Topaz Photo AI switched to a subscription model, and the backlash was predictable. Now a developer named Vinicius Egidio is building Open Photo AI (OPAI), a free, open-source tool that runs entirely on your machine with no cloud uploads or recurring fees.
This isn't a ChatGPT wrapper. All the AI inference (the actual computation that enhances your photos) runs locally using ONNX Runtime, with optional TensorRT acceleration on Nvidia GPUs. The project ships six models, each named after a city and tuned for a specific job: Tokyo and Kyoto handle upscaling for photos, Saitama specializes in cartoons and illustrations, Athens and Santorini do face recovery with different aggressiveness levels, and Paris fixes lighting problems like backlit or underexposed shots. There's a GUI with batch processing and an "autopilot" mode that picks enhancements automatically. A CLI is coming.
The models are pulled from existing open-source AI research and documented on Hugging Face. It runs on Windows, macOS, and Linux across both AMD64 and ARM64 architectures. The whole thing is built with Go and Wails 3+, licensed under AGPL-3.0.
Is it going to match Topaz's quality right now? Probably not. Topaz has a full engineering team and years of model training behind it. But for anyone who resents paying a monthly fee to upscale photos they already own, this is a real option that's actively developed (141 commits and a Kickstarter campaign running). The local-only approach also means your photos never leave your computer, which matters if you're working with client images or sensitive material.