unnatural perspective

[StevenJohn]

Is it just me, or does using shift to get perfectly straight vertical building seem a little unnatural? To me, a taller building should have some vertical convergence because the top of the building is further away.

[Louie Powell]

It is certainly easy to overdo vertical shift. I've seen a few instances when it seemed to me that the image appeared to be top-heavy even though close examination showed that the verticals were absolutely vertical.

However, the vertical distortion that one gets by NOT using shift is very uncomfortable to me. I find myself making perspective corrections in my digital images to restore what I believe to be a natural perspective.

Ultimately, I suspect that this is something that large format photographers train themselves to see. When I used only 35mm, I rarely noticed vertical distortion. But having used large format almost exclusively for the last 5-7 years, it's something that really bothers me if it isn't corrected.

[StevenJohn]

I agree that an uncorrected perspective can be bothersome as well. Although in the past, I have used the convergence to my advantage . I have a medium format shot of a lighthouse where the uncorrected perspective really adds to the feeling of height in the tower.

[Preston]

Maybe I'm confused, here, but wouldn't vertical shift simply adjust the position of the image on the film, rather than correct perspective? It seems to me that in order to correct convergence of vertical lines, would not one tilt the camera back so that it is parallel to the subject?

--P

[Nathan Potter]

For the situation described, perfectly straight vertical buildings, front and rear standards need to be parallel to the subject plane. The degree of departure from parallelism is a matter of personal preference and the visual statement that the photographer is trying to make.

Nate Potter, Austin TX.

[Bruce Watson]

Is it just me, or does using shift to get perfectly straight vertical building seem a little unnatural?

Camera movements are tools. Use them any way you want to make the image you want to make.

You aren't required to set the film plane exactly parallel to the face of the building. You can, but you don't have to. The joy of LF is you can typically dial in the amount of convergence you want to use.

What I do is first set up with the film plane parallel to the subject plane. Then I'll tilt the back a bit to induce the amount of convergence I want in my photograph. I like a bit of convergence -- way less than I'd get from just tilting the whole camera back, but more than none. Clearly, YMMV.

[N Dhananjay]

I think the reference here is to pointing the camera upwards (or more generally, at an angle to a receding plane). Doing so causes the lines in the plane to recede to a vanishing point. An LF camera gives you ways to avoid doing this - wither using tilt or by keeping the camera square to the plane and using rise/fall.

Convergence that is significant does not bother me but becomes part of the abstract structure of the image. But mild recession can be quite problematic. If a building facade shows mild convergence, it will appear to be falling over backwards. Another way of illustrating this - if you stand across the street from a skyscraper and want to get the whole building in the frame, you would point the camera up at a significant angle and the top of the building will be substantially smaller than the bottom, but that is fine because our eye-brain combination sees things this way and accepts it. But if you are standing across the street from a small building and point the camera up slightly to get the full building in the frame, you have mild convergence. the top of the building seems slightly smaller than the bottom. But at such small angles in normal vision, your eye-brain combinations corrects and presents you with a mental image of the top and bottom being the same size. So, when you see a picture of a building with mild convergence which from past visual experience you know you would see without any convergence, there is a tendency to interpret the top as being further away (i.e., the building is faller over backwards).

Hope this helps.
Cheers, DJ

[Joanna Carter]

Basic rule of thumb: if you are selling the image to an architect, the sides of a building should be perfectly parallel, no matter what the final appearance; if you are selling to a designer, then convergence is an advantage to accentuate the impact of the shot.

Think of it as function vs art.

There is also another guideline that, if you have to tilt your head back to see the top of the building, then you should allow some convergence (unless your client is an architect :rolleyes: ).

[Andrew O'Neill]

I agree that when making sure the front and back of the camera are parallel to the subject, in this case a building, the perspective looks a bit off. I like to leave a wee bit of keystoning when I shoot buildings, as that is how our eyes see...

[J. E. Brown]

Good discussion here. To add to it:

Is there a priority of when you should use rise/fall versus vertical shift with film/lens planes to keep things square? Does it depend on the scenario you are faced with? I generally set up the tripod, level it, then raise or lower the camera as needed, and fine tune things with rise/fall in order to keep things parallel. I assume when you have a very tall building, that would not be practical, though...hmm.

Anyway, I think that convergence can definitely be an effective tool in a photo, if that is what you want (think of all the effects we have at our disposal with various lenses and film and lighting, just another tool in the bag). The fantastic thing about most LF is we don't have to put up with it if we don't want it. :)

[Jack Dahlgren]

[QUOTE=J. E. Brown;617626]Good discussion here. To add to it:

Is there a priority of when you should use rise/fall versus vertical shift with film/lens planes to keep things square? Does it depend on the scenario you are faced with?QUOTE]

Rise and Fall are the same as "vertical shift". Typically shift refers to horizontal movement and rise and fall are vertical.

Rise, fall and shift are all planar to the lens board.

[EdWorkman]

When one looks at a building one knows that it has vertical faces [assuming a regular bldg blah blah, -a box] When one looks at a picture of that building made by tilting the camera-expecially using a wider view lens -it looks like the building is pointy or falling backward. But assuming the same viewing point the perspective is exactly the same. So unless you are depicting a Frank Gehry creation, minimize the convergence or we'll assume you took the picture with your phone, or worse, or make it reeeeeeeeally artsy.

[EdWorkman]

expecially ?????

[Toyon]

Andreas Feininger said that you should correct about 70% of converging verticals, otherwise it looks unnatural.

[StevenJohn]

EdWorkman-
One also knows that lines converge at infinity from the viewer. Railroad tracks converge to a point on the horizon, but we know they are a constant width apart. This is why, in my opinion, I think perfectly vertical lines on a tall building looks unnatural.

[EdWorkman]

Yahbut- when you look at a box building you, or most people, don't think it's pointy or falling over backward, as your brain should compensate and tell you that the building is just fine. On a 2D print your brain sees triangles, and then should ask "what the heck happened to the building?" Note I didn't say "perfectly" vertical or parallel, I said minimize convergence. Sure if you are really close to a really tall building a little convergence is okay. If you are looking at a 4 story building not very much convergence is NOT okay, make it almost none. Above, Architects were mentioned- I am a Structural Engineer and am familiar with buildings- I about puked when the Orange County Register published a "phone" picture of a UCI bldg I participated in- supposedly a documentary photo but it totally destroyed information about the structure.
Like I also said , if you wanna destroy reality, do it thoroughly so it's "art" and can't be confused with the portrayal of a building as a building. But hey, no reason for you not to do what you want to, and if you make it look like it's falling backward, I'll laugh and point.

[StevenJohn]

Hi Ed,

I'm an engineer too, but a little different strain (ME).

I mostly agree. A four story building wouldn't show much convergence if you were standing a little ways away from the building. Now if I were standing a foot away from the base of the building and looking up, I would expect to see just a little convergence. This is of course exaggerated with wide angle lenses.

Another example would be a skyscraper. Taking a picture of the TransAtlantic building from across the bay, it should appear vertical. Stand at the base and shoot up, and it would converge.

What determines the convergence is where you're looking to make your infinite point. If you're across the bay and making the horizon your infinite point, the skyscraper's vertical lines should be parallel and vertical. If you're standing close to the skyscraper, looking up, and making the top of the building (or a projected point in the sky beyond the building) a infinite point, then the vertical lines of the building would start to converge on the infinite point.

I guess what I see as being unnatural isn't just a non-converging building, but a non-converging building when it should converge...a closely shot wide angle looking up.

[StevenJohn]

Actually, one of the most impressive optical illusions I ever saw was aided by convergence. I once stood under the St. Louis Arch and looked up. Clouds were passing overhead parallel with the two legs of the arch. The arch thins at the top, but with the convergence and clouds passing over, it made the arch appear to be falling over.

[Robert Hughes]

with the convergence and clouds passing over, it made the arch appear to be falling over.
I've noticed that too on the St Louis Arch, also the Sears Building in Chicago. I get a feeling similar to that nauseated loss of balance I feel standing on a cliff edge.

[John Rice]

Like I also said , if you wanna destroy reality, do it thoroughly so it's "art" and can't be confused with the portrayal of a building as a building. But hey, no reason for you not to do what you want to, and if you make it look like it's falling backward, I'll laugh and point.
That's my general approach. I've shot a lot of Architecture and I always strove to get verticals absolutely correct, except when they were far from correct. I have always viewed slightly off verticals as sloppy work. Having said that, there is more to it. The problem I see an awful lot, especially with house exteriors in the newspaper is that the photographer used too short a lens and typically shot the house from the sidewalk on the same side of the street as the house. This looks awful and unnatural to me. It does make the house look "big" because it is looming over you, but I still think it looks awful.

If you shoot the building from the right location, you can keep verticals correct and have it look great. This might not include extremely tall buildings. But I believe any architectural subject can be photographed best with verticals correct, so long as all other aspect are handled well. Too many photographers get wrapped up in the ability of the camera to correct verticals without thinking beyond that.

[Kirk Gittings]

Andreas Feininger said that you should correct about 70% of converging verticals, otherwise it looks unnatural.

And yet a quick review of his images on the web shows nearly perfect correction on the vast majority of his images that could be corrected (barring images that are pointed down or up radically).

http://farm3.static.flickr.com/2594/4129369914_6b0c1f73ca_b.jpg


There are "conventions" in professional architectural photography that generally call for perfect correction even when it makes the top of the building look odd. Feninger in the final analysis, trained as an architect, followed convention.

[tgtaylor]

Maybe I'm confused, here, but wouldn't vertical shift simply adjust the position of the image on the film, rather than correct perspective? It seems to me that in order to correct convergence of vertical lines, would not one tilt the camera back so that it is parallel to the subject?

--P

Since the top of a tall building is at a greater distance from the lens (or the human eye) than the bottom of the building, you see convergence. To correct for that convergence you tilt the back of the camera to change the magnification on the top or bottom of the building on the ground glass and, hence, film plane. Vertical rise or fall has - as you surmised - no effect on convergence; it simply allows you to place the whole building on the film plane without tilting the camera and introducing a greater convergence than what already exists.

Although a simple to understand concept, in practice it becomes complicated to apply since you have many choices of how to handle the convergence with a camera with ample movements. it boils down to how you like your convergence? Extreme, as when pointing a camera straight up or even amplifying with movements, or none. Or somewhere in between?

It’s quandry that's Guarantee to keep you under the focusing cloth composing and focusing.

Thomas

[John Rice]

I'm a little baffled where all the fancy explanations are coming from. They seem to lend more confusion than answers. Buildings are generally vertical. 90 degrees to a horizontal line. If the the film is vertically parallel to the building, there will be no vertical convergence, even though one end of the building will almost always be further from the film than the other end. End of story. Rise and fall of the lens relative to the film (you can raise or lower either or both) allows you to place the building where you want on the film.

EDIT: As I read again and again over these posts, I'm getting the impression we are talking about difference of technique. When shooting "correct" architecture, you simply never tilt the camera. Normally (again, talking about shooting technically "correct" architecture) you set up the camera, level it front to back and side to side and leave it that way so you have no vertical convergence and the plane of focus is also perfectly vertical. Then you use rise and fall to place the building where you want within the frame.

[Emmanuel BIGLER]

Hello from France (where we do have some sky-scrapers now ;) )

I'm coming late to this discussion, and I agree with Louie Powell :
It is certainly easy to overdo vertical shift. I've seen a few instances when it seemed to me that the image appeared to be top-heavy even though close examination showed that the verticals were absolutely vertical.

This illusion of the "ice-cream cone" and "top-heavy" building is documented in some classical courses on LF photography, for example in the book by Pierre Groulx (in French, the author is a Frenach-Canadian professor of photography) there is a nice example of this.
Pierre Groulx : Photographie en grand format,
Modulo, Québec, 2001, ISBN 2891135059

A perfectly "corrected" image of a tall building, taken from a close point of view with a wide-angle lens can look un-natural, "top-heavy" is really perfect to explain the effect (I cannot find a good equivalent in French)
The solution to this is to avoid wide-angle lenses to document tall buildings by taking a point of view from far away :D : but of course this is not always practical....

Since Joanna from the LF-UK group has contributed to his discussion, may be we can point to another discussion on the same suject. There is an interesting example of the illusion by Charles Twist on the LF-UK group, where the lens is not a wide-angle !
The title of the thread is ; Asymmetric movements - gadget or gimmick? (http://www.lf-photo.org.uk/forum/viewtopic.php?f=3&t=819)
See some examples of the effect are shown in the thread, with WA lenses, vaults inside churches,
http://www.lf-photo.org.uk/forum/viewtopic.php?f=3&t=819&p=5399&hilit=St+Pierre+de+Montmartre#p5399 (photo by Charles Twist)
http://www.lf-photo.org.uk/forum/viewtopic.php?f=3&t=819&p=5416&hilit=You%27d+be+hard+pushed+to+spot+things#p5416 (photo by Andrew-of-the-LFUK group)
and an example without a WA angle by Charles Twist
http://www.lf-photo.org.uk/forum/viewtopic.php?f=3&t=819&p=5455&hilit=Andrew%2C+you%E2%80%99re+suggesting+that+the+vignetting#p5455

[StevenJohn]

John-
I agree with your edited post. It is the "technique" which is causing confusion. The top heavy look in the photos that look unnatural to me are from tilting the camera up.

It is sometimes impossible to get the entire building in the frame. I suppose many photographers try to solve this by using a wide angle and tilting the camera up. This changes the building lines from straight 2 point perspective ("correct" technique in your post) to 3 point perspective.

Emmanuel-
Thank you for the examples. The two images in the last link is precisely what I find to look unnatural.