Inpaint masked image regions via RunComfy CLI with JSON input.
npx skills add https://github.com/runcomfy-com/skills --skill image-inpainting
Please help me install this Agent Skill. Skill: image-inpainting Source: https://github.com/runcomfy-com/skills/tree/main/image-inpainting Command: npx skills add https://github.com/runcomfy-com/skills --skill image-inpainting
It solves the problem of precisely editing or removing unwanted regions in an image, such as deleting objects or watermarks without disturbing the rest of the photo.
Use the image-inpainting skill to replace a masked object by providing the source image URL, a grayscale mask URL, and a prompt describing what should appear in the masked area.
High-intent search queries and answers about installing and using this skill.
You can remove unwanted objects by providing the source image URL alongside a valid grayscale mask URL, where white areas indicate regions to inpaint and black areas preserve original content, routing the payload via the runcomfy CLI.
Yes, you can edit a specific region without a mask by using the description-based fallback. This allows you to target a region described in words, automatically routing the request to an appropriate edit model instead of relying on a grayscale mask.
The best way to remove watermarks programmatically is using mask-driven inpainting via the runcomfy CLI. You provide a JSON input payload containing the image URL and a grayscale mask URL defining the watermark area to clean up while preserving surrounding content.
Yes, you need a valid grayscale mask URL for image inpainting. The mask must use white to define the regions you want to inpaint or replace, and black to indicate the surrounding content that should be preserved during the local edit.
Model selection works through controlled routing based on input availability. The system automatically selects the best inpainting model when a grayscale mask is provided, or switches to a description-based edit model when masks are unavailable.
The limitation of description-based image editing is that it relies on a fallback model when a grayscale mask is unavailable. Without a precise binary mask, the edit model targets regions based on word descriptions, which may offer less controlled local replacements compared to exact masking.
prime-skillsRoute image editing requests to the appropriate RunComfy model and execute them.
prime-skillsRemove or replace unwanted image regions using masks or descriptions.
s1366560Generate and edit images with OpenAI's GPT Image 1.5 model.
agentspace-soPerform mask-driven image inpainting to remove or replace image regions.
