• Is the theme/content direction known? ✓ Skip
• Is the video purpose/publishing platform mentioned? ✓ Skip
• Is the duration/aspect ratio specified? ✓ Skip

• Simple Requirements (Clear Theme & Purpose): Ask all questions at once, 3-5 questions are sufficient
• Complex Requirements (Commercial Projects, Multi-scenarios): Split into two rounds, first ask core questions, then details
• When Users Are Clearly Confused: Ask at most 2 questions + provide guiding options. For remaining unknown information, make bold assumptions and mark them in the output, it's more efficient for users to revise after seeing the effect than to fill in blanks out of thin air

Handling Principle for Confused Users: Don't pile up questions due to incomplete information. First ask the most critical 1-2 questions, derive the rest based on the theme, and mark assumptions with "⚠️ Assumed to be X here, let me know if you need adjustments" in the output.

"What does this video mainly want to tell the audience? / What feeling or action do you want the viewers to have after watching it?" (Internal understanding: Narrative core, CTA, emotional goal)

"What kind of people will probably watch this video? Which platform will it be mainly published on?" Platform examples: Douyin/Kuaishou / WeChat Video Account / YouTube / Bilibili / Brand Official Website / Internal Presentation (Internal understanding: Target audience, platform tone, preference for vertical/horizontal screen)

"What is the expected total duration of the video?" Reference options: 15 seconds (ad hook) / 30 seconds (short commercial) / 60-90 seconds (standard short video) / 3-5 minutes (in-depth content) / longer (Internal understanding: Number of shots, narrative rhythm, duration budget per shot)

"Is the video vertical (for mobile scrolling) or horizontal (for computer/TV viewing)?" (Internal understanding: Aspect ratio 9:16 / 16:9 / 1:1, which affects composition and density of screen elements)

"Are there any videos that you think have the right vibe? Or any visual imagery in your mind?" (Internal understanding: Visual language reference, color tone, camera movement style)

1. What is the emotional state of the target audience? (Relaxed / Serious / Curious / Touching)
2. What kind of trust does this brand/content want to build?
3. How does the platform tone affect visual density?

Practical Mnemonic: Default to 3.5 characters/second for Chinese narration, which is the general benchmark for corporate promotional videos.

⚠️ This is the upper limit, not the target. Leave 20% breathing room: the actual number of line characters should not exceed 80% of the capacity, and the remaining time is for pauses, emotions, and screen breathing.

1. Check the lines first — Read the lines aloud and time them, it's more accurate than any formula
2. If lines exceed time: Choose one of two options
   • Cut lines: Remove modifiers, keep core information ("This product uses the latest advanced technology to provide you with an ultimate experience" → "This product uses the latest technology for an ultimate experience")
   • Extend shot: If the screen information is sufficient to support it, extend the shot duration
3. If lines are too short: Don't forcefully extend the duration — Short lines + silence + screen breathing are often more powerful than forcing extra words
4. Cross-shot lines: If a piece of narration spans multiple shots, clearly mark which part of the lines corresponds to which shot in the storyboard design to avoid audio-visual misalignment during editing

1. Screen Content — What is in this shot, what is the subject doing (specific, not templated)
2. Shot Distance — How much of the scene is shown
3. Shot Angle — From which angle to shoot
4. Camera Movement — Whether the camera moves, how it moves
5. Duration — How many seconds this shot lasts
6. Lines/Narration — What is said during this shot, check if the word count is within the duration capacity (required, write "Screen only, no narration" if there are no lines)
7. Atmosphere/Emotion — What feeling this shot wants to convey

• "The camera slowly pushes in, showing the overall environment"
• "Demonstrate the product's core features and reflect brand value"
• "The character's expression is natural, conveying positive emotions"

• "The thin spout of the pour-over kettle aligns with the center of the filter cup, water falls vertically from a height of 15cm, and the coffee powder swells into a small dome after being soaked"
• "The file size changes from 4.2MB to 312KB, this number change is stretched to 3 seconds in slow motion"
• "He stares at the terminal output showing successful deployment, his mouth doesn't move, but there's a glimmer in his eyes"

Usage: Calculate ASL first, then select BPM from the corresponding range. Do not reverse the order.

• Opening (Hook Section): Don't use full energy, leave room for escalation, or use silence + sudden entry to create impact
• Information/Content Section: Music recedes, serves as a background track, prioritize voice/content, lower the volume appropriately
• Climax/Turning Point: Music and screen advance simultaneously, drum beats or strings build up emotions, hit points align with editing cuts
• Closing/CTA: Volume fades out or ends with a clean sting, don't cut abruptly

Hit Point Principle: For emotional burst shot switches, product appearances, title entries, align the music's downbeat/drum hit with the editing cut — this is the core of professionalism.

Complete Suno prompt guide can be found in the "Suno AI Prompt Special Guide" section of

references/music-design.md

. The following is a quick operation framework for generating music scoring plans.

• Number of shots ≤ 8: Card-style shot-by-shot description (clear and easy to read)
• Number of shots 9-20: Markdown structured table + per-shot prompts
• Number of shots > 20: Group by narrative paragraphs, each group has a summary + shot details

• Total Duration: XX seconds / X minutes
• Aspect Ratio: 16:9 horizontal / 9:16 vertical
• Total Number of Shots: XX shots / Average Shot Length: X seconds
• Overall Visual Style: [One sentence describing the visual atmosphere]
• Music Scoring Direction: [Style + BPM range]
• Narration Word Count Budget: Total duration XX seconds × 3.5 characters/second × 80% ≈ Upper limit of XX characters (20% breathing space reserved)

• Equipment/Lens: [Recommended focal length, e.g., 85mm prime / 24mm wide-angle / macro lens]
• Lighting: [How to set up lights or use natural light, number of lights, direction, soft/hard]
• Shooting Key Points: [Key operations needed during actual shooting, e.g., focus tracking, keep stabilizer balanced, actor guidance]
• Post-Production Tips: [Color grading direction, speed adjustment, alternative angles to shoot as backups]

@Material Name

⚠️ Important Limitation: Seedance does not support negative prompts, don't write "don't include", use positive descriptions instead.

[Subject + Action] + [Scene/Environment] + [Lighting] + [Lens Language] + [Style/Texture] + [Image Quality Constraints]

A male independent developer wearing a white linen shirt, sitting in a dim coffee shop corner,
staring at the success prompt that just appeared on his MacBook screen, the corner of his mouth slightly lifting,
neon lights from outside the window filter in, mixing warm and cool light, close-up shot, camera slowly pushes in,
screen is stable without shaking, face is clear without distortion, cinematic feel, 4K HD.

Refer to the camera movement trajectory and rhythm of @Video 1,
place the product in @Image 1 into the same scene,
replace the background with a minimalist white workbench, cold white light shines directly from above,
camera slowly orbits around the product to emphasize craftsmanship details,
screen is stable, details are clear, Apple keynote-level product texture.

Extend @Video 1 by 10s, continue showing the side of the product,
camera slowly moves from the side to the back, lighting remains exactly the same as the previous segment,
movement is smooth and coherent, no frame jumps, natural transition with the previous segment.

Complete Seedance 2.0 usage guide (@ syntax, multimodal combination, long video workflow, troubleshooting checklist) can be found in

references/seedance-jimeng.md

1. Subject first, technical parameters later — AI models assign higher weight to words at the beginning, the more specific the subject description, the better
2. Avoid conflicting instructions — Don't write "handheld" and "perfectly stable" at the same time
3. Emotional words are effective — "melancholic", "euphoric", "tense" have actual impacts on AI generation
4. Numbers are more accurate than adjectives — "15cm high pour" is more accurate than "close to"; "drops from 4.2MB to 312KB" is more accurate than "file size decreases"
5. Avoid vague words — "beautiful" is ineffective, "warm golden backlight creating rim lighting on subject's hair" is effective
6. Speed must be clear — "slow motion 120fps" is clearer than "slow"; "real-time" is clearer than writing nothing
7. Negative prompts — Attach when necessary: "no text overlay, no watermark, no camera shake, no cartoon style"
8. Style anchors — Use film/brand aesthetics as anchors: "Wes Anderson symmetry", "Wong Kar-wai color grading", "Apple keynote aesthetic", "A24 film texture"

1. Must specify focal length — Don't say "use telephoto", say "85mm or 135mm prime lens, standing distance about 1.5 meters"
2. Lighting must be operable — Don't say "use warm light", say "One LED soft light placed at 45 degrees to the left, about 80cm away, with a softbox"
3. Actor/Subject Guidance — Describe expressions/actions specifically, e.g., "No need to smile, look at the upper right corner of the screen, stay still for 2 seconds"
4. Alternative Plans — Recommend one alternative shooting method for each key shot to prevent on-site accidents

• references/shot-types.md

  — Complete shot type library + prompt examples

• references/music-design.md

  — ASL/BPM quick reference, genre × usage mapping, Suno AI prompt special guide

• references/prompt-examples.md

  — Industry-classified video prompt examples (general AI video tools)

• references/seedance-jimeng.md

  — Complete Seedance 2.0 guide: multimodal @ reference, prompt formula, scene templates, long video workflow