Face distortion is not due to lens distortion

Here is a series of photographs of my face, taken from different distances, using lenses with different focal lengths (see here and here for more examples). Because I covaried distance and focal length, my face appears about the same size in each image. However, the relative size and positions of my various facial features changes very markedly – in the last photo I have no ears! Why does this happen?

Photos of me using different focal length lenses (85-8mm on an APS-C sensor, so equivalent to 127.5-12mm on 35mm film) from five different distances (200-20cm). Note that the wide angle lens used for the last three photos is not a fisheye lens (it’s one of these).

Many people would probably tell you this was due to ‘lens distortion’ – implying that the wide angle lens used for the photos on the right somehow distorts reality, creating an imperfect image. This is totally incorrect. Actually, the only ‘distortions’ are caused by geometry, they are nothing to do with the lens or the camera.

In the leftmost image, the subject is far away (2 metres) from the camera. At this distance, each of my facial features is a similar distance from the camera – within a few percent of the total distance – so my face appears flat. In the rightmost image, I’m about 20cm from the camera. Because my nose is about 10cm away from my ears on my head, this means that there is a large proportional difference in the distance from the camera to my nose, and the distance from the camera to my ears. My nose appears much larger, because it is proportionally closer to the camera than the rest of my face.

The crazy thing about this is that it happens in real life too, we just don’t often notice it. If you look at yourself in the mirror from very close up (or get close to someone you’re intimate with), you get exactly the same distortions (closing one eye helps with this, as most people can’t maintain vergence that close). I find the middle image above to be the closest to how I think I look, probably because the distance of 40cm means it’s approximately what I see when I look in a mirror normally. Portrait photographers usually use long focal lengths (at least 85mm) because this is thought to produce a more ‘natural’ and flattering portrait.

All of this is important, and not just to make yourself look hot on Facebook (a hint ignored in both those links: don’t get too close to the camera!). Every time we go through passport control at an airport, the photo on your passport gets compared to the real life you, by someone who never met you before. Similarly, if you ever end up in court for something, CCTV evidence – usually shot from many metres away – might be used to identify you. It turns out that people are surprisingly bad at correctly identifying strangers in this way (e.g. see this paper). I wonder how many false convictions this has resulted in over the years….

36 Responses to Face distortion is not due to lens distortion

  1. Anon says:

    I saw a poster at a conference a while ago that was studying this (or, more precisely, they were studying the effect those distortions had on social judgments of those faces)–I think they published their results in PLoS One recently.

  2. rika says:

    Makes alot of sense, but isn’t there a lens distortion-effect also?
    Check this out; http://verybadfrog.com/36785/photo-stories/how-camera-lies-lens-distortion-portraits

    • Bill says:

      No, that’s not the lens, that’s the proximity effect. However, the idea that the lens is the cause is frequently repeated and is firmly in people’s heads.

  3. david says:

    Why did you not photograph your face from the SAME distance with different lenses? Then you would have seen that ALSO the lense size has an effect on the picture. Incorrect article.

    • bakerdh says:

      Hi David. Yes, that’s a good idea, I might try it sometime.

      • Corwyn says:

        Do. You will discover that David is mostly wrong. Lenses are admittedly not perfect and will give you some distortions, but if those distortions were large the lenses wouldn’t sell (excluding fisheyes, and other lenses *intended* to be distorted)

    • Bill says:

      David, I assume that by “lens size” you mean focal length. Yes, changes in lens focal length will change the field of view and the apparent sizes of subjects within the photograph.

      However, using different lenses in the same position relative to the subject, you will always get the same geometry of viewing and thus the same perspective, even if the field of view and apparent size of the subject both change due to using a different lens.

      That’s not the same thing as the distortion caused by perspective as you move closer to the subject to compensate for the angle of view of a wide-angle lens. You get very different distortions with different lenses precisely *because* of the photographer compensating by moving the camera, thus changing the perspective.

      In other words, it’s not using a different lens that changes the perspective and causes distortion, rather it’s the lens enticing the photographer to change the perspective to get the framing that he wants, and that action by the photographer then changes the distortion.

  4. […] There’s also the technical issue of our close-range camera phone lenses distorting our faces. […]

  5. […] There’s also the technical issue of our close-range camera phone lenses distorting our faces. […]

  6. […] There’s also the technical issue of our close-range camera phone lenses distorting our faces. […]

  7. […] on the left, then I would propose that the helmet is inaccurate. Here is an additional read: https://bakerdh.wordpress.com/2012/05…ns-distortion/ Okay, let's get back on-topic. Last edited by CSMacLaren; 1 Day Ago at 12:50 […]

  8. […] Mirror vs Front camera? i think the camera can fuck with your face Face distortion is not due to lens distortion | Daniel's Visionarium whether it's lenses or the geometry thing or whatever ^ i mean are you gonna tell me all of […]

  9. […] There’s also the technical issue of our close-range camera phone lenses distorting our faces. […]

  10. […] easier to take selfies of the ‘acceptable’ type I described above; they also reduce the facial distortion that close-up photos inflict on my flawless […]

  11. David says:

    For a curvilinear lens that has no distortion this effect is entirely a result of geometry. I wrote up an analysis of this effect as it relates to movie prop helmets and quantified it in terms of differences in angular magnification along the depth of the object. Longitudinal magnification is another method for analyzing this apparent distortion which quantifies how much an object is magnified in length along the viewing axis as a function of distance. It will appear stretched at close range. When we see an object in 3D our brains compensate in judging the correct look of an object along its depth dimensions, but in 2D the distance from the object does affect our perception because of the lack of that extra information of having two viewpoints from both eyes.

  12. […]  Face distortion is not due to lens distortion by Daniel Baker […]

  13. Manny says:

    “it happens in real life too, we just don’t often notice it.”
    The reason is we have two eyes who see two different images and a brain that puts together these two images into one with less distortion.

    This begs the question: why don’t camera makers sell cameras with two lenses and a big brain?

  14. […] comme ce poste en Visionarium de Daniel notes, cela se résume à la géométrie et non une sorte d’effet de distorsion optique. Cela […]

  15. NewEden says:

    Hello.
    Do you know the wide-angle distortion correction Android smartphone applications?

  16. Mw says:

    I do wonder, if getting a photo id done in a booth is technically more accurate, but distorts things, so you don’t quite look familiar.

  17. No wonder so many people struggle with their self image, it is literally being “modified” on different reflective surfaces and different types of media. Even mirrors are often telling a lie, depending on lighting and mirror surface.

  18. […] There’s also the technical issue of our close-range camera phone lenses distorting our faces. […]

  19. […] of York writes that people incorrectly assume that selfies exaggerate specific facial features thanks to lens distortion, but it’s simple geometry that’s to blame. The parts of your face that are closer to the camera […]

  20. […] Daniel Baker, a psychology lecturer at the University of York in England, examined this issue in a blog post. Baker explains that tight shots can distort your facial features. He makes his point with five […]

  21. Lily says:

    Not true. Standing close to the mirror in real life does not create a exaggerate perception as the camera does. Not only that, there’s plenty of people who has done the same thing as you have, except they keep the subject at the same distance as the camera- and every photo that comes out ends up looking distorted. The finale photo came out closer to real life/proportionate. It’s pretty obvious from your expirement, if you keep moving yourself, you’ll get different results. I would be impressed with your finds had you kept yourself at the same distance for all photos using different lens (like many other photographers have done). The fact you chose the middle photo to be “true” to real life, sounds as though you are picking based on which photo you feel you look best in- even though the background in the middle photo looks distorted. The background on the far left photo looks way more accurate to how a building structure looks. Even at the same distance looking into the mirror, the background shouldn’t distort like the middle photo you used.

  22. […] To see the Note click here.To hide the Note click here. As it turns out, I am not the first to call out the lens distortion theorists. See Daniel Baker’s blog Face distortion is not due to lens distortion. […]

  23. […] is more a matter of geometry, as Daniel Baker, a lecturer in psychology at the University of York, explains on his blog. The parts of your face that are closer to the camera seem larger than other features in comparison […]

  24. Lilly,

    Daniel clearly shows he looks the best and most accurate in the first picture on the left and his ears are fully showing,his nose is small and his face isn’t long,in the last picture his face is totally distorted and his ears don’t show at all.

    • Eric says:

      This is correct. You do not get a distortion effect from looking into a mirror at any distance (unless the mirror is bent /curved). And yes, the left-most pic is most accurate

  25. […] his blog, a senior lecturer at York University determined that how far you are from the camera massively impacts the photograph. You can see in his pictures how different he looks each time, and […]

  26. […] people you wouldn’t normally date, and that could lead to something magical. Especially as photos don’t always look like their recipients. And as traditional online dating revolves around pictures, being less dismissive of potential […]

  27. helen says:

    The degree of distortion is greatest for faces that deviate from the others in the set on a particular dimension

  28. Fred Rample says:

    The two most common types of lens distortion are barrel distortion and pincushion distortion.
    Barrel distortion is where straight lines bend outward from the center of the image.
    Pincushion distortion is where straight lines bend or “pinch” inward from the center of the image.

Leave a comment