Near perfect lenses are easy to find these days. The problem is… I don’t want them. I want lenses with character.
A while back I wrote this article on my lens likes and dislikes. Consider this an update. There may be some repetition, but hopefully I’ll have some new tricks and tips to add.
FLARE
Not long ago I found myself shooting some shots on a slider looking out a window at a brightly lit desert. My lenses of choice for this project, TLS-rehoused Cooke Speed Panchros, didn’t like this, and rebelled by milking up to the point where I probably lost 50% of the contrast in the shot. This shouldn’t be surprising, given how old these lenses are, but I hate it when technicalities get in the way of an artistic plan.
Hard mattes have largely gone out of style. Part of this is due to how they change the shape of the bokeh in the background, where point sources that would have been soft, round circles without hard mattes become hard-edged squares with them. (Out-of-focus highlights take on the shape of the hard matte opening, as the matte becomes an additional aperture in front of the lens.) Hard mattes can also reduce exposure, particularly if the opening in the matte is significantly smaller than the front element of the lens. I learned this as a young first camera assistant working on a night shoot. The “B” camera assistant had a 200mm Nikkor prime on his camera, so he put an 85mm hard matte in front of it to cut flare. The DP had him take it off, pointing out that the front of the lens was around 5″ in diameter while the opening in the 85mm hard matte was only a couple of inches. Even though it appeared that the lens’s view was unobstructed, adding that hard matte eliminated about a stop of exposure. (In the days of film, this resulted in an unpleasant surprise in dailies.)
I’d never tried putting a hard matte in front of Cooke Speed Panchros, and as video gives us instant dailies I gave it a try. The image cleaned up immediately: blacks were instantly restored to true black as all of my contrast came back. The flare within the frame wasn’t the problem; it was the flare from outside the frame, that struck the lens at an obtuse angle, that caused all the problems.
Suddenly I remembered that I’d run into the same issues with modern Cooke S4s. On a recent project I’d lit someone from below while bouncing light off a desk, and as the camera passed over the desk the lens flared dramatically. There was nothing bright in the frame at the time, and when the top of the desk came into frame it didn’t cause any flare at all. I’d run into flare of this type on longer Cooke S4 lenses while shooting out of windows into daylight (mostly on the longer lenses, like 75mm and 100mm) but this was the first time I’d seen this on a wider lens (35mm).
It seems that Cooke lenses, both old and new, are susceptible to soft veiling flare from outside the frame, even though the same source within the frame won’t cause any noticeable flare at all.
I’ve seen this happen to a lesser extent with Zeiss Ultra Primes. I shot an exterior spot against overcast skies and we had to dodge the shadows a bit due to veiling flare originating from outside the frame.
My new rule of thumb: if I’m shooting out a large window, or an open overcast sky, and light can get into the lens at a glancing angle, throw a hard matte on. Contrast snaps right back.
TLS Cooke Speed Panchros create beautiful flares if you really hammer them with light (such as shooting into the sun), but the lens housings are so deep that it’s hard to create a hard flare using a light that’s outside of the shot. It’s impossible to get the light down the lens without making it part of the picture. I’ve learned to use lenses with more elements and less depth between the front element and the end of the lens for this effect. (The old Cooke 5-1 zooms that I used in my earliest days in the industry work spectacularly for creating long, complex lens flares.)
Recently I shot a project using Zeiss Master Primes as we wanted to make the background as soft as possible while giving post enough room to reframe interviews as needed. I never closed the lens down beyond T1.3, and no matter how many bright sources I put into the background the lenses never showed any sign of veiling flare. It did, however, show some unpleasantly sharp blue flare if the background sources were too bright. Where most lenses would show flare as a “bloom,” Master Primes showed it as a point. They did a spectacular job of eliminating flare in general, but when it did show up it wasn’t pleasant.
My assistant thought the issue was in the lens itself. I think we may have been seeing light reflecting off the deep blue anti-reflective coating on the Arri Mini sensor back onto the rear element of the lens. Either way, I had to tone those highlights down considerably.
LENS COVERAGE
I’ve been shooting a lot of work on the Alexa Mini lately, and one of them initially called for using a zoom lens. None of my projects have required Arri Raw lately, so I haven’t captured anything in 3.4K Open Gate. That’s good, because there aren’t a lot of zooms out there at the moment that will cover sensor size. One can run into issues with zooms even at 3.2K. I like to use primes more than zooms, so large sensor zoom coverage wasn’t really on my radar. Fortunately, one of my local rental houses tested all of their zooms and captured reference images, so I was quickly able to see which zooms would work at 3.2K and which wouldn’t. Many would work at 3.2K, but would fail at 3.4K.
Surprisingly, all of the old Cooke Speed Panchros work at 3.2K except for the 16mm, which cut off the corners of 16:9. Fortunately I learned about this issue on a 2.4:1 project, and that aspect ratio was completely safe.
PERFECTION
I’m not a fan of optical perfection. Shooting with Master Primes was fun because T1.4 yields super smeary dreamy backgrounds that are hard to get any other way (the differences between T2.8, T2 and T1.4 are startling: T2.8 backgrounds can still be relatively sharp, while T2 blurs them out nicely and T1.4 obliterates them). At the same time, the lenses were too sharp. I added a 1/8 Hollywood Black Magic filter to take the edge off as I didn’t feel comfortable photographing faces any other way.
Zeiss Ultra Primes are my “workhorse” lenses. When I need something inexpensive, reasonably sharp, and lightweight, these lenses deliver—and they still have some character to them. They’re the modern versions of Zeiss Super Speeds, which are very funky lenses indeed, and they retain some of their qualities. Backgrounds soften nicely and faces don’t photograph harshly. They open up to T2, which is fine for most everything I shoot. (I used to light to T4 out of kindness to my camera assistants, but now that they all pull focus off 17″ monitors I’ve embraced my inner low light evilness.)
Speaking of Zeiss Super Speeds, their bokeh is truly unique among motion picture lenses. Modern lenses are designed so that the bokeh is evenly exposed all the way across, such that an out-of-focus highlight becomes a flatly illuminated disk. Super Speeds create bokeh with hot centers that bleed off, which makes for very interesting backgrounds. Unfortunately they tend to show color fringing in areas of high contrast, and the entire image can turn green or magenta depending on where focus is placed in the scene. Lenses are really only optimized for the point of focus, and objects in front or back of that point tend to shift in hue (green or magenta, depending on whether they are in front or in back of the point of focus). This is true of all lenses to some extent, but it’s very true of Super Speeds. (Canon CE primes will show high contrast color fringing as well, when they are severely stressed.)
Cooke Speed Panchros do the opposite, where the bokeh looks like donuts with dark centers and bright edges. This should be more distracting than it is, because rings don’t blend well together, but it gives the lenses a “vintage” feel that I happily excuse. I hate this look in cheap still lenses, but in old motion picture lenses it often adds character.
I haven’t used modern Leica primes in the field yet, but my tests have shown that they are much like Master Primes: sharp, almost to a fault, with very little “character” to them. For modern lenses with “character” I go for Cooke S4s, which are very kind to faces and have some visual oddities that aren’t as obvious as the Panchros but still give the image a nice abstract feel. Focus shifts result in an almost 3D effect as if the foreground is “lifting off” the background, although I don’t see the same wonderful interference pattern effect that I see on the Panchros, where out of focus background textures appear to “ripple” around sharper foreground objects. (For more on that, see my article on the new Panchros and watch the exterior video clips.)
Strangely, I don’t like zooms that much, even though they are nothing but heavy, time saving compromises. Both the Angenieux 24-290 and the Canon 30-300 show pincushion distortion in the middle of the zoom range (it starts at around 50mm on the Angenieux). This isn’t noticeable in most shots, but if you’re shooting architecture or trying to frame between symmetrical lines then you’ll be in for a surprise. I don’t mind barrel distortion, where vertical lines bulge away from the center of the lens, but the opposite effect disturbs me.
The new Angenieux 25-250 looks promising, although it only opens up to T3.5 or T3.8. That’s not very useful for shallow focus or interior shooting except at high ISOs. Still, it does cover Arri Mini’s 3.4K Open Gate.
VIGNETTING
Pick up a lens. Any lens. Open the aperture all the way wide. Look through the back, and then move your head slightly off to one side until the image seen through the lens takes on a cat’s eye shape. Now stop the lens down a stop or two. The aperture is smaller, so the image seen through the lens is smaller, but you probably won’t see the cat’s eye shape anymore as the aperture is small enough that the end of the lens doesn’t cut into the image circle from the same perspective.
Now you know why lenses vignette when they are wide open.
All lenses do this, including Master Primes—which I witnessed on the shoot referenced above, when I used them wide open all day. The bokeh at the edges of the frame become ovals instead of circles, because the sensor “saw” the light at the edges of the lens as coming through an oval opening. Bokeh takes on the shape of the smallest lens aperture, so highlights at the edge of the frame become cat’s eyes.
It’s interesting that all lenses, even the most expensive ones, show some of this effect wide open, although it usually goes away when stopped down about one stop. (The old rule of thumb for capturing the sharpest image with reasonable vignetting was to close the lens down 2 2/3 stops from wide open, but these days closing down one stop seems to improve things dramatically.)
While rereading Canon’s TV Optics III book for clues as to why this would happen, I came across an interesting note about something that I’d seen personally but never quite understood. In a zoom lens, the entrance pupil recedes back into the lens as it zooms in. The entrance pupil is the image of the aperture as seen from the front of the lens. As a film camera assistant, the best way to check for hairs in the gate was to zoom the lens all the way in and look into the front while holding a flash light: the aperture image moved to the back of the image and grew larger, so it was easy to see around the edges of the film gate to check for hairs. It turns out that this is also the reason why zoom lenses vignette: when the entrance pupil is so far back in the lens, the front of the lens causes that cat’s eye shape to appear very easily at wide open apertures. If you look through the back, you don’t have to move your eye too far off axis to make make that shape appear. Closing the aperture down reduces that vignetting, as the film/sensor only sees down the center of the lens, where the front of the lens doesn’t encroach. This is easy to see just by looking through the back of the lens.
Now you know why zoom lenses get darker when they are zoomed all the way in.