2 & 3 point perspective, Barrel & Pin distortion, and natural perception!

[Deleted member 11105]

Hi,
I have been thinking about the way we see things naturally and how photographs relate to this.

I was looking at a photograph taken by Ezra Stoller of the Seagram building by Mies Van *** Rohe in the book Minimum published by Phaidon (http://www.egodesign.ca/en/article.php?article_id=363 you can see a small version here)

I noticed how the photograh has been taken at a level of about a 1/4 of the way up the seagram building. I think the camera has been focused about 1/4 of the way up the seagram building and on the right edge of it also. I'm asuming the photograph has been cropped, almost half from underneath and a bit off the sides perhaps, depending on the format.

I have notice in this photograph that all of the sides of the buildings run parrallel to the plane of focus and due to the lens used having no barrel or pin distortion all of the sides are dead straight. This is when i realised that photographs shot like this display a 2 point perspective.

The photo has also been created using a wide angle lens and naturally we never look at a whole scene like this but instead only focus on elements of it at a time. Also I have learnt from the books I have studied on perspective for designers that the way we actually see is closer to 3 point perspective.

I think the photo of the seagram building is a near perfect photo of such a subject. I can't think of any instances when pin cushioning might be desirable. But isn't barrel distortion similar to 3 point perspective. Might it sometimes be desirable if the subject of the photograph is in the centre of the frame and we want to create an image for which the experience of viewing aspires to becoming as close as possible to how we percieve naturally?? Perhaps in portrait photography?

Please let me know if the asumptions i have made about the photography by Ezra Stoller are correct and what you think about my suggestion that barrel distortion is closer 3 point perspective, and only when combined with the focal point of the image coinsiding with the center of the frame, becomes closer to how we actually see,

Regards,
Duncan

 

[tbm]

Camera location looks about right; can't tell where the focus is from the small image linked to.. somewhere on the building would obviously be the natural choice.

What's your point?

 

[tbm]

Ahh, cool.

Can kind of see where you're coming from, after 5 minutes researching perspective on wikipedia

I think perspective type, perspective expansion / compression and lens distortion have to be considered separately:

We see in 3 point perspective as we have the ability to rotate our heads about all three axes; which we can also do with a camera so it can also see in 3 point perspective.

Perspective expansion and compression is a function of the distance between the subject and the camera, so changes with respect to focal length - hence the term "standard" lens (c. 50mm on full frame and 30ish on crop) as it's closest to the perspective of the human eye. Portraits are usually shot with slightly longer than standard lenses to compress perspective slightly and reduce / flatten the effect of facial features.

Finally barrel and pincushion distortion are more of a function of the lens design and are not strictly related to focal length (distortion can vary between different types of lenses of the same focal length). These distortions serve to bend parallel lines, rather than alter their positional or rotational relationship.

I'd have thought that the only difference in perspective between the camera and eye is that resulting from expansion and compression, so in summary the only thing you can control (and that needs controlling) to mimic the perspective of the eye is distance to the subject and corresponding appropriate focal length.

Not sure this answers your question, but it's the best I can come up with :bonk:

 

[HoppyUK]

I'm really not sure what you're getting at here, but the terminology is very confusing.

There is no evidence of either barrel or pincushion distortion in that picture but the converging verticals have been corrected, both up and down from the centre. That appears to be the only major change from a straight camera capture, and is quite normal practise. I'm not sure if perspective has been changed in any other dimension, but if it has it's not by much.

Focus looks like it's very close to infinity anyway, with plenty of depth of field, even on a large format camera. It's hard to tell from that pic, but it all looks reasonably sharp (or equally unsharp, from that rather poor reproduction).

The perception of persepctive is a function of distance from the subject, and is unrelated to the camera, or lens, or eye, but what we 'see' is hugely modified by the brain, to the point where comparisons between cameras and the way we see are meaningless.

The camera is a two-dimensional imaging device, that captures only a small part of the eye's field of view, for one brief moment in time. Our eyes are 3D devices, constantly scanning a moving scene occupying only a small area of sharp focus at a time, within a very wide 180 degree frame. In addition, they are coupled to a post processing computer, the brain, that can do things Photoshoppers can only dream of. Not the same at all.

Here's a famous one for you

http://images.google.com/imgres?img...rosoft:en-us:IE-SearchBox&rlz=1I7GGLR_en&um=1

 

[tbm]

Surely if the verticals were converging and had to be corrected this would mean either they were not perpendicular to the viewing angle or the lens had barrel distortion?

This seems like a logical assumption.

Just out of interest, assuming the original shot was taken on medium format film (as opposed to digital media, given it's age) how whould one go about correcting converging verticals?

I assumed that it was shot with the image plane dead vertical (to keep the lines parallel), and cropped to achieve the desired composition..

 

[HoppyUK]

Yeah same,

i was wondering if in the case of distortion and converging verticals if this could have been corrected by adding some kind of correction lens to the enlarger when printing? i don't know if such a thing exists?

Kind of.

Terminolgy. Barrel and pincushion distortion are lens aberrations. The effect refers to the bending of straight lines towards the edges of the picture, particularly noticeable in architectural subjects. This is generally very undesirable and lens designers strive to minimise it. Neither kind of distortion is evident in the OP link.

It is better not to talk about perspective distortion, but either exagerated or compressed perspective. Perspective is exagerated by a close viewpoint, and a wide angle lens allows the camera to capture a wide field of view which tends to look unnatural in this situation. It is of course not unnatural, it just usually looks that way because when we are close, we are used to seeing a much narrower field of view. Basically, the spacial relationships between objects, and their relative size, is exagerated.

By the same token, by standing further away and using a telephoto lens to draw the subject apparently closer, this compresses perspective and the relative size of objects remains similar. Imagine the beginning of a Formula One race, with the camera at the far end of the straight and a telephoto lens, all the cars look like they are on top of one another, because they all look the same size. But if you physically move in close and stand directly at the front of the grid with a wide lens, the cars at the front are large, and those at the back much smaller.

Parallel lines, such as the sides of a building, naturally converge as those parts of the subject that are further from us are seen as smaller. The camera naturally records this. But since we 'know' that buildings are in fact the same size both at the bottom and the top, this 'looks' unnatural. This can be 'corrected' in order to 'look' more natural by moving the lens in the camera, or darkroom enlarger, away from the square-on position. The eye readily accepts a small degree of correction like this (actually it's not a correction, it is in fact an optical trick) but when it is taken too far, often deliberately, it looks wrong, but can be visually appealing.

That is how the image in the OP was produced - a combination of camera and darkroom manipulation. The camera was a large format type (as per the pic of the photographer) which allows both the lens and the film to be moved every which way to achieve various optical effects. In normal cameras, both the lens and film/sensor are fixed. With digital today, it is possible to do all these things, and many more, with post processing software.