Vanishing points & wide angle lens.

[Brian Bremer]

Hello everyone!

I've been practicing the fundamentals tonight. After working on linear perspective, I came across a photograph on Wikipedia that I wanted to use as a study.

Original Photo

I began extending parallel lines to vanishing points and estimated a horizon line. Keeping in mind that all parallel lines have the same vanishing point, I was surprised to see that different structures, though likely to have the same orientation, had clearly different vanishing points. I'm not a photography guy, though I wondered if the appearance of a very wide angle field of view had something to do with this. Also, the photo appears to be cropped together from two separate photographs, or perhaps pages of a book.

I've looked through photography websites and I've only learned that vertical lines converge sharply, when viewed down field, in a wide angle shot. I've also attempted the same process with other wide angle photos and found similar, though not as extreme, results.

So what's the take away lesson from all this? This doesn't reconcile well with what I've been practicing and understand about linear perspective. Did I make a fundamental mistake in drawing these parallel lines or setting up vanishing points? Or am I to understand there is something different in this view, something creating an exception to the rules?

Thank you!

Brian

[Brian Bremer]

Hello everyone!

I've been working on the fundamentals; drawing cubes and spheres with attention to values, figure studies and drawings that emphasize perspective. Coming across an intersting photograph on Wikipedia, which you can find here, I felt compelled to extend parallel lines to vanishing points and approximate a horizon. Attached is that product.

Well, I've somewhat confused myself. Everything I have read about linear perspective is that all parallel lines share a common vanishing point. In this photograph, individual structures that I would expect to be oriented in the same direction sometimes vanish to very different points. I can see that the picture is distorted, possibly cropped together, potentially scanned from a book and maybe taken with some kind of wide angle lens. Can those circumstances achieve the noted effect? What I'm seeing here doesn't reconcile with what I have been learning, and I'm just trying to develop my understanding to explain why.

I did some reading on photography websites about wide angle lenses, though didn't find much relevant to the matter at hand. I did play around with some other photos that were said to be taken with a wide angle lens. Ultimately, I wasn't able to find such extreme outcomes. So have I made a fundamental error, or does this photograph bend the rules of linear perspective?

Thank you for any wisdom, experience or advice!

Brian

** Edit 2-10-12 8:50 pm EST - I understand the statement that "all parallel lines share a common vanishing point" is not accurate. I wanted to mention this and spare the time of a well-intentioned reader from posting a detailed explanation that lines drawn in linear perspective that appear parallel to the viewing plane of the observer will not converge towards vanishing points.

[J Wilson]

...individual structures that I would expect to be oriented in the same direction...

That is the first place I would look for my answer. This is a presumption that is probably just not true. Those buildings are OLD. It might surprise you to know just how not "square" (as in the corners being nice 90 degree angles) many old buildings are, let alone expecting outlaying buildings to remain in the same exact orientation relative to each other.

Also, things don't have to be "off" by a whole lot to make a pretty distinct difference when you are talking about extending lines to the horizon.

[kevinwueste]

Actually, this looks pretty ok for 2 pt. The two things you are calling out are in effect - the distortion of the lens - which is HUGE! And, the two pages of this book are not laid flat enough - they are offset ( see roof of church-type building in center).

-Kevin

[Black Spot]

If you look at the well with the figure standing in it, you can see they're different sizes. This being such a small object renders the rest to picture quite useless for this this kind of exercise. And the roof would be straight across. The two halves don't match. Find a better example to work on.

[jetpack42]

The lesson here is that you're wasting your time.

Instead of over analyzing a shitty photo, go outside and try to observe and draw perspective.

Even if you gained a theoretical understanding, what's the point? You've got to be able to draw it. So get out there and draw. So many people (and I used to be guilty as well) waste so much time trying to learn something without doing it.

[Elwell]

Two issues:
1: That picture is obviously composed of two photos taken from two different vantage points.
2: Medieval and other pre-modern towns and cities very rarely have their buildings laid out on a regularly oriented grid.

EDIT: merged threads.

[[email protected]]

[...] maybe taken with some kind of wide angle lens. Can those circumstances achieve the noted effect? What I'm seeing here doesn't reconcile with what I have been learning, and I'm just trying to develop my understanding to explain why.

A wide angle lens reduces the distance between vanishing points, causing noticable distortion. In extreme cases, the resulting perspective will be curvilinear, like in a fish eye lens. It does not cause the phenomena you just noticed.

I understand the statement that "all parallel lines share a common vanishing point" is not accurate.

It is accurate. Can you show me an exception to the rule?

[cro-magnon]

the perspective will always be slightly curvilinear, unless we make cameras the size of the object we're photographing.

[Koji Bryant]

Keeping in mind that all parallel lines have the same vanishing point, I was surprised to see that different structures, though likely to have the same orientation, had clearly different vanishing points.

You can have MULTIPLE vanishing points on the same horizon line. You can have multiple horizon lines too (but that's advance stuff )

[Brian Bremer]

Just wanted to say thank you for all the replies, most appreciated.

Jwilson and Elwell's posts encouraged me to double check; 55°53'33" N by 11°27'43" E on Google Earth, if curious. The city streets do meander everywhere. Wish I looked before posting, would have saved the mods (sorry Blackspot!) the time of moving the thread. Attached is the aerial view; the red and blue buildings are the only two square structures, which vanish accordingly. While the original photograph is indeed poor, that doesn't seem to cause the questioned effect, outside the water fountain. I guess the apparent distortion made me begin to ask questions about curvilinear perspective and works like Escher's hand holding the metal sphere.

A wide angle lens reduces the distance between vanishing points, causing noticable distortion. In extreme cases, the resulting perspective will be curvilinear, like in a fish eye lens. It does not cause the phenomena you just noticed.

It is accurate. Can you show me an exception to the rule?

Thanks for the information about photography lenses. It's not an exception per say, but just how something like railroad ties in a one point perspective appear. I know you already knew all this. It's challenging for me to write clearly about multiple lines, the viewing plane, etc. while describing what is parallel and what isn't.

Koji Bryant, thanks for the cool example. I will try to spot more complex potential studies when I'm out and about!

Brian

[Black Spot]

I use Google Street View as well. I'm not happy with the side road on the left, which I will change, but it does show how much you have to change when doing a scene.

[treeshark]

Part of the problem I find is that straight lines are not perceptually straight except at very narrow fields of view. As soon as you get at all wide they curve, cameras of course show straight lines as straight at wider angles because the camera receptor is flat, unlike your eyes which have a spherical receptor. What this means is that how you project your image on to a flat surface just depends on which geometrical (or combination of methods) you choose to use to flatten your image on to paper. Spherical and Cylindrical projections are usually shown in extreme "fisheye" situations, but they can be used much more subtly than that so in practice the viewer will be unlikely to notice any degree of warping.
Rob

Here's a link where I rabbit on about the subject at greater length.
http://www.treeshark.com/treeblog/?p=301

And another where I deal with some of the unavoidable issues with linear perspective.
http://www.treeshark.com/treeblog/?p=307