Gear
The reciprocal rule is lying to you — in both directions
1/focal-length was written for 35mm film and grain. On a 61MP sensor it's optimistic; with modern stabilisation it's three stops pessimistic. Here's the corrected version.
5 min read
You know the rule: with a 500 mm lens, don’t handhold slower than 1/500s. It’s the first thing anyone learns about long lenses, it’s easy to remember, and on a modern camera it is wrong by about four stops — in whichever direction you least expected.
Where it came from
The reciprocal rule dates to 35mm film, and specifically to what “sharp” meant when you were looking at a 6×4 print. The tolerance built into it is the tolerance of film grain and a small enlargement. It was never a statement about physics; it was a statement about a viewing condition that no longer applies.
Two things have changed since, and they push in opposite directions.
Correction one: your pixels are smaller than film grain
Camera shake produces a blur circle on the sensor. Whether that blur is visible depends entirely on how it compares to the size of one pixel.
A 24 MP full-frame sensor has pixels about 6.0 µm across. A 45 MP sensor has pixels around 4.4 µm; a 61 MP body is nearer 3.8 µm. The same physical hand tremor that was invisible at 24 MP smears across two pixels at 61 MP.
This is not the “more megapixels means more shake” myth — the shake is identical. It’s that you’ve bought the ability to see it. Print both files at the same size and they look the same. View both at 100% and only one of them is sharp.
The correction: below 4 µm of pixel pitch, add half a stop. Wingbeat applies exactly this — a 1.5× bump to the handhold floor once your calculated pixel pitch drops under 4 µm. Note where that line falls: a 45 MP full-frame body sits at 4.4 µm and stays clear of it. A 61 MP body doesn’t.
Correction two: stabilisation is worth about three stops
Modern in-body and in-lens stabilisation genuinely works. Manufacturers will claim five, six, sometimes eight stops; those numbers come from a bench test with a static subject and a cooperative operator.
In the field, holding a long lens on a moving bird, three stops is the honest working figure. That takes your 500 mm lens from a 1/500s floor to about 1/63s — before the bird gets a vote.
The correction: divide the floor by 8 when stabilisation is on.
Correction three: panning is harder than standing still
Stabilisation systems are at their best when you’re trying not to move. Panning asks them to distinguish deliberate rotation from unwanted shake, and they only partly manage it. Meanwhile you’re rotating a heavy lens smoothly, which is its own source of vibration.
The correction: multiply the floor by 1.5 when panning.
And the tripod case, which isn’t what you think
A tripod doesn’t stack with stabilisation — it replaces it. Wingbeat treats a static rig on a tripod as worth four stops (÷16), and a panning rig on a gimbal head as worth three (÷8), because the moment you start rotating the setup you reintroduce most of the shake you’d eliminated.
One counterintuitive detail: a static tripod is exempt from the pixel-density penalty entirely. There’s no hand tremor there for those small pixels to magnify.
And if it’s gusty, a 600 mm lens on a tripod is a sail. Give a stop back.
The corrected rule, for a 500 mm lens
| Setup | Corrected floor | Rounded to a real stop |
|---|---|---|
| Naive reciprocal rule | 1/500s | 1/500s |
| 24 MP, stabilised, static | 1/63s | 1/100s |
| 24 MP, stabilised, panning | 1/94s | 1/100s |
| 61 MP, stabilised, panning | 1/141s | 1/160s |
| 61 MP, unstabilised, panning | 1/1125s | 1/1250s |
| 61 MP, tripod, static | 1/31s | 1/60s |
(The app always rounds up to a stop your dial actually has. The figure is a minimum, so rounding down would be worse than useless.)
Three stops separate the naive rule from the stabilised 24 MP case, and another three and a half separate that from the unstabilised 61 MP one. That’s the difference between ISO 800 and ISO 12800 — between a photograph and a texture study.
The part that matters most
All of the above is a floor, not an answer. It’s the speed below which your hands ruin the shot. The bird has its own, usually much higher, requirement.
Wingbeat calculates both and hands you whichever is faster. When the floor is the binding constraint — when the bird would happily sit still at 1/250s but your lens won’t — it tells you, so you know that a tripod, not a faster shutter, is the thing that would actually help.