Love, Rendered is a documentary short that used artificial intelligence to reconstruct a personal memory for Burt and Ethelle Shatz, a couple married for more than 70 years. The filmmakers worked with Google DeepMind engineers and Primordial Soup, Darren Aronofsky’s creative venture, to animate a day the couple never recorded on film.
Recreating an unrecorded moment
The film follows the couple as they attempt to recover the memory of the day they met at a student co-op in Cleveland. Director Liz Garbus and producers Dan Cogan and Darren Aronofsky collaborated with technical lead Michael Chang to translate the couple’s recollections into a visual sequence. Ethelle participated directly in the process, correcting details such as the curve of a staircase or the shape of a shoe heel to align the reconstruction with her memories.
Technical approach and human guidance
The team combined two AI-driven methods to preserve what they describe as emotional truth. First, generative models were used to restore and colorize black-and-white photographs of Burt and Ethelle from their youth. Those restored images served as visual references for subsequent work.
Second, performance-capture models mapped present-day micro-mannerisms—head tilts, hesitations, and subtle facial crinkles—onto the younger likenesses. Michael Chang, the project’s technical lead, said mapping those traits onto the restored images helped bring the recreation to life. A colleague, Jess Gallegos, explained the workflow in more technical detail during production.
The creative team framed the project within reminiscence therapy practices, noting prior encounters—such as Garbus observing fMRI responses to familiar voices—that influenced their approach to memory and sensory cues. Darren Aronofsky emphasized that machine learning served as a tool guided by human direction throughout the process.
The filmmakers report that combining image restoration with pose and performance control generated a “memory” that Burt and Ethelle described as feeling authentic. The project was presented as an exploration of how emerging AI tools can help people reconnect with personal histories.
To try related tools, the Gemini app can restore and colorize family photographs. Users are advised to upload an image and ask: “Can you restore and colorize this photo? Preserve the appearance, expression, and pose of the people.”
Original source: Google AI