Why cheap lens adapters lose infinity focus

You buy a twelve-pound adapter so you can put a 1970s lens on your camera. It arrives, it screws on, the lens turns smoothly, and everything within a few metres comes into focus beautifully. Then you point it at a building down the street and nothing happens. The focus ring hits its hard stop at the infinity mark and the building is still soft.

The adapter is not faulty. It was never going to work, and the listing had all the information needed to know that in advance. It just didn't put it anywhere you'd look.

The one measurement that decides everything

Every lens mount has a fixed distance from the flat metal face of the mount to the film or sensor plane. It goes by flange focal distance, register distance, or flange back distance depending on who is writing, and it is a property of the camera body that cannot be changed.

A lens is designed around that number. An M42 lens is built so that when its rear face sits 45.46mm from the film, and the focus ring is at its infinity stop, parallel light lands in focus on the film. Move the lens further forward and the focused image forms behind the sensor. It never arrives.

That gives you the whole rule for adapting lenses, and it is subtraction:

adapter thickness = lens mount register − camera mount register

If the answer is positive, the adapter is a spacer and the job is easy. If the answer is zero, you need a mount face that is basically a shim. If the answer is negative, you cannot do it with an empty tube, no matter how much you pay.

The numbers

Read down that column and the last fifteen years of camera design make sense. Mirrorless bodies removed the mirror box, the register distance collapsed to somewhere between 16 and 20mm, and suddenly every SLR lens ever made sat comfortably above every mirrorless body on the list. A Nikon F lens on a Canon RF body needs 26.5mm of empty tube. Nothing clever, no glass, no compromise — the adapter for it is a spacer with a bayonet on each end. Same story for F to C-mount at 28.97mm and Nikon F glass onto Sony E at 28.5mm.

The cases that cannot work

Now do M42 onto Nikon F. 45.46 minus 46.5 is −1.04mm. The adapter would have to have negative thickness. Even a perfect adapter of zero thickness leaves the lens a millimetre too far from the sensor, and a millimetre is an enormous distance in this context.

This is the single most common adapter disappointment, and it catches people because Nikon F and M42 are both extremely common and the adapters are extremely cheap. Canon FD onto anything with a mirror is the other classic: at 42mm it sits below EF, below K, below F, below everything. FD lenses were essentially stranded for two decades until mirrorless arrived.

Sellers get around it in two ways.

The first is a corrective lens in the adapter. It works, in that infinity focus comes back, but you have quietly bolted a weak teleconverter to the back of your lens. Magnification goes up by roughly 1.1x to 1.2x, you lose a fraction of a stop, and you have inserted a cheap element into a system you presumably chose for its optics. The good ones are acceptable. The eight-pound ones are two pieces of window glass and a spacer.

The second is to say nothing and let you find out. These are the adapters where the reviews are a long argument between people saying it works fine for portraits and people saying it is broken.

Recessing the mount: how the impossible ones get close

There is a third option that only really exists if you make the adapter yourself, and it is the reason the M42 to Nikon F project on this site claims what it claims.

The register distance is measured to the flange face — the flat ring the lens seats against. But the camera's mount throat is a hole, and behind that flat ring there is empty space going back towards the mirror box. Nothing says the adapter has to live entirely in front of the flange plane. If you build the M42 threads so they sit down inside the throat rather than on top of the flange, the lens's rear face ends up below the camera's mount plane and you claw back part of that missing millimetre.

You are limited by what is actually in there. On an SLR the mirror needs its swing, and a lens with a protruding rear element will foul it. The recess also has to clear the aperture coupling and any screw drive. This is why these designs are described as minimal rather than universal: the tolerance for error is a fraction of a millimetre, and a design that clears the mirror on one body may not on another.

The honest way to state it is that a recessed adapter can restore most of the focus range, that many M42 lenses have a little slack past their infinity stop from the factory, and that between the two you often get usable infinity focus. Not always. Test before you trust it on a job.

How much error actually matters

Depth of focus — the tolerance at the sensor, as opposed to depth of field out in the world — is roughly 2 × N × c, where N is the f-number and c is your circle of confusion.

On full frame with the usual 0.03mm circle of confusion, a lens at f/2 gives you about 0.12mm of total tolerance. That is your entire error budget for adapter thickness, and a printed part that comes out 0.15mm thick from over-extrusion has already spent it.

At f/8 the same sum gives 0.48mm, which is why an adapter that seems fine on a stopped-down landscape lens turns out to be visibly wrong on a fast fifty shot wide open. Fast lenses are the test. If you want to know whether an adapter is right, put the fastest lens you own on it, open it up, and photograph something a few hundred metres away.

Diagnosing yours

Point the lens at something genuinely distant — not across a room, not the far end of a garden. Several hundred metres, or the moon.

If you turn the ring to its infinity stop and the image is soft, and turning further is impossible, the adapter is too thick. The lens is too far forward. This is the failure mode you cannot fix by shooting differently, only by removing material.

If the image comes sharp before the ring reaches the infinity stop, the adapter is too thin, or exactly right. A small amount of travel past infinity is normal and deliberate on many lenses: manufacturers leave it so that thermal expansion in the cold never strands the lens short of infinity. If you have a lot of travel past infinity, and close focus has got noticeably worse, the adapter is under-thickness.

For a printed adapter that is too thick, the fix is boring and effective: a sheet of 400 grit on a flat surface, the adapter face down, and small circles with light pressure. Check often. Removing 0.1mm takes less time than you expect, and you cannot put it back.

Getting the fit right is a separate problem from getting the thickness right, and if the threads are binding or rattling rather than sitting at the wrong depth, that is covered in printing threads that actually fit.

The part nobody tells you

Adapting a lens costs you the automation. Depending on the combination, that can mean no autofocus, no aperture control from the body, no EXIF record of what lens or aperture you used, and no image stabilisation that knows the focal length. Some adapters restore some of it with electronics, at ten times the price of a tube.

None of that is a reason not to do it. Manual focus on a mirrorless body with focus peaking and magnification is genuinely easier than manual focus ever was on a film SLR. But it is worth knowing before you order that the pleasant part of adapting old glass is the optics, and the tedious part is that your camera no longer knows what is attached to it.