Talk:Perspective distortion (photography)

From Wikipedia, the free encyclopedia
Jump to: navigation, search
WikiProject Film (Rated Start-class)
WikiProject icon This article is within the scope of WikiProject Film. If you would like to participate, please visit the project page, where you can join the discussion and see lists of open tasks and regional and topical task forces. To use this banner, please refer to the documentation. To improve this article, please refer to the guidelines.
Start-Class article Start  This article has been rated as Start-Class on the project's quality scale.
Taskforce icon
This article is supported by the Filmmaking task force.
 

STOP - We're forgetting viewing the presentation image as a necessary component of system perspective distortion[edit]

Nice first paragraph, containing:

"... Perspective distortion is determined by the relative distances at which the image is captured and viewed, and is due to the angle of view of the image (as captured) being either wider or narrower than the angle of view at which the image is viewed, hence the apparent relative distances differing from what is expected ..."

That is, according to the definition of distortion, we are comparing input to a system, to output from a system, and in this case, the input is the photographer's viewpoint or perspective, and the output is the audience's viewpoint or perspective of the presentation image.

BUT no one is following that criteria when subsequently exploring conditions where system perspective distortion may be observed, explained, and intentionally employed or avoided if desired.

An image captured with a wide angle lens is not distorted, it can't be, because it is only the system input.

It must be compared to the system output before assessing system distortion, if any.

It must be compared to the experience of the audience of the resulting presentation image, the system output, in order for there to be a discussion of system distortion, in this case, system perspective distortion.

If the audience positions themselves such that they see the presentation image from proportional distance resulting in an identical angle of view to the photographer's capture angle of view, then there is no distortion.

"... But wide angle lenses distort everything, especially close up ...", someone might say.

No, they do not.

If you get close enough to the resulting printed image, relative in distance that the photographer was from the subject, such that the photographer's angle of view and the print-viewer's angle of view are the same, then the audience will have the same experience of the subject, NO DISTORTION.

Please edit subsequent sections from the first paragraph to ALWAYS consider the definition of system distortion as a comparison of input to output.

See also https://www.edmundoptics.com/resources/application-notes/imaging/distortion/ and http://www.parallelhomeaudio.net/TypesAudioDistortion.html which sadly do not grock that they are comparing input to output, duh.

http://www.encyclopedia.com/science-and-technology/computers-and-electrical-engineering/electrical-engineering/distortion DOES mention system input versus output, but cites no author or published source other than the Columbia Encyclopedia, below, I will edit it slightly to be non-system specific ( audio versus imaging ):

"... [ System ] Distortion ... [ is ] undesired change in [ information ] as it passes from the input to the output of some system or device ... [ and ] results in poor reproduction of [ the original information ] ... ..." -- The Columbia Encyclopedia, 6th edition, The Columbia University Press

I will keep searching academic and research materials that are not readily published or accessible on the web.

But, c'mon, folks, stop thinking that something in reality that we can observe without a camera -- someone's nose looking way-big when we put our eye right up to it -- is "Perspective distortion ( photography )" -- we don't need no camera to see big noses up close, that is NOT distortion, that's reality.

Thank you, carry on -- Photographic Fallacy Hunter --2600:8806:2400:7E00:2CA4:94D1:6BC:80F1 (talk) 05:22, 25 June 2017 (UTC)

Old talk not in a section[edit]

This article previously claimed that perspective distortion was caused by lens focal length, and it had the title "Perspective distortion caused by lens focal length".

This is factually inaccurate, since perspective distortion of the sort described here ( big noses in portraits, and as shown in the images included in the article ) is a function solely of camera to subject distance. Anyone with a camera and lenses of at least two different focal lengths can do an experiment to demonstrate this.

Confusingly, later material in the article (the text describing the photos included in the article) correctly explained that the effect was caused by camera to subject distance, and that reframing identically with a wider angle lens required placing the camera closer to the subject, thus causing this perspective distortion.

I altered the introductory text to correct the factual inaccuracies. I slightly altered the later text describing the images to increase clarity a little. Then I moved the article so that it would not have a title that was factually incorrect. Then I went through the rest of Wikipedia and updated links so that they point to the new location. I've left a redirect at Perspective distortion caused by lens focal length for the usual reasons, and also because the myth that focal length affects perspective distortion is widely believed, and the article might well be searched for under these terms. Jeff Medkeff | Talk 16:11, Jan 31, 2005 (UTC)


Discussion prior to article move follows:

A close up portrait shot with a, say 24 or 28 mm, compared to a portrait of the same person using a 105 is in my opinon the best example of perspective distortion -- all assuming a 35 mm camera ;-). -- Egil 18:52 Jan 29, 2003 (UTC)


nice addition, Egil. could you explain about "pillow" distortion? I've never heard of it. And, yes, it is a 35mm camera. I'll try the portraits when I next get a chance (probably in late february!) That amateur photographer, Koyaanis Qatsi

It is usually called cushion, mea culpa. A perfect wide angle shall never produce curved lines. But even the best of extreme wide angle lenses always produce curved lines to a certain degree, esp. close to the image boundaries. So if you take a picture of a rectangle, it ends up looking like a cushion. A fish-eye is a lens where they have given up correcting this completely. See Aberration in optical systems.-- Egil 19:14 Jan 29, 2003 (UTC)
Right, I knew about the fisheyes. For 35mm cameras, doesn't the designation "fisheye" lens start at 18mm? I had a (one) photography class in June 2002 and then I bought a camera & started taking photos. I'm trying to fact check. Do you think that the bit about curved lines should be removed or rephrased? I take it, it seems misleading. Anyway, I hope I'm not harming the project adding this stuff. Best, and thanks, Koyaanis Qatsi 19:19 Jan 29, 2003 (UTC)
No, no, no! On the contrary, this is extremely useful! In the Wiki world, the initial version of an article is never like the latest - things are developed! I think the distinction wrt. fisheye is the distortion, not the focal length. There are fisheyes > 18 mm and wide angles < 18 mm. But the shorter the focal length, the more difficult it is to get rid of the cushion 80.202.80.14
Interesting stuff. I'd love to see a wide-angle lens >18mm which doesn't cushion the corners. Koyaanis Qatsi

I'm taking the pics back out of the table, as for people with a narrow browser window the table is too wide. By simply sticking them next to each other, we let the browser wrap them into a column -- which is not as satisfying as seeing them in a quad arrangement, but vertical scrolling is much more comfortable for the web reader than horizontal scrolling. --Brion 19:29 Jan 29, 2003 (UTC)


Please note that the link to "Perspective Distortion: Source," added to the bottom of this article points to a new article in which the Conclusion is at variance with the conclusion with this article. Suggest this page is too flawed for minor editing and be replaced with Perspective Distortion, Source. See link at bottom.Pat Kelso 22:57 CST Jan 30, 2004


The title isn't OK this article is about perspective distortion and not lens distorsion (barrel distortion for instance). Ericd 09:51, 7 Mar 2004 (UTC)

Perspective distortion, p e r - s e, has nothing to do with lenses. Please see Perspetive distortion, Conclusion. Thank you. Pat Kelso 18:46, Mar 7, 2004 (UTC)
I will fully disagree, there are several form of perspective distortion, each time perspective is distorded for any cause there is perspective distorsion. The expression clear enough in itself to say what is "THE REAL" perspective distotion. This article deals about the influence of focal length on perpective representation in photography, it's a a important topic in photography, and this kind of distorsion has to do with camera lens. I'm was'nt a great fan of the previous title for the reason exposed before. But I find the present title worse because they are form of distortion (barrel & pillow) caused by the defect of the lens.
BTW what is your new article about ? Seems to be about 3D to 2D projection isn'it ?
If talking photography distortion w/photos, it is not my article ... Pat Kelso 21:09, Mar 7, 2004 (UTC)
And please next time you want to change the title of and article use "Move this page" this will cause less confusion and keep the talk page relevant.
As wikipedia is a collaborative effort can you help me to give accurate titles to the articles as my English is somewhat limited I have to idea how to solve this issue.
Ericd 19:19, 7 Mar 2004 (UTC)

Sorry saw the 'move this page' suggestion too late. Will use in future if my poor brain can retain it that long. But have moved my part of Perspective distortion via cut and paste to Perspective projection distortion ... this to accomodate objection that 'persepctive' can refer to many other things.... Pat Kelso 19:47, Mar 7, 2004 (UTC)

Please in the future use the "Move this page" method.
-- Egil 20:07, 7 Mar 2004 (UTC)
Yes sir. Pat Kelso 21:26, Mar 7, 2004 (UTC)

OK I move this one to "Perspective distortion caused by lens focal lenght" Ericd 19:56, 7 Mar 2004 (UTC)


Could also call it "Perspective compression and extension", since this technique can be applied equally well for paintings (although not seen very often). Perspective distortion should be a pointer to the two articles. Ideally, both should be combined in one article. -- Egil


Or could the article be called "Perspective change"? -- Egil 20:35, 7 Mar 2004 (UTC)

I don't know all this needs some serious brainstroming to give some structure to the question. I remember an interresting article when I was much younger and new to photography. Some paintings have true perpective according to optical laws while other have faked perspective. Aùmong the interresting case in paintings is "La última cena" from Leonardo da Vinci that as an accurate super-wide perspective did Leonardo get this by geometry reasoning or by a "camera obscura" view ?
Ericd 21:32, 7 Mar 2004 (UTC)


________________________

re "Some paintings have true perspective according to optical laws while other have faked perspective. Among the interesting case in paintings is "La última cena" from Leonardo da Vinci that as an accurate super-wide perspective did Leonardo get this by geometry reasoning or by a "camera obscura" view ?"

  • There is no such thing as "true perspective according to optical lays", except as noted in Perspective projection distortion. It has no application here.
  • "Leonardo da Vinci that has an accurate super-wide perspective." It may be an accurately drawn perspective projection - but which has innate distortion. "Camera obscura"?? Please!
  • The "fake" perspective used by Leonardo was used to ameliorate the natural distortion of perspective projection and which becomes more pronounced as the picture widens. Pat Kelso 09:34, Mar 22, 2004 (UTC)

The focal length of a lens has no dilatory effect on perspective, p e r s e. The perspective is always correct for the focal lengths involved. The focal point of a lens is analogous to the station point of perspective projection - a point that may be moved without changing the correctness of a perspective projection. Will leave this posted for few days for comment before incorporating rationale into the article. Pat Kelso 01:53, Mar 25, 2004 (UTC)

Need better examples[edit]

Wow, no talk in over 2.5 years. This article could use better photo examples, esp. including something forward of the main subject and some more clearly identifiable background objects whose sizes can be easily noticed. A set of only 2 or 3 in more normal layout should be fine. Dicklyon 18:11, 9 December 2006 (UTC)

Also need photo examples of the distortion appearing in many pictures shot with telephoto lenses and a couple of photo examples of the artistic use of perspective distortion that aren't copyrighted. Anoneditor 02:05, 11 December 2006 (UTC)

Dependence on viewing distance[edit]

I'm not sure I like this new treatment. I'm quite familiar with the traditional 19th-century analysis of "correct perspective" based on the viewing distance compared to the focal length times enlargement factor. But it's been presented here without the enlargement factor, so it makes little sense. Even with enlargement factor, it made only theoretical sense. The alternative is to say that perspective is just the relative size of objects in the scene, and that it can be exaggerated by shooting very close, or distorted the other way by shooting from far away. You might still want to invoke a "normal" viewing distance, but of course it will depend on print size if you do. In practice, I don't think you'll find that the perception of perspective distortion will abate much at all by changing your viewing distance. Geometrically, it makes theoretical sense, but the perception of photos doesn't necessarily follow the simple-minded rules that would make it look that way. Is this treatment following some book? If so, need to cite it. If not, let revert it. Dicklyon 07:23, 11 December 2006 (UTC)

Before I wrote this piece, I tried shortening the viewing distance for wide angle pictures with closer objects and telephoto shots of objects at a distance, and it seemed to me that the apparent distortion lessened as stated in the article. There were two problems, however, which might be mentioned: First, if you move back far enough for the telephoto shots, the picture gets awfully small. A practical problem but not a theoretical one. Second, on the wide angle shots, the problem was that with really wide angle lenses, you can't get close enough and still have your eyes focus and be able easily to view the perimeter of the picture. Again, a practical but not theoretical problem.
Exactly. It works in theory, but not so well in practice.
Have you tried it? Though I read about this issue long ago (I was one of the users of Kodachrome 10) I'm not quoting a source on this; I resorted to the experimental method and, as a result, was satisfied that my memory was correct on the point.
I have. I guess I'm a youngster, since Kodachrome II was my thing.
As to the magnification factor, I completely agree that it should be included to have a rigorously correct article, but I felt that what we're doing here is imparting general knowledge. I can't image anyone trying to calculate the distance from which the picture should be viewed to ameliorate or eliminate the distortion in ordinary circumstances, especially if the point of the picture is to create the apparent distortion. However, I certainly won't be offended if you want to include a more mathematical treatment of this issue.
Well, the viewing distance stuff doesn't really work as general knowledge without the theoretical background and the other relevant factor. I'd rather leave it out, which is more conventional in treating perspective. That is, treat perspective as determining relative sizes of objects in images based on their relative distances, and "distortion" in a qualitative way as meaning when those relationships look odd. Those relationships don't change with viewing distance, so that can be safely omitted, as it usually is these days.
Just as an aside question, why do you feel that the geometry isn't particularly relevant to the perception issue?Anoneditor 17:38, 11 December 2006 (UTC)
I think that when people look at a photo they see it as a photo. Viewing it closer or further doesn't change the relative sizes of things, and we're used to allowing the absolute sizes of things in images to vary with viewing distance. People are not perceiving a photo as if they're looking through it at the world, usually. Just my impression. Dicklyon 22:23, 11 December 2006 (UTC)

This article is less correct than before.[edit]

To say that perspective distortion is dependent solely on subject distance is less correct than to say it depends on focal length. Although relative distances of various parts of an image are relevant to perspective distortion, they do not change it. If you change distance to subject and change nothing else about an image, you will not change the perspective distortion (you will change the perspective, but not the distortion), while if you change focal length and change nothing else, you will change perspective distortion (although the effect might now always be obvious).

What perspective distortion is actually dependent on is solely angle of view. When the angle of view is greater than what your eyes would see from the same position you get extension distortion. When it is less than what your eyes would see, you get compression distortion. Changing distance to keep any object in the photo at the same apparent size as it is in another photo with a different angle of view will make the distortion more obvious, but the effect exists with or without the comparison shot. Shooting a photo with a "normal" length lens will make it pretty much impossible to create perspective distortion without changing angle of view somewhere in the post processing or print process.

If perspective distortion were caused solely by distance to subject, you would not need to experiment with two different focal length lenses or cropping images to see a difference in it. Rodrin (talk) 04:34, 19 November 2008 (UTC)

I agree. Perspective just depends on camera position, but perspective distortion depends on the camera angle of view compared to the viewing angle of view. So fix it. Dicklyon (talk) 07:10, 19 November 2008 (UTC)
OK. Article is partially fixed, but it really still needs more work. I will add another example later. Rodrin (talk) 17:11, 19 November 2008 (UTC)

I see that someone made another edit, and they are on the right track to simplify the first paragraph under the Cause heading. However, I am still not satisfied with it. Now it's nice and simple, but it doesn't really explain why distortion is taking place. "Perspective distortion is caused when the apparent relative distances of objects from the viewer compared to their actual distances change (what it says now, but more specific), but their apparent relative distances from the viewer compared to their surrounding objects remain the same (a simpler and therefore somewhat clearer version of what it said before)." That's probably about as clear as mud. I think I need better wording. Any thoughts? Rodrin (talk) 18:53, 20 November 2008 (UTC)

I think I came up with something better, so I put it in. I'm still open to suggestions if you think you have something clearer. Rodrin (talk) 20:28, 20 November 2008 (UTC)
Your edits are not totally correct. The problem is that people seem to believe that the perspective distortions at short focal lengths and long focal lengths are caused by the same thing. They are not. Granted, wide angle distortions are caused by the angle of view of a lens, it's also important to note that the extreme distance between the center and sides of a frame is also frequently caused by rectilinear conversion.
However, telephoto "compression" is more about the eye's inability to resolve depth of objects from afar. This can be easily seen in any photo lens of moderate focal length (or even your own eyes). An interesting experiment to try: Place a camera on a tripod in an area where you can see far away unimpeded with several objects on the path. Around half a mile, at least. Flat land or a hill will both work. Now, using a short focal length and a long focal length (doesn't matter if they're primes or zooms). Center on an object and take a photograph at a short focal length and one at a long focal length. Be sure to maintain centering on the same object. If you crop the image at the short focal length, you should be able to match the framing of the photo at the long focal length. Given enough resolution of the camera and lens, you'll be able to see the same compression on the short focal length as the wide focal length. I will post pictures at a another time demonstrating this phenomena, but as it stands, the page is incorrect. Kakomu (talk) 22:27, 23 March 2009 (UTC)
Well, the article is by no means perfect still (the caption under the animation near the top of the page is still not really good, for one thing). However, it is not essentially incorrect in the way that you seem to think. In general, perspective distortion at short focal lengths and long focal lengths is caused by the same basic thing, that is, making the viewpoint in a photo appear to be something different than it actually was. You brush on the topic that extremely wide angle lenses have their limits, but that is not really what this article is addressing.
It's not so much the eye's inability to resolve depth of objects far away, but rather the fact that the geometry of linear perspective cannot convey the same sense of depth for objects far away as it does for objects close up. In other words, linear perspective is inherently relative, which is the way that the article addresses this issue. The article does not ignore the point that your experiment demonstrates. In fact, it specifically mentions it in the section about compression distortion. Your demonstration does not make the page incorrect, but rather validates it. I will try to explain further below.Rodrin (talk) 19:43, 16 August 2009 (UTC)

Changes[edit]

Distortion at a wide angle of view occurs with a short focal length, as I've enumerated in the intro paragraph. If you apply the the thin lens (1/X + 1/Y = 1/Z where X is object distance, Y is image distance and Z is focal length) to any focal length, you'll note that objects that are close to the lens will project images much less distance than objects that are far from the lens. For instance, if you were to solve for Y with the formula -1/(1/X - 1/Z) for any focal length, you'll note a curve at closer distances which levels off after a certain distance. A short focal length (such as 10mm) has an extreme curve, meaning that the image of an object .5 feet away will be projected much closer than the image of an object 1 foot away. As this demonstrates, the distance from the lens is one of the contributing factors to perspective distortion. If an object looks normal in a 50mm lens, the object will continue to look normal in a 20mm lens, if the camera and object remain in the same position. It's the repositioning of the camera to maintain the same framing of the object that produces the distortions.

Wide angle lenses (short focal lengths) allow you to see an entire object that is close to the lens, giving the foreshortened appearance. The short focal length is not the direct cause of the distortion, but allows the viewer to readily see the distortion. I want to note that cropping a photo to simulate a longer focal length will produce a more "normal" image. I also want to point out that focal length and angle of view are closely related. Wikipedia's page on angle of view is quoted as saying: "For lenses projecting rectilinear (non-spatially-distorted) images of distant objects, the effective focal length and the image format dimensions completely define the angle of view." Because this is a photography page, it's important to recognize this fact and start to involve the issue of focal length with angle of view. I would like someone to actually post some scientific information or rebut what I'm writing. As it stands there are absolutely NO citations and no consensus in the discussion.

Finally, compression of objects that are far is independent of focal length. The focal length will simply "magnify" what is seen afar and allows someone to see it more readily. As I mentioned above, compression of far objects can be seen on a photo taken with a normal lens, just at a lower resolution.Kakomu (talk) 20:22, 24 March 2009 (UTC)

You are starting out with sound facts, but some of your conclusions are not correct. You demonstrate that linear perspective exists, and that linear perspective is relative. Basically, three dimensional objects look different at closer distances than they do at further distances. Really, people are generally aware of this, whether it's ever been pointed out to them exactly how it works or not. The distance from the lens is one of the contributing factors to perspective distortion the same way that wood being flammable is one of the contributing factors to a building burning down. It makes it possible; it doesn't make it happen. Your conclusion that, "If an object looks normal in a 50mm lens, the object will continue to look normal in a 20mm lens," is not correct. What I think that you actually mean is that an object will look the same, that is, have the same linear perspective geometry, through either lens. That is correct, but that is not the same thing as looking normal. In fact, the linear geometry being the same is the reason that the object will not look normal through a 20mm lens. When you switch to the 20mm lens, you change the apparent viewpoint. That is, you make the object appear to be further away than it actually is, but it retains the linear perspective geometry of its actual distance from the viewpoint. It's the disparity between the actual viewpoint and the apparent viewpoint that causes perspective distortion. The repositioning of the camera in no way produces distortions, although it can make them more obvious.
It's not that wide angle lenses allow you to see an entire object that is close to the lens that gives the foreshortened appearance; it's that wide angle lenses allow you to see an entire object within a much smaller viewing angle than it covers in real life at the same distance that gives the foreshortened appearance. If the viewing angle matched the capture angle, the object would appear normal, not foreshortened. The short focal length causes the wide angle of view, and the wide angle of view directly causes the distortion (technically, the capture angle of view being wider than the presentation angle of view). Having a really close object makes the distortion more obvious; it does not cause it. Cropping a wide angle photo to simulate a longer focal length will produce a more "normal" image because cropping the photo changes the angle of view to a more "normal" angle of view. Yes, focal length and angle of view are closely related, and this page links to Wikipedia's page on angle of view. Perhaps there could be more here on focal length's relationship to angle of view, but I think at some point you have to just link to the other page to explain that relationship fully; I'm just not entirely certain exactly where that point should be. I agree that citations within this article would certainly be an improvement.
The flatness of the appearance of objects that are far away is independent of focal length. However, this flatness does not become compression distortion until you make the objects appear to be closer than they actually are. Otherwise it's normal for the objects to appear relatively flat (beyond a certain point, pretty much completely flat). You make the objects appear closer than they actually are by capturing their images at a narrower angle of view than you present them at. One way of doing this is by capturing them at a long or telephoto focal length. Another way is by cropping a photo (usually accompanied by blowing it up or sometimes by viewing it from a closer distance).
Again, you explain well enough how linear perspective works, or how the appearance of objects changes with distance, and this is relevant because it makes perspective distortion possible, but it does not cause perspective distortion. Perspective distortion is caused by a disparity between the actual viewpoint and the apparent viewpoint of a captured image. Otherwise, the effects of linear perspective appear normal rather than as distortions. This disparity is in turn caused by a disparity between the angle of view of the capture and the angle of view of its presentation. Thanks to Zeeman's Paradox and the "normal viewing distance," the angle of view of the presentation can be considered, for the most part, near to a certain constant, making the angle of view of the capture the more crucial consideration.
Explaining perspective distortion by explaining only how linear perspective works is, again, like explaining why a building burned down by explaining only that wood is flammable. If you say that a building burned down and someone asks, "What happened?" and you reply only, "Wood is flammable," (even if you go on to explain how combustion works) then you have merely explained something that they were already aware of, even if they didn't know exactly how it worked. You have not told them what started the fire, whether it was an electrical short or a cigarette butt or whatever. For perspective distortion, whatever alters the angle of view is what 'starts the fire,' whether it is a wide angle lens, a telephoto lens, or a crop of a larger photo.
Another reference that might interest you is the first post here: http://www.openphotographyforums.com/forums/showthread.php?t=7427 Doug came to the same realization that I just explained above, and expresses it with a formula.Rodrin (talk) 22:13, 16 August 2009 (UTC)