The correction the thread made
Two replies made the same point independently: the system responds to the camera rotating, not to its distance from anything. That is why there is no subject distance at which it becomes appropriate, and why the useful question is a different one — how probable is unwanted camera movement here, and will the stabiliser actually help with it.
When to turn it off
- On a solid, heavy support, where there is no movement to correct.
- When panning, on lenses whose documentation says so, since a deliberate sweep can be read as shake.
- When the shutter is short enough that shake cannot register anyway. One reply gave their own thresholds as roughly 1/60 at the wide end of a kit zoom and 1/200 at the long end.
- The reasoning for bothering: stabilisation works by shifting the optical path slightly, so switching it off can give a marginally sharper frame — but only when it was not going to help in the first place.
The tripod question, which is not settled
- Early stabilised lenses were known to fight a tripod, and the standing advice was to switch off. One reply thought this no longer applies to modern lenses.
- Another disputed the premise that a tripod counts as a solid surface at all, describing their own as springy with a resonant frequency that a stabiliser hunting for movement could excite.
- The reply everyone can act on: behaviour varies by lens, so read the documentation for the specific one, because some must be off on a tripod and for others it does not matter.
What people actually do
- Leave it on for handheld work generally, and for travel, where it was called too convenient to give up.
- For a commercial shoot with time available, one contributor would use an unstabilised lens and a tripod instead, and rely on the support rather than the correction.
- Long lenses do not change the rule. A bright subject such as the moon can be handheld on a long lens with a fast enough shutter and stabilisation on, though a tripod is better whenever time allows.