AI Restores 70-Year Love Story in New Documentary “Love, Rendered”
The short documentary Love, Rendered follows Burt and Ethelle Shatz, a couple married for more than seven decades, as they confront Burt’s worsening memory loss. Google DeepMind teamed with filmmaker Liz Garbus, producer Dan Cogan and Darren Aronofsky’s Primordial Soup to recreate the day the pair first met—a moment that was never photographed. Using Google’s Gemini‑powered image‑restoration…
Key points
- Google DeepMind and Primordial Soup used Gemini image‑restoration and performance‑capture models to recreate the Shatzes' unrecorded past
- The workflow combined black‑and‑white photo restoration with AI‑driven pose mapping to preserve emotional truth
- The film illustrates AI‑enabled reminiscence therapy, showing a consumer feature in the Gemini app for photo restoration
The two‑step workflow—photo restoration followed by pose and performance control—allowed the AI to fill visual gaps while preserving emotional truth. The result is a frame‑by‑frame “memory” that the Shatzes describe as authentic, illustrating how generative tools can support reminiscence therapy and personal storytelling. The film also showcases the Gemini app’s consumer‑facing feature that lets users restore and colorize old photos with a simple prompt.
Beyond the personal narrative, the project demonstrates a practical use case for emerging generative models as creative mediums, highlighting how human guidance remains essential when AI is used to preserve fading histories.
Recreating a 70-year love story frame by frame
Google AI Blog · 9 September 2026
Recreating a 70-year love story frame by frame
“Love, Rendered,” a new documentary short, explores memory loss and the power of storytelling — and it showcases how technology can help rekindle memories that have begun to fade. The film follows Burt and Ethelle Shatz, a couple married for over 70 years, as they navigate Burt’s cognitive decline. Among his fading memories is one of their most precious: the day they met at a student co-op in Cleveland. Because it wasn’t photographed or filmed, the day existed only in their minds.
When I stepped in as the film’s technical lead, I knew the project would be as emotionally demanding as it was technically complex. Memory loss is personal to me. My grandfather suffered a stroke and memory loss before he passed. The last time I visited him, I was in my thirties. He was convinced I was still in college and was so happy I was graduating. That bittersweet memory never left me.
The experience made me want to see if this technology could help me connect more deeply with my family through the same medium. When this project began, I asked my father for old family photos. I tested our image restoration on photographs from when my parents met and used our video models to animate them. Watching my parents move as twenty-somethings really drove home how these tools could help keep what matters most from slipping away.
“Love, Rendered” was directed by Academy Award–nominated filmmaker Liz Garbus, who produced it alongside Dan Cogan and Darren Aronofsky. It was created in collaboration between Google DeepMind and Primordial Soup, Aronofsky’s creative venture.
Exploring the mystery of memories
Liz and Darren both came to the film curious about the resilience of memory. While directing the documentary “Coma,” Liz saw fMRI scans light up when patients in minimally conscious states heard familiar voices or were shown images of loved ones. Years later, Darren encountered footage of a former ballerina with Alzheimer’s who, upon hearing “Swan Lake,” instinctively danced the choreography from her wheelchair.
These touchstones led the creative team to reminiscence therapy, the clinical practice of using sensory cues like songs, family stories, and old photographs to stimulate memories, spark conversations, and rekindle emotional connections. But what happens when a memory isn’t attached to such cues?
To recreate Burt and Ethelle's unrecorded past, the filmmakers partnered with my team. Sitting alongside us, Ethelle became an active co-creator, correcting the curve of a staircase or the shape of a shoe heel.
Capturing human essence with AI
We worked alongside the team at Primordial Soup, leveraging emerging AI models as an art medium and toolkit, guiding the technology with human direction every step of the way.
Bridging the gaps between the missing details required a two-part technical approach designed to preserve emotional truth.
Image restoration: The team used generative models to restore black-and-white photos of Burt and Ethelle from their youth. The restored photographs helped to ensure that subsequent recreations remained faithful to their subjects.
Pose and performance control: Engineers used our performance capture models to map Burt and Ethelle’s present-day micro-mannerisms — the specific tilt of Burt’s head, a brief hesitation in his speech pattern, the subtle crinkle around his eyes. Mapping those traits onto their younger likenesses brought the recreation to life.
Combining these two mediums allowed us to intertwine Burt and Ethelle’s past and their present. In doing so, we generated a “memory” that they told us felt authentic.
See my colleague Jess Gallegos explain the workflow in more detail.
As Darren noted during production, a tool — like a paintbrush or a hammer — does nothing until it’s guided by human hands. In “Love, Rendered,” machine learning was the tool we used to guide Burt and Ethelle back through time, helping them hold onto a vanishing moment.
You can watch the full film here to experience their love story, frame by frame.
Preserving your family’s story
Whether you’re looking to connect with a grandparent or preserve your family’s core memories, you can restore old photos in the Gemini app. Upload an image of the photo and ask Gemini: “Can you restore and colorize this photo? Preserve the appearance, expression, and pose of the people.”
This text was published by Google AI Blog and written by Michael Chang. It is reproduced here with attribution so you can read it in full; the rights remain with the publisher. Read it at the source ↗
The headline, key points and digest above were generated by Digest AI's editorial model from the linked sources. Automated summaries can contain errors: the sources are the record. Spotted a mistake? Tell us.
Comments
via GitHub Discussions