tilted up to capture the top of the building, the lines convergence.
The fix for such situations is usually just to get a wide lens and point it straight ahead. But you get too much of the ground that way.
Solution: Only use the mid and top of a wide lens.

Yes, agreed.

Now, with the "wide lens pointed straight ahead imagine hypothetically removing the film/camera from the situation. This leaves only the lens plus a flat image plane onto which the image is projected. Obviously the image of the building now continues to occupy the same part of the image plane as previously. If you could now place a piece of film to cover only that area of interest, then no additional film need be wasted.

This is, in essence, what the view camera with a rising lens does. It sets a relationship between the lens and the piece of film such that you don't use the entire image circle of the lens, but only the portion that contains the image of the building.

Rectilinear correction would indeed seem to have something to do with it.

Well I think that most people would agree that a pinhole lens is always rectilinear. But with respect to the sides of the building converging (or not) a pinhole lens will behave the same way as a regular sort of lens - if the film plane is parallel to the face of the building the sides of the building will not converge. But if the film plane is tilted back or forward, then the sides of the building will converge or diverge on the image.

 

[flavio81]

That is, you simply get to see a diffrent part of the projected image?

Yes!

This is exactly what happens when you use shift lenses (on 35mm and medium format cameras). It's equivalent to rising or lowering the front of a view camera. Or laterally shifting it.

Geometry distortion can't be that great with a normal lens as to warrant such a feature.

But the goal (well, one of the possible goals) is not to correct the distortion generated by the lens (i.e. barrel/pincushion), but the perspective distortion.

Of course it wouldn't be like just moving the camera a few inches higher. That much is empirically evident and logically deducible. But is it equivalent to moving the camera up a meter or three (or lowering it)?

Yes, it is equivalent to choosing a different, physical location of your camera!!

That's why I like shift (or tilt/shift) lenses. They achieve impossible viewpoints such as simulating taking the picture as if you were located under the ground.

 

[flavio81]

No. A 50 is a 50. No matter where you use it. Coverage or what is considered wide on a certain format is something completely different. Two lenses of the same focal length can have different angle of coverage.

Just to nitpick, and to make sure nobody is confused...

One can have two focal length lenses of 50mm. But their coverage and their angle of view can be completely different.

For example a 50mm for the 135 format will have a coverage of at least 24x36mm, and the angle of view at such coverage will be 46 deg. diagonal.

a typical 50mm for the 6x7 format will have a coverage of at least 55x70mm, and the angle of view at such coverage will be 84 deg. diagonal.

And yes, the latter 50mm could in theory be mounted in a 135-format camera and be used as a "shift" lens of 46-deg angle of view, due to the greater coverage.

 

[Sirius Glass]

I, for one, am only 6 feet tall.
To get the same effect from moving the camera higher, you would have to move it several/many feet - at least with a building.
By the way, the angle of view of a lens is related to both the focal length and the size of the film/sensor being used with it.
An ideal lens (which doesn't exist) would have a magnification that is set by the focal length, and a field of view/angle of view determined by the film size. Given a single film/sensor to lens distance (the focal length), a big sheet of film would give you a wide angle field of view/angle of view. A little tiny sensor would give you a narrow field of view/angle of view, and a a middle size of film or sensor would be in between.
Real world lenses have physical and optical limitations that mean the big sheet of film would only have an image on the centre part, and only have a sharp image on the centre of that.

Hey, I feet cheated! I am height disadvantaged at 5'6" and the difference is even bigger.

 

[Doremus Scudder]

It's not nearly as complicated as it may seem, there's just a lot going on.

Some fun facts:

The focal length of the lens is independent of the size of the image circle projected. Lens design makes the difference. For a given focal length you can have a lens that just barely covers the film format or one that is much, much larger. With view cameras, we like lenses with lots of coverage when we need to use movements. That doesn't change the size of the image (magnification) at all.

"Wide angle" is ambiguous. It can refer to a wide angle of view (think short focal length for the film format) or it can refer to the angle of projection. Lenses with wide angles of projection have larger image circles than ones with smaller angles of projection. I like to think in terms of "short," "normal" and "long" focal lengths for a given film format and leave "wide angle" alone unless I qualify it with "angle of view" or "angle of projection."

Distortion usually refers to defects in a lens' rendering. Barrel distortions makes a square look convexly rounded, pincushion distortion makes it look concavely rounded. Chromatic abberation means different colors don't focus at the same place, making a softer image, etc. Perspective is a different thing, and converging verticals falls under that category.

Perspective we can define as the relative sizes and positions of objects projected on a two-dimensional surface. Viewpoint determines perspective. Changing the angle of the film plane relative to the subject also changes perspective; the part of the film that is moved farther from the subject will have a smaller image than before it was moved.

Now, on to rise/shift: Think of the lens as a projector and the film as a screen. With a view camera we usually have a lens that projects an image that is significantly larger than the "screen." That means, we can move the projector (or the screen) around to use the portion of the image circle we want. Cameras that don't have movements don't need lenses with large image circles. View cameras, with all their movements, can take full advantage image circles that are much larger than the film format.

Let's look at the classic scenario. We have a tall building. If we tip the whole camera to point up to get all the building in, the verticals will converge because the film plane isn't parallel to the building façade. The part of the film that's farthest from the plane of the façade will have a smaller image than the part that's closest. (Note, this isn't "distortion," it's the perspective we should expect when the film plane isn't parallel to the subject plane).

If we want the verticals to be parallel, the solution is to keep the film (camera back) parallel to the façade, but then, we don't have the top of the building on the ground glass when the camera is in "zero" position. But, remember all that extra coverage the lens has? There's lots more image to use; the top of building is there, we just have to put it on the film. So we raise the lens (the projector) to put that part of the building on the ground glass (the screen).

Note that we do need a lens that is the right focal length and has a generous enough image circle to be able to do this from any given camera position. Still, you can do this with short, normal and long focal length lenses.

What's the difference between changing camera position by moving the camera up and using front rise with a lower camera position? In the first case, the optical center of the image will be centered on the film, since the lens axis intersects the film in the center. With front rise, we're moving the lens axis, so the optical center will appear lower in the final image. The looks are very different. The same happens with shift; if you shift lens or back, the optical center of the image (perspective wise) is no longer in the center of the film.

Best,

Doremus

 

[greg zinselmeier]

Jolly good show!! Doremus

 

[grat]

Addressing solely the perspective shift of shift/rise, picture the lens as having two image circles, with their "cones" converging in the middle of the optical system. One image circle is the small one projected onto the film plane. The second image circle is the plane of sharp focus.

By shifting the lens relative to the film plane, the effective movement of the field of view is magnified far more than you could achieve by simply moving the camera.

I'm not sure I explained that well-- in my head, it makes perfect sense.

 

[Sean Mac]

Draw the ray path.

Remember the image projected is inverted.

Left for right and up for down.

It's obvious how the displacement from center expands the upper part of the image.

Thereby restoring the parallels we know exist.

 

[DREW WILEY]

Don't confuse wide-angle lenses with larger image circles per se. Generally, longer focal length lenses have bigger image circles, and nearly always with less distortion, but it depends on the specific design. Half an hour hands-on with a view camera and a relevant selection of lenses, and an experienced operator, would teach you far more than reading a hundred posts about it. Seek out that kind of opportunity if you can.

 

[Sirius Glass]

Posts, descriptions, books are not very helpful. I only really figured it out by going out in the field with a camera and playing with it.

 

[Kilgallb]

I'm a bit confused about the exact physics/geometry/optics of this often tooted feature of LF cameras.
It's not very well explained anywhere I can find. And the explanations that is there, is often contradictory.

- Is it due to geometric "distortion" of the lens. As in that it gets easier to do with wideangle lenses?

- Or is it due to actually moving the image circle? That is, you simply get to see a diffrent part of the projected image?

Some old folding cameras have front rise with a normal lens. Patent Etui for example.
That speaks for the latter. Geometry distortion can't be that great with a normal lens as to warrant such a feature.

I have a hard time wrapping my mind around what is going on.
Of course it wouldn't be like just moving the camera a few inches higher. That much is empirically evident and logically deducible. But is it equivalent to moving the camera up a meter or three (or lowering it)?
Any helpfull visualisations or analogies?

I'm a bit confused about the exact physics/geometry/optics of this often tooted feature of LF cameras.
It's not very well explained anywhere I can find. And the explanations that is there, is often contradictory.

- Is it due to geometric "distortion" of the lens. As in that it gets easier to do with wideangle lenses?

- Or is it due to actually moving the image circle? That is, you simply get to see a diffrent part of the projected image?

Some old folding cameras have front rise with a normal lens. Patent Etui for example.
That speaks for the latter. Geometry distortion can't be that great with a normal lens as to warrant such a feature.

I have a hard time wrapping my mind around what is going on.
Of course it wouldn't be like just moving the camera a few inches higher. That much is empirically evident and logically deducible. But is it equivalent to moving the camera up a meter or three (or lowering it)?
Any helpfull visualisations or analogies?

Or is it due to actually moving the image circle? That is, you simply get to see a diffrent part of the projected image?

thats it

 

[brbo]

Draw the ray path.

Remember the image projected is inverted.

Left for right and up for down.

It's obvious how the displacement from center expands the upper part of the image.

Thereby restoring the parallels we know exist.

Nothing gets "restored" when you move the sensor/film into different part of the image circle. Parallels are always parallel (assuming rectilinear lens and parallel film to subject plane) in the image circle.

 

[Sean Mac]

You are making the assumption.

 

[Doremus Scudder]

Addressing solely the perspective shift of shift/rise, picture the lens as having two image circles, with their "cones" converging in the middle of the optical system. One image circle is the small one projected onto the film plane. The second image circle is the plane of sharp focus.

By shifting the lens relative to the film plane, the effective movement of the field of view is magnified far more than you could achieve by simply moving the camera.

I'm not sure I explained that well-- in my head, it makes perfect sense.

I think maybe you're talking about the distance at the plane of sharp focus that is displaced when you use lens shift/rise? Magnification may be the wrong term here; focal length and subject-to-film distance determine magnification ratio.

But yes, you'd have to move the camera a whole lot to get a similar (not identical, though) framing to what you get when shifting/raising the lens just a small amount.

Doremus

 

[Alex Benjamin]

The reason why the lines are converging when you are shooting a tall building is that the top of the building is further from the film plane then the bottom of the building. Therefore, it appears smaller, as objects that are further always do. If you simply move your camera higher, you still get the same problem, so not a solution.

Same thing, by the way, happens on the side walls of the building, noticeable if the building is long.

What happens when you use front rise is that you are correcting that aberration by reducing the distance between the farthest aspects of the building—the top—, equating it with the closer one—the bottom—in relation to the film plane. Why do you need so little rise to achieve this? Because all distances are relative to each other, i.e., the relation is the same from the object to the lens, and from the lens to the film plane (always remembering that the image is upside down on the film plane).

 

[Saganich]

You can take advantage of lens vignetting when moving the circle to the top of the frame...I read that somewhere.

 

[xkaes]

It's not nearly as complicated as it may seem, there's just a lot going on.

Some fun facts:

The focal length of the lens is independent of the size of the image circle projected.

Doremus

Exactly. The focal length, the image circle, and the picture angle are three different things.

You can have a wide angle lens (a wide angle of coverage) and a long lens (a narrow angle of coverage) that both have the same image circle.

You can also have two different lenses of the same focal length with completely different image circles.

 

[flavio81]

Or is it due to actually moving the image circle? That is, you simply get to see a diffrent part of the projected image?

thats it

Yes, you move the film position so it sees a different part within the image circle.

 

[AgX]

What is the ratio from camera rise to moving the whole camera higher?

At the typical employement of rise, to avoid converging lines at architecture:


At raising the complete camera you still look rectangularly at the facade, but have lesser foreground in view. As typically you have not much means for such raising the gain though would be small.

At rise just of the front standard, which can be more effective concerning avoiding foreground, the perspective changes, at same point of view, to looking obliquely at the facade. This becomes visible at protruding parts of the facade as sills. The same time keeping both standards parallel to the facade evokes the impression of looking rectangularly at it. These contradicting impressions are not natural.

In general shifted architectural photos may look not natural, even worse if taken not rectangalarly to a facade but along a corner of a building. A good architectural photographer still may find a pleasing outcome by emplyoing camera movements,

 

[David A. Goldfarb]

Doremus and Flavio81 have the clearest explanations thus far.

Yes, rise/fall/shift are just moving the film to a different part of the image circle relative to the lens. Lenses for fixed-lens cameras usually try to restrict the image circle as much as possible to maximize contrast by eliminating almost all non-image light. Large format camera lenses can have larger image circles to make use of the camera's movements, and after the movements are applied, you can use a compendium shade to block off any non-image light while accounting for camera movements to maximize contrast, if that's a concern.

A good way to imagine the difference between moving the camera several feet vs. using rise/fall/shift, is to imagine taking a photograph of a field freshly ploughed in straight rows perpendicular to the film and lens plane. The row that coincides with the optical center of the lens will be straight, and the other rows will converge. By using shift (no ladders required) and changing the camera position, you can control the composition and location of the vanishing point in the image while keeping the horizon straight. You can decide whether the lines should converge in the distance in the center of the image by keeping the camera in the zero positions, or move the camera to one end of the field or the other, right or left, to change the vanishing point, but use shift to keep the same composition and horizon as you had with the camera centered in the field.

Go out and experiment with the camera, and you'll see what I mean.

 

[AgX]

These are the issues Helge seemingly has problems with (and some textbook authors too...)


The various forms of distorsions:

-) perspective distorsion
--) due to point of view (basically scale issues)
--) due to direction of view (basically angle issues)

-) optical distorsions (straightness of lines at 2D-subject seen rectangularly)


-) the relation between FL, lens design (angle of coverage), effective angle of view and format

 

[Doremus Scudder]

These are the issues Helge seemingly has problems with (and some textbook authors too...)


The various forms of distorsions:

-) perspective distorsion
--) due to point of view (basically scale issues)
--) due to direction of view (basically angle issues)

-) optical distorsions (straightness of lines at 2D-subject seen rectangularly)


-) the relation between FL, lens design (angle of coverage), effective angle of view and format

I think the ambiguity of the word "distortion" is confusing the discussion here somewhat. Strictly speaking, changes in perspective due to changes in viewpoint are not "distortions." Changing the orientation of the film plane relative to the subject causes differences in the relative sizes of objects on the ground glass. This is a "projection distortion" similar to that in Mercator-projection world maps (where Greenland ends up being much larger than Australia). Correcting or enhancing converging parallel lines falls under this category as well; we're really just changing the relative size of the image from one end/side of the film to another.

Lens distortions like a lack of rectilinearity or the distorted (oval) rendering of circular objects at the edge of the field of view with short focal-length lenses are really outside of the discussion.

I find it helpful to think about manipulating "projection distortion" (usually dealing with parallel lines in architectural work) in terms of positioning the camera back relative to the subject first, and then moving the lens around (shift/rise) to frame the image as a second step.

Deciding on the position of the optical center of the image, as you and David discuss, is often an image-making goal in itself, but often ends up being a side effect of correcting convergence. Anytime we use rise or shift, we end up displacing the lens axis from the center of the image, thus moving the "optical center" of the image away from the actual center of the frame.

Here's an image where I intentionally used shift to keep the left arched doorway on the lens axis. Note the "straight-on" viewpoint of the door compared to the obvious side view of the middle door. The result is an intentional "distortion," if you will, which gives the rendering a more geometric-drawing feeling than a straight-on image would.

The same thing happens when using rise, just displaced 90°; the "optical center of such images is always lower than the center of the image. This is the main difference between using front rise as opposed to just moving the whole camera up to frame the image while keeping the verticals parallel.

Best,

Doremus