Image editor

This is a Replicate pipeline that edits and transforms images using simple text instructions.

https://replicate.com/pipelines-beta/image-editor

Features

• Find objects: Locate specific things in your images
• Remove backgrounds: Get rid of backgrounds automatically
• Fill in areas: Replace or add content to parts of an image
• Generate image descriptions: Get text descriptions of what’s in your images
• Improve resolution: Make images larger without losing quality
• Crop: Cut images to specific dimensions
• Convert formats: Change between different image file types
• Add text: Put text on images with various fonts
• Apply filters: Use ImageMagick for different image effects

Usage

Try prompts like:

• “Remove the background”
• “Crop to show just the dog”
• “Add ‘Happy Birthday’ in Roboto font at the top”
• “Make the image twice as big”
• “Turn this photo black and white”

You can combine multiple operations in one go:

• “Find the cat, crop around it, and add a colorful border”
• “Remove the background, make the image 1.5x bigger, and add ‘Summer Vacation’ text”

Models

Under the hood it uses these models:

• anthropic/claude-3.5-sonnet: Understands your instructions and writes code
• adirik/grounding-dino: Finds objects in images
• black-forest-labs/flux-fill-dev: Fills in or replaces parts of images
• cjwbw/rembg: Removes backgrounds
• nightmareai/real-esrgan: Upscales images

Tools

• Pillow: For basic image handling
• ImageMagick: For advanced image transformations
• Google Fonts: For adding text with different styles