YouTube Image Generation
Inputs
Use script.md, brand style, aspect ratio, visual references, and any restrictions on people, logos, or copyrighted material.
Process
- Build a visual style guide for the video.
- Convert each scene into one or more image prompts.
- Separate final-frame visuals, B-roll, diagrams, icons, and thumbnail candidates.
- Include negative prompts and consistency notes.
- Generate images when an image tool is available; otherwise output prompts.
Prompt Pattern
Use this pattern:
[subject], [action or composition], [environment], [mood],
[camera/framing], [lighting], [style], [color notes],
16:9 YouTube video frame, high detail
Avoid: [negative prompt]
Output Format
Create image-prompts.md:
# Image Prompts
## Style Guide
## Scene Images
| Scene | Purpose | Prompt | Negative Prompt | File |
|---|---|---|---|---|
## Thumbnail Candidates
## Consistency Notes
## Checks
Quality Check
Every image must support a specific narration moment. Avoid decorative visuals that do not clarify, intensify, or advance the video.