Make exactly the same framing on a face with different focal lengths. With a zoom lens, use at least three: at either end of the zoom scale and in the middle. If you have more than one lens, use that, too. You will need to move the camera towards and away from your subject to keep the framing consistent.
Subject: this is long-suffering wife and reluctant model Ann.
I’ve converted the focal lengths into 35 mm equivalent, as my camera works to a 1.5x crop factor. If you’re peeking at the EXIF data, yes I have rounded up a couple of the numbers for neatness.
24 mm Equivalent Focal Length:
Way too wide for a normal portrait, this has the distorted ‘back of a spoon’ look to the nose in particular. Quite unflattering!
35 mm EFL:
Better, less obviously distorted, but still not wholly natural. Still a little bit ‘looming towards you’.
50 mm EFL:
This is the first one that looks reasonably natural to my eyes. The proportions of the facial features are quite normal.
75 mm EFL:
Of the focal lengths I tried, this is the one that is meant to be closest to the optimal one for portraits. It does look natural and flattering, so the theory is sound in my experience thus far.
100 mm EFL:
To my eyes this looks equally fine – no obvious distortion, nothing that looks too odd.
150 mm EFL:
This is where slight visual oddities started creeping in. The face shape is starting to look marginally too wide to be a true likeness. The features are starting to look flattened out.
200 mm EFL:
The flattening effect is getting more noticeable. The facial features are starting to converge on the centre of the face and the space around them is getting disproportionately broad.
345 mm EFL:
The features are flattened out to an almost comical degree here. Compared to the more optimal ‘middle’ focal lengths, this squeezes the features further into the centre of the face, and the whole face shape looks like it’s been stretched out on a rack. Very unflattering! (sorry Ann)
What I’ve learned
A great technical exercise! It’s unlikely that anyone would knowingly take portraits at the two extreme ends of the scale here as they each in their own way distort the subject’s features in a very unflattering way. It is very enlightening however, to see the extremes as they help to illustrate exactly why there are accepted standards on the ‘best’ focal lengths for portraits. Interestingly, viewing each one with the adjacent versions either side it’s hard to spot the differences – but when you look at one from each extreme, they are very different indeed.
One point of note is that I found quite a few online sources ‘correcting’ the commonly-quoted belief that it’s the focal length that causes the distortion – technically it’s the distance not the focal length… but if you’re framing the image in the same way, that requires to you compensate with your distance from the subject, which is what causes the distortion. I recall an exercise on Art of Photography that showed that an equivalently-framed crop from a wide shot would match a telephoto shot taken from the same distance.