From the mailbag:
I want to buy a lens in the future for portraits primarily for self portraits. Most people want portrait lenses to produce pleasant images instead of realistic ones. For this, with my facial features favor a longer focal length. But it got me wondering how far away we have to be away from the subject for the subject’s face to look “normal”. Is there any good research on this? People have an image of what a person looks like in their head (presumably?), so how far do I have to be to replicate that?
The reason why I have such a question is because sometimes I want to take pictures to make things look nice (art) but sometimes I want to document reality (high resolution reality measurement device). But with portraits it’s very confusing to me how to shoot “accurate” portraits.
To reproduce the taking perspective in the viewing environment, it is sufficient to make the viewing distance the same as the capture distance. By viewing distance, I mean the distance from the viewer to the photograph. By capture distance, I mean the distance from the camera to the subject. So If you photograph a person from 1.5 meters away, if you view the image from that distance you’ll see a “natural” perspective.
Once you know that, the selection of the lens becomes a matter of knowing the viewing distance and picking the focal length that gives the desired field of view.
You may have noticed that I said the above is a sufficient condition. It is possible to relax the constraints somewhat. Note the above formulation means that the subject is life sized in the photograph. But you’d have the same geometry if you made the subject half sized in the photograph and viewed it from half the subject distance. However, in that situation, higher order psychological effects may come into play.
What size of the photograph did you imply for the same viewing distance?
Depends on the field of view that you want. For a straight head shot, at least 15 inches high, I’d think. For head-and shoulders, larger. The person will be life-size in the image.
Thanks, does it mean that viewing distance is proportional to object size on the photograph to have the same viewing angles? This is a bit confusing.
I think if you consider that the image in the photograph is life-sized, you can answer your own question. I hope so, because your question is not sufficiently detailed that I can answer it myself.
Note that is a sufficient condition, not a necessary one.
I forgot to consider viewing distance in the original question and in retrospect it’s important. I’m thinking about what looks right for a person on their desktop computer maybe 20 inches away from a 32-27in 4k monitor. The viewing distance will be shorter than the shooting distance. Let’s say subject size is same as real life size. Then I think to compensate I’m supposed to shoot a little farther away than I would if the viewing distance was the same as the shooting distance, right?
If shooting distance is 6ft and viewing distance is 6ft then it’s accurate.
If shooting distance is 6ft and viewing distance is 3ft then viewing distance is too close.
If shooting distance is 12ft and viewing distance is 3ft then is it accurate again?
When does this rule break down? If I shot 1 foot away and viewing distance was 1 foot away it might be accurate in terms of what the eyes would see in real life, but nobody would say the image looks “right” because that’s not the image people have of my face in their minds.
I’m deeply confused. 🙁
If subject is life size in the photograph, then viewing distance should be the same as shooting distance.
Jim, perhaps I’m misunderstanding the misunderstanding, but perhaps Erica is somewhat conflating the issue of size as seen on the photograph with the issue of perspective distortion. IIRC (can’t check it because my copy isn’t handy), Leslie Strobel’s “View Camera Technique” book had some good diagrams/photos showing perspective distortion, including the difference between photographing planar objects vs. 3D ones.
Oof, I will list all the understandings I have about this whole thing so anyone can pick them apart if they feel like it.
Perspective distortion is a function of distance to the subject when taking a picture. Too close and distances between close and farther away objects are exaggerated. Too far and everything is compressed. Being too close or far causes noticeable distortion.
A friend might look at such a picture and think it looks “wrong” somehow even if they can’t explain why. Technically speaking it’s not “wrong” because if a person was actually camera distance away from me that’s probably akin to what they’d see.
But subjectively they might feel it’s “wrong” because what they see in the picture does not match the mental image in their head of how I look when they interact with me day to day. So somewhere between shooting too far away and too close is a range of distances that are “correct” or “normal” looking.
People who look at me closely and interact with me tend to be within normal speaking distance… so maybe 2-4 feet away. So intuitively I would’ve thought to capture an image that accurately shows what people see when they talk to me, I would take a picture 3 feet away. But nobody ever recommends doing it so closely. Online I see people saying to always shoot 6 feet away to “avoid distortion”. Is that because the way the camera sees the world is very different than the way humans perceive the world? Perhaps a friend 3 feet away sees what the camera might’ve seen, but their brain processes the image and corrects the perspective a bit. If that’s the case then it would be wrong to shoot 3 feet away.
The brain undoubtedly always does a lot of things processing visual data, and from what I know/have gathered so far, the analogy between a camera and human vision (as opposed to just the eye as an optical instrument) is troublesome some to begin with. But independently of that, consider two things:
1. Even though a person looking at you from 3 feet away will “see” the same perspective distortion as a camera (in the sense that the image on each eye’s retina will have the same perspective distortion), neither you nor your observer will ever be perfectly still. Thus, even during very short timespans, the observer will perceive more visual data than a camera would during a single exposure. And the longer you remain at that viewing distance, the more movement will occur and the bigger the difference will be between what a single exposure contains and what a human oberserver sees.
2. Humans, unlike photographic cameras, have stereoscopic vision. This also significantly increases the amount of visual data perceived by a human—and that’s not just a mere increase in the amount of detail but a fundamental change in the type of seeing. Three-dimensional versus two-dimensional, if you will. (Or, if you extend upon my first point and include time, you might even say four-dimensional.)
Given that, it would seem unlikely that perception of perspective by human beings (of three-dimensional objects) is identical to perception of perspective in photographs (which are effectively two-dimensional and thus deprive the viewer of the additional perceptual data mentioned above).
Also, speaking from an evolutionary standpoint, faces are probably the single most important perceptual object for the human visual system. There is most certainly a hell of a lot of further processing going on in the brain a far as faces are concerned.
In the end, no portrait—no matter the perspective—could ever be entirely natural in the sense that it is “identical to how humans perceive faces”. I guess the general wisdom to use short-to-medium distances that avoid both strong compression and strong emphasis of near-far relations just produces a result that is a good compromise fitting decently with how we ususally perceive (and remember!) faces (or heads, or whatever). But I don’t think equating distances/perspectives between two very different visual systems will yield one “right” answer for your question. (Nor will anything else, most likely. 😉 Though it would still be very interesting to see a study comparing psycho-visual perception of portraits and actual faces at certain viewing distances.)
And then, of course, there’s the matter of perspective “distortions” sometimes actually bringing out important features in portraits… 😉
Nate Weatherly says
I’m not sure it needs to be this complicated! For portraits it really has almost nothing to do with your lens’s focal length and everything to do with the distance from the camera to your subject. My rule of thumb for new portrait photographers is to choose where you’re going to stand first and THEN choose your lens. Think about the effect you want the image to have and how that relates to the physical distance between you and your subject. If you want your image to have a personal, relational, peer to peer feel, then you’ll want to be standing at a normal, comfortable conversational distance from your subject (maybe 3-4 feet) when you take the image. If you want a sense of deeper intimacy between the viewer and subject then you want to be within arms reach. If you want to give a more professional feel, a sense that this is a person who owns their space and is confident and maybe cooly detached, then back up to 6-10 feet or so. If you want to give a sense of stark detachment and distance then back up even more.
Then, after you’ve chosen the subject to camera (or really, the subject to VIEWER relational distance) choose your focal length and aperture based on how much of the person/scene you want to show, whether you want them to be integrated with or detached from their environment.
The fact is, if you’re standing the same distance from your subject every focal length will render their features with the same amount of perspective “distortion”. If you maintain the same subject distance and a photo taken with a 200mm and a tightly cropped 35mm frame will show the exact same amount of “distortion” (apart from the barrel/pincushion distortion which are artifacts of the lens designer’s choices, not inherent to focal length).
But I would wager 10:1 that most of the good folks that come in to have their portrait made would rather have it done with a lens possessing a proper focal length for the given camera format. After all and understandably so – they just want to look their best and not be caricaturized. With that said, let’s dive into the nitty-gritty of portraiture and lens selection.