Google AI Recreates Pelé's Greatest Goal From 1959 That Was Never Filmed

Google AI Brings Lost Football History to Life

On August 2, 1959, Pelé scored what he considered the greatest Pelé goal of his career at Rua Javari stadium in São Paulo, Brazil. The "Gol da Rua Javari" featured three consecutive sombreros over defenders, a knee flick past the goalkeeper, and a header into the net—all without the ball ever touching the ground1

1

. For 67 years, this legendary unfilmed goal existed only in the memories of fans who witnessed it and a single photograph by Raphael Herrera capturing the header moment2

2

. Google DeepMind has now completed a historical reconstruction of this lost moment using advanced generative AI tools.

Building the Foundation Through Historical Research

The project, developed in partnership with Pelé's family and the Pelé Brand, combined traditional filmmaking with three Google AI models: Veo 3, Gemini Omni, and Nano Banana Pro1

1

. Historian Anita Lucchesi gathered nearly 2,000 historical records, including stadium blueprints and family albums, while conducting eyewitness interviews with surviving fans who saw the goal decades ago1

1

. The team used Gemini Omni, a multimodal AI model launched at I/O 2026, to create an accurate picture of what 1959 São Paulo looked and felt like2

2

. This approach demonstrates how Google AI can serve cultural preservation rather than just synthetic content creation.

Recreating the Scene With Period-Accurate Details

Google sent a film crew to Javari Stadium, which still stands in São Paulo's Mooca neighborhood, and hired local actors to don period-accurate 1950s soccer jerseys. The team sourced a heavy leather ball matching the type used in the mid-20th century, along with Pelé's dark boots and authentic uniforms3

3

. Instead of traditional motion-capture suits, Google used a model called Performance Control, which takes input video and produces 3D blue-mesh renderings of people, then transfers that motion onto the AI video. "All these AI tools, they're such amplifiers of our imagination," says Tom Murray, a research engineer at Google DeepMind. "It allows us to really explore and experiment in a way we've just never been able to".

How AI Models Transformed Footage Into Vintage-Style Footage

The live-action footage was fed into the AI models, which replaced the stunt player with Pelé's digital likeness, restyled the modern stadium to match 1959 architecture and weather conditions, and generated period-appropriate crowd scenes1

1

. To ensure the AI-generated footage looked authentic, the digital output was run through a filmout machine to capture the distinct look and feel of 1950s cinema, then refined with traditional VFX for ball compositing and color grading1

1

. Nano Banana Pro, designed for precision image generation, worked at its most ambitious level—not generating content from nothing, but rebuilding a real event from fragments of evidence1

1

.

A New Application for AI That Recovers Rather Than Creates

The reconstruction now displays at the Pelé Museum in Santos, with Pelé's daughter Flávia Kurtz stating: "He would be so proud to see all this happening. He'd always say it was a shame that the goal was never recorded"1

1

. This project represents the most culturally significant application of Gemini Omni to date, showing that AI video generation can serve preservation purposes1

1

. In a year when AI has mostly been used to flood the internet with synthetic content, this stands as a rare case of it being used to recover something real that was lost1

1

. The fact that it happened with the consent of Pelé's family, and with support from footballers, fans, and real historians, separates it from typical AI deployments3

3

. This approach to recreate Pelé's greatest goal opens possibilities for using AI to preserve other lost cultural moments, suggesting a future where generative AI tools help document history rather than replace human creativity.