CLI Usage¶

Direktor provides a command-line interface for generating videos from text files.

Basic Usage¶

direktor <input_file> [options]

Arguments¶

Argument	Description
`input_file`	Path to the input text file (required)

Options¶

Option	Default	Description
`--stage`	`6`	Stage to run up to (1-6)
`--output`, `-o`	temp dir	Directory for the final `output.mp4`
`--temp-dir`	`temp/<hash>`	Temporary working directory
`--keywords-file`	-	JSON file with keyword overlays
`--resume` / `--no-resume`	`True`	Skip stages whose outputs already exist
`--clean`	`False`	Remove the temporary directory before starting
`--no-optimize`	`False`	Skip narrative content optimization
`--verbose`, `-v`	`False`	Enable DEBUG logging
`--version`	-	Show version and exit
`--help`	-	Show help message

Examples¶

Run Complete Pipeline¶

direktor article.txt

Run Specific Stages¶

# Only generate script (stage 1)
direktor article.txt --stage 1

# Generate script and audio (stages 1-2)
direktor article.txt --stage 2

# Generate through transcript (stages 1-3)
direktor article.txt --stage 3

Resume from a Stage¶

If you've already completed earlier stages, use --resume (the default):

# Resume from image prompts (stage 4)
direktor article.txt --stage 4

# Resume from image generation (stage 5)
direktor article.txt --stage 5

# Resume from video creation (stage 6)
direktor article.txt --stage 6

Custom Output and Keyword Overlays¶

# Write final video to a specific directory
direktor article.txt --output ./videos/

# Add keyword overlays from a JSON file
direktor article.txt --keywords-file keywords.json

The keyword file format is:

[["Introduction", 0, 10], ["Main Topic", 10, 60]]

Output Location¶

By default, outputs are stored in a temporary directory based on the input file's hash:

temp/<md5_hash>/
├── podcast_script.txt
├── audio.mp3
├── transcript.json
├── image_prompts.json
├── images/
└── output.mp4

Use --output to copy the final output.mp4 to a different directory.

Exit Codes¶

Code	Description
`0`	Success
`1`	Input file not found, missing environment variables, or pipeline failure

Environment¶

The CLI automatically loads environment variables from a .env file in the current directory.

# Create .env file
cp sample.env .env

# Edit with your API keys
nano .env

# Run
direktor input.txt