Your photos. True colors. No cloud.
FlipsiColor corrects and restores photos and videos with local AI models — RAW support, lens correction, upscaling and face restoration, all running offline on your own machine.
Free & open source (GPL-3.0) · v0.8.1 · nothing leaves your machine
What it does
A photo pipeline that stays home
FlipsiColor bundles everything a photo or video needs after the camera: correction, restoration, enhancement. The AI models run on your hardware — your images are never uploaded anywhere.
AI color correction
Automatic color restoration powered by NAFNet, Restormer and AiLUT models — natural-looking results.
RAW support
Canon CR2/CR3, Nikon NEF, Sony ARW and DNG via LibRaw, with Lensfun lens correction built in.
Video color correction
Frame-accurate grading with scene detection and preserved audio — VapourSynth or FFmpeg backend.
Low-light enhancement
Automatic brightening of dark images and videos with 11 classic algorithms and a smart auto mode.
AI upscaling & face restoration
RealESRGAN upscales up to 4x, CodeFormer brings faces back — both individually switchable.
13 languages, dark & light
DE, EN, ES, FR, IT, NL, PL, PT, TR, RU, ZH, JA, KO — theme switching with live preview.
Downloads
Latest release: v0.8.1
Older releases & portable versions: GitHub Releases
Get involved
Testers shape the roadmap
FlipsiColor improves fastest with feedback from real users. The Linux build especially needs eyes on it right now — it is actively developed, but barely tested outside my own setups, so edge cases are almost guaranteed to exist.
Every bug report, edge case or "works on my setup" helps. Testers and feedback are explicitly welcome — whether you find a crash, a visual glitch, or just want to suggest a feature.
In short
What is FlipsiColor?
FlipsiColor is a free, open-source photo and video editor (GPL-3.0) for Windows and Linux that corrects, restores and enhances images using AI models that run entirely on your own computer. It reads RAW files from Canon (CR2/CR3), Nikon (NEF), Sony (ARW) and generic DNG, applies lens correction via Lensfun, grades videos frame by frame with scene detection, upscales images up to 4x with RealESRGAN, restores faces with CodeFormer and brightens dark shots with automatic low-light enhancement. All AI models run offline on the user's own hardware — no upload, no account, no subscription.
FAQ
Common questions
Does it upload my photos anywhere? No. Corrections run on your own hardware. The only network traffic is the one-time download of the AI models from the project's GitHub Releases.
Is it really free? Yes, free and open source under GPL-3.0. There is no paid tier, watermark or usage limit.
Which file types are supported? JPEG, PNG, WebP and common RAW formats (CR2, CR3, NEF, ARW, DNG) for photos; common video formats for grading with preserved audio.
Do I need a powerful GPU? No. The CPU-mode handles the pipeline; an NVIDIA/AMD GPU only speeds the AI steps up.
How is it different from Darktable or Lightroom? Those are manual editors with full control per slider. FlipsiColor automates the correction with AI and stays fully offline — a complement, not a replacement.
Context
Where it fits
Cloud AI editors. Convenient, but your photos travel to someone else's server — often with compression and subscription fees.
Manual editors (Darktable, Lightroom). Full control, but every correction is manual work.
FlipsiColor. The middle path: AI does the heavy lifting, everything runs locally and offline, RAW and video included — free under GPL-3.0.