LESSON 5 · PROMPTING

The prompt formula: Subject → Action → Camera → Style

A good prompt isn't a pile of pretty epithets — it's a precise technical brief for the model. Seedance 2.0 has a working structure it uses to "read" your scene. Master it once, and you'll get predictable results instead of a lottery.

The base formula of 7 blocks

Every strong prompt is built from seven meaningful blocks. They follow a logical order — from the main subject to the technical constraints. You don't have to fill in all seven for every clip, but the first two are the foundation, and without them the scene drifts.

1
Subject
Who or what is in frame: age, material, type of object
2
Action
One action, a verb in the present tense
3
Environment
Where the scene takes place: location, background, atmosphere
4
Camera
Shot size, movement, angle, lens
5
Lighting
Source and quality of light, contrast
6
Style
Visual language: cinematic, commercial, color grade
7
Constraints
Quality and stability constraints
🔑 The first 20–30 words rule
The model gives the most weight to the beginning of the prompt. The first 20–30 words decide what the scene will be. So put Subject and Action right at the start — those are the two accented blocks. Add the technical side (camera, lighting, constraints) afterward: it refines, but it doesn't define the essence of the frame.

Four rules that always work

  • One action per shot. One verb — one movement. If your heroine "walks, turns, and smiles," that's three actions and three blurry results. Pick one.
  • Verb in the present tense. "rolls," "falls," "turns" — the model reads the present tense as something happening right now, in frame.
  • Clarity beats intensity. A stack of adjectives ("dramatic, stylish, premium, beautiful") doesn't amplify the scene — it confuses the model. One precise detail is stronger than five generic ones.
  • Phrase it positively. Describe what should be in frame, not what shouldn't. Save the prohibitions for the quality-constraints block.
✅ The 3000-character limit
The prompt field holds up to 3000 characters. That's plenty — but don't stretch the text just for length. A long prompt without structure works worse than a short, focused one. Spend your characters on specifics, not on filler.

A compact template — copy it and fill it in

Keep this skeleton handy. Fill the square brackets with your own details, and you'll always have a working prompt with the right structure.

Subject: [who/what — age, material, type]
Action: [one present-tense verb — a single motion]
Camera: [shot size] + [movement] + [angle] + [lens]
Style: [visual anchor] + [lighting] + [color grade]

Weak vs. strong: one scene, two results

Let's compare two prompts for the same task — a commercial spot with a perfume bottle. The difference isn't length, it's the presence of workable directions.

A cool cinematic ad for a perfume bottle, dramatic, stylish, premium, beautiful lighting. Wide camera, realistic.

Why it's weak. There isn't a single workable direction here. "cool," "dramatic," "stylish," "premium" are judgments, not instructions: the model doesn't know what to actually do. There's no action in frame, no specific camera movement, no description of light or background. "Wide camera, realistic" is far too vague to set anything. The result will be random from attempt to attempt.

✓ Strong — result
A clear glass perfume bottle sits on a black stone pedestal in a dark studio. Condensation rolls slowly down the glass as a single drop falls onto the surface below. Medium close-up with a slow circular dolly move. Soft side lighting, high contrast reflections, luxury beauty-ad look. Subtle room tone and one sharp glass tap. Stable product shape, clean label detail, natural motion.

What changed in the strong version. The Subject became concrete ("clear glass perfume bottle" on a "black stone pedestal"), one clear action in the present tense appeared ("condensation rolls… a single drop falls"), the camera got a shot size and a movement ("medium close-up… slow circular dolly"), the light is described by quality ("soft side lighting, high contrast reflections"), the style is set by an anchor ("luxury beauty-ad look"), and at the end come the stability constraints ("stable product shape, clean label detail").

Block-by-block breakdown

BlockIn the strong prompt
Subjectclear glass perfume bottle on a black stone pedestal
Actioncondensation rolls slowly, a single drop falls
Environmenta dark studio
Cameramedium close-up, slow circular dolly move
Lightingsoft side lighting, high contrast reflections
Styleluxury beauty-ad look
Constraintsstable product shape, clean label detail, natural motion
✅ Do
  • Put Subject and Action in the first 20–30 words.
  • One verb in the present tense per shot.
  • Describe light by quality, not with the word "beautiful."
  • Close the prompt with stability constraints.
  • Use concrete nouns and materials.
⚠️ Avoid
  • Stacks of adjectives without specifics.
  • Multiple actions in a single shot.
  • Judgments ("cool," "premium") instead of instructions.
  • Vague camera directions like "wide camera."
  • Long text just for the sake of length.

This formula is the foundation for the whole course. In the coming lessons you'll build on it: adding references by tag, splitting a scene into several shots, and constructing transformation arcs. But the foundation is always the same — Subject → Action → Camera → Style.

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