Frozenpepper Carves Out FP-Background_Obliterator For Local AI Cutouts

    
        By vramkickedin    
     | 
    
            May 28, 2026 at 10:58 am        
    
     | 
    
        2 min read

FP-Background_Obliterator is a new open-source tool that combines advanced AI background removal with a full layer-based editing environment, all running locally on your machine. It provides both a rich desktop GUI and a headless MCP server, making it accessible for manual editing or automated agent workflows. Powered by the InSPyReNet neural network and OpenCV, it delivers professional-grade cutouts with real-time edge refinement controls.

Created by developer Frozenpepper, the project was originally a personal workflow tool for high-quality background removal that outperformed local Photoshop. After using it privately for nearly a year, Frozenpepper repurposed the tool as a headless service for AI coding agents and has now released it to the open-source community. The source code and documentation were cleaned up with help from Google’s Gemini Flash model.

Two modes for every workflow

Key Features

Professional AI masking via InSPyReNet.
Live edge adjustment: threshold, dilate, blur.
Layer stacks, folders, and opacity management.
Rectangular selections with copy/paste.
MCP server connects Claude, Cursor, Windsurf.
Accelerated on Apple MPS, NVIDIA CUDA, or CPU.

Photographers, designers, and small agencies who need rapid local background removal can use the desktop app for precise manual edits. Privacy-conscious professionals benefit from never sending images to cloud services, while developers and AI enthusiasts can integrate the MCP server into agent-driven image pipelines. Even hobbyists with consumer GPUs will find the hardware-accelerated processing fast and accessible.

First open-source release notes

Frozenpepper notes that this is their first open-source release, and the project’s README and code cleanup were generated with the help of Google’s Gemini Flash model. The tool’s InSPyReNet model weights are subject to their respective creators’ licensing, though the project code itself is under MIT. While no formal releases have been published yet, the repository includes automated setup scripts for macOS, Linux, and Windows to streamline local installation.