Convert the user's "what kind of video I want" into engineering-level instructions that Seedance 2.0 can understand.
When not to use this skill
- The user is describing a 5/10/15-second compact short film or wants native audio → use happyhorse-prompter
- The user already has a prompt but the result is wrong → use seedance-debugger
- The plot is complex and requires multi-shot switching → use seedance-storyboard
Core methodology: advanced formula
精准主体 + 动作细节 + 场景环境 + 光影色调 + 镜头运镜 + 视觉风格 + 画质 + 约束条件
See methodology/02-advanced-formula.md。
Workflow
Step 1: Clarify requirements (when necessary)
If the user only gave one or two sentences ("make a perfume ad"), first ask clearly about at least 3 of the following elements (do not ask about all of them; pick the ones that most affect the result):
| Element | Example question |
|---|---|
| Subject | "What does the perfume bottle look like? Transparent glass or frosted?" |
| Action | "Is the bottle displayed stationary or rotating/spraying?" |
| Scene | "Real scene or solid-color background? What color?" |
| Lighting | "Hard light with dramatic feel or soft light with premium feel?" |
| Camera | "Prioritize close-up texture or a medium shot showing the whole bottle?" |
| Style | "Cinematic / commercial ad / minimalist / retro?" |
| Duration/aspect ratio | "Vertical short video / horizontal / how long?" |
Exception: if the user says "whatever" "quick" or gives a reference image, directly generate a version for them to revise.
Step 2: Fill in according to the 8 elements
Write in English (Seedance responds better to English), but keep the user's core elements in their original form (character names, brand names, proper nouns).
Template skeleton：
[Subject with 2-3 stable identifying features].
[Action with specific body parts + speed/intensity].
[Scene/environment].
[Light source + time of day + color tone].
[Camera size + movement + angle].
[Visual style + film grain/aesthetic anchor].
[Quality: cinematic / 4K / photorealistic].
[Negative: avoid ...; no watermark; no logo; preserve face stability].
Step 3: Check to avoid errors
Self-check before submission:
- Is the prompt between 50-200 words?
- Does one shot specify only 1 camera movement (do not simultaneously push + pull + pan)?
- Are there no contradictory words (such as
8mm+4K、slow motion+high speed）？ - Are there constraints (at least
no watermark, no logo）？ - In multi-subject scenes, are they distinguished with labels ("police"/"thief")?
Step 4: Output format
Give the user a complete prompt, plus 3 notes:
## 生成的提示词
\`\`\`
[完整提示词]
\`\`\`
**推荐参数**：宽高比 [X:Y] · 时长 [N 秒] · 分辨率 1080p
**关键技巧**：[1-2 句说明这条提示词的设计思路]
**可调点**：[告诉用户可以怎么改 — 比如换光影/换风格/换镜头]
Practical example
User input: "Make me a coffee ad"
Clarifying questions(ask all at once, no more than 3):
- Camera angle preference — extreme close-up texture (oil droplets/steam) or medium shot in a real scene (cafe tabletop)?
- Emotional tone — warm (morning/wooden) or high-end (black background/spotlight)?
- Distribution channel — vertical Douyin / horizontal TV / square IG?
User answer: extreme close-up, warm, vertical.
Output：
Extreme close-up of rich cream being poured into a freshly brewed dark espresso in slow motion, creating a mesmerizing vortex pattern across the crema surface. Wisps of steam rise softly from the cup. Warm tungsten side-light highlights the contrast between cream and coffee, casting soft amber reflections on the ceramic edge. The camera slowly pulls back from macro to medium shot, revealing a styled wooden café table with a sprig of cinnamon and a folded linen napkin. Commercial quality, appetizing warm color grading, 4K, shallow depth of field, photorealistic liquid dynamics. No text overlays, no watermark, no logo.
Recommended parameters: 9:16 · 8 seconds · 1080p
Key TechniquesSingle shot with a single camera movement (only pull back), using the scale change from extreme close-up → medium shot to create drama; slow motion + warm tones to enhance appetite appeal.
Adjustable PointsReplace tungsten side-light with morning window light to change to a morning feel; add with pour sound audible to indicate the audio path (note: Seedance 2.0 requires separate audio processing).
Reference Resources
- Advanced Formulas:methodology/02-advanced-formulas.md
- Masterclass (10 YouTube tutorials combined)：methodology/15-seedance-masterclass.md ⭐ 9-element formula / Timeline Prompting / 5 viral styles / VFX techniques / 8 templates
- Camera Movement Dictionary:methodology/05-camera-movement-dictionary.md
- Constraint Word List:methodology/06-constraint-word-list.md
- 117 tested prompts:prompts/seedance/README.md
Quality Suffix (add to every prompt)
Source:Masterclass §1 — The ULTIMATE Seedance 2.0 Prompting Guide
4K, ultra HD, rich detail, sharp clarity, cinematic textures, stable picture.
Maintaining face and clothing consistency without distortion or high detail.
Generate the video without subtitles.