Build a short video around one clear idea

Goal: prepare a readable message that works with the kit's voice-first production sequence. Time: about 25 minutes. Before you start: complete lesson 00.

Your first video needs a clear direction before it needs a striking visual style. The Lumio example gives you three spoken lines: an opening, a practical step, and an invitation to act. Use that small structure to learn how a message moves.

1. Write down the single idea your viewer should remember.
2. Read the opening aloud and separate the supporting ideas.
3. Plan a review with sound and another without it.

Vocabulary

• Hook: The opening phrase giving a reason to keep watching.
• Shot: A visual unit shown during part of the message.
• CTA: A call to action, such as discovering Lumio.
• 9:16: A vertical frame, taller than it is wide.
• Voice-first: Building video timing from recorded speech.
• Subtitles: Text synchronized with speech and also delivered as SRT.

1. Give the video a specific job

Start by reading this fragment from the English example:

{
  "title": "Lumio",
  "objective": "Present a simple method.",
  "audience": "Small teams",
  "cta": "Discover Lumio",
  "url": "https://lumio.example.invalid"
}

This is a field-reading exercise, not a complete runnable brief. The objective describes what the video should accomplish. The audience tells you who needs to understand it. The call to action, or CTA, gives the ending a direction.

Give each field its own job. An objective such as presenting a simple method helps you decide which details to include. An audience such as small teams helps you choose familiar language. A CTA such as discovering Lumio tells the viewer what to do after the explanation.

On paper, finish this sentence: “After watching, the viewer should remember…” For this exercise, an idea can move forward through a clear step. If your answer introduces several unrelated benefits, choose one and save the others for another campaign.

This is an editorial exercise. The software can validate a brief's structure, but the quality of an intention and the results of publication remain yours to assess. There is no promised retention rate, sales result, or ideal total duration attached to the example.

2. Make the opening understandable immediately

The example begins with:

{"id":"hook","text":"An idea deserves a clear path."}

A hook is the first phrase that gives someone a reason to keep watching. Aim for an understandable opening from the first second. Use that as a writing goal rather than a timing constraint enforced by the software.

Read the sentence aloud on its own. Could a viewer recognize its subject without an explanation from you? Does it lead naturally into the next line? The Lumio hook introduces a clear path; its next line can therefore explain a step along that path.

⚠️ Use numerical promises only when you can justify them with authorized facts.

Post-copy checks limit some invented numbers, claims, and URLs in caption.md. Brief validation checks structure without establishing the truth of the spoken script or comparing it with your facts. Lumio is fictional, and the example attributes no actual customer result or measured performance to it.

You can make an opening more useful by removing a vague introduction or choosing a concrete verb. Keep the claim faithful to the information you have. Stronger wording still needs the same evidence as quieter wording.

3. Separate the spoken ideas

The complete example script is:

[
  {"id":"hook","text":"An idea deserves a clear path."},
  {"id":"step","text":"Define a step. Stay on course."},
  {"id":"cta","text":"Discover Lumio."}
]

The first line opens the subject, the second develops it, and the third closes it. The IDs connect text to technical files and processing. They are not spoken aloud. Keep them unique, and keep an ID stable when you are simply correcting the wording of that same line.

One idea per shot does not mean one script line always becomes one shot. The kit can divide material into sentence units and can also use explicit scenes. The step line already contains two sentences. Short units of meaning give the animation something clear to accompany.

Try underlining the verb in each sentence and naming its purpose. If one sentence asks the viewer to understand several different actions, simplify it before adding effects. The schema allows up to 12 lines of 500 characters each. These are technical ceilings; filling them is not a writing target.

4. Let the speech establish the timing

Voice-first means that speech comes before the final animation timing. The pipeline synthesizes each line, aligns its words with local Whisper, then builds the timing used by the rest of the production. The standard brief does not accept an arbitrary duration for each phrase.

Read your script aloud before synthesis. Listen for breathing space, awkward phrases, and words that are hard to say. This will not predict Gemini's exact duration, but it helps you find sentences worth dividing. The campaign also rejects individual voice lines longer than 30 seconds.

The code works at 30 frames per second and keeps about three seconds after the final speech. That closing hold gives the CTA and address time to remain visible. The actual total duration depends on the generated audio, so the course gives no guaranteed running time for this brief.

Changing the voice or text can change the audio and its timing. Keep your review notes tied to the actual version you heard. That makes it easier to tell whether a later difference came from wording, delivery, or visuals.

5. Write for the vertical frame

The demo omits format, so the schema uses 9:16. Compatible base styles also accept 1:1 and 16:9; the six brag tones require vertical video. Stay with the default for your first exercise.

A long phrase wraps across more lines in a narrow frame. Shorten the wording before trying to make every character smaller. The template fits text inside a readable layout box and can reject a composition that cannot remain readable.

Logo, title, illustration, and subtitles share that frame. At each scene, identify where the viewer should look first. A useful visual supports the spoken idea without requiring a second complicated reading task. Check the actual result; the course does not prescribe a universal margin percentage.

6. Review through two different channels

Watch without sound to check that the visible text and CTA still make sense. Then listen without watching to check that the script forms a complete message. These are editorial checks, not an automatic accessibility certification.

The templates display subtitles synchronized with timing. The package also includes subtitles.en.srt, which stores spoken text and time references separately. caption.md contains suggested social post copy. Keep those purposes distinct when reviewing or correcting files.

Finish with a reference note containing your objective, audience, three spoken lines, and exact CTA. Use it when inspecting the first package in the next lesson. Remove repetition to keep the video concise, then judge its length against its actual message.

If you get stuck

See Composition and checks, Whisper and synchronization, or Brief validation. Start with the wording when a phrase is too dense for the frame.

Key takeaways

• Give each shot one clear idea.
• Speech provides the timing reference; schema limits are ceilings.
• The SRT follows spoken words, while caption.md prepares the social post.