SuperAPI AI Directory
Partner中文AI newsFavorites
Skillsseedance-prompt

seedance-prompt

Agents & models

View repository

Updated 10-16 02:39

About this skill

Seedance 2.0 Product Promo Animation Prompt Generator

Interaction Flow

Step 1: Collect Information

Use AskUserQuestion to ask the user for the following information (all at once):

  1. Product Information: What is your product called? Please briefly describe the product positioning and core features.
  2. Specific Requirements: What do you want the video to focus on? (e.g., a certain feature, brand tone, technical characteristics)
  3. Image Assets: Do you have any image assets to provide? (product images, logo images, interface screenshots, etc.)
    • If so, ask the user to send the images directly or provide the image paths
  4. Voiceover Requirements: Is voiceover needed? If so, please specify the language and the general content direction

Step 2: Analyze Assets

  • Read the product description provided by the user and extract the core features
  • If the user provided images, use the Read tool to view the image content and analyze the logo structure, product interface layout, brand colors, etc.
  • Based on the product positioning, select the most suitable style (A/B/C/D) from the aesthetic style matrix

Step 3: Generate the Prompt

Read references/prompt-system.md to get the complete prompt system rules, including:

  • Detailed description of the aesthetic style matrix
  • Strict writing rules (word count limits, camera constraints, morphing transitions, etc.)
  • Timestamp structure and output template

Generate the final prompt according to the rules and template in prompt-system.md.

Step 4: Image Reference Guidance (if reference images are involved)

Seedance 2.0 supports the "image reference" (reference image) feature. If the generated prompt needs to use the images provided by the user as references:

  1. Tell the user to select "Omni Reference" mode in Seedance 2.0
  2. Clearly annotate the placeholder and purpose of each image, for example:
    • [图片一] → Logo image, used for the logo form reference at 9-12s
    • [图片二] → Product interface screenshot, used for the interface form reference at 2-5s
  3. Explain that the user needs to drag the corresponding images into the corresponding positions in the Seedance interface

Step 5: Output and Check

  • Output the prompt in pure English
  • Confirm the character count is between 1000-2000 (verify with wc -c )
  • If it exceeds 2000 characters, condense it and output again
  • The prompt should include Music and Sound Design descriptions; if the user needs voiceover, it should also include a Voiceover script
  • Attach a Chinese explanation: the chosen style, the design intent for each time segment, and the image reference position explanation

Step 6: Dreamina CLI Video Generation

After the prompt output is complete, check whether the user has installed the Dreamina CLI:

  1. Check installation status: Run which dreamina or dreamina --version to check whether it is installed
  2. Guide based on the result：
    • Not installed: Use AskUserQuestion to ask the user whether they want to install the Dreamina CLI to generate videos directly. If the user agrees, execute the installation command: 
      curl -fsSL https://jimeng.jianying.com/cli | bash
      After installation is complete, guide the user to use the CLI to generate videos.
    • Installed: Use AskUserQuestion to ask the user whether they want to directly use the Dreamina CLI to generate this video. If the user agrees, use the dreamina CLI together with the prompt just generated to generate the video.

Note: If the prompt involves image referencing, the image path provided by the user needs to be passed as a parameter to the CLI.