LESSON 14 · REMIX

Adapting other people's prompts with AI

You don't have to write a prompt from scratch. The fastest working trick is to take a ready-made prompt from a public library and rework it for your own task using an AI assistant (Grok, ChatGPT, Claude, etc.). You get the proven structure of a viral clip and simply change the goal.

🔁 The idea in one sentence
The AI carries over the STRUCTURE (timecodes, camera movement, sections, rhythm), while you change the GOAL (subject, genre, product, platform, duration). It's faster than starting from scratch, and it preserves logic that has already proven itself.

How it works — 4 steps

  1. Find the source prompt. Take any example from a library of ready-made prompts (see the link below) — ideally from the same "weight class" as your task (commercial → commercial, action → action).
  2. Give the AI a meta-instruction. Clearly state what to keep (structure, camera, length) and what to change (product/character/genre/location/format). A ready-made instruction template is below.
  3. Get the adapted prompt in English and drop it into generation.
  4. Refine by iteration: change one axis at a time (first the product, then the palette, then the platform), comparing the results.

Meta-instruction for the AI (copy it)

Paste it into Grok/ChatGPT/Claude, plug your goal into the [brackets] and attach the source prompt at the end.

You are a prompt assistant for the Seedance 2.0 video model.
Below is a source reference prompt. Keep its structure, rhythm, timecodes,
camera movement, lighting and overall production look — but change the goal:
• product/hero → [your object]
• location/scene → [your location]
• mood/genre → [optional]
• format → [16:9 / 9:16 / duration, if needed]
Keep the length and number of shots. Return ONLY the finished prompt in English.

Source prompt:
[paste the library prompt here]

Example: before → after

We take the popular "car reveal" prompt and rework it into a sneaker reveal — the structure, the arc of light and the premium look stay the same; only the product and the location change.

① Source (from the library)

Car reveal in an underground parking garage.

A matte black sports car sits in an empty underground parking garage as overhead lights flick on in sequence down the length of the car. Slow low-angle arc shot circling from the front wheel to the headlight. Cool blue practical lighting, wet concrete floor with sharp reflections, premium commercial look. Low engine hum and a soft electrical click as each light turns on. Stable car proportions, clean paint surface, natural reflection motion.

② Result after the AI remix

Same skeleton → sneaker reveal in a studio.

A pair of pristine white sneakers sits on a matte concrete pedestal in a dark minimalist studio as overhead strip lights flick on in sequence along the shoe. Slow low-angle arc shot circling from the heel to the toe cap. Cool blue practical lighting, polished floor with sharp reflections, premium commercial look. Soft fabric rustle and a subtle electrical click as each light turns on. Stable product proportions, clean material texture, natural reflection motion.

What else you can "dial" with a single instruction

  • Genre/mood: "cinematic" → "cozy lifestyle", "documentary", "retro 90s".
  • Platform/format: 16:9 for YouTube → 9:16 for Reels/Shorts, trimmed to 8–10s.
  • Camera style/reference: plug in a film stock/lens (Fujifilm, anamorphic) or an engine (UE5).
  • Localizing the lines: ask the AI to translate the dialogue into the language you need, preserving timing and lip-sync.

Where to get sources

Large public libraries of ready-made Seedance 2.0 prompts with preview videos are a handy "raw-material base" for remixing. For example: youmind.com/seedance-2-0-prompts (filters by category, sorting by popularity).

✅ Do
  • Change one axis at a time — that way it's clear what affected the result.
  • Ask to keep the timecodes and camera vocabulary — that's the "skeleton".
  • Hold the length/number of shots, otherwise the pacing will drift.
⚠️ Remember
  • Respect the source and the rules of the platform where you generate.
  • Don't reproduce the likeness of a real person without consent.
  • AI makes mistakes — reread the result and remove contradictions.

Try Seedance 2.0 yourself

Generate your first clip in ZenCreator — Seedance 2.0 is available in the Image-to-Video tool.

Start Creating
PreviousNext lesson