I've been told (so to speak) that 90mm to 100mm is a "ideal" portrait lens, which is to say for pics of people. A better way to think of it is that you want to be a certain distance away from the subject(s) for the nose/face/ears perspective to look right, or at least not strange. For that distance ask yourself what lens will make the subject(s) fill up the frame. That's the ideal portrait lens. For that distance you could chose any lens you want but you'd have to throw away part of the captured field if you got too much. The same goes for everything else. The perfect lens fills the frame for whatever subject you want for the distance you are from the subject. Pic your distance, then pick the lens which fills the frame with your intended subject.
JMHO of course but the geometry works out that way.
jimmuller wrote:
The perfect lens fills the frame for whatever subject you want for the distance you are from the subject. Pic your distance, then pick the lens which fills the frame with your intended subject.
Traditionally many people had photographs taken at typical interaction distances, about 5-6 ft. (1.5-1.8m). More formal portraits were at 3m or 10 ft. Now the peoples are using wide lenses and capturing images of themselves on the Samsung and Iphones. So the standard of acceptable human perspective had changed. I think it looks awful to have the hugely chin and tiny ears of a camera so close, but human arm length is short.
I suppose it is an ill defined term these days, but I’m pretty surprised no one has posted what I at least learned from multiple photography books published across multiple decades:
Normal focal length is the focal length that will produce no perspective distortion for the viewer. Thus it depends not just on film/sensor format, but also on the size of the print and the distance it is viewed from.
Because people usually hold or stand about the diagonal away from a print that leads to the rule of thumb of the same as the diagonal of film/sensor format. And it has zero to do with the eye…
However, that’s typically not true for rather small or rather large prints. Postcard size print is usually held further than the diagonal away and thus “normal” for postcard size is notionally longer than the imager diagonal. Giant wall size prints are usually viewed from closer than the diagonal and thus shorter focal length than the imager diagonal is “normal”.
Obviously there is quite a bit of wiggle room before perspective distortion becomes noticeable and quite a bit of variation in how people view prints, so any precise definition of “normal” is fairly pointless. That also means one can develop all sorts of false narratives as to “why” such and such a lens is “normal” based on some particular anachronistic market force from the past. Usually people who have shot many different formats immediately see through such appealing if misleading definitive narratives while those for whom photography has always been 35mm 3:2 fall victim to them easily.
kwalsh wrote:
Normal focal length is the focal length that will produce no perspective distortion for the viewer.:
Ah, but that is only half the equation, and as sometimes understood is incorrect. Perspective and perspective distortion are determined solely by the position of the camera w.r.t. the subject, person or otherwise. You can choose your shooting position and pick a lens to fill the frame or you choose your lens and stand where the subject will fill the frame. Regardless of how you do it, perspective is determined by your camera position. All the lens does, long or short, is fill the frame or not.
Perspective is solely a result of camera position, but perspective distortion is significantly more visible in wide angle lenses due to the field of view captured. Perspective distortion is in a normal lens as you can see converging lines when tilting the camera up or down, but those are exaggerated further by the stretching that happens at the edges of the frame in wide angle lenses.
And I think you’ll see that if you do perspective distortion correction from a fixed position with a 50mm lens and a 14mm lens you’ll need to stretch that frame a LOT more when correcting the UWA, so I think it would be safe to say there is more perspective distortion with a wider lens, even if the perspective itself is the same (as in, the relationships between positions of objects in the frame are the same).
Jman13 wrote:
Perspective is solely a result of camera position, but perspective distortion is significantly more visible in wide angle lenses due to the field of view captured.
I think I see what you are saying. Correct me if I'm wrong. I was using the term distortion for things like noses too large because the camera it closer to a face that one would typically see the subject. You are talking about the radial scale of the image not being constant. With a pinhole camera the radial (horizontal or vertical or any combination) of any point on the subject is produced exactly in scale in the image. Any point on the subject is a distance h away from the center line as sine of the angle off center. That line cast onto the image is scaled exactly the same, the two triangles being geometrically similar.
I *think* you are talking about the radial scaling not following that similarity relationship, i.e. with a non-constant radial scaling. If that is in fact the case, then it is a weakness of sorts of the lens.
As far as distortion from noses too large, etc, then yes, you would be correct - that stays consistent regardless of focal length and is dependent only on camera and subject position.
Perspective distortion is also often used to describe the exaggerated change in angles from tilting a wide angle lens up, which is how I was meaning it, but yes, two different things.
Perspective distortion is dependent on the ratio between FoV of the captured image and the angle the viewed print occupies. If they are the same there is no perspective distortion. If there is a substantial difference there is perspective distortion.
If the print occupies a much larger viewing angle than the FoV the camera captured we get “compressed” perspective distortion. This of course happens with long focal lenses (e.g. “telephoto” lenses) unless we decide to view a small print from across the room. The camera captured say 5 degrees but when we view the print at a typical viewing distance it occupies say 50 degrees. We appear to have brought the scene “closer” by making everything in the scene larger but the relative sizes of objects in the scene are still those of something far away. That’s the origin of “perspective distortion” - the size relationships in the print no longer behave as we expect.
Explained better than I can, with example images, here:
kwalsh wrote:
Perspective distortion is dependent on the ratio between FoV of the captured image and the angle the viewed print occupies
That's exactly what I was saying, that the distance from which one observes a picture needs to produce a viewing angle that matches how the picture was taken.
One could argue that the perspective distortion described isn't really distortion but rather the way things actually are. If you stood in the the same place and took shots with two lenses, one long, one short, they would capture images of different sizes so that the long lens lost the edges. But if you the made the captured frame larger with the long lens (and no part of the lens mount or camera got in the way), the edge effects on its image would be the same as with the short lens. The geometry is the same. You just don't see it with the long lens because its image is so large it is truncated by the edges of the frame.
A thought experiment: Imagine you are viewing three vertical telephone poles, each nearly infinitely long. One is right on front of you, the others off to the side on a line from the first that is perpendicular to a line from the first pole to you. In other words, the poles and you form a right angle with the pole in front of you at the apex of the angle. Now tilt your head back slowly, eventually looking straight up. The pole in front of you will always appear totally vertical. The poles to the side will tilt forward so that by the time you are looking straight up it will appear nearly horizontal to you. The pole closest to the center pole will tilt forward more slowly. (I could work out the 3-d trig for you but I really don't want to!) That tilting is real and the amount depends solely on the position of the pole w.r.t. to you. If you took a pic with a short enough lens you might capture both "horizontal" poles. If you used a longer lens the image would be larger and might not capture that far pole to the side at all, and maybe neither side pole. But the change in apparent angle isn't distortion. It's geometric and a matter of what your frame can capture. The angle change of the pole you can't capture is the same whether you capture it or not.
philip_pj wrote:
50mm cemented its position because back then (ah, the halcyon days of the 20th century) camera users took photos of people. It used to be estimated around 90% of what they shot were of fellow human beings. Now it's more like 5%, if that. And it's easy to design and produce 50s. cheaply, as small light lenses.
EB-1 wrote:
Traditionally many people had photographs taken at typical interaction distances, about 5-6 ft. (1.5-1.8m). More formal portraits were at 3m or 10 ft. Now the peoples are using wide lenses and capturing images of themselves on the Samsung and Iphones. So the standard of acceptable human perspective had changed. I think it looks awful to have the hugely chin and tiny ears of a camera so close, but human arm length is short.
EBH
The phone cam has made wide angle lens closeups of faces common. A regrettable trend, and I'm usually a wide angle enthusiast for most types of scenes! When snapping a photo with my 3-lensed smartphone, I often pick the "2X" magnification so as to avoid the omnipresent 'smartphone wide angle' look.
A normal lens presents an angle of view similar to that of human vision. (46 degrees). With full frame that is a 50mm, with APS-C it is 35mm, with 2 1/4" medium format it is 75-90mm and with 4x5 large format it is 150mm. It isn't all that complicated.
