Content To Video

Content to Video

Convert various content sources (articles, papers, GitHub repositories, etc.) into professional videos with structured research, scriptwriting, and Remotion-based generation workflow.

Workflow

1. Gather Requirements

You need to first check the remotion-best-practices skill, which is the core skill for video generation

Ask the user:

• Content Source: URL, file path, or topic description
• Target Audience: Developers, general public, students, etc.
• Video Style: Educational, promotional, documentary, tutorial
• Visual Style: Apple keynote style (minimalist tech), Tech startup pitch style (TED / startup launch), Internet product promo style (Figma / Notion style), Documentary style (Netflix style), Fast-paced short video style (TikTok / Reals style)
• Format: Landscape (16:9), Portrait (9:16), Square (1:1)
• Language: For voiceover and subtitles, whether multiple language versions of the video are needed

Create project directory: <topic-slug>/ Write all user requirements to <topic-slug>/video-style.md

2. Deep Research

Create <topic-slug>/deep-research/ directory to store:

• Source content (downloaded articles, scraped web pages)
• Resources extracted from source content: images, videos (downloaded, AI-generated, screenshots)
• Additional resource generation and creation
• BGM options (search and download, see free-resource skill for details)
• Key insights and data points
• Finally generate a Readme.md for this directory for later use

Processing by Source Type:

• URLs: Fetch full content, extract key points, download related images
• Papers: Parse PDF, extract abstract/methods/results, generate visualizations
• GitHub: Analyze repository structure, README, code highlights, generate diagrams

See references/research-workflow_CN.md for details.

3. Plan Video Structure

Create <topic-slug>/video-sessions/ directory with subdirectories:

• session-01/, session-02/, etc.

Each session contains:

• config.json - Session configuration (see schema below)
• assets/ - Session-specific images, videos, audio

Session config.json schema:

{
  "title": "Session Title",
  "scripts": [
    {
      "type": "composition_type",
      "content": "Text or content description",
      "transition": "fade",
      "resources": ["path/to/image.png", "path/to/video.mp4"],
      "narration": "Voiceover text",
      "prompts": "AI visual layout prompts, style to display, elements and text to generate, detailed prompts for resources used"
    }
  ]
}

See references/session-config-schema_CN.md for full schema.

4. User Review

Present complete video structure:

• Total duration and session breakdown
• Session titles and key points
• Sample voiceover text
• Expected visual style

Get user approval before proceeding. Allow iteration on:

• Session order and content
• Voiceover tone and length
• Visual style preferences

5. Generate with Remotion

Create <topic-slug>/remotion-video/ directory:

• Initialize Remotion project
• Create session compositions: src/compositions/session-01/, etc.
• Generate audio voiceover (TTS or user-provided), use qwen3-audio skill, generated audio returns duration, note to reserve blank transition time before and after each session switch, must wait for audio to finish before playing next session or animation
• Add assets to public/assets/
• Configure transitions and effects

Session Structure:

session-01/
├── index.tsx          - Main composition
├── components/        - Session-specific components
├── assets/           - Audio, images
└── config.ts         - Session configuration

See references/remotion-patterns_CN.md for common patterns.

Directory Structure

Final project structure:

<topic-slug>/
├── deep-research/
│   ├── content/
│   ├── images/
│   ├── bgm/
│   └── notes.md
├── video-sessions/
│   ├── session-01/
│   │   ├── config.json
│   │   └── assets/
│   ├── session-02/
│   └── ...
├── remotion-video/
│   ├── src/
│   ├── public/
│   └── package.json
└── output/
    └── final-video.mp4

Key Considerations

• Token Efficiency: Research phase may consume many tokens. Use incremental saving and summarization.
• Parallel Work: Can prepare Remotion templates while researching.
• Iteration: Expect 2-3 rounds of user feedback on scripts before video generation.
• Quality vs Speed: Ask user preference upfront - quick prototype or polished production.