Robert Capa is quoted as saying “if your pictures aren’t good enough then you’re not close enough”, but if you’re using a wide-angle lens for close-up portraits this may not be the best advice.
Most photography forums, at some time or another, end up having a long thread on how perspective, depth-of-field, and/or background compression work. Usually you can find a mix of both well informed and not so great suggestions. I’m sticking to a simple example and relatively simple math.
My model for homing in on the sweet spot for portraits with a wide lens was Spotty the Seal. He was the only one willing to stand still long enough to get it done.
I cropped and resized the following pictures to standardize the framing and to fix the absolute size of Spotty’s rather generous nose. His schnoz will be our reference point.
All theses pictures were snapped with the same 23mm lens at f/2 in front of an APS-C sensor. I tried adjusted exposure, WB and levels so that minor changes in color balance or tonal range wouldn’t be too noticeable. Since there are slight angle changes due to hand-holding for each shot, there may be small variations in each that are not due to the distance from Spotty’s nose, but this should have a relatively minor impact compared to moving up to 3 times farther away.
The gallery below combines shots from some key distances.
From left to right, these shots were from 2, 4 and 6 foot focus distances. The photos in the bottom row had one additional processing step added – the perspective was adjusted by using the adaptive wide-angle lens filter in CS6.
Two feet is in-fighting distance. It is where you could, if you were a street fighter and not a street photographer, get your thumb into someone’s eye by extending your arm all the way. It is not my favorite position for a portrait due to practical problems with reproducing the image for display in a way that the viewer does not perceive a warped perspective.
Let’s take a closer look at the extreme ends of the focus distance series. Here are the 2 foot and 6 foot images displayed at the same size on screen.
Notice how Spotty looks a little chubbier in the second shot. Also, notice that his nose seems to take up less space on his face and the background behind him looks bigger, as if the building across the parking lot is getting closer to him.
The reason the background looks bigger with respect to the foreground is because I moved back a couple steps and then cropped the second image more during editing. His face looks bigger compared to the nose because it is also “background” that’s being compressed in with the foreground, it’s just not as far back as the building is.
In theory, you could also make the equivalent crop by switching to a smaller sensor with the same lens.
For an explanation of the “telephoto compression” effect you get when selecting longer lenses, and the math behind it, try this page on Wikipedia. It adds some additional factors to the equation compared to cropping using the same lens.
Here is the full series of photos from 2 feet to 6 feet for reference.
While the difference between the uncorrected 2 foot and 6 foot shots is pretty easy to see, moving out to 3 or 4 feet produced smaller relative changes.
Here are the same photos after applying the adaptive wide-angle lens filter.
I don’t know the exact algorithm used for this filter, but it looks like the original image is mapped to what appears to be a sphere that is sized according to the focal length of the lens and the focus distance. The wider the lens and the closer the distance, the more extreme the spherical mapping.
Bottom-line is that it is intended to make an image taken from a close perspective look more natural when reproduced for display. It is not the same as lens correction, which only corrects for distortions due to design shortcomings of any particular lens.
Here is a direct comparison of the 2 foot distance perspective with and without the adaptive wide-angle lens filter in CS6.
The change in relative size of the background vs. foreground is exagerated in a way that is similar to moving away from the subject and cropping. People, as well as small aquatic mammals, tend to look more pleasingly proportioned in smaller prints from a more distant perspective image. This makes the adaptive filter a decent tool in the digital tool box, but it has at least one major limitation.
Here is what the full uncropped frame looked like before and after the filter was applied.
Notice how the window frames and rooftop change in the background in the second image. This appears to be due to the software equivalent of printing the photo on a ball – the original straight lines buldge outward. There was also a crop involved. The default settings I used indicated that a 23mm lens adaptation with 1.54x crop was used.
So, if you are taking people pictures with clean or organic backgrounds, no problem. You can use either the “step back and crop” approach, or the “stay close and fix it in Photoshop” method. However, straight lines may become an issue in the second case.
Using a 35mm full-frame equivalent lens at 2 feet, filling the frame for a head and shoulders (landscape orientation) or half-torso (portrait orientation) portrait requires you to cut the resolution down only a little. You lose approximately half the original file size if you step back to 3 feet, and even more at 4 feet to keep the frame filled.
If I change this 23mm f/2 lens to 33mm using Fuji’s TCL-X100 teleconverter everything will step up a notch by starting with a 50mm lens FOV before cropping.
A camera like this Fujifilm X-100T works well for one or two person snapshots with surrounding context at around 3-4 foot distances, and you can usually correct the perspective for closer shots if you need to. For more flexibility though, the teleconverter is a useful option, and after this quick experiment, I already ordered it for the holidays.
I like to get the limits of my equiptment figured out before I need to use it. Sometimes I don’t get a second chance to make a memory that will last.