Generate cinematic video prompts from text, books, and images using AI
npm install scenecraft
Transform any content into cinematic video prompts
Installation •
Quick Start •
Features •
Usage •
API
---
SceneCraft generates professional video prompts optimized for AI video generators like Veo, Runway, Pika, and more. Feed it text, books, or images and get back detailed, cinematic scene descriptions ready to paste.
`` | _| _|
___ ___ __ _
/ __| __ ___ _ _ ___ / __|_ _ __ _ / _| |_
\__ \/ _/ -_) ' \/ -_)| (__| '_/ _
|___/\__\___|_||_\___| \___|_| \__,_|_| \__|
``
| Feature | Description |
|---------|-------------|
| Multi-Input | Text, PDF, EPUB, Markdown, JPG, PNG, GIF, WebP |
| Smart Prompts | Camera, lighting, mood, dialogue included |
| 4 Styles | Cinematic, Documentary, Artistic, Minimal |
| Interactive | Browse scenes, one-click copy to clipboard |
| CLI Mode | Pipe-friendly, JSON/Markdown output |
| Gemini 3 Pro | Powered by the latest AI model |
`bash`
npm install -g scenecraft
Or run directly:
`bash`
npx scenecraft
- Node.js 20+
- Gemini API Key - Get one free
`bash`
scenecraft setup
`bashInteractive mode (recommended)
scenecraft
$3
In interactive mode, press
C to copy any scene prompt to your clipboard.Usage
$3
Just run
scenecraft with no arguments for the full interactive experience:`bash
scenecraft
`You'll be guided through:
1. Choosing input (file or text)
2. Setting scene count (1-20)
3. Picking a visual style
4. Selecting prompt elements
Then browse your scenes and copy with one keystroke.
$3
For scripting and automation:
`bash
Basic generation
scenecraft generate -f book.pdf -s 10Specific style
scenecraft generate -f story.txt --style documentaryJSON output for processing
scenecraft generate -f input.epub --format json -o scenes.jsonMinimal prompts (no camera/lighting)
scenecraft generate -f text.md --no-camera --no-lightingInclude character dialogue
scenecraft generate -f script.txt --dialogue
`$3
| Flag | Description |
|------|-------------|
|
-f, --file | Input file (txt, md, pdf, epub, jpg, png, gif, webp) |
| -t, --text | Direct text input |
| -s, --scenes | Number of scenes (1-20, default: 5) |
| --style