Micro focus adjustment usually does not physically realign the AF sensor or image sensor. In practice, it applies a small offset to the focus position the camera tells the lens to move to.

So if the camera’s AF calculation says “go to position X,” MFA makes it command something like “X + offset” or “X - offset” instead. That correction can be stored per lens because each lens/body combination can have slightly different tolerances.

A key reason this is needed is that many phase-detect AF systems behave more like open-loop systems than perfect closed-loop systems: the camera estimates the required focus movement and sends the lens there, rather than continuously verifying final focus on the imaging sensor. Even if some feedback is involved, real systems still have manufacturing tolerances and small alignment/mechanical variances.

Those variances can come from the camera body, the lens, or both. A body-side error may be similar across lenses, but each lens can also have its own focus travel and positioning differences, which is why the correction may not be the same for every lens.

So, MFA is best understood as a per-lens focus position compensation for AF calibration errors, not a physical sensor realignment.