“Vertical convergence” – You want truth, or drama?

[Heroique]

I recently posted this comparison in Leigh’s on-going “tilts and swings” thread, but I thought a quick survey of your opinion would be fun (Seagram Building, NYC, by Mies van der Rohe).

The poll addresses “vertical convergence” as it applies to a specific style of architecture (“International Style”), but I don’t mean to limit your thoughts to that. (For example, you’ll note the photos might demonstrate horizontal convergence, too.)

Better, even if you’re a landscape photographer – or portraitist – please share your experience on the matter. We want to hear...

(Note: My apologies, I just noticed you can pick only one option.)

;)

[EdWorkman]

I cannot abide buildings that "fall over backwards" in photos.
If one does that the photo better be surrealistic.
Or - there's a building falling backwards at sundown- be under it.
I could entertain a variation of the left pic, if the camera were turned to the right to capture parts of several buildings all tilted toward each other as hinted on the right border- give it a go.

[Roger Cole]

In this case, I like the first photo with the converging verticals. But that's because it's a dramatic, impressive shot where the converging verticals "work" - we know why it looks that way and it adds to the impression of size. The second is just a shot of a building.

Oddly enough, I think moderately converging verticals may often be more objectionable and shocking than such radically converging ones. The latter are clearly not exactly reality.

[Ash]

Waaaaay prefer the 'correct' parallels. Convergence is icky.

[Vaughn]

Shot #1 is reality. From the viewpoint at the camera, the top of the building is significantly farther way than the base -- and thus should appear to our eye as much smaller than the base. To show this building from this vantage point with parallel sides would be greatly distorting reality. Or perhaps better said as distorting our visual perception of reality.

[Heroique]

BTW, I’ve always thought the tiny, toothpick-like “stilts” at the base of Mies buildings (like this one) could not possibly “support” the weight of a skyscraper.

I think shot 2 makes it look too top-heavy for those stilts – so unstable, it might collapse at any moment, straight down into the plaza. (Hmm, shot 1 looks like it’s already falling over backwards.)

If you’re visiting NYC, you might think, “We’re not going anywhere near that building!”

[Richard M. Coda]

It all depends on the image. I prefer #2 in this case. #1 looks too claustrophobic to me.

These two images below, convergence is absolutely necessary, as they are more about graphic design that the actual building.

http://www.pctype.com/rcphoto/test/rcoda_WTC.jpg
World Trade Center, Sept. 13, 1981

http://www.pctype.com/rcphoto/test/Waterfront.jpg
7154, Scottsdale, AZ, 2007

[Lachlan 717]

Convergence screams "tourist snap shot" to me.

[Oren Grad]

Parallel verticals scream "large format snap shot" to me.

Exaggerated stretch to preserve textbook-correct verticals is icky.

Seriously: in this case, I don't think either works terribly well. It's a challenging subject; sometimes it's difficult to make a picture that doesn't scream "look how contrived I am!"

[Gem Singer]

Shot #2 demonstrates the reason that I use a large format camera instead of a small or medium format camera. I can correct for convergence.

[Kirk Gittings]

You have picked an extreme example to discuss. Trying to correct perspective in skyscrapers tests the best lenses, cameras and photographers. And there are limits-always-in terms of what you can correct. Notice in the second image that the photographers location is nearly at 1/3 of the height of the subject building. He/she is shooting from a window, terrace or roof of an an adjacent building with a very wide lens. Imagine the distortion of the top of the building if this had been shot at ground level with the perspective corrected (if even possible).

Both images are part of the "normal" coverage of a skyscraper that I would do on assignment given the limitations of equipment and points of view.

[Kirk Gittings]

Shot #2 demonstrates the reason that I use a large format camera instead of a small or medium format camera. I can correct for convergence.

Given the elevation of the camera position, this shot could just as easily be done with a DSLR and tilt shift lenses.

[Gem Singer]

Kirk,

I didn't state that it can't be done with a DSLR or a medium format camera. I am aware of the use of tilt shift lenses.

I merely stated that I believe it is easier and less expensive to correct convergence using movements on a large format camera.. That's why I prefer doing it that way.

[Brian C. Miller]

Besides "dramatic convergence," one of the reasons that the first photograph works is because there is another skyscraper framed within it, which lends to the building's immense size.

I must agree with Kirk, there is an enormous limitation to squaring up the first building using LF. I don't think any of my lenses would do that on 4x5, and I'm sure they wouldn't on 8x10. Perhaps the convergence could be corrected during enlarging.

[Robbie Shymanski]

I merely stated that I believe it is easier and less expensive to correct convergence using movements on a large format camera.. That's why I prefer doing it that way.

Not to nit pick, "easier and less expensive" method is a digital point & shoot and correcting perspective in PS or Elements, if those are your reasons.

[Gem Singer]

Robbie,

Perspective correction with Photoshop has not been an easy thing for me to master.

The results are not as pleasing to my eye as when I use camera movements.

"Less expensive" is referring to the additional cost of a small or medium format camera with a tilt shift lens. since I already have LF cameras with the capability of making perspective correction.

[Sirius Glass]

Not to nit pick, "easier and less expensive" method is a digital point & shoot and correcting perspective in PS or Elements, if those are your reasons.

Photo$hop cost more than either of my 4x5 cameras. Then I would also have to buy a larger, faster computer with more memory. Hence, digital is never the answer. :eek:

Steve

[Robbie Shymanski]

Photo$hop cost more than either of my 4x5 cameras. Then I would also have to buy a larger, faster computer with more memory. Hence, digital is never the answer. :eek:

Steve

Photoshop Elements 9.0 is listing between $90-$100. (mine came free with my printer) A decent digital pointer could be had for less than $200. Up the quality if you are shrewd on craigslist or eBay. All of this runs on even the most basic of computers.

Granted, I received my Calumet C-1 as a gift and my Toyo 45G, I picked up for $200. My SA 90 cost me $300 2nd hand. I think the recessed board for said Toyo was $30 plus $90 for a bag bellows. I was lucky that I was also gifted with a ton of 4x5 holders. Still had to pick up a change bag because I will go through 24 sheets when shooting in the field: $20. Even if I am going cheap, that's still gonna cost me .$75 a shot just for film (BW, by the way). I also require a tripod, but I can use one I've been gifted with or one that also works with my smaller format rigs. (Gosh, I wish I had an assistant just to move all this around!) So, I still have to take all this back to my darkroom to process. It's gonna take me, let's say 10 minutes to mix my chemistry and get everything up to temperature. So, let's wet those sheets! And then into the soup (personally, I prefer Pyro PMK, but Rodinal 1+25 is fast) so 6-8 minutes. 60 secs in the stop. And after a minute or so in the fix, we can flick the lights see the negative image that still has to finish processing. If I were to place a dollar amount on my time, this is getting pricey fast.

Now, it doesn't take 5 minutes to download, title the file, open PSE, open "Lens Correction", determine the horizon line, throw some grid lines up, adjust vertical, adjust horizontal, crop, done. All in the time yr Tri-X is still going through developer.

Don't get me wrong, I love LF and spending days in my darkroom. But digital is fast closing the distance where analog used to lead. The mechanics and the optics of shooting tilts and shifts, plus the hardware and software is getting better and cheaper. Film and chemistry isn't. In-expense of equipment and ease of movements aren't going to stand as reasons to embrace LF and reject current and future developments. There are always more than one way to take a picture. It all comes down to the person doing the shooting.

[Heroique]

...“Lens Correction”...

30 seconds of mischief...

Photo 1 w/ -150 in PS
(Now the photographer needs a 47mm XL.)

Photo 2 w/ +150 in PS
(Now the photographer needs an Aspirin.)

-----
This will help Photo 1 catch-up in the votes. ;^)

[Bruce Watson]

The truth often has more drama than most people want.

[SamReeves]

I liked #2 better, but I wonder what converging boobs would look like?

[rdenney]

There are times when we use convergence to establish the third dimension, and times when we eliminate it to represent the structure in its native rectilinear form.

Architectural photography grew from architectural renderings. It is the very rare rendering that shows a building with vertical convergence. The reason is that renderings and photographs are two-dimensional representations, and when verticals converge, it doesn't really tell us what the building actually looks like, as a building. As an image, we might have a different objective. Most of the time, though, a rectangular building looks as it is when verticals are (nearly) vertical.

The only "accurate" representation would be provided by a fisheye projection onto the inside of a sphere, with the eye placed at the center of the sphere so that all parallel lines would appear parallel when viewed from that spot. Anything else is a distortion of one sort or another. Those distortions are used to portray straight things as straight, tall things as tall, wide things as wide, deep things as deep, light things as light, heavy things as heavy, and on and on.

Of course, when the vertical vanishing point is within the image, it will be impossible to eliminate the convergence. Looking up at buildings from within them is an example of that. A rendering from the side that allows the verticals to be rendered vertically requires a different vantage point, so it's not really a proper comparison. I think keeping the verticals vertical is a method by which a structure can be shown as if it is viewed from an infinite distance even from a close vantage point. I've seen similar effects from a photographer who photographed a multi-hundred-foot-tall tree by photographing it from the side while hanging from a rope, and then stitching all the images together to retain vertical rectilinearity.

Rick "thinking each choice might be valid for different applications, but an architect would normally prefer the second" Denney

[David Low]

Just by coincidence someone posted some digital colour town centre/architectural photos on a folder at work today. From the properties they were taken with a Nikon D700 and the majority were with either a 17 or 19mm lens. I thought they looked awful with angles all over the place and everything seeming to topple backwards or into the centre. This falling over effect completely dominated the images for me, but maybe that's just because I now understand a little of how this can be avoided with camera movements??

[engl]

That is the curse of the large format photographer. Neither (moderate) barrel distortion nor converging lines used to bother me, I'd happily look past these technical details if the image carried itself.

Now I immediately notice it and usually find it very distracting, even if there is no reason to "fix" it for that particular image.

I don't consider there to be any "true" or "correct" way though. The world is not a photograph, and it is not experienced like one. Finding the "true" way to depict reality in a photo is like finding the "true" way to eat an apple like an orange.

[Heroique]

I noticed a few voters requested samples w/ modest vertical convergence.

Photo 1 is the Empire State Building, NYC.

Photo 2 is the Governor’s Palace, Williamsburg, Va.

In each case, I think the modest vertical convergence offers a very “natural” look. If the verticals were perfectly parallel, I suspect the “feel” would be very different – some might say “unnatural,” despite the architectural reality (See Vaughn’s post #5). It’s interesting how applying just a touch of convergence can create a very different reaction in the viewer.

[Roger Cole]

I noticed a few voters requested samples w/ modest vertical convergence.

Photo 1 is the Empire State Building, NYC.

Photo 2 is the Governor’s Palace, Williamsburg, Va.

In each case, I think the modest vertical convergence offers a very “natural” look. If the verticals were perfectly parallel, I suspect the “feel” would be very different – some might say “unnatural,” despite the architectural reality (See Vaughn’s post #5). It’s interesting how applying just a touch of convergence can create a very different reaction in the viewer.

I dunno, I like the shot #1 in the OP, but don't care for these. These are too close to the way we normally perceive them to be taken as artistic effect, yet different enough to be distracting. I like the first photo in the original post because it's dramatic and doesn't really try to look like a building seen from far away. Looking up like that, you notice converging verticals even visually, or at least I do.

But these - meh.

[Andrew O'Neill]

You can't compare those images as they are different. Now if you took image 2 and added convergence it would be disturbing to me.
Looking "up" at a tall building with our eyes, there is convergence. Why would you even bother to correct it?? When I photograph buildings, slight convergence gives "lift" to the image. I don't photograph for an architect, so I don't give a damn.

[goamules]

I kind of agree with Andrew. It seems to me if the purpose of the shot is to relate how tall the building is, and how strange it is to look straight up at it from directly in front, the converged image looks "real". If the purpose is to relate precisely what the building looks like, from a half mile away, I suppose there should be no convergence.

I recall in my drawing classes in grade school learning about converging lines such as when you stand on a railroad track and the rails appear to connect in the distance. If that's what we see, why try to fake it so it doesn't appear in a photograph?

Now, perhaps the amount of convergence is "wrong" with a camera with no movements. Too much? Too little? It seems there has to be some because that's what our eye sees.

[E. von Hoegh]

I pretty much agree with the last 2 posts, and I'll plump for mild convergence with the caveat that it depends on the shot. Sometimes a "perfectly" corrected building tends to look topheavy. Exaggerated convergence like you'd get with a 35mm lens on a Nikon is a no-no, a little convergence is natural.

[rdenney]

I pretty much agree with the last 2 posts, and I'll plump for mild convergence with the caveat that it depends on the shot. Sometimes a "perfectly" corrected building tends to look topheavy. Exaggerated convergence like you'd get with a 35mm lens on a Nikon is a no-no, a little convergence is natural.

Hmmm. "Natural." I wonder what that means.

Does that mean that the convergence no longer becomes noticeable one way or the other? In that case, to whom? We've already seen how our perception changes as we understand the effects at play.

I repeat that when the vanishing point is within the frame, it becomes impossible to remove the convergence without repositioning to remove the vantage point from the frame. There was a photo posted here once (which I can't find) of a skyscraper under construction photographed from above using a 72mm SAXL. I don't remember who made it, but it was breathtaking in its three-dimensional sense. The vanishing point was within the frame, or very nearly so, and obviously the convergence was required.

But there have also been a number of architectural images where the verticals have been restored to vertical, and in these cases the height of the structure seems to read more clearly. The leaning back effect, especially with a wide lens, serves to make the building look shorter. It depends on what the photographer is trying to portray.

Even with natural scenes, though, it's an important consideration with different solutions in different situations. I've made fisheye images that don't look fishy because the subject was sufficiently round. In those cases, the fisheye projection avoided the usual rectilinear distortion of round objects in the corners. But I've also made images of vertical cliff faces that would have lost their verticality altogether without the correction for convergence. I can't imagine this picture looking more natural with the trees leaning over into the waterfall. I fully corrected the convergence because confers don't look right if they aren't growing straight up.

http://www.rickdenney.com/images/bridalveil_falls_062892-8_lores.jpg

I left a bit of convergence in this image, but any less and the work of those frontier stonemasons would have lost the loftiness they were seeking.

http://www.rickdenney.com/images/Concepcion022793-8_lores.jpg

I made no correction for convergence in this photo, because doing so would have ruined the effect of looking up into the tree. In fact, I think I leaned the image frame over as well.

http://www.rickdenney.com/images/japns_maple_scan0015_lr.jpg

Maybe not great photos, but anything they lack is because of how I visualized the image--I dialed in just the amount of convergence I wanted in each case.

Rick "thinking 'natural' will never be precisely defined" Denney

[Kirk Gittings]

There is no truth in photography. There is only personal opinion striving for or masquerading as truth.

[paulr]

As others said, it all depends on the image. Personally I gravitate toward parallel lines, but this can be seen as habit as much as anything else.

Not just my habit, but cultural habit. The reason convergence looks "wrong" goes back to renaissance era debates on how to depict three dimensional space on a plane. There are many possible ways to do it, none of which actually looks like the 3 dimensional world.

The winners in the debate were Alberti and Columbus, who proposed the set of rules we are now inclined to think of as natural law: that horizontal lines may converge but vertical ones may not. This became the standard for western painting ever after (not counting the many avant garde movements) and informed the basic design of the view camera.

If one approach looks more right or wrong to us, it's because of conditioning ... much like the way in music Western tonality when sound more right than Asian tonality, if it's what you grew up with.

[Vaughn]

Hmmm. "Natural." I wonder what that means....I fully corrected the convergence because confers don't look right if they aren't growing straight up.Rick "thinking 'natural' will never be precisely defined" Denney

I have been guilty of straigthening up leaning trees using the rear tilt. As I wrote earlier, it is all about managing the image -- making it what the photographer wants it to look like.

If left to themselves, those dang trees just don't grow very straight!

[Robbie Shymanski]

There is no truth in photography. There is only personal opinion striving for or masquerading as truth.

Good one Kirk. It's all just information that we all digest in our own way.

[Heroique]

...If one approach looks more right or wrong to us, it’s because of conditioning...


There is no truth in photography. There is only personal opinion striving for or masquerading as truth.


Good one Kirk. It's all just information that we all digest in our own way.

This makes for an interesting sequence of remarks – Paul’s claim about “conditioning” helps explain Kirk and Robbie’s claim about the non-existence of truth in representation. That is, Kirk and Robbie may have been conditioned to think so.

So shouldn’t we be careful w/ unyielding judgments? Our modern period makes it quite fashionable to tout conditioning & doubt truth when it comes to artistic representation. In other words, let’s not forget that we’re culture-bound, just like people in earlier periods, and be self-critical about our claims. (There may be more Truth in that photograph of the tall building than you’ve been conditioned to believe!) I say: read the ancients, read the Renaissance masters, read the moderns. It’s instructive to hear the best minds disagree about whether parallel (or converging) vertical lines can capture visual Truth.

-----
Paul, I’m certain many of us would love to see a pull-out quote from Alberti about vertical lines, if you have his writings!

[paulr]

Paul, I’m certain many of us would love to see a pull-out quote from Alberti about vertical lines, if you have his writings!

I'll take a look ... everything I've read comes from art history essays. I should at least be able to find 2nd hand quotes by historians who know this story a lot better than I do.

I'm not quite sure I understand all your other remarks. Your point about us all being the products of cultural conditioning I agree with ... not that we are inescapably bound by it, but that its force is pervasive, and that much of what we perceive as "natural" or self-evident is really better understood as received ideas.

I think we're at our best (or most interesting) when we're aware enough to at least question these ideas.

I have little patience for the postmodern notion that all truths are cultural constructs. Gravity, for example ...

But some, like the relative correctness of one kind of convergence over another, can be easily seen in a historical light. Our current consensus is the product of an old debate.

[Heroique]

...I think we’re at our best (or most interesting) when we're aware enough to at least question these ideas...

Yes, that’s what I meant.

Thanks for looking-up Alberti on vertical lines – I suspect he’ll please some, displease others, and enrage a few.

;)

[Kirk Gittings]

http://www.photoeye.com/auctions/img/205/Large_H300xW300.jpg

I always got the idea that Strand could give a s__t for convention, but followed his visual instincts. Because of that, IMHO he was the most interesting of the early modern architectural photographers.

[paulr]

Kirk, I dare you to post that Strand pic on photo.net to see what suggestions you get for improving it.

Heroique, I did some preliminary snooping around on Alberti and the story seems more complicated than I remembered. A couple of articles suggest that he might have been instrumental in the development of the camera obscura, but they don't cite their sources ...

[Kirk Gittings]

More Strand
http://masters-of-photography.com/images/screen/strand/strand_church.jpg

[Maris Rusis]

A couple of careers ago I actually got paid to do philosophy. I vaguely recall a few of the principles. Some extremely condensed abstract philosophising follows:

Both views of the Seagram building, the one with vertical convergence and the one with parallel sides, are "true". Truth exists and is a concept that applies only to propositions. So what propositions are implied in the two views of the Seagram building?

An easy way into unfolding the problem is to borrow a concept from cybernetics - the black box. A black box is a system that converts an input into an output via a fixed set of rules. We don't need to know the rules, just that they are fixed. Gratifyingly, the system works just as well backwards. If we have an output we can predict the input exactly. Photography works like a black box.

In both pictures of the Seagram building there is a one to one correspondence between points in the picture and points on the actual Seagram building. That correspondence is the transfer function that happens inside the photographic black box. Given either photograph and the transfer function used to make it then it is possible to reconstruct the actual appearance of the Seagram building itself. The two views, convergent and non-convergent, are the outputs of two different but self-consistent black boxes.

The basis of photographic truth lies in the fix-ed-ness of the transfer function that works in the heart of the photographic black box. Conspicuously, a fixed one to one correspondence of points in the picture to points in the subject is not guaranteed in painting, drawing, or digital picture making. These processes may support some true propositions but photographic truth isn't one of them.

[paulr]

Both views of the Seagram building, the one with vertical convergence and the one with parallel sides, are "true". Truth exists and is a concept that applies only to propositions.

And if you take the most common (unstuddied) proposition, that a photograph should look the way the world looks to our eyes, both views are necessarily false.

[peter ramm]

Maris, when I taught clinical neurobiology I would start the first lecture with "Everything I am about to teach you is false. However, it is the best I can do at the moment." This really pissed off some of the students (particularly med students) while others accepted that what is practical and what is true are not the same - and that this matters.

Today's untruth: Primary leverls of the visual system maintain a highly distorted representation of the external world. The ordered structure we experience is rendered and perceived, not seen. Perception is different for each of us. There is no single path to a good photograph.