Some Things I Know About Lenses

I don’t know a ton about lenses. I’ve always focused on cameras and film technology, but I’m slowly building a library of information in my head—partially from what I’ve learned over history, and partially from new experience. Here’s a short list.
Nearly every lens vignettes, where “vignette” means the edges of the frame darken when the lens is used in certain ways. This can generally be fixed by stopping down the aperture. Vignetting is very easy to see on a waveform monitor, and it used to drive me nuts when lighting green screen in video: I could never get the line of exposure flat all the way across the screen! It took me a while before I discovered that stopping down a bit fixed my lighting considerably.
Speaking of which, I was once taught that lenses generally look best stopped down 2 1/3. I’ve noticed that lenses generally don’t look very good wider open than T2, and closing down a stop or two works wonders for increasing contrast and resolution, as well as reducing vignetting.
There is only one point of focus. Depth of field is an illusion. It’s a really useful illusion, but it’s all about finding the limits at which a person or a camera system renders a soft image sharp enough to be mistaken for being sharp.
I hate playing focus splits. The point of focus should always be on something.
When possible I give my camera assistants a stop of T4. That’s a good compromise between a soft background and having a reasonable chance at keeping a moving object or person in focus. When I worked as “B” camera assistant on a TV series toward the end of my camera assisting career I found myself working daily with a Cooke 10-1 zoom with a maximum stop of T4. I could do anything at that stop.
I try not to shoot wider open than T2.8. At T2.8 a camera assistant has a fighting chance; at T2 life becomes very difficult. T1.4 is nearly impossible in a number of situations.
I never shoot above f/11 as that’s where diffraction sets in on 35mm-sized sensors. f/16 starts looking a bit soft, f/22 looks really soft.
The highest contrast occurs when shooting down the middle of the lens, typically at f/4-5.6 or so. The lowest contrast happens near wide open, when the full surfaces of the glass elements are used and a light of light bounces around internally. There was a school of DPs in the 1980s who shot wide open nearly all of the time because the light appeared to “fill itself in.” Also, at very low light levels, what you see is roughly what you’ll capture on film: you can light entirely by eye, take one meter reading and shoot. Above about 100fc you can’t do that anymore. I heard a story about a DP who had to light a feature to f/11 for depth of field. He shot a test with his gaffer where they attempted to fill a shot by eye and the shadows turned out black. They had to work completely by meter as they couldn’t see what they were doing at those light levels: their eyes fooled them.
T-stops are “transmission stops.” They account for light loss, due to scattering and internal reflections, as it passes from the front element to the rear element. T-stops are generally around 1/3 stop slower than the actual f/stop, but not always.
f/stops are the fraction of the diameter of the aperture over the focal length of the lens. They measure a physical thing. For that reason they are always used when calculating depth of field, as a stop of T2 may actually be f/1.8, so depth of field is slightly reduced.
The front entrance pupil (mistakenly called “the nodal point”) is the point around which a lens rotates perfectly. It’s generally where the image flips on its way to the sensor. If the camera rotates around that point then everything in the frame will track perfectly: objects in the foreground won’t appear to be moving faster than the background. Two C-stands set in a line directly out from the lens will block each other no matter how the camera is panned or tilted. To find the horizontal position slide the camera backwards and forwards until the C-stands track together. To find the vertical position you’ll need to raise and lower the camera using a specialized head as most tripod heads put the camera well above the center of rotation, which is often in the head itself. (When is this trick helpful? If you’re replacing the background of a shot with a matte painting, or you want to block one thing with another during a panning shot and have them track perfectly.
Anamorphic lenses are two lenses in one: a long one in the vertical direction and a wide one, typically half the focal length of the other, in the wide dimension. It’s difficult to get these two lenses to focus in the same place at the same time, so lens designers rely on their overlapping depths of field to make them appear sharp. That’s why some anamorphic lenses never look sharp wider than T4 or T2.8: both lenses are focusing in slightly different planes and there’s not enough depth of field overlap for them to look sharp in the same plane.
I prefer barrel distortion (verticals bulging outward) to pincushion distortion (verticals bulging inward). Most zoom lenses show one or the other, and they often shift back and forth across the range of the zoom. The Angenieux Optimo 12-1 shows barrel distortion at the ends and pincushion distortion in the middle range. Most primes are either rectilinear (show perfect verticals, or close to it) or show a little bit of barrel distortion.
Years ago hard mattes were in vogue, although I haven’t seen them in a while. These are plastic covers that snap onto matte boxes with holes cut in them that are the exact angle of view of the lens. They’re very helpful in keeping flares out of shots, but they also affect the quality of the bokeh (how the out-of-focus portions of the image look). Highlights in the background can take on a square shape when the hard matte blocks some light around the edges of the lens at wider apertures. In fact, we used to think that we could use 85mm hard mattes on 200mm prime lenses because it appeared that the lens could see through it even though the front element was so much bigger than the hard matte opening. It turned out we were effectively adding another aperture to the front of the lens, reducing exposure by as much as a stop. That’s not good for night work.
I was once told by an optical designer that there’s a trade off between contrast and resolution: you can’t have a lot of one without giving up a lot of the other. I don’t know exactly why that is. What I do know is that Zeiss lenses tend to look sharp due to contrast, not resolution, as they emphasize contrast in their lenses and that makes them look sharper than they are. Cooke lenses, on the other hand, emphasize resolution over contrast, so they are sharper while appearing to be much lower in contrast. (I need to research this area more thoroughly.)
For some reason primes can focus down to about 2′ and no further. When researching how close I can focus on a specific lens I know 2′-3′ is a safe assumption. Zooms are a toss up: the short ones can focus that close, but the longer the range the farther away close focus is. The Angenieux Optimo DP 30-80 will focus down to 2′, but the Optimo 24-290 focuses down to 4′.
If you want to make something look really big you need a small lens. I had a shot where a baby had to appear to swallow the camera, and while we could get close enough with a small handheld camera the lens diameter was too wide. We used a GoPro instead: the baby bit it and we had the shot. This works with other objects too, for example if you need something small to wide the frame then the front element of the lens should be smaller than the object.
That’s about all for now. More as I learn it or remember it.