Image Stabilization Tested: Sharp Handheld Shots at 2 Seconds

The old rule says never hand-hold slower than one over the focal length. A new test kept a full-frame camera sharp at two full seconds handheld, a shutter speed that would have needed a tripod on any camera a decade ago. The result says less about steady hands than about how far stabilization has moved. It also comes with a limit that stabilization marketing rarely mentions.

What They Tested

The Phoblographer’s Feroz Khan used a Nikon Zf, which Nikon rates at 8 stops of 5-axis in-body stabilization, with a Nikon Z 24-120mm at f/4 and about 85mm for every frame. Auto ISO held exposure roughly constant, and no sharpening was applied. The rig weighed about 1.44kg with a grip attached, and Khan shot without bracing the camera strap, with a loaded crossbody bag on one shoulder. He shot each speed twice, once with VR off and once on: 1/60s, 1/25s, 1/4s, 1 second, and 2 seconds.

What It Found

At 1/60s, the slowest speed Khan recommends for beginners without stabilization, the VR-off frame was fairly sharp but showed slight softness on horizontal building lines. VR cleaned it up. At 1/25s the unstabilized frame looked acceptable until a 100% crop. With VR on, distant facade detail on the Burj Khalifa resolved clearly.

At 1/4s, the VR-off frame was visibly blurred without zooming, and the stabilized one looked tripod-mounted. At 1 second, the unstabilized frame was unusable, while the stabilized frame held up at full magnification. At 2 seconds, the stabilized frame came back sharper than the unstabilized 1/60s shot. From 1/60s to 2 seconds is about seven stops, close to Nikon’s 8-stop rating. Nikon claimed that rating as a world first when the Zf launched in 2023.

The result is consistent with Photography Life’s testing of the Nikon Z6 III, which shares the Zf’s stabilization. It got sufficiently sharp handheld shots at 2 seconds at 120mm, but only down to about 1/10s on live birds at 400mm.

The Limit Nobody Puts in the Spec Sheet

Stabilization corrects camera movement, not subject movement. Khan’s FAQ is explicit that VR does not freeze subjects. A 2-second exposure of a city skyline works. A 2-second exposure of a person, a bird, or traffic produces motion blur no stabilizer can remove. The 8-stop figure is a static-subject number. Khan also notes sensor size changes the math, and his test applies to full frame. He recommends turning VR off on a tripod, since the mechanism can introduce its own movement in long exposures.

For readers weighing a camera on stabilization performance, the Zf used in the test:

Source: The Phoblographer

Content created with AI assistance and reviewed for accuracy.