One of the common misunderstandings in photography has to do with the focal length of a lens, or its optical distortion properties. Many photographers claim that a wider angle lens will distort facial features either because of the lens distortion, or the focal length of the lens being too short. In this article and the accompanying video (which is extracted from our upcoming Photography Life Basics Video), we will prove that focal length has nothing to do with distorting a subject’s face and the additional information on lens distortion will explain in detail exactly what gets impacted by lens distortion.
What is Face Distortion?
Before we discuss the impact of lens distortion and focal length, let’s first take a look at what face distortion is. While the word “distortion” here sounds similar to what we see on wide angle lenses that suffer from optical distortion, the two are not the same, as facial distortion is the result of perspective distortion, a phenomenon that occurs due to subject being too close to the camera. Take a look at the below subject, photographed with a short focal length lens at close distances:
We can see that in the first photo the subject is too close to the camera, which completely distorted his facial features – the face is stretched across the frame. In the vertical photo, the camera was moved away from the subject, which resulted in a lot less face distortion. However, the subject is still too close and the area outside the center frame is stretched heavily, causing the subject’s shoulders to appear very disproportional compared to the rest of the body.
While getting close to subjects can result in interesting and funny-looking photos like the ones above, this is not the way to photograph people normally.
Lens / Optical Distortion
Lens distortion, also known as “optical distortion” has nothing to do with perspective distortion mentioned above. Most lenses suffer from some type of optical distortion, but the particular offenders are wide angle lenses, which can show pronounced “barrel” distortion in images, as shown in a diagram below:
Seeing such distortion can easily trigger one to assume that facial distortion is caused by barrel distortion, since the effect is similar – the frame is stretched in a barrel fashion just like what one would see in the image samples above. However, that’s definitely not the case, because even lenses that exhibit the worst amount of barrel distortion won’t distort the subject as long as the subject is kept away from the camera.
Now, does this mean that optical distortion doesn’t play a role in distorting subjects? No, it sure does! If you use two similar focal length lenses with completely different distortion characteristics, the lens that exhibits more optical distortion will definitely make your subject appear more distorted. However, the effect is a lot less pronounced than you might think and can really only be seen at extreme close-ups. The effect of perspective distortion is far more substantial on a subject than optical distortion. For example, the Nikon 50mm f/1.4G suffers from approximately 1.4% barrel distortion, which can be easily seen when photographing anything with straight horizontal lines:
The barrel distortion on this lens practically does nothing in terms of introducing distortion to subjects. However, if the subject is placed at a very close distance, say if the subject’s face is tightly framed, we see two effects at the same time – perspective distortion and optical distortion, both of which are going to make your subject’s features appear slightly disproportional.
This leads us to focal length. Let’s take a look at what focal length does to our subjects in terms of distortion.
Focal Length
Focal length affects how tight our subjects appear in the frame, but it does absolutely nothing to affect the perspective. If you stand in the same spot and use a zoom lens to shoot at different focal lengths, the perspective will remain the same. This can be easily proven by photographing the same subject at different focal lengths, then analyzing the relative sizes of objects in the background. Take a look at the below comparison, shot at 24mm and 50mm focal lengths:
While the two different focal lengths resulted in differently framed images, the size of the subject relative to the background is the same. Here is the same image on the left, except this time it is cropped to have the same framing as the image on the right:
Note that using two completely different focal lengths also did nothing to distort the subject’s face or body, which proves that focal length has nothing to do with face distortion.
Take a look at another image, which I captured at 11mm on a full-frame camera (which is considered to be super wide), then tightly cropped:
At 11mm, if I placed the subjects at a very close distance to fill the frame or have similar framing as in the above image, they would appear extremely distorted. And yet at this particular distance and with this heavy of a crop, the subjects appear normal. This again proves that focal length has nothing to do with facial distortion!
To further solidify this concept, take a look at the below video:
P.S. This video was extracted from our upcoming Photography Life Level 1 Basics video product, which we are planning to launch very soon. Since it was a more advanced topic, we decided to post it for our readers in this article.
In the video, we demonstrated that at the right distance, the model’s facial features appeared normal. But once we moved closer to the model, we saw the immediate effect of perspective distortion on the subject. Here is the same model at 35mm at two different distances:
We moved the camera, but the model remained in the same place. You can see a drastic change in the way her face appears in the two images, with the one on the right showing a much more stretched face, bigger nose and eye. Putting the two images together and aligning them reveals this drastic change in facial features:
This is something to expect when the subject is framed too closely with a wide angle lens.
Some people might argue that the focal length impacts distortion indirectly, because a wider lens requires one to get closer to the subject to keep the same framing on the subject. That’s definitely true, but that again has nothing to do with the focal length of a lens itself – it is the changed distance that causes such distortion issues. Here is another example, this time wih the subject framed the same, but captured at different focal lengths:
Now this is a classic case of perspective distortion – look at how going from 300mm to 14mm changed not only the perspective, but also the model’s facial features and her body.
What is the point to use wide angle lense if you crop you image so much to get no distortion? What about dof? If you want to get the same dof as a 85mm you will get distorsion as you will have to get closer to your subject. So no you won’t get distortion with a wide angle lense is you don’t mind to crop a lot and have your subject fill a small part of your frame. It is like to say I can use my 14mm for wildlife! Yes I can but I will crop a lot and lose resolution! That is just my opinion! I like to get as much details when I take my picture, framing as I want my picture with minimal cropping.
Eric, the point of the article was not to show that wide angle lenses are better or worse than other focal length lenses – it is simply to show that focal length has nothing to do with face distortion. I would never want to use a wide-angle lens for tight portraiture work, as wide angle lenses only work great for environmental shots.
but it is useful demonstration for beginners that will make them step back a little to get a nice (body not head) portrait with 35mm for example…
I’m with Eric – this point is moot. If we wanted to argue technicalities I would say focal length *does* distort your subject, viz: stand in one point, shoot your subject at two focal lengths without moving your camera or subject – say the woman in front of the arch in your article. Now – are you telling me you have two identical photographs? No. The wide angle presents more of the scene while the longer focal length does not. What happened? It zoomed in. Your model is larger on the long lens. Sorry, but your photos are not identical and that’s due to the “distortion” introduced by your zoom – one subject is much larger than the other, distorted by scaling up or down. I think my point is about as useful as saying using different focal lengths doesn’t distort your subject – we might both be technically right, but practically pedantic.
Chris, the model is larger in the image with longer focal length, but the perspective did not change, which is what I was trying to illustrate. See the second image, which shows the cropped image from the shorter focal length lens. I believe your definition of “distortion” does not really fit here, because you are talking about field of view differences, while I am talking about perspective distortion…
Nasin, your point on this arcticle is perfect. The thing is that this is a more advanced concept and it’s hard for beginners to understand it.
Thanks for the great examples; clear and precise explanation of the concepts.
The best way to understand Perspective is that it a function of Viewpoint only. (How close you are)
However, What we see is also the result of field of view.
Field of view is the image circle that a lens can cover.
At any given viewpoint the perspective drawn by any rectilinear lens is identical.
However, the included field of view, varies with the field of view of a particular lens.
Wide field lenses are normally short focus, but not necessarily so, large format lenses of all focal lengths, normally cover a large image circle to accommodate camera movements. As do perspective control lenses.
Thanks for this Nasim – I’ve been trying to tell people this for years. So many so called ‘experts’ say in their books that ‘I used a long lens to compress the perspective’ when in fact they simply moved further from the subject!
Andy, I have seen many similar references from other photographers too, which is why I wanted to clarify the subject in a little more detail with illustrations. Glad you found the article useful!
Well, they *do* compress perspective by using a longer lens. But if they cropped a wide-angle shot TAKEN FROM THE SAME PLACE to get the same coverage as a telephoto they’d have the same effect… and a much lower resolution image. :)
I take onboard what you say about wide angle, but clearly, when a face is close to a superwide, it is perceptively distorted by elongating the distance between nose and back of head, whatever the physics of it.
Of more interest was your sample image of the newlywed couple on the bridge. If this is the happiest day of their lives, I dread to think what they look like when an unexpected bill arrives! They need to ‘wear that frown, upside down’ in the words of ‘Elf’
:)
Ross, thank you for your feedback as always! Yes, you are right, at very close distances it is the perspective that impacts face distortion, not the tools or the focal length. Where the confusion arises is in framing – with super wide angle lenses, we tend to get closer to subjects, because we want them to take up more space in the image, which in turn creates perspective distortion.
LOL for your comments on the newlywed image – that was literally the first image in my catalog shot at 11mm, so I extracted it out without paying attention to details and expressions :)
Nasim, is that bridge in Keystone? Looks very similar to a bridge where I shot a small wedding last summer.
I know the comment was made in jest, but as far as the artistic merits of the sample photo goes, it just goes to show you that persons will judge your overal work by the worst photo you present, even if it was only presented for demonstration purposes in a technical article. We cannot let any mediocre work out for any reason or risk it be held against you in another context.
DVDMike, I get what you are driving at regarding the comment I made about the ‘couple on the bridge’ and Nasim himself LOL’d it, understanding perfectly what I meant. I have seen more than enough of Nasims’ work to know that this was purely a test shot not intended for public or client consumption, and the fun of photography (yes, it can be a business, but it should always be fun too) is that it is a visual medium, and sometimes hilarious things happen, outside of the control of the photographer.
In point of fact, I stumbled across a ‘photobomb’ selection of images online last week where a guy was sitting against the railings of a small boat at sea, and posing for a photo, and in the background, immediately behind him, a shark had leapt out of the water, and had struck a pose which, for all the world, looked as if it was smiling for the photograph – such things do make me collapse in giggles, and I know Nasim took this in the spirit it was intended. I hope any casual readers of this thread will also understand the very high professional standards of Nasims work here, and accept my unreserved apology if my comment confused or upset anyone at all… no way I would disrespect such an interesting and talented photographer like that!
The fun, and joy that we get with perfect images can oftentimes be equalled, or even surpassed, when something amusing emerges from the click of the shutter – let’s embrace that with equal enthusiasm, otherwise, I’m selling my gear and taking up golf! (only kidding).
:)
nikkor 24mm 1.4 for sale for just this reason …
bought it for dslr video which i realized i didn’t want to do … for events, weddings, 24-70 is great, and i go to 24 only in emergency … once tried a 16-35 in a small space, my god, aliens on the edges …
LOL, aliens on the edges :) I have seen too many of those!
Nasim & Gregory,
i was thinking of buying a Sigma 24mm f/1.4 Art for my Canon 6D for Street photography.
There are some shooters who use just one prime lens – 24/1.4
should i buy or not ? or after a while i will grow bored of it and start craving for the 24-70 or 24-105 ?
are you kidding?
i have ended up doing a lot of events and weddings where a zoom is great to use for fast framing/shooting … the 24 is a very cool lens, for street, landscape, environment .. i find it not so useful for people .. or food or interiors …
but my love lately for street has been nikon’s 28 1.8 .. light, relatively small on a d600 .. 24-70 is both handy and heavy .. and i wish it was a bit longer for true portrait coolness ..
one lens? 24-105, can do a LOT .. adjust distortion in lightroom/ps etc
Depends on your intention. In a church to get the bride together with the beautiful surroundings, group shots in tight spaces or on the dance floor. But it is not a widely used lens for us, about 5% of images. I prefer 35mm.
Thanks very very much for you helpful views, dear Gregory, ….. HF, …. and … Mit
For street photography you should look at 50mm. This focal length is a good all rounder and much more useful than a 24mm. I own a Canon 24-105 f4 L which I took on a recent holiday and it to me is a bit bulky and intrusive for a street photography lens. The fact that I can quickly select different focal lengths without swapping lenses is definitely a plus in certain situations. It’s only an f4 lens and I did miss having a shallower depth of field. I do prefer primes overall though.
I’ve shot a few aliens myself. Many years ago, before Nikon had FX, I shot a small wedding reception at a persons home. It was pouring rain and they wanted a group photo. The living room was small, but the exact dimensions escape me. I had to shoot close and wide to fit everyone in frame. I used the 12-24DX. How could I have shot this without perspective distortion?
Years later, I routinely used the 24-70 on FX to shoot the historical family photos inside of churches. Mothers used to always insist on these traditional photos. I’d always go back as far as I could and shoot the group shots at least 35mm. But often in small churches, there isn’t enough space between the pews and alter and or height / room so that you can get far enough back to avoid perspective distortion. As everyone knows who shoots weddings, you have to do your best to make do with the environment that you are given, and often in a very short timeframe.
My point here is suggest a follow up article on how best to overcome situations where perspective distortion of persons would otherwise occur. This is a good article in demonstrating that proximity causes prospective distortion. But the next logical discussion I’d find more useful is how best to mitigate these situations when the obvious solution is just to move further away and adjust your focal length accordingly.
i have used only one lens in the last 2 years, the 40mm f/2.8 pancake. Now, I don’t prefer, in fact i hate the 35-45mm range, not because i have used it for so long, but it is Compositionally a No-Man’s Land. I am of the opinion that the 2 best focal lengths for composition are the 28mm and the 50mm, and anything inbetween is perfectly boring, “neither here, nor there” kind of feel. Due to this, i didn’t buy the 35mm f/1.4 Art Sigma, and yesterday Canon announced a new 35/1.4 II L
28 is a perfect wide (widest with minimal distortion) ; & 50mm is a proper tele (mid-telephoto).
I am not impressed with 35mm as such. So what to do ? i am in a fix. We don’t get a Nikkor 28mm f/1.4D anymore, and on ebay it is expensive as hell. Also, Leica 28/1.4 Summilux is exorbitantly priced. Nikkor 28/1.8G is not highly rated as well, like its Canon couterpart. Hence, the next best option is the 24/1.4 and Sigma provides at less than half rates than OEM lenses. Nasim, i am reading your reviews for the last 3 years, but commenting here for the 1st time only.
Thanks, Warm Personal Regards
Chirag, I’m not sure that I follow your comment and how it relates to perspective distortion. However, I find it interesting as I’ve been shooting the majority of my photos overvthevlast couple of years with 3 focal lengths. I use the 28g for street, walking around, indoor as well as landscapes. I also use 50mm D and both G for walk around and many portrait scenes. The third lens is the 85G.
Unlike yourself, I’ve come to the conclusion that I need a 35mm prime. 28mm is too wide for Too much of what I shoot. And 50 is often not wide enough. It all depends on what scenes you shoot of course, but I was thinking of going to the 20g for landscapes and replacing the 28g with the 35g for basic walking around photos.
Excellent article! BTW this raises a very interesting possibility with high resolution sensors: use a wide angle and plan to crop as a replacement for carrying a heavy zoom. Instead of carrying a 3 pound f2.8 zoom, or sacrificing fstop and resolution by carrying a “bridge” camera, carry a high resolution (full frame?) body with a lighweight prime as the shorter end of the zoom and plan to crop.
Art, Using a prime and cropping instead of using a zoom, is a technique I use quite often because I have some primes that perform better than zooms. On thing to keep in mind with this technique is to compensate the f-stop.
Suppose our scene requires a 100 mm lens, set at f/4 for our desired depth of field. If we use a 50 mm lens we must set it at f/2 in order to obtain the same entrance pupil diameter of 25 mm. Likewise, if the 100 mm lens would be set at f/22 for maximum DOF, the 50 mm must be set at f/11 to give the same DOF and the same level of diffraction after cropping the image by a factor of 2.
The above scenario is equivalent to using a Four Thirds System (which has a crop factor of two) for capturing the same image as an FX system.
We often hear the myth that fast lenses, when used wide-open, have a paper-thin DOF. They don’t: it is only very large entrance pupil diameters that produce a paper-thin DOF. E.g. a 500 mm f/4 has an entrance pupil diameter of 125 mm; the same as 50 mm f/0.4!!
That’s interesting you say that about the thin depth of field coming from a wide opening regardless of aperture. Aperture is, of course, a relative measurement, but you’re saying it’s a property of the absolute size of the aperture. I have a 120 – 300mm f2.8 lens that produces beautifully thin focal planes compared to my 80 – 200mm f2.8. I guess this must be why.
If you want a wide angle lens that renders straight lines take a look at the Nikkor 28mm f/2.8 AIS manual focus. Still available new from B&H. http://www.bhphotovideo.com/c/search?N=10036925&InitialSearch=yes&sts=pi
I am not convinced! … Actually, I think that the demonstration is plain wrong and the “proof” (24 vs 50mm) is only showing a part of the image were the effect is minimal. The main point to understand is that the distorsion is not linear: it increases from the focal point to the border of the frame. Cropping around the focal point is a way to show only the effect at its minimal impact but it stills there. Imagine that you inflate a little bit a ballon. Put randomly stickers in all its surface. Now you inflate more. The surface increases. In other words the curvature decreases: the angle formed by the surface and its tangente (a virtual line with only one point of contact with the surface) is more flat=wider angle… just like a wider lens. Now look at the relative distance between stickers. If the balloon were a sphere then the distances between them would increase at the same rate wherever are the stickers. A balloon is not a perfect sphere, it has a focus point (a zone with more material that stretches less!). It is a cone. Then you can see that 2 stickers close to the focus point stay closer together than 2 stickers elsewhere when you inflate the balloon. In photography the light comes on a focus point, so we capture a cone of light. The “perspective” is due to this difference in distance relation when the angle between the axis of the cone (that pass by the focus point) and the borders of the cone change. Changing the angle have a huge impact on the border. But when you look only close to the focal point and along the axis – as you did in your “proof” – there is virtually no visible effect. Also keep in mind that everything is in 3D, which you did not discuss! That means that the foreground/subject/background relationship is also impacted. Considering that the sensor surface is perpendicular to the cone axis, the point located on the border seems to not move that far away when zooming (changing angle) but what you see in the center “move” a lot! Think again with the balloon in mind. When you blow, the distance from your mouth and the center of the balloon (its edge or focal point) increases very fast, but the border of the balloon in contact with your mouth do not travel that much, isn’t it? This is why the long focal (small angle) compress the picture. Also, what if the sensor surface is not perpendicular with the optic axis? This is how you achieve tilt/shit effect! Actually, she should use angle of view instead of focal lens. Now, you are right about the fact that optic distorsion and perspective distorsion are different. Both impact the final result but not necessary at the a comparable level. It would have been easier to explain all this with diagrams…. Sorry I am an engineer nerd. Now as photographers, we just have to learn how to take advantage of our lenses particularities too have fun.
The article is about standard rectilinear camera lenses, it is not about non-rectilinear lenses such as fisheye lenses.
See “rectilinear grid” in this article:
http://en.wikipedia.org/wiki/Regular_grid
The focal length of a rectilinear lens, combined with the sensor size, determines its angle of view therefore its image scaling. The central image field of a 25 mm lens is exactly the same image as the field from a 50 mm lens, once any minor barrel/pincushion distortion has be corrected.
The image space field is flat whereas the object space field of the lens is part of a sphere — hence the widest possible angle of view is less than 180 degrees (even with a 1 mm lens).
http://en.wikipedia.org/wiki/Gnomonic_projection
http://www.mir.com.my/rb/photography/companies/nikon/nikkoresources/fisheyes/6mmf28.htm
220º :-)
Ok, you were talking about a rectilinear lens!
Marcos, That fisheye lens is awesome!
Pete, thank you, you are absolutely correct – the article is is indeed about rectilinear lenses only. For anyone who is interested in finding out more about rectilinear vs curvilinear lenses, please see my previous article on distortion: https://photographylife.com/what-is-distortion
Patrice wrote “The main point to understand is that the distorsion is not linear: it increases from the focal point to the border of the frame. Cropping around the focal point is a way to show only the effect at its minimal impact but it stills there.”
I thought the same thing but I was out of breath after noting that the focal length did distort the image in a scalar way. (The title, by the way, only refers to distortion. Only the article text then clarifies “perspective distortion”).
I think the title is designed to shock and make people say “Huh? What are you talking about?” Then a technicality is explained but for all pragmatic purposes people will still, by and large, do what they always do, which is use a short focal length for wide scenes and a longer one for distant objects or cropped scenes because doing it any other way *does* result in either perspective distortion (as you fill the frame on a shorter lens) or loss in resolution (if you go the way of cropping a tiny portion of the centre of your image. By the way, I so don’t understand why people would do this – I’ve done it with bird photography and when you’re cropping 90% of the image to get the 10% in the middle, the bird looks pretty awful. If you explicitly don’t want a distorted image, why would you use a wide angle lens for a distant object and crop? Ok, if it’s a life and death situation, or the only way you can prove the existence of the Tasmanian tiger, then *maybe* – but so much compromise on image quality! To save weight? Yes, heavy gear sucks, but again, what quality image will you end up with?)
Your inflated ballon is genius, and your comment should be the main article
I already new this instinctively when for instance shooting with my Coolpix A which has a 28 mm lens. Never get to close to faces with this lens and crop if needed. But thanks for the clarification, it was an interesting read.
Patrice – What you describe is optical distortion in reference to photography (you balloon example) and not perspective. Perspective is what is being demonstrated in the first 24/50 example. Try it for yourself – stand in one place and photograph what is in front of you with different focal length lenses. Enlarge each picture to make the subject size the same in each picture. What you’ve now got is several identical pictures, at least in terms of their perspective. This is a fundamental fact of lenses, distance to subject and perspective. It is a common misconception that focal length controls perspective; especially when ‘so called’ experts continue to propagate that error.
Nasim,
When I first read your assertion that focal length does not distort faces, I was curious to see how you’d prove that.
Then you meticulously showed the effects of focal length and subject distance, I was like… Sure, if you want to put it this way, but the result is the same – for all practical purposes faces taken with wide angle lenses are more distorted than those taken with tele lenses.
By the time I finished the article and saw your examples I agree with you that it helps for photographers to have a finer and more accurate understanding of the physics behind what is causing which kind of distortion.
Theoretical? Yes. Pedantic? Maybe. But that doesn’t mean useless. If one understands what is causing a particular effect one can have a better control of the art that is photography.
An analogous misconception is that taking photos in low light environments result in noisier photographs. We could easily prove that light has nothing to do with the noise observed and that the real culprit is the high ISO which is required for shooting in low light.
The benefit of understanding what exactly causes noise can help us reproduce the effect of high noise, say for artistic purposes, in an outdoor brightly lit scene (by dialing in high ISO and perhaps using plenty of filters).
Thank you for an illuminating read.
KL,
Being technically correct is in no way being pedantic: it is simply being correct; as opposed to being erroneous and propagating endless myths.
Science is not based on popular opinion, it is based on independently verifiable empirical evidence. Nasim’s technical articles are superb resources of reliable information because, if his original version contains a scientific error or ambiguous wording, he updates his article. The fundamental principle that makes science the most reliable source of information that humans have ever had, is that science is self-correcting: it continually adapts to newly discovered empirical evidence and it is highly resistant to opinions and dogma.
You wrote: “An analogous misconception is that taking photos in low light environments result in noisier photographs. We could easily prove that light has nothing to do with the noise observed and that the real culprit is the high ISO which is required for shooting in low light.”
The increased noise in low light environments is cause by the reduced number of photons per second illuminating the scene, reflecting them off the scene towards the camera, and reaching the film/sensor through the lens. ISO has nothing to do with it (especially as many recent sensors are ISO-less). The image noise is caused by photon shot noise, and of course the quantum inefficiency of the sensor (modern sensors a very efficient). If the photographer selects a shutter speed that allows only 100 photons to be counted by a sensor pixel then that pixel will have a terrible signal-to-noise ratio of only 10 — totally irrespective of the ISO speed setting of the camera! A noisy image results from a combination of poor light and a poor photographer (who doesn’t begin to understand the basics of quantum mechanics).
Hopefully, I’ve adequately debunked your myth.
Pete
Ok – next time I say “focal length affects perspective” I will word it “For all practical purposes, focal length affects perspective” :D
Why not say: It’s where we position our eyes/camera in the scene that determines the perspective.
The focal length just changes the angle of view / magnification ratio / image scaling.
Is there any link to any webpage you’d recommend to understand the basics of quantum mechanics of which you talk about, in respect to noise of a digital image sensor?
Expecting that a photographer understands quantum mechanics to improve his/her skills is a bit too much, even for a pro, I think. Even though I’m not a pro I am eager to learn about… well nearly about anything you throw at me, if it catches my interest, and this surely does. ;-D
Marcos, I think the Wikipedia article titled “Shot noise” is an excellent place to start because it includes dozens of links to related topics and other sources of information.
Many many thanks Pete!
Well put.
I think when most people talk about different focal lengths changing perspective they are assuming different focal lengths but the same subject magnification in which case that would be correct.
Hi Nassim,
A wide angle lens pulls in details from a wider field of view than a narrower angle lens. In the subsequent flat 2D image these details are placed side by side. If your lens is 10 cm away from somebody’s nose and 20 cm away from his or her ears the details will be represented as seen by the lens; a large nose and more distant and therefore smaller appearing ears. To say that ” … we will prove that focal length has nothing to do with distorting a subject’s face … ” is an inaccurate oversimplification.
I agree that we are not talking about distortion, we are talking about perspective. However, it is also true that the typical close-up, wide angle perspective produced by a wide angle lens is a product of the focal length and can not be produced by a lens with a longer focal length.
Bengt, Nasim’s statements are correct, and he has backed them with verifiable empirical evidence.
Most photographers learn photography from people who do not understand the fundamental principles of art, projection, and perspective — which is why they are confused and they believe in the common myths that are endlessly regurgitated across the World Wide Web.
A 50 mm lens on a 135 format (FX) camera gives roughly the same angle of view (AOV) as does human vision when we are focussed on a central subject within the scene (rather than moving our eyes and head around to explore the scene). A 50 mm lens gives a diagonal AOV of 47 degrees (when focussed at infinity). Now, if we print or otherwise display the image such that it subtends a diagonal AOV of 47 degrees at the eyes of the viewer, then the perspective will be portrayed to the viewer exactly as it was captured by the photographer’s camera.
A 14 mm lens has a diagonal AOV of 114 degrees. If we display this image to the viewer such that it subtends a diagonal AOV of 114 degrees at their eyes, then the perspective will again be portrayed to the viewer exactly as it was captured by the photographer’s camera. When viewed from the same distance as the image from the 50 mm lens, this requires an image that is linearly larger by a factor 50/14 = 3.57 — nearly 13 times the area. If we instead display this image to the viewer with a much smaller AOV than 114 degrees then the viewer will observe increasing stretching of the scene towards the corners of the image. This type of perspective distortion (caused my a gross mismatch between captured AOV and the displayed AOV) is very different from the facial features perspective distortion that result simply from the photographer being far too close to the subject.
People who have fully functioning 3-D visual perspective compensation do not observe the facial feature distortions that result from being very close to someone’s face (even our own when we look in a mirror). The camera records only a 2-D image from 4-D object space [time being the fourth dimension], therefore our 3-D close-range perspective compensation is unable to provide automatic correction. This is very easy to demonstrate: look at your face when it is very close to mirror then alternately cover and uncover one of your eyes with your hand. The difference in perspective is remarkable!
CONCLUSIONS
1. The focal length of a rectilinear lens cannot, and does not, change the perspective of a captured scene. The captured perspective is determined entirely by the photographer’s placement of the lens entrance pupil in the 3-D object space of the scene. In order to change the captured perspective you have to move your feet, not move the focal length control on your zoom :-)
2. When we are restricted in the placement of our feet, we have to suffer the unwanted perspective distortion, e.g., when photographing a group of people in a small room. In these situations, it would be very nice to have a lens that has its entrance pupil located many metres behind its focal plane (well behind the photographer’s head). Such lenses do exist — object space near-telecentric lenses — but they are hideously expensive and bulky; affordable only by the movie industry.
No ! Focal length is a big part of perspective control. Statements to the contrary may be fun and provocative but they are not correct.
What can be asserted without evidence can be dismissed without evidence — Epistemic Logic 101.
Nasim has provided empirical evidence to support his scientifically correct assertions. You have provided only hot air.
I’ve read that if we view a print at a distance equal to the focal length of the taking lens multiplied by the print magnification, it will look “normal”.
I really like your reply Pete, just one thing… you say “135 FX format..” … I guess you must have meant 35mm FX Format, right?
Kind regards,
Marcos
Hi Marcos, Thanks very much for your comment.
The term “135” was introduced by Kodak in 1934. It is a film gauge of 35 mm, having a full frame of 36×24 mm; half-frame cameras (18×24 mm) were popular during the 1960s.
Digital sensors of approximately 36×24 mm are referred to as “full frame”; “full-frame digital”; and “FX format”.
I tend to write “135 format” or “135 (FX) format” simply because there are 35 mm gauge films that are not 135 format, e.g., 35 mm gauge movie films. My old habits are hard to break!
Kind regards,
Pete
Thanks!! You deserve your own blog! ;-)
> To say that ” … we will prove that focal length has nothing to do with distorting a subject’s face … ” is an inaccurate oversimplification.
I think the point is that it’s the point of view not the lens that distorts the image. The statement “wide angle lenses distort faces” isn’t wrong, it’s also an over-simplification. Try to fill up a frame using a wide angle lens and things tend to look weird. People look weird from up close (according to Ken Rockwell — pause while you laugh — our brains record faces as they look from a distance of about ten feet, so that’s what looks normal to us).
Another way of looking at it: if you took a picture of a person from up close using a pinhole camera — whether projecting onto a flat or curved sensor — it would also look distorted.
“if you took a picture of a person from up close using a pinhole camera — whether projecting onto a flat or curved sensor — it would also look distorted.”
I don’t believe so, because optical and perspective distortions are a product of the lens. It would be interesting to try out on, say, a mannequin head (which would remain perfectly still throughout the exposure) and a pinhole setup on a DSLR.
Spy Black, When we photograph a millimetre ruler that is angled away from the camera, the closest lines are longer and spaced further apart than the distant lines. This difference increases as the ruler is placed closer to the camera — irrespective of the focal length of the lens being used.
It is the trigonometric relationship between the 3-D z-axis (depth/distance axis) object space and the 2-D image space that causes the increasing facial distortions as the photographer gets closer to their subject.
If you attempt to draw a picture of a ruler angled away from you then, before you start the drawing, you have to establish your desired vanishing points — these vanishing points entirely determine the perspective of your drawing. Likewise, the location of your lens entrance pupil entirely determines the perspective of your photographs — the focal length of the lens changes only the overall magnification of the scene. Of course, the whole scene is clipped [only partially rendered] by the size of your image sensor or the size of the paper on your drawing board.
I hope that makes sense. Experimenting with photographing a mannequin head is by far the best way to properly learn and understand the points made in Nasim’s article.
And i always thought barrel distortion was a property of the lens barrel. Thanks for the clarification and this great article Nasim
Brilliant article Nasim. (I won’t get into details about “well, if you said this or that, the other way around” … we understand how it’s meant. Being picky is for lawyers.)
Marcos, thank you for your feedback and I am glad you enjoyed the article. I tried to keep things simple to understand, so if I oversimplified some concepts or perhaps stated something wrong, please accept my apologies. If there is something that stands out, please let me know and I will work on updating the article.
Hi Nasim, thanks for getting back to me! I think Pete A and Bengt Nyman both have some reasonable PoVs (pun intended) which could get some coverage in a sequel to this article.
It sounds reasonable to me that a wide angle lens (without being ‘fishy’) has to introduce more “light information” as one gets closer to the edges of the lens/sensor/film, therefore making things thinner/smaller as we get to those edges.
On the other hand I think Pete A actually brought up a different topic which is the “distortion” of printing or simply viewing things at a different AOV than the one at which the picture was taken. I guess also that this is one of the reasons the bigger cinema screens are slightly bent — just guessing, as i said.
Best regards,
Marcos
Marcos, There are mathematical [trigonometric] reasons why planetariums have to use a domed building rather than a flat ceiling for displaying their images. A flat ceiling would have to be infinitely large in order to provide a complete hemispherical view (180 degrees angle of view). Using a domed building can provide the audience with an angle of view that is equal to or greater than 180 degrees.
The reason for using curved screens in cinemas and on large screen televisions is completely different to the planetarium reason. Curved screens are used to reduce light fall-off towards the corners that occurs due to: the highly reflective, therefore directional, coatings applied to cinema screens; and the inherently directional nature of LCD screens. Screens that are completely non-directional require a much higher level of illumination than directional screens because they are distributing the light power over a 180 degree field: most of which is wasted because the viewers are occupying only a much smaller field than 180 degrees.
I hope that makes sense,
Pete
Makes perfect sense! Top explanation, thanks!
Pete A, thank you for injecting rigour into this discussion.
Sir Reginald, Thank you for your kind comment. I usually spend many hours writing [as in: editing, rewriting, and creating many typos during the process!] each of my long technical comments.
The reason that I do this because I think Nasim has created a website that might qualify as being one of the best [as in both technically accurate and easy to understand] encyclopaedias of photography that is available on the World Wide Web — it’s a gold mine of interesting information and education.
When I read comments on articles claiming that the author is wrong, or is just promoting a product, I think it is important to challenge such claims, rather than ignoring them. Leaving such claims unchallenged will likely cause non-expert readers of this website to doubt its authenticity.
Best wishes,
Pete
Another way to demonstrate that distortion is a factor of distance to the object instead of focal length is to look at very small sensor cameras. These cameras use shorter focal distances and are essentially like extreme “crop factor” situations. My phone has a 4mm focal length, much shorter than anything available on a full frame system. But, it only uses a tiny sensor behind the lens, effectively an extreme crop of only 17.3 sq mm (compared to 864 sq mm on full frame). Essentially the phone crops the center 2% from the area of a full frame sensor. The only real difference in the lens system is the size of the image circle (it’s extremely difficult to make a lens system capable of producing a full frame sized image circle at a distance of only 4mm from the aperture). Any DSLR photographer would say that you’d never take a portrait photo with a 4mm lens, but the majority of photos of people taken each day are with lenses with very short focal lengths and extreme “crops”.
You’ve made excellent points. Likewise, using a 50 mm lens on an 8×10 inch format camera gives a very wide angle of view, but the perspective of the image is exactly the same as using a 50 mm lens on a 135 (FX) format camera located in the same position.
These are the abject proofs [evidence beyond a shadow of a doubt] that the focal length of a rectilinear lens does not, and cannot, change the perspective of the captured scene.
Also, Patrice is correct that there is a form of distortion on rectilinear lenses as you move from the center of frame outwards. It’s the result of projecting what is essentially curved visual information onto a flat surface and works the same as map projections (https://en.wikipedia.org/wiki/Map_projection). Photography is similar to a Mercator map projection, which maps the curved earth onto a flat surface (I’m ignoring some important differences here on purpose). You end up with a map that distorts the size of land masses near the north and south poles, creating a map we are all familiar with that makes Greenland appear almost as large as South America (even though SA is about eight times larger). The notable difference between this and photography is that map projection is a cylinder, so this area distortion is only at the top and bottom. Photography is a sphere, so the area size distortion is all around.
This distortion is dependent on the field of view of your lens. Think of the world you see as an image painted on the inside of a bowl. If you have a narrow field of vie and “cut out” a small piece of the bowl to flatten, it doesn’t distort much. If you attempt to flatten the entire bowl, you will need to either make cuts to get it to lay flat (looking a bit like a flower or star) or stretch the bowl (distorting the surface area of objects on the outside edges more than the center).
This is separate from the distortion due to perspective that occurs from objects being closer/farther from the lens.
Our eyes project the image on a curved “sensor”, so we do not see such distortion when we look around.
I think this discussion would make a good follow-up article with some images to illustrate.
Please read and understand my comments numbered 8.1 and 13.1. If you would like further clarification, then by all means ask.
Oops, and 16.1.1.1
I don’t think anything I said disagrees with your notes. As noted in the gnomonic projection article, there is no distortion at the tangent point (center of the frame), but it increases greatly as you increase the the angle from that point. This type of projection is essentially like the painted bowl analogy. To flatten a 3D sphere (the light information coming into an aperture) to 2D (the sensor), the image will be stretched to greater degrees the closer you get to the theoretical maximum of 180 degrees. At which point we start dividing by zero. :)
Yes, and this type of distortion is always caused by inept photographers who have failed, dismally, to consider the intended purpose of their image before they pressed the shutter button. There are very good reasons why both rectilinear and non-rectilinear lenses are made available to photographer.
However, this has nothing whosoever to do the facial distortion that always results from being too close to the subject. This distortion is caused by a failure to understand how the 3-D object space z-axis maps to the 2-D object space x-y axes.
You are conflating two entirely different types of perspective distortions: those that result from a gross mismatches between the captured and the projected spatial mappings (which is zero in the centre of the image and increases towards the corners); and those that result irrespective of those mismatches because they result from only from the z-axis location of the lens in 3-D object space; distortions that are manifest in the centre of the image.
Please read my comment 14.1.1 in which I discuss photographing and drawing a millimetre ruler.
I’m not conflating them, just pointing out that Patrice was accurate to say that photography with wide angle lenses does include another form of distortion. I have made no claims that this distortion affects facial features in the center of the frame.
This form of distortion has nothing to do with the skill of a photographer – all rectilinear lenses distort in this way (although it isn’t noticeable for narrower FOVs) and non-rectilinear lenses distort in different ways. They are simply realities for photographers to understand and choose how to address in their work (some avoid it, some embrace it).
The ruler example can end up combining these two forms of distortion (perspective lines and stretched area) if a wide angle lens is used and the ruler extends from the center of the frame to the edge of the frame. This will exaggerate the perspective effect.
While discussing each issue in isolation can be useful in understanding them, a photographer must combine all of this information when framing an image. Significant differences in distance from aperture to objects causes perspective distortion (the topic of this article), using a very wide angle lens causes area distortion around the outer part of the frame (mapping a wide FOV onto a flat surface).
Hey, why are having this discussion if we both agree with accuracy of Nasim’s article?
Yep, we both agree the article is accurate.
Here’s a fun (okay, odd-looking) image I did a few years back that’s my extreme version of this. This was taken using a medium format pinhole camera (25mm focal length, around 12mm eviq on full frame sensor, 6×6 film, something like 116 degree FOV on the diagonal, 100% rectilinear by definition) with the tip of my nose at the aperture plane. My face gets very warped from the combination distortion:
https://bradjudy.smugmug.com/photos/i-vswcjbv/0/XL/i-vswcjbv-XL.jpg
Probably the most disturbing self-portrait I’ve done.
The lighting required to make the exposure time (at around f/200) anything I could manage to sit through also bled in a bit and there’s some motion blur from the fact that I had to open and close the shutter by hand, but you get the idea. Given the size of most 12mm lenses, it’s an image you could probably only take on a pinhole camera.
Thank you very much for your reply and for sharing your self-portrait, which I found to be usefully educational rather than disturbing.
One of the problems I suffer is my frequent failure to understand whether a commentator is adding useful information to the posted article, or is in some way (perhaps covertly) trying to discredit it.
Best wishes and apologies for my terse replies to you,
Pete
NOW I AM PRETTY UPSET WITH BOTH OF YOU!!! (Nah, just joking ;)
Well, maybe I am, but only a bit… now you talk about “perspective distortion”… and “area distortion”… what the hell you mean? Is “area distortion” a synonym for “optical distortion”?
If it is, I could sleep again.
Kind regards,
Marcos
Marcos, Rather than talk about perspective distortion and area distortion, I shall just talk about the distortion of facial features as mentioned in Nasim’s article.
The distance from the tip of the nose to the ears is roughly 15 cm. If we photograph a face full on, or nearly so, we can calculate the distortion of the nose relative to the ears from the two axial magnification ratios. Objects that are near to us look much bigger than objects that are far away. If we double the distance, the object halves in size.
When the camera is very close to the nose it will look huge compared to the ears, in other words, the face will look very distorted. Let s = the camera to nose distance in metres; and MR = the magnification of the nose relative to the ears.
MR = (s + 0.15) / s
I’ll show MR as a percentage increase because this indicates the amount of distortion of the nose.
s MR
0.5 30%
1.0 15%
2.0 7.5%
3.0 5%
5.0 3%
10 1.5%
Even at 2 metres there is 7.5% distortion of the nose, which is easy to see in a photo. Notice that this type of distortion has nothing whatsoever to do with the focal length of the lens or the captured angle of view. It is caused entirely by being too close to the face to capture a natural/undistorted look.
Hope that helps,
Pete
Had to reply to my own post, I guess the CMS here hit the ‘sublevel limit’. ;-)
Thanks Pete, simple calculation, easily explained and makes totally sense. Would be a very nice addition to the original article, since it exemplifies perfectly the strong effect a small change in distance has on the subjects’ distortion.
Like 0-1 m RED-zone, 1-2 m ORANGE, 2 and onwards GREEN, you’re good to go!
Long before Nasim wrote this article I read about perspective distortion on the attached site. To sum it up: “So you may ask when do you use the 50mm? I follow a simple guideline when making portraits. 200mm for face only, 135mm for entire head, 85mm for head and shoulders, 50mm for upper body, and 35mm for full length.”
http://blog.sebstudios.net/for-photographers/portrait-lens-50mm-85mm/
Marcos, if you do the math on that rule of thumb from the article you reference, you get an even easier rule of thumb to remember: stay at least 2m away from a person when photographing them if you want to minimize facial distortion, the more the better.
Brad, I totally agree.
My personal preference (which is not meant as a recommendation to others) is to work much further away than 2 metres from my subjects, when space allows. Not just because it often produces more flattering images, but because many people find it difficult to relax when a black FX-size camera plus fat lens and hood is being pointed at them and intruding into their personal space. I detest being photographed therefore I’m acutely aware of even minor tension that it often causes in other people.
Another issue to always consider is that some of the people who have a prominent nose most certainly do not appreciate having it exaggerated by the photographer!
Nasim, great writing as always! Just wonder what is the rule of thumb when you want to minimize the distortion? For 35mm and 20mm lens, what is the distance you want to keep?
Another awesome article. I feel frustrated to see that so many people has totally incorrect idea of how focal length, FOV and perspective works (especially with small sensored system)
Taking a picture with wide angle, cropping it to match the FOV of longer lens and comparing perspective/DOF is an easy experiment anyone with a zoom lens can do. Few people do that. Too bad.
Nasim, very good article. I did not get a chance to read all the comments, but one problem with wide angle lenses you don’t mention in article is composition issues. to take wide angle lens to shoot a portrait of a typical female and make her happy with the photos (i am thinking my wife who is very critical) , not only does your subject has to be a distance from you, because of wide angle lenses design, the person better be somewhere in the middle of your frame. You put her at the very edge and distortion is unexcusable. So your framing becomes critical, you must be quite a distance from the model and also keep her in the middle. With tele lenses you can have your subject anywhere in the frame including the edges and the face still will look good….. maybe I am oversimplifying things…..
David B, You are indeed oversimplifying things. You “did not get a chance to read all the comments”, which is a pity because I put many hours of effort into explaining the things that you have not yet understood. The design of wide-angle lenses is NOT the cause of perspective distortion — the distortion is caused by camera users who cannot be bothered to spend time learning perspective, geometry, and reading through the explanatory comments on blog articles.
Hi Nasim,
I am confused about getting the right lens for portrait.
I use D3200 and i’ve bought nikor 50mm f1.8g .
And also i want to buy FX lenses so that after i upgrade i don’t have to change lenses..
tell me a longer focal length around 150-300mm and Sharp lens.
(exceptional lenses: 70-200 f2.8 and 24-70mm f2.8)
Or should i get 85mm 1.8 .. can it give the best background blur on my dx??
i am confused and my budget is also limited
This is silly. Of course your lens causes the distortion, try your first example with a 300mm lens. You won’t be able to, because the lens will prevent you from getting that close.
If you take 2 lenses, stand the same distance from your subject, the wider lens will distort, the other lens will not.
That’s a fact.
Now you can claim you can solve this by zooming with your feet, and that’s true, but the fact is, that it is in fact the lens creating the distortion.
I will be going to Glacier NP in a few months and am considering a new lens. I googled this topic for that very reason. I understand the close-up concept, but for my purposes how does this effect landscapes? If I shoot a scene in 14mm then multiple images in 35mm and stitch them together, then do the same for 50mm and 100mm so that I get the same area of coverage in each panorama, will the overall distortion be the same or different at each of the lengths? I tried stitching using 5 and 10 photos shot at 55mm covering ~45 and ~90 degree horizontal angle of view. The 45 dregree was ok. The 90 degree was too much for my taste. It looks like about 25 degrees (2-3 photos) is all I can get without the distortion being too much for my preference. I’m trying to determine if it is worth the outlay to get a wide angle lens if it is going to distort the landscape.
If you’re shooting a glacier who cares? As long as you keep the camera level when it’s supposed to be level the results with a good 14mm should be fine. I max at 16mm (Nikon 16-35) and any distortion is looked after fine in software, but that’s only necessary if there are straight lines around the edges. If you’re happy with a bit of quality drop-off at the edges I’d keep life simple and go with the ultra-wide angle (but sub-fisheye).
Thanks. After posting this, I went to the park and took a series of photos at 18 35 and 50mm on a DX then created a pano of each set. I don’t like how pushed back everything is with a wide angle. I like the 35mm (53mm FX equivalent) best. I’ll go with creating the pano by stitching 3-5 images together.
That’s the important thing, doing what produces results that you like. I’m not sure what you mean by ‘pushed back’ though – if you crop the top and bottom of a wide-angle shot you can produce a panorama effect and make the subject appear closer. In fact you can produce a panorama by doing the same with a telephoto shot!
If I make the photo with a wide angle lens, there is way more foreground than I need and the subject is smaller, making up less pixels. If the subject is mountains in the distance then this is not desirable. I could crop and enlarge, but at that distance it will become pixelated. I’d rather use a lens that produces a larger image to start with to better fill the sensor then make less of a crop. It will be a better image quality. I just have to photo merge multiple images together and that is no big deal.
Quite true. All depends on what you want to do with the final image. If you’re shooting with a 24MP camera and not printing bigger than A2, for example, I think you should have enough quality with a wide-angle shot, but if your pano prints are wider than that then you’d need more quality, as you describe. IMHO.
Perhaps this will help:
You spoke of perspective distortion.
What is “perspective”? Another expression is “point of view”. I think this expression will help people understand what you are saying:
If I stand in this spot and photograph a subject over there, then I am seeing the person from this (literal) point of view. This is my perspective. Whether I use a telephoto lens, or a wide-angle lens, my point of view stays the same – it is just that the width of the angle of view is wider, or narrower. The telephoto enlarges the subject but my point of view is constant = my perspective is constant = there is no perspective distortion.
Now, if I go and walk right up to that subject’s face, and I put my face right before hers – one centimetre away – I cannot view her entire face without turning my gaze in all directions: up, down, left, right, to take in the entirety of the view of her face from that ridiculously short distance of 1cm. But some lenses are so wide that even from 1cm they can capture that entire view. The image produce using a lens like this *from this point of view = from this perspective* will be very distorted. It is not the lens choice which causes this – we demonstrated that with our first example from a distance – it is the change in perspective; it is the movement of my point of view from afar to near.
Finally, some wide lenses are fish-eye lenses – you can identify them because straight lines near the edge of frame are rendered as curves. In this case, the lens is clearly (and intentionally) distorting the image. Most lenses are designed *not* to distort views but despite best intents, will have some distortion, especially at frame’s edge. This is the barrel distortion you mentioned and is unrelated to the distortion caused by changing your perspective / changing your point of view.
Beautifully put.
At what focal length does lens distortion no longer play a part in the image? ie, at what focal length do you no longer get lens distortion?
.
Nasim, thanks again for another great article. I just purchased a Tamron 15-30 for my Nikon D-750 and to me it presents kind of a Catch 22. In order to frame the subject properly at an extremely wide setting, I have to get too close resulting in a lot of distortion. If I stand at a far enough distance to eliminate the distortion, the subject matter is much too small. Either I don’t need that wide of a lens or I just need to learn how to better utilize it. My previous experience has been with a Nikon 24-120 which seems perfect for about 90% of my shots.