Group photos fail in a predictable way. Everyone smiles, the shutter goes, and one person in the back row has their eyes shut. The more people you put in the frame, the more certain that becomes, and it is a genuine probability problem rather than bad luck. This blink-free photo calculator works out the chance that a single frame catches nobody mid-blink, and how many frames you would need for a stated level of confidence.
Arb Digital builds free tools, including a fair number that are simply enjoyable. This one is honest about what it is: an estimate built on an average blink rate, with a published model behind it and a large amount of variation between real people that the model cannot see. It sits beside our camera field of view calculator and depth of field calculator, which answer questions about the optics rather than the subjects.
Where the Model Comes From
The calculation implemented here follows the work of Nic Svenson and Piers Barnes at Australia's CSIRO, published as "Blink-Free Photos, Guaranteed" in Velocity in June 2006. It won the 2006 Ig Nobel Prize in mathematics, awarded, as Improbable Research records, for calculating the number of photographs you must take to almost ensure that nobody in a group photo has their eyes closed. Barnes was approached by Svenson, a photographer, with a problem she had met repeatedly in practice, and turned it into a probability.
The reporting at the time, including the ABC Science account of the award, gives the rule of thumb the work is remembered for: for groups of fewer than twenty people in good light, divide the number of people by three and take that many photographs. In poor light, where the shutter is open longer, divide by two instead. Above about fifty people the number of shots needed climbs steeply enough that the researchers suggested giving up on a perfect frame.
The blink figures the model needs come from ordinary vision science. A review by Willett, Maenner and Mayo in Frontiers in Systems Neuroscience on the perceptual consequences and neurophysiology of eye blinks puts a typical rate at roughly fifteen blinks a minute, with a whole blink lasting about 250 to 450 milliseconds and the period of actual visual occlusion nearer 40 to 200 milliseconds. Those are the defaults on this page, and every one of them is adjustable because none of them is a constant.
What This Calculator Does
It reports four things. First, the chance that one named person has their eyes open in one frame. Second, the chance that everybody does, which is the first figure raised to the power of the group size. Third, the chance of getting at least one clean frame in the number of shots you plan to take. Fourth, the shot count that reaches your target confidence, which is the headline.
It also prints the original rule of thumb alongside its own answer, so you can see where the two agree and where they part company. They match well for small groups in decent light, which is the case the rule was written for, and diverge for large groups where the rule was never meant to apply.
How to Use It
- Count the faces. Only people whose eyes are visible and open-able count. Someone turned away or in dark glasses is not in the calculation.
- Set the blink rate honestly. Fifteen a minute suits a relaxed adult indoors. A tired wedding party at eleven at night, or a group squinting into low sun, blinks far more.
- Set the exposure time. This is the one photographic control you actually have. A faster shutter narrows the window in which a blink can spoil the frame.
- Enter the shots you plan to take. The third grid item tells you what that plan is worth.
- Read the headline as a count, not a guarantee. There is no number of frames that makes a clean shot certain, which is why the tool asks for a confidence level rather than promising success.
The Formula, Worked Through
Treat each person's blinking as independent and randomly timed. If someone blinks r times a minute, that is x = r/60 blinks a second. If a blink spoils the frame whenever it overlaps the exposure, the vulnerable window t is the blink duration plus the exposure time. The chance that a given person is mid-blink at the moment of exposure is then xt, and the chance their eyes are open is 1 − xt.
For n independent people, the chance that everybody is open in one frame is (1 − xt)n. Call that p. Across s independent frames, the chance of at least one clean one is 1 − (1 − p)s, and inverting that for a target confidence C gives s = ln(1 − C) ÷ ln(1 − p), rounded up.
Work the defaults. Fifteen blinks a minute is 0.25 a second. A 250 ms blink plus an 8 ms exposure is 258 ms, or 0.258 s. So xt = 0.25 × 0.258 = 0.0645, and each person is open with probability 0.9355. Twelve people give p = 0.935512 = 0.4494, so slightly under half of all single frames are clean. Five shots give 1 − 0.55065 = 0.9494, about 95 per cent. Reaching 99 per cent needs ln(0.01) ÷ ln(0.5506) = 7.72, so eight frames. The paper's rule of thumb for twelve people in good light says four, which on this arithmetic is worth about 91 per cent — a fair description of "almost ensure".
The Assumptions, and Why This Is a Novelty Rather Than a Rule
The model is elegant because it assumes a lot, and each assumption is somewhere it can be wrong.
It assumes blinks are independent between people. They are not entirely. Blinking is partly social and partly stimulus-driven: a flash makes people blink, a shouted instruction makes people blink, and a group all looking into the same low sun blinks together. Correlated blinking makes clean frames rarer than the arithmetic says, and it makes a burst of frames much less useful, because a flash-provoked blink can spoil several consecutive shots.
It assumes a single blink rate for everybody. Published rates for resting adults span roughly 6 to 30 a minute, and the spread is not noise — it reflects real differences in age, alertness, medication, contact lens use and how dry the room is. One heavy blinker in a group of twelve dominates the result, and averaging them all at fifteen hides that.
It assumes the vulnerable window is the blink duration plus the exposure. In reality a partly closed eyelid at the start or end of a blink may or may not spoil a picture, depending on how much of the iris is covered and how forgiving you are. That judgement moves the effective duration between roughly 40 and 450 milliseconds, and the answer moves with it by a factor of several.
And it assumes frames are independent, which requires a real gap between them. A five-frame burst at ten frames a second spans half a second, which is shorter than some blinks, so it is closer to one attempt than to five.
None of that makes the model useless. It is genuinely predictive about the thing that matters — that difficulty grows fast with group size — and it explains why photographing forty people is a categorically different job from photographing four. It is a novelty estimate, in the spirit of the prize it won, not a rule you should plan a wedding schedule around.
What Actually Helps, Beyond Taking More Frames
The single most effective lever in the model is exposure time, and it is also the one you control directly. Dropping from 1/30 s to 1/250 s removes about 29 milliseconds from the vulnerable window, which on the default blink duration is a little over a tenth of the total. That is why the original rule has a bad-light variant: slow shutters are not just a blur risk, they are a blink risk.
Reducing the number of independent attempts by provoking a synchronised blink is the trick most professionals use without calling it that. Asking everyone to close their eyes and open them on a count means the whole group is at the start of an open period at the same moment, which is a much better position than a random one. Our exposure value calculator helps you find the fastest shutter your light allows.
Splitting a large group into two or three overlapping photographs is the other honest answer. The chance of a clean frame falls exponentially with head count, so two photographs of twenty are enormously easier than one of forty. If the arrangement is yours to design, the event seating table calculator is a more useful planning tool than this one.
How This Page Sits Beside Our Probability Tools
The boundary in one sentence: this page applies a specific published photographic model, while the binomial distribution calculator is the general machinery for "at least one success in n trials" and the conditional probability calculator handles dependence between events.
If you want the underlying maths in a general form, the odds probability converter and the coin flip probability calculator cover the same ideas without the photography. On the camera side, the camera field of view calculator tells you whether the group fits in the frame and the depth of field calculator tells you whether the back row will be sharp.
Arb Digital builds free calculators because genuinely useful pages earn attention and links. If you want tools, calculators or content built for your audience, we can help.
Browse All Free Tools Talk to Arb DigitalCommon Mistakes to Avoid
- Treating the answer as a guarantee — the output is a confidence level, and no shot count reaches certainty. The paper's own phrasing was "almost ensure".
- Counting a burst as separate attempts — frames fired inside half a second are not independent, because a single blink can span all of them.
- Ignoring the flash — a flash provokes blinks, so consecutive flash frames are correlated in exactly the way the model assumes they are not.
- Using one blink rate for a mixed group — published rates span roughly 6 to 30 a minute, and a single heavy blinker dominates a group result.
- Shooting one huge group instead of two — difficulty grows exponentially with head count, so splitting the group is far more effective than taking more frames.
Related Free Tools From Arb Digital
Work the general probability with the binomial distribution calculator, the conditional probability calculator and the odds probability converter. Set the camera up with the camera field of view calculator, the depth of field calculator and the exposure value calculator, plan the room with the event seating table calculator, and browse the rest in the free online tools hub.
Frequently Asked Questions
The rule from the original CSIRO work is to divide the group size by three in good light, or by two in poor light, for groups under about twenty. This calculator gives a fuller answer for your own blink rate, exposure and confidence level, and for twelve people at the defaults it suggests eight frames for 99 per cent confidence against the rule's four.
Nic Svenson and Piers Barnes at CSIRO in Australia, published as "Blink-Free Photos, Guaranteed" in Velocity in June 2006. It won the 2006 Ig Nobel Prize in mathematics.
A novelty estimate. It comes from a real published model and it captures something true — that difficulty grows fast with group size — but it assumes independent blinking, a single average blink rate and a fixed vulnerable window, and none of those holds exactly. Treat the number as an indication, not a plan.
A commonly quoted figure is around fifteen times a minute for a resting adult, but published ranges run from roughly 6 to 30 a minute. Tiredness, dry air, contact lenses, bright light and flash all push it up, which is why the rate is an input here rather than a constant.
Yes, though less than people expect. The vulnerable window is the blink duration plus the exposure, and a blink is far longer than most exposures. Going from 1/30 s to 1/250 s removes about 29 milliseconds from a window of roughly 280, which is a useful improvement rather than a fix.
Because a blink lasts a quarter of a second or more, and a fast burst can fit entirely inside one. The model assumes independent attempts, so frames need enough separation that a single blink cannot spoil all of them.
The chance of a clean frame falls exponentially with head count, so beyond about fifty people the required number of shots becomes impractical. The original researchers suggested abandoning the perfect frame at that point. Splitting the group in two is far more effective than shooting more.
If the blink rate and duration you enter multiply out to a vulnerable window longer than the average gap between blinks, the model says a person is always mid-blink, and the probability of a clean frame collapses to zero. That is a signal that the inputs are outside the range the model can describe, not a result.
This page is a novelty estimate based on a published probability model, not photographic or scientific advice. Blink rates and durations vary widely between individuals and with conditions, and real blinking is not fully independent between people, so treat every figure here as an indication rather than a prediction.
