Explain Why a Short Film Needs Its Running Time

A useful runtime claim has a reason attached. Describe what must happen, what needs time to become noticeable, and which part of the estimate you have actually tried. “Approximately five minutes; the final wait is still being tested” gives a collaborator more to assess than a confident number selected before the film has a temporal shape.

Do not begin by defending every second. Begin by finding out what those seconds are for. Some may carry a physical action. Others may allow an expectation to change. A transition may remove time without removing anything the film needs. The work is to distinguish those cases, then make the estimate as firm as the available material permits.

Put the number’s origin beside the number

A screenplay, a recorded rehearsal, a timed storyboard, and an assembly offer different kinds of information. Label the basis rather than letting all of them become “runtime.” An assembly has an elapsed duration, though further edits may change it. A proposed hold in a storyboard records an intention. A spoken read-through measures that reading, not every silent action the film will include.

Screenwriter John August makes a useful distinction in his discussion of page count and running time: page count is a crude approximation, while estimating a particular movie calls for attention to the time its individual scenes take. His discussion draws on his experience and research into feature screenplays; it does not establish a conversion formula for shorts.[^1]

A one-line action such as “They wait for the dryer” contains no fixed duration. Nor does filling a page with a detailed account of their stillness necessarily add screen time. Typography is the wrong measuring instrument for this part of the film.

An external limit belongs in a separate field. A recipient may require a particular maximum, including specified material such as credits. That creates a delivery condition to check against the actual instructions. It does not establish the best duration for the experience. This article uses no named festival or eligibility rule.

Break the film where something changes

Our original fictional example, Spin Cycle, follows a laundromat attendant trying to close while a customer waits for a blanket to dry. The attendant eventually puts away her keys and sits nearby. The story ends while the dryer continues; it does not wait for the practical task to finish.

A first temporal outline could distinguish four passages. Closing tasks establish the attendant’s intention to leave. Successive announcements bring that intention into conflict with the customer’s wait. The stopped machine appears to end the problem, until the blanket proves damp. The restarted machine accompanies the attendant’s choice to stay.

These are useful boundaries because the film’s question changes at each one. Dividing an intended five minutes into five equal blocks would not explain those changes. It would merely distribute a number.

Under each passage, describe the action that requires duration. Emptying a basket may be compressed if nothing changes during the task. Touching the damp blanket may need a continuous exchange if the recognition passes from one person to the other. Once the attendant sits down, the practical choice is visible; the remaining interval has to justify what it offers beyond confirming that fact.

That last passage is our uncertainty. Does the short end on a decision, or does it give the two people time to inhabit the decision? Those are different ending experiences. Neither can be selected by counting scene headings.

A timing sample with its limits visible

For this article, two original schematic animatics were made from the final passage. They use abstract figures, a fixed view, a rotating shape representing the dryer, and a synthetic hum. They are real timed files, but not filmed performances. Their purpose is to compare proposed intervals before asking actors or a physical machine to supply them.

The held sample lasts 48 seconds. The compressed sample lasts 30 seconds. Both run at 24 frames per second; their decoded video counts are 1,152 and 720 frames respectively. Their video and container durations were checked, and both files decoded without reported errors. The accompanying timing log preserves those checks.

Those numbers describe the files. They do not measure emotional effect, demonstrate the time a performer needs, or establish how long a real dryer runs. The durations were deliberately authored, then checked against the exported objects. That distinction matters: a precise export can faithfully preserve a poor timing proposal.

Both versions start when the dryer restarts and end while it is still turning. The action and its order stay the same. Only two intervals change:

This is a deliberately narrow comparison. No faster chair move, extra close-up, new dialogue, or music cue has been added to make the compressed version persuasive. Its reduction comes from eight fewer seconds before the keys go away and ten fewer seconds after the attendant sits.

The samples do not reproduce the earlier announcements or the damp-blanket exchange. A viewer meeting these diagrams without the preceding scene would not have the same context as a viewer of the intended short. The text description and construction notes make the sample readable without video or sound.

Ask what the shorter version changes

The physical decision remains in both samples. The keys disappear into the pocket; the chair moves; the attendant sits. Shortening therefore does not remove the action by which she agrees to wait.

What changes is how much of the passage remains unresolved before that action, and how long the image continues after it. The held proposal spends more time with leaving still possible, then more time after leaving has been set aside. The compressed proposal brings the decisions closer together. These are descriptions of the authored timing, not measured findings about what viewers feel.

A useful viewing question would be: when did you understand that the attendant was staying? A second would ask what, if anything, changed after she sat. “Which is better?” skips the distinction the comparison was built to examine.

There are several plausible editorial outcomes. You may decide the longer pause with the keys calls too much attention to a choice the performance should make quietly. You may prefer its hesitation but find the final hold repeats an already complete ending. Or the seated interval may be where the film’s attention shifts from the attendant’s problem to the shared room. Each would support a different edit. None follows automatically from 48 being larger than 30.

The comparison can also reveal that neither proposed version is useful. Perhaps the important change happens when the attendant touches the blanket, making the later keys redundant. Then the next experiment belongs earlier. A timing study should be allowed to move the question, rather than become a demonstration staged to defend the first estimate.

Give the next trial a defined beginning and end

To move from these schematics toward a performance estimate, keep the sample’s boundaries and record the final passage with stand-ins or performers, using material you are entitled to make and share. Agree whether the trial begins at the restart sound, the customer’s hand leaving the controls, or another observable event. Choose an equally clear stopping point. Otherwise two timings may differ because the stopwatch included different material.

Record the whole attempt before trimming it. Note the task, the distance the attendant travels, the sound available, and whether the person performing the customer can affect the timing. A rehearsal using an empty room and an imagined machine cannot answer every question about a noisy location, but it can answer more about a chair move than an unperformed stage direction.

Do not force the rehearsal to match the animatic. Its value is partly in exposing the diagram’s convenience. A performer may put the keys away during the move, not in a separate three-second block. Carrying the chair may be awkward or conspicuous. The machine’s sound may cover an intended change in breathing. Revise the proposal around what the trial shows.

Keep observations separate from interpretations in the log. “The keys were pocketed before the move began” is an observable description. “The decision felt reluctant” is an interpretation requiring its own context. “Every viewer understood” would require evidence a single rehearsal does not provide. For the samples here, no live rehearsal or audience comparison was performed.

Build a whole-film estimate without multiplying the sample

The last 48 seconds cannot establish the duration of everything before them. The opening may compress readily; the blanket exchange may need more time than the script suggests. Treat the other passages separately.

Here is a planning calculation for the fictional project. The earlier encounter and credits are assumed ranges for the exercise, not recorded observations. The final-passage figures come from the authored animatics, not a live-action test.

The arithmetic gives 3 minutes 40 seconds to 5 minutes 3 seconds. Its endpoints are not a confidence interval or a prediction accurate to the second. They are simply the sum of the listed assumptions. Before putting a rounded number in the pitch, keep that distinction in the working document.

A defensible statement at this stage would be:

Proposed duration: approximately five minutes, including titles and credits. The current outline and two schematic versions of the final wait give a rough planning envelope of 3:40–5:03. The ending compares a more immediate commitment with a longer shared wait; its live-action duration remains to be tested. Timing the earlier blanket exchange is the next step toward a firmer estimate.

The supporting detail does not all need to sit under the title. The headline estimate can remain short, with its basis in the accompanying note. What matters is that a recipient can distinguish the proposal from a completed film’s duration.

Make the decision the evidence permits

When a shorter delivery is required, identify the passage you would change and what that change costs. Removing a duplicate announcement may preserve the movement. Removing the only interval in which the two people sit together would create a different ending. The number alone cannot distinguish those edits.

For Spin Cycle, the useful result is not “48 seconds wins.” It is a more specific decision: the ending’s duration depends on whether the film needs to stay after the attendant commits. The sample makes that question inspectable, while leaving performance and audience response unclaimed. A runtime statement becomes stronger when it can name that remaining uncertainty instead of hiding it behind a neat number.

Sources

[^1]: John August, “How accurate is the page-per-minute rule?”, March 9, 2020. Locators: the discussion after the screenplay-length/runtime comparison, the recommendation to consider scene timing, and the feature-film scope note. Inspected September 19, 2026. Used for the bounded distinction between a rough page-count estimate and the duration of particular scenes, not a short-film conversion rate.

Frequently asked questions

How can I make a short film's runtime claim useful?

Attach a reason: describe what must happen, what needs time to become noticeable, and which part of the estimate you have tried. 'Approximately five minutes; the final wait is still being tested' gives a collaborator more to assess than a confident number chosen before the film has a temporal shape.

What kinds of evidence can inform a runtime estimate?

A screenplay, recorded rehearsal, timed storyboard and assembly offer different information, so label the basis rather than calling all of them runtime. An assembly has an elapsed duration; a storyboard hold records an intention; a read-through measures that reading. A page count is a crude approximation, and an external maximum belongs in a separate field as a delivery condition.

What do the Spin Cycle timing samples actually compare?

Two original schematic animatics use abstract figures, a fixed view, a rotating shape and a synthetic hum; they are timed files but not filmed performances. The held sample lasts 48 seconds and the compressed sample 30 seconds, both at 24 frames per second. Both start when the dryer restarts and end while it turns, and only the pause with the keys and the final seated interval change; the physical action remains.

What can the timing samples not show?

They do not measure emotional effect, demonstrate the time a performer needs, or establish how long a real dryer runs. Shortening does not remove the action of staying; it changes how long the passage remains unresolved before that action and how long the image continues after it. No live rehearsal or audience comparison was performed, so viewer response remains unclaimed.

How should I build a whole-film estimate from a final-passage sample?

Treat the other passages separately; the last 48 seconds cannot establish the duration of everything before them. In the planning example, the earlier encounter is an untested outline estimate of 175–235 seconds, the restart through the ending uses the two schematic proposals at 30–48 seconds, and titles and credits are an assumed 15–20 seconds. That sums to 220–303 seconds, or 3:40–5:03, which is a provisional envelope, not a confidence interval or second-accurate prediction.

Further reading

• The Film Needs to Feel Slow. How Can the Pitch Show Movement Without Promising Speed? — Read this next when the runtime must support a deliberately slow experience, so the pitch shows what the time makes available rather than asking readers to be patient.
• Pitch a Film Made From Still Photographs — Read this when the final wait will be pitched through stills or a fixed view, where hold lengths and sound cues need explicit temporal relationships.
• One Continuous Shot or Hidden Cuts: What Does the Feature Actually Need? — Read this when deciding whether the final shared wait needs one continuous take or can be shortened with hidden cuts.