Google Reconstructs Pelé’s Legendary Goal Using AI
It’s a goal that was never filmed, until now
Google DeepMind has successfully reconstructed Pelé’s iconic 1959 goal using Veo, Gemini Omni, and Nano Banana Pro AI models. The project, a collaboration with Pelé’s family, brought together historical records, eyewitness interviews, and traditional filmmaking techniques.
Pelé scored a remarkable goal in 1959—a three-sombrero sequence over defenders, followed by a knee flick and a header into the net, all without the ball touching the ground. This "lost goal" remained in the memories of those who witnessed it for 67 years.
The reconstruction project:
- Involved nearly 2,000 historical records and interviews with surviving eyewitnesses.
- Combined live-action footage shot at the original stadium with AI models to recreate Pelé’s likeness and the 1959 atmosphere.
- Resulted in a film that captures the essence of the "Gol da Rua Javari," now on display at the Pelé Museum.
Google AI models used:
- Veo 3: For generating realistic footage.
- Gemini Omni: A conversational video-generation model, demonstrating its cultural significance in preserving historical moments.
- Nano Banana Pro: Designed for precise image generation, crucial in rebuilding the real event from available evidence.
"He would be so proud to see all this happening," said Flávia Kurtz, Pelé’s daughter, referring to her father’s absence due to his passing but his enduring legacy.
In a year marked by AI-generated content, this project showcases a unique application: recovering a lost historical event rather than creating new content.