The restoration of old photographs has long challenged AI systems. While tools have excelled in creating realistic images from text prompts, restoring and enhancing aged photos with fidelity remained elusive. That’s changing with the arrival of Flux.1 Kontext, a new generative AI model developed by Black Forest Labs.
A generative model focused on image context
Unlike traditional text-to-image generators, Flux.1 Kontext uses a combination of reference images and text promptsto create new visuals that respect the original context. This approach allows the model to preserve key features—faces, objects, landscapes—while correcting damage or improving detail. By simply uploading an old or damaged photo and providing a basic instruction like “restore and colorize,” users get results in seconds.
Stunning results with some trade-offs
Restorations are often visually striking, adding color and clarity to black-and-white or degraded images. In a test with a 1934 photograph of the Real Madrid football team, the AI added lifelike tones and enhanced detail without compromising historical authenticity. However, zooming in may reveal subtle distortions, such as altered facial proportions or invented features—an issue common to generative models known as “hallucinations.”
Available now on Freepik and other platforms
Flux.1 Kontext is already integrated into platforms like Freepik and Fal.ai, offering free trials and paid options. On Freepik, a premium subscription gives users access to up to 180 photo restorations per month for just €10. While the model is not yet open source like other Black Forest Labs tools, its accessibility makes it one of the most powerful restoration solutions available today.