Architecture: Is Removing Converging Verticals Overdone?

<p>There seems to be a consensus among traditional or commercial exterior architechtural shooting that lens shifts are necessary for removing keystoning effects. I understand the motivation, and I think the resulting shots can look "natural" (or at least fit what our eyes have been trained to see), but I feel this is only the case with focal lengths on the normal-to-longer side of things.<br>

<br />When it comes to wider lenses, say 90mm and wider, with buildings filling the majority of the frame, I think the perspective "correction" actually looks like a <em>distortion </em>and seems unnatural. Particularly with tall buildings and ultrawides. It seems to me there is nothing about what the eyes actually see that resembles this type of look, where ultrawide shots nonetheless have parallel verticals. The natural perspective differences between the tops of buildings and the ground (i.e., when you are also on the ground) results in a natural keystoning, and I'm not sure why many feel it necessary to remove this.<br />I realize I'm painting this in somewhat all-or-nothing type of language, which it certainly is not. But it does seem most pronounced with the commerical shooters. Can anyone explain the reasoning behind this practice?</p>

<p>Keystoning is very useful when attempting to call attention to the height of the structure, or even make a short structure appear taller/higher, but I find that most of my clients (architects, real estate professionals, builders) want shots that are "squared-up".</p>

<p>Wide angles make things look stretched in the corners without any lens shifts... but shifts make it worse. I don't think there is one orthodox approach to perspective control, particularly when it comes to wide angle pictures. One thing that seems to suit some compositions is "mild" keystoning - in other words, only partially correcting and leaving the verticals not-quite-square.<br>

When you get down to it, a lot of architectural pictures IMO would look better with the camera higher up, but that's not always practical.</p>

<p>It depends a lot on who the photo is being done for. Architects want it straight, builders sometimes don't care as much and when it comes to expressive photography--express yourself!</p>

<p>Although we see keystoning with our eyes, we actually mentally correct it and rarely, unless we think about it, will remember the building that way but rather as being straight. Longer lenses actually correct much of the issue naturally because of the distance between the vantage point and the subject--you can lay on the ground or shoot from a helicopter, but a building that is off in the distance will not show keystoning.</p>

<blockquote>

<p>Although we see keystoning with our eyes, we actually mentally correct it and rarely, unless we think about it, will remember the building that way but rather as being straight. Longer lenses actually correct much of the issue naturally because of the distance between the vantage point and the subject--you can lay on the ground or shoot from a helicopter, but a building that is off in the distance will not show keystoning.</p>

</blockquote>

<p>I think you're speaking to my point about lens-shifting looking better with the normal-to-longer lenses. Your point about mentally correcting the keystoning seems right, but I'm not sure its true with the extreme perspectives ultrawides can fit in. I know that when I look at ultrawide exterior architecture shots with parallel verticals, I immediately notice that it doesn't look "right" to me. I don't notice this discomfort with shots taken with longer focal lengths, even though analysis of the shot still reveals lens shifting. The further perspectives associated with normals and teles by their nature require less shifting, and I might be reacting more favorably to the fact their is less manipulation.<br>

<br />I also might be reacting against my day job. I look at blue prints all day, and they are always drawn from orthographic perspective... which literally doesn't exist anywhere in the universe. In orthographic perspective, both horizontal and vertical vanishing is taken away. With some exterior archictecture shots, the same thing happens except only in the vertical plane: the horizontal perspective is allowed to perspective.</p>

<p>I wonder if you are referring to those shots where the lens also does have a lot of internal distortion, like most 35mm camera lenses. I shoot a lot of architecture on jobs, use a 75mm on 4x5 fairly often and rarely get residual distortions--especially on exterior shots. I think most of the shots look pretty "normal" to the eye. I am not so sure that is always the case with images shot with 35mm cameras. But, many do correct images in post and the problem there is that you not only need to correct the canting, but also the vertical presentation. PS corrections will "stub" a building and not correcting the vertical correctly really gives some strange effects, let along the barrel distortion that is prevalent with the ultrawides.</p>

<p>As and architect an an amateur architectrual photographer I have to agree with you - sort of. I think it's really about camera location realtive to the building and how tall the building is. For skyscrapers, a corrected image is has more infromation and shows the building more clearly but has an artificiality to it - but one most people have learned to accept. On a shorter building, if the camera position makes you feel your are looking <strong>at it</strong> in it's surrounding, converging verticals just look amateruish and sloppy. If you get close and create a feel that you are looking <em><strong>up at it</strong></em> then non-converging verticals look strange. Watch North by Northwest sometime and see what a master does with converging and nonconverging lines to get the point across (it even starts with the opening credits).</p>

<p>I love LF, almost never correct the verticals. It is too "required" for my taste.</p>

<blockquote>

<p>I wonder if you are referring to those shots where the lens also does have a lot of internal distortion, like most 35mm camera lenses. I shoot a lot of architecture on jobs, use a 75mm on 4x5 fairly often and rarely get residual distortions--especially on exterior shots.</p>

</blockquote>

<p>I was actually speaking about large format photgraphy, as this is where the practice originates and still dominates, but I suppose the format doesn't matter. When I said 90mm wides, I guess you could speak about 28 or 24mm wides on 35mm film. I posted in the large format forum because of the large propensity for this practice with large format, but there's no reason it can't apply to the smaller formats.<br>

I was playing with the word distortion. Lens shifting is often said to "correct perspective distortion," but with ultrawides I said it introduces its own kind of distortion: the kind where the photograph doesn't match what the unaided eye ever sees. In reality, there is no such thing as parallel verticals with near-to-distant objects, such as as with tall buildings. Perspective requires that verticals converge as distance falls away. Incidentaly, this happens in both the vertical and horizontal planes, yet in the kind of shooting I am talking about - wideangle exterior architectural shooting - the horizontals are left alone while the verticals are forced to parallel. There is a strangeness to this while also being accepted practice.</p>

<blockquote>

<p>I think most of the shots look pretty "normal" to the eye.</p>

</blockquote>

<p>My conclusion thus far is that "normal" fits with lens shifting on the longer focal lengths, like say 120mm and up. With the wider focal lengths, and particularly with tall buildings closeup, taken from (or near) ground level, the forced parallels on the verticals just strikes my eye as very strange. I think its one of those things where, once you see that you are being tricked, you can't ever get tricked again and you always see the "trick." Of course I dont' reall mean people are being "tricked," but you get what I mean: the manipulation of perspective.</p>

<blockquote>

<p>As and architect an an amateur architectrual photographer I have to agree with you - sort of. I think it's really about camera location realtive to the building and how tall the building is. For skyscrapers, a corrected image is has more infromation and shows the building more clearly but has an artificiality to it - but one most people have learned to accept.</p>

</blockquote>

<p>That may be the main thing I'm missing: how tall the building is. Skyscrapers with parallel verticals taken from ground level with ultrawideangle lenses definitely appear more artificial looking than more modest buildings, mainly as the intellect tells you something isn't right. The shorter buildings have less to be wrong about, so to speak, so there's less offense.<br>

But then again, the shorter buildings let you get closer, which some shooters take advantage of... allowing them to exploit the lens shifting thing some more.</p>

<blockquote>

<p>On a shorter building, if the camera position makes you feel your are looking <strong>at it</strong> in it's surrounding, converging verticals just look amateruish and sloppy. If you get close and create a feel that you are looking <em><strong>up at it</strong></em> then non-converging verticals look strange. Watch North by Northwest sometime and see what a master does with converging and nonconverging lines to get the point across (it even starts with the opening credits).</p>

</blockquote>

<p>Good point. Now I gotta check out that scene.</p>

<p>I think a lot of people do go overboard -- as referred to above.</p>

<p>The main difference to me is whether everything is correct as regards geometry and perspective effect, which are different than the effect you get when you take the film plane out of vertical.</p>

<p>When I tilted my old Yashica-mat years ago to get the whole front of a 3- or 4-story building in the view, the result in the print was that the building looked like it was falling over backwards.</p>

<p>If I were to take that same shot with another camera with front rise, keeping the film plane vertical, I might get some converging verticals because of perspective effect, but not to the extent that the building appeared to be in danger. And I wouldn't worry about that at all.</p>

<p>Additionally, some people will swing the back to parallel with the plane of a building that's a block long, even though they're standing at one corner of it; this results in what I think is distorted perspective and doesn't do anything good for the photo.</p>

<p>Maybe the distortion that you're noticing is due to the rectilinear design of many large format wide-angles to some degree? They keep straight lines straight but elongate them and compress the image at the edges of the frame. I say this because lens shift does not alter image shape or perspective. Perspective is changed by camera position and image shape is changed by the angle of the film plane.</p>

<p>Bravin, if the building is central to the frame and the camera is inclined upwards, then the resulting keystoning looks like it does "to the eye" and therefore appears natural. However, if the building is off-centre and the camera is inclined (even just a little) then the building will appear to be falling over; an effect that I think looks ridiculous. However, I am surprised just how many times such photographs appear in (non-technical) publications and I have to suppose that the editors - and most people - are so accustomed to seeing it that they just don't notice it.</p><p>A year ago I bought a Cambo Wide DS and 6x12 back for its 2:1 format panoramic potential but was impressed by the front rise capability for architectural work. Using a 47mm or 58mm Schneider lens, the resulting wide angle views are terrific and buildings remain vertical. To me, except for some minor stretching of elements that extend into the extreme corners, the effect is beautiful, elegant - and totally realistic.</p><p>Although I do not have a 4x5 background, it seems to me that the term "perspective correction" is often used in a misleading manner. Before using my Cambo Wide, I understood from written comments that it was achieved by turning a knob and watching tilted verticals become vertical. In reality, perspective correction, or vertical verticals, at least, are achieved by keeping the camera horizontal so that the verticals simply stay vertical. The problem is then that the top of the building is above the frame and the lower half of the frame contains foreground; front rise is applied and brings the building down into the frame. Just like magic.</p><p>For the last few weeks I have been working on the 45th floor of a high-rise building; I was captivated by the view of the surrounding buildings and was struck by the realisation that EVERY vertical line appears vertical, even when looking downwards at a sharp angle. At first the effect looked quite UNnatural. That view told me that my expectation was wrong; perhaps my wrong expectation arose from having, over the course of my life, repeatedly seen (crappy) photographs with non-vertical verticals!</p>

<p>all this is a subject that was extensively discussed in the literature on photography...<br>

for me architectural photography of quality has vertical verticals. Especially using wide angles working with a monorail camera with all movements can make the critical difference compared with photographs taken with a digital camera. As Rod wrote: every day we see crappy photographs - so a carefully taken picture with a lf camera can pop out.<br>

In my work I am often in the situation that I have to use my shortest lens (Schneider SA 47 XL) to get all I have to on 6x9 (and sometimes even 4x5). And very rarely I have the impression that the result is a 'unnatural'. Sometimes 'unnatural' perspectives can result from the fact, that the camera position is very close to one angle of a whole complex, that has to figure on the picture. But for these cases we can work with horizontal tilts to get better proportions.</p>

<p>The photographic image, like a drawing, may either be a statement, a document, or an image comprised of both.<br /> Just as one may select a focal length to describe scale, and aperture to either isolate or embrace the whole, so we also use camera movements set the scene.<br /> Once we understand method, such as the Scheimflug rule and everything else from camera to darkroom, any "rules" as such are only applicable to any given project. It's about want you want to say, and who you are speaking to.<br /> Using pencils, I have studied the rules of draftsmanship for technical and geometric drawing, using drawing boards and an array of equipment, compasses, rulers, the whole kit. As well there have been countless hours working freehand.<br /> Whilst camera work has it's own set of disciplines, the freedom to apply is up to you.</p><div></div>

<p>Perspective correction is a matter of doing justice to the architecture. Who would want to portray a portrait subject less than ideally? </p>

<p>There is more to the image than just the look.<br>

Alberto Pérez-Gómez has analysed the phenomenon of the perspective view as a legitimate representation of architecture, but also as a representation of world view introduced by the starting modernism in renaissance. I think that technological contingencies have a big role in this tradition of vertical verticals. Anyone who has tried constructing a perspective knows that a 2-point perspective is far easier to manage than a 3-point one. So the images without converging in that one direction became predominant before photography became the ultimate medium for architectural representation and dissemination. And photography just appeared to be technologically capable of proceding this tradition.<br>

I tend to think that the ability of photography to provide adequate representation of spatial conditions is actually over-rated. The convention of the image of an architectural journal is actually setting limits to architecture itself. The idea that everything has to 'look' is piling stress for architects and driving them away from actual spatial issues.<br>

But I do use a 58mm SA XL on a Cambo Wide DS for the looks, so I contribute to this system. Sorry, if off-topic, this has been an interesting issue for me.</p>

 

<p>Indrek, I don't think you are off topic at all.</p>

<p>David L., to your comparison: "Who would want to portray a portrait subject less than ideally?"<br>

.... I wish I had a spare lifetime to walk you through the history of art.</p>

<p> </p>

<p>I have misused some words in this thread, and some folks have picked up on it. So I should clarify some of my points.<br />1. It is true perspective is a function of distance only. I have been ambiguous on this thread and have used the term in multiple senses. Perspective is controlled by the distance between between subject and camera (or subject and eyeball). There is no actual, technical, "perspective correction" other than physically moving your location, despite the fact many of us fall in to the trap of using this "perspective correction" language, which seems to permeate the photographic literature. There is optical manipulation via moving lenses in relation to each other, and there is digital manipulation using software, but this is not the same as correcting perspective. Photoshop even uses the term "perspective" when cropping, despite the fact it can't actually do anything about the perspective.<br>

<br />2. The converging verticals thing is a function of perspective just as horizontal vanishing points are. The two are exactly the same phenomenon. One way of looking at this is that reality doesn't distinguish between "vertical" and "horizontal." There is just distance, in 3 dimensions, and with distance comes perspective. If buildings were made with crazy anglular structures in between horizontal and vertical, those structures would be subject to the same physics of perspective and distance as everything else is.<br>

<br />3. The practice of straightening out lines in wideangle exterior architectural photography (and certainly not all of it - I'm overgeneralizing here) seems only to apply to verticals, not horizontals. This is clearly a manipulation for psychological reasons. Point #2 states that perspective affects all dimensions, not just the vertical ones, yet the vertical ones are the province of exterior architecture manipulation, or so it seems to me. Again, generalizing here.<br>

<br />4. I'm guilty of this stuff too. I haven't been shooting 4x5 for very long, but my very first shots got me almost immediately in the habit of keeping the camera level to ground and shifting the lens up with buildings. It's almost hard to fight the habit, I have found. Maybe I'll get better at fighting it.<br>

<br />5. I'm not complaining about this, but remarking that it looks stranger and stranger the wider the angle of view. With the normal and tele lenses, sometimes it is hard to even see the manipulation. With wider lenses, it is fairly obvious. With ultrawides, it looks, dare I say, um, sometimes I think it looks ridiculous. But that's just me. And if I had an ultrawide lens for my view camera, I'm sure I'd be tempted to do it anyway. It's almost like it feels "expected" when buildings are in the shot.<br>

<br />6. The statement that you should just do whatever you want is obviously correct. For artistic expression or whatever, do whatever makes sense to you. But I have the sense there is a commercial expectation from clientele, or from professional peers, or for reasons relating to professional reputation, or whatever, when it comes to commercial architectural shooting. At least from what I can tell, it seems most prevalent in the commercial architectural world, though I could be wrong about this. Amateurs like me seem to just go about doing whatever they want, but once you cross that line into professional architectural shooting, it seems like an expectation.</p>

<p>Bravin, you will never be bored.</p>

<p>Indrek, if I'm understanding you right, the idea of parallel verticals pre dates photography, but that once photography came around, it kind of picked up where painting (drawing, etc.) left off?<br>

<br />Are you also saying that the "look" of accepted architectural images (presumably including the parallel verticals thing) is acting as a hindrance to progress in architecture itself?</p>

<p>Kevin, I understand your point. It'd be a tragedy if art (and photography) were all the same. And I'm a defender of anyone doing, practicing, their own approach. I'm an engineer, not an architect, and believe form and function should be appreciated. </p>

<p>Hi David, well I totally agree with, and join you in the belief that form and function should be appreciated.<br>

Some of the most famous so-called "Modern Masters" in the art world attended schools where form and function were at the foundation of their studies. These were painters who went on to produce iconic abstract artworks. But every oblique line, every curve, placement of an object or slash of colour was placed with an absolute understanding of form and structure.<br>

....<br>

Just for fun, pictured here is my first ever large format exposure. The camera was a home-made pinhole camera with a movable front standard and a sliding pinhole panel, permitting 20mm of shift. The film was trimmed from 5x7 to fit a couple of 1920 wooden holders. The challenge of this exercise was to get the image as upright as possible, because with many such old buildings, a truly vertical line of reference is pretty hard to find.</p><div></div>

<blockquote>

<p>Just for fun, pictured here is my first ever large format exposure. The camera was a home-made pinhole camera with a movable front standard and a sliding pinhole panel, permitting 20mm of shift.</p>

</blockquote>

<p>Could you explain how moving a pinhole replicates lens-shifts? By the way, I'm always surprised at how good pinholes can be. Do you recall your exposure time?</p>

<p>> Are you also saying that the "look" of accepted architectural images (presumably including the parallel verticals thing) is acting as a hindrance to progress in architecture itself?</p>

<p>I think that photography is too much concerned with the object instead of the space. This has to do with what the community thinks of 'what is architecture?'. And the community consists of an entire row of education, criticism, journalism, photographers, common people and whoever. The culture has been developed into a global unity which sets rules for localities that could have a totally different scope.<br>

For example here in Estonia we have had periods of different economic standard and still been designing and building things. The buildings of poor time have tried not to give up on the look of the object but for lack of money they have been built out of crappy materials. Once finished we get a nice set of images without converging verticals (which has nothing to do with my argument), and the next day everything starts to deteriorate but we (the architects) are still happy as the beautiful imagery is already disseminating. And everybody takes it for architecture. We are designing with the image as an ultimate goal (to some extent).<br>

So, I'd say that it would contribute to the practice of architecture if technical ways suitable for the specific spatial conditions were selected instead of a commonly presumed standard look. But it certainly will not be easy to break the wall of the whole tradition as the editor has to accept Your image.</p>