How Adotopia works

There are two different scripts in this product, and mixing them up is the single most common confusion.

• The context is the film, series or recording your advertisement will be shown inside. Adotopia reads it to find the right moment to interrupt.
• The advertisement is what you are making. Its words are yours, or the agent writes them.

Anywhere a screen says "context", it means the first one.

Four ways in, and the first is the quickest:

• Topic only and optional concept — no film. You get a cast, a script and a video, but no insertion slot, because there is nothing to insert into.
• Upload context media — a video, an audio track or a script file. A video is read through its frames, audio through its soundtrack.
• Paste context script — the film's script, subtitles or transcript.
• Reuse analysed context — something you brought before, without paying to analyse it again.

Every mode has a box for your concept for the advertisement. Leave it empty and the agent writes the ad. Fill it in and the agent stages what you wrote, keeping your wording and writing only the visuals.

The agent reads your whole source and proposes the people who will appear. They are always original characters, never the film's own: same roles and energy, different people, so nothing copies the source.

Each one is yours to change, two ways:

• In words — type an instruction and press Apply.
• With a photo — upload a picture of who you have in mind. A vision model reads it and rewrites the character to match. The photo never appears in the video.

A character you edit is kept. Pressing Regenerate rewrites only the others, so your work is never thrown away.

Topic, tone and length. Length is the one that matters for cost: rendering video is almost the entire bill.

This screen also holds Models & predicted price. Open it and you choose which model writes the ad and which one renders it, and the predicted price updates as you choose.

• Writing — DeepSeek, ChatGPT, Grok, Mistral, Gemini or Kimi. They differ by more than fifty times on price, but writing is pennies, so a dearer one barely moves the total.
• Rendering — the cheap models are billed per clip and the Seedance ones per second. This is where your money actually goes.

Each model has a backup provider behind it, so a bad day at one company does not stop you working.

If you brought a context, Adotopia gives you the second to insert at, and says why: how well the ad matches the mood there, how strong the natural break is, and whether the surrounding scenes are brand-safe.

It never puts an ad mid-climax. The heat map shows the emotional shape of the whole title with the chosen moment marked.

Below it is the script, shot by shot, and every field is editable. There is also a prompt box: type “make it slower and set it at dawn” and the whole script is rewritten for you to review.

The ad is rendered with audio, in clips that are joined into one file. Longer ads take longer, and you are billed for what the run actually uses rather than the prediction.

Afterwards you can burn an on-screen message onto the video, placed where it will not cover a face or the main subject, and moved when the content underneath changes.

One credit is one US cent of usage, and pricing follows the country on your account. Every job shows its price before it starts.

Top up with any amount you like from the Credits page, or redeem a voucher code there. Your balance is always in the top bar, and the Brief step refuses to start a job your balance cannot cover rather than failing halfway.

• Video Ads — upload your own video and have an ad spliced into it at the recommended second, at the source resolution.
• Static Ads — text and image ads for a web page, with a placement recommendation and a preview of how it will sit.
• Engine — the analytics behind it all: the emotion timeline, ranked slots, and a yield optimiser across several advertisers.