image

Generate images from a text prompt using AI models. Supports Imagen models (dedicated image generation) and Gemini multimodal models (text+image output). The model type is auto-detected from the model ID: imagen-* uses the image API, everything else uses the text API with image output enabled.

Ask AI about this page

Usage

1egaki image <prompt>

Arguments

Argument	Required	Description
`<prompt>`	Yes	prompt

Options

Option	Default	Description
`-m, --model [model]`	-	Model ID for generation. If omitted, shows an interactive picker (or uses default in non-TTY mode)
`-o, --output [path]`	`egaki-output.png`	Output file path (index suffix added when generating multiple)
`-n, --count [n]`	`1`	Number of images to generate
`--aspect-ratio [ratio]`	-	Aspect ratio for the generated image. Imagen supports: 1:1, 3:4, 4:3, 9:16, 16:9. Gemini supports additional ratios: 2:3, 3:2, 4:5, 5:4, 21:9. OpenAI models (gpt-image-, dall-e-) have no native aspect ratio parameter, so the ratio is snapped to the closest supported size and stated in the prompt text
`--seed [seed]`	-	Seed for reproducible generation. Same seed + same prompt = same image
`--image-size [size]`	-	Output resolution for Gemini text-model image generation. Only applies to gemini--image models
`-i, --input [file]`	-	Reference image for editing or variations (repeatable). Accepts local file paths or URLs (http/https). Pass one or more images along with a text prompt to edit them
`--mask [file]`	-	Mask image for inpainting. Accepts a local file path or URL (http/https). White areas in the mask are replaced with generated content. Used together with --input
`--quality [level]`	-	Image quality level. xai: low, medium, high (default: medium). openai: auto, low, medium, high (default: auto)
`--resolution [res]`	-	Output resolution. xai: 1k, 2k (default: 1k)
`--output-format [format]`	-	Output image format. xai: png, jpeg (default: png). openai: png, jpeg, webp (default: png). fal: png, jpeg
`--negative-prompt [text]`	-	Describe what to avoid in the generated image (Fal models)
`--allow-people`	-	Allow generating images of people (Imagen blocks people by default)
`--json`	-	Output result metadata as JSON to stdout (model, usage, warnings, file paths)
`--stdout`	-	Write raw image bytes to stdout instead of saving to a file. Useful for piping to other tools

Global Options

Option	Default	Description
`-h, --help`	-	Display this message
`-v, --version`	-	Display version number

Examples

1# Generate a simple image

1egaki image "a sunset over mars"

1# Use a specific model with aspect ratio

1egaki image "cyberpunk city at night" -m imagen-4.0-ultra-generate-001 --aspect-ratio 16:9

1# Edit an existing image

1egaki image "add a wizard hat to the cat" --input cat.jpg -o cat-wizard.png

1# Edit an image from a URL

1egaki image "make it pop art" --input https://example.com/photo.jpg

1# Inpainting with a mask

1egaki image "fill with flowers" --input photo.jpg --mask mask.png

1# Generate with Gemini multimodal at 4K

1egaki image "dreamy landscape" -m gemini-2.5-flash-image --image-size 4K

1# Route through Vertex AI (Google Cloud billing)

1egaki image "product photo on marble" -m vertex/imagen-4.0-generate-001

1# Generate multiple images

1egaki image "abstract art" -n 4 -o art.png

1# Pipe to another tool

1egaki image "logo design" --stdout | convert - -resize 512x512 logo.png