There is a moment everyone who works with manual lenses runs into sooner or later: you look at the monitor, the image is sharp, and yet something isn't right. You couldn't say what. One corner that stays mushy, sharpness that shifts when you change aperture, one lens in the set that seems to breathe differently from the others.
Nine times out of ten it isn't the sensor and it isn't the operator. It's collimation.
What collimating a lens means
A lens is a stack of optical groups that all have to share the same axis. One axis, perfectly aligned, from the first element to the last. Collimating means bringing every group back onto that axis and checking that the image plane falls exactly where the system expects it to. If you want the full picture of what it is and when it's needed, we covered it in Cinema lens collimation: what it is and why it matters.
When a group shifts by even a few hundredths of a millimetre, the axis breaks. The lens still works, the focus ring still turns, but the result is no longer symmetrical. And it's that asymmetry which is the symptom to look for.
Decentring almost never comes from one dramatic event. It comes from years of transport, from a knock absorbed by the hood, from a fixing screw that has given a tenth, from a group put back badly in an earlier service. The vintage lenses that reach the workshop often have forty years of small impacts behind them.
The four symptoms you see on the monitor
One corner softer than the other three. This is the clearest signal. On a healthy lens sharpness falls off towards the edges symmetrically: the four corners look alike. If only one lags behind, or if the right side is visibly less defined than the left, the group is off axis. Don't confuse it with the normal edge fall-off of vintage glass wide open: there the fall-off is there, but it's the same all round.
The focus plane is tilted. Shoot a perfectly flat wall head-on and focus on the centre. If one side of the frame comes into focus and the other stays out, the image plane isn't parallel to the sensor. It's the most treacherous symptom, because on set you read it as a framing error rather than a lens problem.
Sharpness moves as you stop down. A decentred lens sometimes seems to improve at f/5.6, and that much is normal: depth of field masks the fault. But if the sharpest point in the frame moves as you close down, that's a real anomaly.
The bokeh is asymmetric. Out-of-focus highlights away from the centre always deform a little, that's the cat's eye everyone knows. The problem is when they deform differently on the left and on the right.
A proper check in ten minutes
You need a flat focus chart big enough to fill the frame — a chart on a stand, or failing that a newspaper page taped flat to the wall. Then:
- Camera on a tripod, sensor parallel to the chart. This is the part everyone gets wrong: if the camera is two degrees off, the test tells you nothing. Use a level, or measure the two diagonals from the sensor to the edges of the chart with a tape.
- Wide open. That's where the faults show.
- Focus on the centre, as precisely as you can. Peaking at maximum and punch-in.
- Without touching anything, magnify the four corners one by one and compare them.
- Repeat at close distance and at infinity. A decentring can show up at only one of the two ends.
If three corners look alike and one is clearly different, you have your answer.
What can be corrected and what can't
Mechanical decentring — a slipped group, a loose screw, a helicoid with play — can be corrected. It comes apart, it's realigned on the bench with a collimator, it goes back together and it's checked. It's precision work, but it's ordinary work.
A flange focal distance error, meaning the right image plane but in the wrong position relative to the mount, is corrected with shimming: calibrated spacers under the mount that bring the distance back to its nominal value. On a set that has to be uniform lens by lens it's all but compulsory after any mount conversion.
What can't be corrected is an optical element that has taken physical damage: a chipped element or a cemented group that has separated has to be replaced, and on a discontinued lens that depends entirely on finding a donor.
A note on sets
One decentred lens is a problem. A set where every lens is decentred differently is a far worse one, because the character changes from focal length to focal length and the edit doesn't hold together. That's why collimation should be checked across the whole set at once, rather than lens by lens when someone complains.
If you have a doubt about a lens or about a set, bring it to us: the bench check is quick and we'll tell you straight away whether the fault can be corrected and what it involves. Write to us.