The Ansel Adams Fallacy: "True perspective depends only on the camera-to-subject distance"

Thanks for your posts, although I confess to not being sure exactly what you are saying.

I rather like the following diagram from Jacopo Barozzi da Vignola (1583) in which a drawing or painting is imagined as transparent and the lines of sight are shown in more detail. When viewed from the centre of perspective, every point in the 2D image lines up exactly with the corresponding point on the 3D object. The position of the viewer's eye is critical.

It is obvious that if the viewer moves either closer to the image or further away, then the image will be seen with a different perspective in which points on the image cannot all line up with the corresponding points on the object.

- from Wikimedia

This diagram works equally well if we imagine that the image is a photograph on a transparent medium. The viewer needs to be at exactly the position of the camera when the photo was taken. It should then be possible to position the image so that everything in the image lines up exactly with the real scene behind it. The viewer's eye is then at the centre of perspective of the photograph.

The viewers position has NOTHING to do with the perspective of the image they are viewing, it changes the perspective they see, but not the perspective of the image.

You may as well claim an image is underexposed because the viewer has the lights off!

Obviously.

Perspective distortions such as telephoto compression do not exist in the image per se, but they may exist in our view of the image.

Actually 'Telephoto compression' is in the image - it is exactly the distance related perspective discussed above. It's not something magical to the lens cropping or enlarging a normal image from the same position will show just the same perspective.

So, you are saying that telephoto compression is in the image itself and does not depend in any way on how that image is viewed, is that correct?

If that is so, please describe precisely how you would determine that one image has telephoto compression while another does not. And how much telephoto compression does it contain, can you quantify it?

I don't think your theory will survive any close scrutiny, but let's see what you can come up with.

For me as a photographer, the only thing that matters is the "feel" of the used 'perspective distortion'. When I want a flat, correct 'look', I will take a long lens; when I want a 'close, spectacular' look, I will use a short lens.

For me it's often decided by how close I can get to the subject. For air shows & motorsports close is rarely an option, and for real estate a wide angle can be need to get enough in - making the rooms look big is a side advantage 😁

The 'compression' - the perspective - is defined by relative sizes of objects in the frame and their relative positions. That's solely defined by the camera position and distance to the objects. It doesn't depend on the distance between viewer's eyes and a printed photo.

I know this argument has been going around in circles for many months now - but nothing can change the fact stated above - maybe a different universe with different physics and optics?..

I think it's helpful to refer (again) to what Adams actually wrote about what happens with the image recorded in the camera vs. what happens when the resulting image is viewed:

He calls the first thing the 'true perspective' and calls the second thing the 'impression of the perspective' or the 'perspective effect'. That works for me.

Well, maybe that universe with different physics and optics exists solely in my, and a few others', heads, as it is clear enough and observable enough to me.

I wouldn't have any desire to determine it let alone do it precisely.
If I needed to do it as an exercise I would shoot a series of items of similar height, in a reasonably straight line & separated by regular distances (Telegraph poles or fence posts are good options) comparing the relative heights of these on the the image would show the perspective.

If you are standing right by one telegraph pole, the next will be x feet away, then 2x, 3x, 4x etc.
This makes the second in view twice as far as the first (and so the second has half the apparent height). A wide angle may be needed to get all of the nearest pole in .
The third is 1.5 times as far as the second (3x/2x) so has 66% of the apparent height of the second. This could be the typical view of a normal lens
By the time you've got to the tenth pole with distances of 9x & 10x the apparent height of adjacent poles only differs by about 10%. This might be the sort of distance typically seen with a telephoto lens, (it depends on the focal length & sensor size or how telephoto the lens is).
(Technically a telephoto is actually one that is physically shorter than it's focal length, but here I'm working with the more common understanding of telephoto as simply a long focal length lens).

Accurately measuring the height of the tenth pole in the series might be a little challenging, but there are ways to do it reasonably well. You will see the same relationship between poles 9 & 10 on a wide angle lens as with the telephoto, its just the telephoto doesn't give you the height of the nearer poles.

What is observable exactly? You mean relative sizes and positions of objects in the frame depend on, errm, imagination/perception when viewing a photo?

This is a reasonable and helpful division. I can control the "true perspective" of the camera by choosing where to move the camera (and in some cases, by moving elements in the scene) for creative impact, then combine that with the desired focal length to ensure my subjects are the intended size in the frame. By and large, I can't control how people choose to view the output. If someone really wants to get their eyeball an inch away from a print for their viewing pleasure, or walk across the room and squint at the distance, I can't really stop them. In my experience, most people view prints/screens from a relatively narrow range of distances based on what's comfortable for them, and thus will see the effects of my camera/subject placement as per my creative intent.

The question the OP is asking here here can be simplified with an example. He is interested in the perspective effect of a print due to viewing distance.

So for example, I view a fisheye shot from a very close distance, the perspective will appear normal. I think this is what the Op is getting at.

Of course we all tend to view prints and screens in certain range of distances between 1m an 30cm, in most cases.

With many fisheye lenses the distortion will NOT look normal viewed close up (I'm not sure any do) There are multiple fisheye designs which map the view in different ways (the mapping function section of en.wikipedia.org/wiki/Fisheye_lens shows this well), some of my 'fisheye' converters cram a significant part of their FOV into the very edges of the image. I think the 'Equisolid angle' function is the nearest for these. Fortunately most of my true fisheye lenses are closer to equidistant or stereographic designs & give a much better image. However fisheye lenses are a rather special case, using distortion to fit in more than would normally be possible!

Changes to the viewing conditions are not the perspective of the print, they relate to a viewing perspective that is not part of the print. There are certainly conditions where the viewing perspective can correct for distortion of perspective in an image. For rectilinear lenses Ansel's claim that perspective of a print is dependant only on camera-subject distance is 100% correct. Not a fallacy as the OP claimed.

At last we are getting somewhere!

You seem now to admit that when the viewer's position changes, the perspective seen by the viewer changes.

Of course, the print itself doesn't change when it is viewed from a different position. The perspective seen by the viewer is a combination of the perspective captured in the image (which depends on the camera position) and the perspective distortions introduced by the viewer's position (relative to the centre of perspective of the image).

The passage by Ansel Adams that I quoted in the OP talks about "wide-angle perspective" and "telephoto perspective". He says that they are caused by the camera-to-subject distance, which is wrong. That is where the fallacy is.

I never denied that and neither did Ansel. perspective that is down to unusual viewing conditions is not perspective of the image.
The perspective of the image is totally down to distance from subject.

IIRC Ansel covered changes in apparent perspective due to viewing conditions in his later book in the series on printing. Viewing conditions of the print being quite irrelevant to 'The Camera' where the passage quoted orginally came from.

I agree with that, except that viewing conditions do not always need to be "unusual" to cause perspective distortions.

If you make images and never view them, what is the purpose of the image?

As soon as you view your images, the viewing conditions need to be taken into account because they can change the perspective that the viewer sees (of course, they do not change the perspective captured in the image).