Seedance 2.5 · video storyboards

Seedance 2.5: how to storyboard a product demo video

A wet bench and a good-looking product is a mood, not a demonstration. A demo has to show something change.

A folded paper mailer on a wet metal workbench in a blue-lit room
A Seedance 2.5 gallery example. A great opening frame. It is not yet a demonstration, because nothing has happened. Existing gallery illustration; model attribution has not been independently verified.

The short version

Show the start, the event and the result, in that order, without cutting away from the moment of contact. Use generated video for the concept, and real footage whenever you make a claim.

A demo has to show something change

Look at the opening frame from this Seedance 2.5 scene. A folded paper package on a metal bench, water droplets catching the light, a blue room out of focus behind it. It looks like a laboratory. It makes you expect something.

And then nothing happens. That is the gap between a nice product shot and a demonstration.

A wet surface next to a package does not show that the package resists water. It shows a wet bench. The demonstration is the change: before, event, after.

So the first question for any demo storyboard is not what does it look like? It is what changes, and how will the viewer see it?

Three shots, in this order

Almost every good product demo is these three shots. Fancy edits usually make them worse.

ShotJobCommon mistake
1. StartShow the product and the surface it sits onToo short, so the viewer never sees the starting state
2. EventShow the thing happening, in one continuous viewCutting away at the moment of contact
3. ResultGive the viewer time to see the changeEnding on the event, so nothing is proven

The middle one is where demos live or die. If your edit cuts away exactly when the water hits, the viewer assumes you are hiding something, even if you are not. Hold the shot through the contact.

The third one is where they get abandoned. A splash is exciting, so people end there. But the splash is the event, not the result. The result is the package afterwards, held long enough to read.

The camera choice is part of the argument

What the camera shows is what the viewer can check. That makes framing an honesty question, not just a taste question.

  • Close-up: shows the surface and the detail. Convincing, but the viewer cannot see where the water came from.
  • Wide: shows the whole set-up, including the source. Less dramatic, much more credible.
  • Both: wide for the set-up, close for the detail, and the viewer gets to see both halves.

The order matters too. Wide first, then close, tells the audience here is the situation, now look closely. Close first, then wide, can feel like a reveal you were forced into.

One unbroken shot of the whole event is the strongest thing you can show. This is exactly why a long take convinces people: there is nowhere to hide a cut.

Write the three shots as separate prompts

Shot 1 · Start
Close product shot of a folded brown paper mailer sitting on a clean wet steel workbench. Soft blue laboratory light from behind, shallow depth of field, fine water droplets on the metal around it. Camera locked off, slightly above the bench. Nothing moves. No text, no logos.
Shot 2 · Event
Same brown paper mailer on the same wet steel bench, same camera position. A steady stream of clear water pours onto the top of the mailer from above, splashing off the surface. Water source visible entering the top of the frame. Continuous shot, no cuts, blue laboratory light. No text.
Shot 3 · Result
Same brown paper mailer on the same wet steel bench, same camera position, water no longer falling. The mailer sits in a shallow pool of water, surface beaded with droplets. Held static for several seconds. Blue laboratory light, calm, nothing moving. No text.

Keep same camera position in all three. A demo where the camera jumps between shots loses most of its power, because the viewer can no longer compare before with after.

Know the line between a concept and a claim

This is the part worth being blunt about. A generated clip can illustrate an idea beautifully. It cannot be evidence.

The clip in this article is a generated concept. It is not a recording of a real product test, and it does not show that any real mailer survives water. Nobody tested anything.

So split the two uses clearly:

  • Concept video. Pitching an idea, showing a design direction, explaining how something would work. Generated video is genuinely useful here.
  • Product claim. Anything that says this product does X. That needs real footage of the real product, and a test a customer can understand.

Guidance from the FTC and the Dutch ACM is short and readable, and neither cares whether a misleading demonstration was filmed or generated.

The practical habit: label concept footage as concept footage, right there in the video, and keep your real test footage for the claims. It costs you one line of text and saves the whole campaign.

Seedance 2.5 scene reference from the gallery. A generated concept clip, not a recording of a real product test. Existing gallery illustration; model attribution has not been independently verified.

Questions people ask

Can I use AI video in a product ad?

For mood, packaging concepts, background scenes and explaining an idea, yes. For a demonstration that supports a claim about what your product does, use real footage of the real product.

Why does my demo feel unconvincing?

Usually because the edit cuts at the moment of contact, or ends on the event instead of the result. Hold through the contact, then hold on the outcome.

How long should a demo clip be?

Long enough to show start, event and result. Often that is eight to fifteen seconds in total. Shorter than that and the viewer misses one of the three.

Do I need to label generated footage?

Check the rules for your market and platform. Labelling is a good habit either way, especially for anything that looks like a test, because it removes the argument entirely.

Where to read more

Outside sources

More from DaFreeAi

Storyboard your demo

Generate the three shots above with the same camera line, then label the result as concept footage.

Open a video model

Updated 12 September 2026.