Agreed - and someone talking again and again about "the appearance of background objects being closer to the foreground objects" is misleading - as is clearly shown when a zoomed out image is cropped to equalize the framing of those objects.
I meant an auto-metering mode, however it was about the visible noise: even if you keep the exposure the same, as you increase the ISO setting, it will gradually reveal more noise in the deep shadows. Although the fixed exposure should be low enough to not get highlights blown at high ISOs. But that's an off-topic I guess... :)
Even with a constant exposure it will increase the visible noise with certain constraints as in my message above.
Anyway the simple fact is, zooming in doesn't change the perspective, it only amplifies it somewhat. Same as with ISO that doesn't create but amplifies the visible noise (makes it more apparent).
This fact doesn't frustrate me.
This is an example of the "nonsense" that I was referring to in the title of this thread.
The argument above is incorrect because it assumes that the perspective captured in an image is exactly the same as the perspective we see when we view that image.
In fact, the perspective we see when viewing an image depends on the position of the viewer relative to the image: both on the distance away (relative to the size of the image) and on the angle of view (relative to an axis through the centre of the image and perpendicular to the plane of the image).
When viewing a photograph (taken with a rectilinear lens), there is only one viewing position that gives exactly the correct perspective, other viewing positions give different perspectives which are distorted (often only slightly, but sometimes more obviously). By "correct" perspective, I mean the perspective seen from the camera position.
If the image is viewed from too far away, then wide-angle perspective distortion may be apparent.
If the image is viewed from too close, then narrow-angle perspective distortion (i.e. telephoto compression) may be apparent.
Whether or not distortion (of either type) is apparent depends very much on the subject matter of the image.
This diagram, from the Manual of Photography (7th edition, 1978) may help. It shows the viewing position needed to see the correct perspective, i.e. exactly the same perspective that the camera saw. For this to happen, the angle of view seen by the camera and the angle of view seen by the viewer must be the same.
Saying "it's not zooming that creates the compression effect" is incorrect.
Saying that "someone talking again and again about the appearance of background objects being closer to the foreground objects is misleading" is incorrect. Compression means that background objects appear closer to the camera; foreground objects also appear closer to the camera; and background objects also appear closer to foreground objects. All appear closer by the same factor.
Cropping a zoomed out image is irrelevant. As I explained, it is magnifying the image (e.g. by zooming the lens; or by looking through a telescope; or by cropping and then enlarging the crop) that causes compression.
The issue is that the objects appear closer to each other without appearing closer to the camera, and it's still perceived as compression.
That's probably the main point of controversy here.
The compression effect is just perspective at a long enough physical distance where the ratio of angular sizes of objects converges to the ratio of their physical sizes. Again the compression can be seen without any zooming/magnification/cropping, maybe if I frame it like this it'll be clearer:
And yet again, compare the image above with another one, where zooming in from a different (closer) perspective doesn't show any compression:
I agree that picking the right viewing distance to the image can amplify the effect, generally classical paintings were designed to be viewed at a certain distance. With digital displays, phones and monitors, the best viewing distance is disregarded most of the time but we still see compression or the lack of it depending on the physical perspective, not the viewing distance.
That is the well-known "dolly zoom" effect. It is a combination of perspective distortion and change of camera position. In other words, the viewing perspective and the camera perspective are changed in such a way that the subject appears to remain at the same distance (the subject is kept the same size in the frame).
So, it is still compression, but done in coordination with changing the camera position. There is a very nice computer simulation of the effect shown in the Wikipedia article. Notice that on zooming in, the background gets closer to the subject while the foreground also gets closer to the subject. In other words, everything becomes compressed closer together.