Ten seconds is a budget, not a duration

Ask somebody how long a product video should be and they will usually answer with a number. The more useful answer is that the number is not the decision. Ten seconds is what you have; the decision is what you spend it on, and that decision is made once, at the start, and shows in every frame afterwards.

The failure mode is easy to describe because almost everybody has shipped one. The clip opens on the product, pushes in, cuts to a detail, pulls back out, and lands on a logo. Five things happen. None of them finishes. It is ten seconds of activity that leaves a viewer knowing roughly what they knew from the photograph, which is the one outcome a film is supposed to rule out.

A camera move is a sentence

The useful unit is not the second, it is the move. A move has a beginning, a middle and an end: it starts somewhere, it travels, and it arrives. If it is interrupted before it arrives, the viewer gets the beginning of an idea and then a new one, and nothing they can carry away.

A ten-second budget holds one of those comfortably. It holds two badly. The arithmetic is not mysterious — a move needs time to establish what it is looking at before the travel means anything, and time at the end for the arrival to register. Cut that in half twice and what is left is motion without a subject.

This is also the difference between a film and an animated photograph. A preset pan across a still you already own is not a move in this sense; it has travel but no arrival, because there is nothing at the end of it that was not visible at the start. That distinction is worth reading separately — it is the whole argument for generating footage rather than animating a photo.

One move, several stations

One move does not mean one thing happens. It means one continuous idea, which can pass several things on its way.

The clearest example in the Nitaris set is the watch film: a rack of focus that travels across the fluted bezel, then the green dial, then the cyclops lens over the date. Three things are shown. It is still one move — the camera never restarts, and each station hands to the next. A viewer reads it as a single look at a single object, and comes away having seen three specific pieces of it.

Compare that with the version that cuts: bezel, cut, dial, cut, date. Same three subjects, same ten seconds, and now it is a list rather than a look. Lists are what the still images are for. The film's advantage is continuity, and cutting is the one thing that spends it.

What the budget is actually buying

It is worth being concrete about what a viewer gets from ten uninterrupted seconds that they cannot get from a photograph:

• Material. Light travelling across a surface settles matte against gloss, metal against plated, glass against acrylic, faster than any macro still does. This is the single strongest thing a film buys, and it needs travel to happen at all.
• Form. A product turning — or a camera moving around it — resolves a shape that one frame flattens. Anything with a silhouette worth seeing from more than one side gains here.
• Scale, sometimes. Not reliably from the object alone, but a move that brings a hand, a wrist or a surface into frame settles it in passing.
• Credibility. A thing that behaves correctly under moving light looks real. This is the quiet one, and it is mostly what a viewer means when they say a film looks expensive.

Every item on that list needs duration. None of them needs five separate shots.

You do not have to spend all of it

The budget framing has a corollary that the fill instinct hides: an unspent second is not a wasted one.

Of the five films Nitaris publishes as worked examples, four run ten seconds and one runs eight. The eight-second one is the caramel popcorn clip below — a widescreen 16:9 shot of coated popcorn tumbling through a swirl of hot caramel. It ends when the tumble ends. There was no reason to hold it for two more seconds, so it does not.

Real Nitaris output · 0:08

That is the honest version of the rule. The move decides the length; the length does not decide the move. A clip that ends the moment its idea completes reads as deliberate, and one padded to hit a round number reads as padded — the last two seconds of a film are the most visible place to be caught with nothing left to say.

How this differs from ordering a set of images

There is a related idea worth keeping separate from this one. A row of listing images is a sequence, and each slot after the first should answer a question the previous slot did not — hero, then material, then scale, then context. That is a decision about what goes *between* the frames.

This post is about what goes *inside* one of them. The two rules point in opposite directions on purpose: across a set of stills, variety is the whole job; within a single film, variety is the failure. A row of six near-identical photographs wastes five slots. A ten-second clip containing six ideas wastes all ten seconds.

The test before you keep a render

Watch it once and try to say, in one sentence, what the camera did. Not what the product is — what the camera did.

If the sentence comes out easily — "it racks focus down the length of the watch", "it circles the bottle once", "it drops through the caramel and settles" — the film has one idea and it landed. If the sentence needs an "and then", you are describing a list, and the render is spending ten seconds to deliver less than its own opening frame already did.

That test costs nothing and it is worth running before the other one, because a clip that fails it cannot be rescued by a better thumbnail, a better crop or a better slot in the row. It is the only judgement in this whole process that has to be made about the film as a film.