This One AI Tool Makes Family Photo Restorations Even Better

Every now and then, a new photo editing tool that I want to highlight on this blog aligns with a real-world need in my life. Recently, I found myself scanning old family photos after the passing of my father, preserving them digitally for my family. While going through the collection, my mom came across a photo of me as a toddler and asked if I could clean it up for her to print. That request presented a great opportunity to explore a new improvement to Gigapixel 7, with its second-generation face recovery model. In this post, I’ll walk you through the steps I took to enhance and restore this almost 45-year-old photo using both Adobe Lightroom and Gigapixel AI.

Preparing the Photo in Lightroom

The first step in restoring an old scanned photo is to give it a solid foundation for editing. I began with a basic crop, adjusting the aspect ratio to suit the 5x7 print size my mom wanted. Even small changes like this can have a big impact on the final result when you’re preparing an image for print.

Once the crop was in place, I focused on cleaning up the photo using Lightroom's Content-Aware Remove Brush. While newer tools like Generative Remove are incredibly powerful, for minor touch-ups like eliminating small blemishes and spots on the photo, the standard brush worked perfectly. After a few minor corrections, I was ready to move on to editing the image's overall exposure and contrast.

Since this was a scan, I didn’t worry too much about getting the white balance perfect, but I did make some adjustments to highlights, shadows, and blacks to bring more life into the image. A quick S-curve helped add contrast, and a slight cooling of the temperature dialed back the warmth in the image.

Upscaling with Gigapixel AI

With the initial edits done in Lightroom, the next step was to upscale the image to print size using Gigapixel AI 7, particularly taking advantage of its Gen 2 Face Recovery model. This AI model offers twice the detail and resolution of the previous generation, which was essential for bringing clarity to my face in the old photo.

When you first upscale an image in Gigapixel, it’s important to balance sharpness with natural detail. In my case, I noticed that pushing the face recovery settings to 100% looked too artificial, so I backed it off to achieve a more natural look. One area where AI struggles is with non-facial elements like hair, and while the face recovery was excellent, there was a slight issue with hair detail that the software didn't fully resolve.

I also experimented with comparing four different models in Gigapixel, ultimately deciding that the Low Resolution v2 model provided the best balance of detail and fidelity for this image.

Final Touches in Lightroom

After exporting the upscaled image from Gigapixel, I returned to Lightroom for the final polishing touches. I added a subtle color grade by adjusting the shadow and highlight hues—cooler tones in the shadows and a hint of warmth in the highlights helped create a balanced, nostalgic feel. A small amount of texture and clarity enhanced the finer details, and a slight vignette brought focus to the center of the image.

Lastly, I finished by applying selective sharpening, zooming in to ensure that only the key areas, like my face, benefited from this additional detail.

Reflections on AI Tools in Photo Restoration

For anyone working with old, low-resolution photos, AI-powered tools like Gigapixel AI are becoming indispensable. This second-generation face recovery model is a perfect example of how technology can make significant strides in photo restoration, preserving important memories for future generations. While it’s not a one-click solution, combining these tools with thoughtful edits in apps like Lightroom can yield impressive results.

For photographers, enthusiasts, or anyone wanting to restore their own family photos, I highly recommend trying out this combination of tools. They offer a level of precision and ease that can really bring old photos back to life. And while AI might not be perfect in every aspect, the improvements are undeniable.