If you are onboarding Seedance 2.0 for the first time, treat this page as a practical SOP. We cover how the product feels, how to write seedance prompts, and how to ship stable clips—plus how NotebookLM keeps your team organized.

Seedance 2.0 tutorial

1) Core experience in Seedance 2.0

Multimodal input: image, video, audio, and text together
All-in-one reference: @asset binding to cut mismatches
Camera control: moves and shot scale easier to specify in prompts
Pacing: better handling of short-form cuts and motion logic

2) A five-step workflow (copy-paste friendly)

Goal: pick use case—UGC, explain, narrative, or ad.
Assets: keep 3–8 core references; avoid overload.
Prompt: order as subject → scene → camera → style → constraints.
Pilot: generate a 5–8s draft to validate direction.
Iterate: change one variable per round for clean reviews.

3) Seedance prompts cheat sheet

Scenario	Keywords to lean on
Product hero	macro close-up, slow push, material highlights, clean backdrop
Talking head	medium framing, stable gaze, lip sync, natural hands
Story beat	shot changes, emotional build, mood light, ambience

Example:

Match @image1 wardrobe and likeness; rooftop city night; medium shot slowly pushes to face close-up; cinematic mood; avoid garbled backgrounds and limb errors.

4) NotebookLM for speed

Store brand docs, script templates, and winning captions in NotebookLM. Export a reusable prompt checklist and a banned-words list, then run final prompts in Seedance 2.0—ideal for teams and batch publishing.

5) Common pitfalls

Warped faces: reference angles and lighting differ too much—standardize them.
Ignored camera: prompts are too abstract—use “first / then / finally” steps.
Noisy frames: too many constraints at once—lower complexity, then layer detail.

Closing

The leap in quality comes from repeatable seedance tutorial habits: tighten seedance prompts, skim credible seedance news, and keep your playbook updated.

立即使用