A longer film moves the hero shot to the end

The length picker is the least interesting-looking control in the Studio and the one that changes the most. It offers ten, twenty, thirty or forty seconds, at eighty credits per ten — so a forty-second film costs three hundred and twenty credits against a ten-second one's eighty. That arithmetic is easy and it is also the wrong thing to think about first, because it suggests you are buying more of the same thing, and you are not.

A long film is rendered in ten-second passes

The first thing worth knowing is mechanical, and it explains everything after it. Nitaris does not render a forty-second film in one go. It renders four passes of ten seconds. The first pass is generated from your product photograph in the ordinary way. Each later pass is a separate request in which the model is shown the preceding ten seconds of footage and asked to carry the shot on from exactly where it stopped.

There is no cut between passes, and the intent is that you cannot tell where one ends. That is a real constraint on how the later passes are written: a continuation that re-introduces the product, the set and the lighting is not giving the model context, it is arguing with footage the model is already looking at — which is how a chained shot drifts or hard-resets in the middle. So continuations are written short, and they describe motion rather than subject. The film keeps going; it does not restart.

The arc is not the same arc

The second thing is the one that should actually decide your choice. The number of passes changes the named beats the film is written against:

• Ten seconds — one beat: hero.
• Twenty — establish, then hero.
• Thirty — establish, detail, hero.
• Forty — establish, detail, texture, hero.

Read that list again with the first post in this series in mind. At ten seconds the hero shot is not the climax of the film, it is the film. Every second of runtime is spent on the one composition that shows the product best. The moment you buy twenty, the hero stops being the whole thing and becomes the destination: the first ten seconds now have the job of establishing, and the payoff is held back until the second half.

By forty seconds the hero composition is the final quarter and three beats come before it. That is a legitimate film and often a better one — it has room to show the thing in its setting, then close on a detail, then dwell on a surface, then arrive. But it is a film that asks for thirty seconds of patience before it delivers the shot that a ten-second version would have opened with.

Which is a thumbnail problem before it is a creative one

A product video spends most of its life paused. It sits as a still in a gallery, a grid, a search result, and only becomes a video for the people who choose to play it. The opening frame is doing the work of a photograph.

Now put the two facts together. The opening frame of a ten-second film is the start of a hero beat. The opening frame of a forty-second film is the start of an establish beat — wider, more context, less product. That is correct film-making and it is a worse thumbnail, and on any surface where most people never press play, the thumbnail is what you actually bought.

This is the trade, stated plainly: a longer film spends credits on beats that are only seen by viewers who stay. If your placement rewards staying — a product page where somebody has already decided to look, an ad slot with a real watch-through — those beats are the point. If your placement is a grid of competing thumbnails, you have paid four times as much for a weaker first frame.

A short test before you pay for forty

Three questions, in order. They are quick and they are the ones that actually separate the cases:

• Does the product have a second thing to say? Establish, detail and texture are three different beats, and each needs something real to be about. A plain object photographed well has one beat in it. Something with a setting, a mechanism and a surface worth dwelling on has three.
• Where will it be seen paused? If the honest answer is a thumbnail grid, the extra beats are being bought for an audience that will not reach them.
• Have you got a ten-second version you are happy with? A film that cannot hold ten seconds will not hold forty — it will hold the same weakness for four times as long, at four times the price.

The third one is the one to insist on. The cheapest route to a good forty-second film is a good ten-second one first: eighty credits buys you the answer to whether the product, the mood and the set are working at all, and if they are not, you have learned it for a quarter of the cost. If they are, the longer run is extending something you already know is good rather than gambling three hundred and twenty credits on it.

What does not change

Worth saying, because it is the part people assume scales with the number: the product fidelity rules are the same at every length. The opening pass is conditioned on your photograph and every pass after it is conditioned on the footage before it, so the shape, colour, materials and logo are held the same way at forty seconds as at ten. The ratios are the same too — vertical 9:16 or widescreen 16:9, your choice before you generate.

And the budget logic from the shorter post survives intact. Each ten-second pass is still a budget, and still buys one idea. A forty-second film is four budgets in a row that happen to be continuous. The failure mode does not change either — it just gets longer, and more expensive to discover.