URGENT UPDATE: Adobe’s Photoshop has just unveiled groundbreaking AI features that can swiftly restore damaged family photos, a process that once required painstaking hours of manual editing. With the introduction of Gemini 3 and Nano Banana Pro, users can now breathe new life into cherished memories with a simple prompt, revolutionizing the way we preserve our family history.
In a newly released video, Photoshop expert Morganti demonstrates how to utilize these advanced AI models through Generative Fill. This method allows users to select the entire frame of their photos and input concise commands like “repair, restore, and remove yellow color cast.” Such simplicity is a game changer for anyone looking to maintain the integrity of their family images.
The video highlights a critical comparison: while using Gemini 3, users can only generate one variation per attempt, requiring them to regenerate if the result isn’t satisfactory. This contrasts with Adobe’s Firefly models, which offer more versatility but may not achieve the same level of detail. Morganti emphasizes that the faces in restored images remain recognizable, a vital aspect often overlooked by other AI restoration tools.
A key moment in the demonstration showcases how sensitive these AI models are to prompt variations, particularly regarding color corrections. Morganti runs the same damaged print twice, first allowing Gemini 3 to eliminate the yellow stain, then adjusting the prompt to maintain the sepia tone. The side-by-side comparison vividly illustrates how neutralizing the yellow cast can alter the overall mood of the image, underscoring the importance of careful prompt selection.
As the video progresses, Morganti pushes the boundaries further by asking the model to “repair, restore, and colorize.” However, this introduces potential inaccuracies; hats change, costume details shift, and even the appearance of a dog can diverge from the original image. This serves as a crucial reminder that restoration and colorization require different levels of intervention, especially when historical accuracy is a priority.
Testing the AI’s capabilities further, Morganti applies the same techniques to a badly faded color photo and a personal family image from the end of World War II. The results are impressive, with Gemini 3 effectively cleaning up scratches and stabilizing tones without the need for extensive masking. However, Morganti urges users to manage their expectations, viewing this AI tool as a strong first pass rather than a complete solution for archival issues.
The emotional impact of reviving these family photos is profound, as these images often serve as the only visual record of our ancestors. As Morganti concludes, knowing when to accept a slightly imperfect restoration versus reverting to manual touch-ups is essential for successful photo preservation.
For those eager to learn more about this innovative process, check out Morganti’s full breakdown in the video linked above. With these advancements, preserving family memories has never been easier or more accessible.
