AWS for M&E Blog

AWS brings generative AI to the FORMULA 1 AWS GRAND PRIX DU CANADA 2024

This year, Amazon Web Services (AWS) debuts the first-ever generative AI-inspired trophy at the FORMULA 1 AWS GRAND PRIX DU CANADA 2024 June 7 – 9 in Montreal, Canada. AWS designers and marketers collaborated with FORMULA 1 (F1) and the AWS Generative AI Innovation Center to explore trophy designs that capture the elegance of F1 tradition while leveraging AWS generative AI models. Using the Amazon Titan Image Generator foundation model in Amazon Bedrock, AWS created hundreds of trophy design concepts in minutes based on various prompts. Let’s explore how the team experimented with different prompts and used generative AI to design a world first for the FORMULA 1 AWS GRAND PRIX DU CANADA 2024.

The data-driven world of F1

F1 is a data-driven sport that blends human, machine, and technology with a single goal: win the race and raise the winner’s trophy on the podium. Across five continents, 10 teams in 20 cars compete in 24 races during an F1 season. At speeds over 200 mph (320 km/h), every blink of the eye creates up to 65 feet (20 meters) of darkness. Drivers experience up to 5g forces during cornering and braking, and train their heart to beat at 160 times a minute for over two hours. Behind each driver is a team of engineers and strategists analyzing data from their team’s two race cars. Each race car generates over 1.1 million data points per second from around 300 sensors. With split-second decision making, danger, and multi-million-dollar implications – a lot is happening on and off the track. And since 2018, F1 and AWS have innovated together to leverage F1 data and AWS’s cloud technologies to transform the sport. The partnership is built on a shared passion for technological innovation, which is why AWS turned to generative AI to create the FORMULA 1 AWS GRAND PRIX DU CANADA 2024 trophy.

Using generative AI on AWS to design the trophy

Because F1 is data-driven, AWS had a variety of unique storytelling opportunities to weave and integrate into the design of the trophy. Should the trophy reflect the Circuit Gilles Villeneuve, the iconic track in Montreal for the FORMULA 1 AWS GRAND PRIX DU CANADA 2024? Can the design represent the significant gravitational forces drivers experience? Can this trophy be a combination of all historic F1 trophies over the last seventy years? AWS designers and marketers collaborated with the AWS Generative AI Innovation Center—mirroring the same path any customer would follow to learn about what is possible with generative AI.

The working group started by using the Amazon Bedrock “Image Playground” in the AWS Management Console. They then leveraged Amazon Titan Image Generator to quickly brainstorm and ideate across multiple internal teams. AWS marketers and designers were able to adopt and immediately start creating designs quickly—even with limited prior experience using generative AI for design. In just seconds, AWS marketers and designers could tweak their prompts and other parameters to get updated and improved generative AI images. As certain prompts showed more promise, AWS leveraged the Amazon Bedrock API to generate hundreds of unique trophy image iterations in a matter of minutes. These images were then shared with stakeholders, enabling a collaborative decision-making process to determine the winning image that would serve as the basis for bringing the trophy to life.

In Amazon Bedrock, the team selected the Titan Image Generator model and chose the following configurations:

Mode: Generate image
Orientation: Landscape
Size: 1024×1024
Prompt Strength: Varied; 4-8
Seed: Varied; 0-2,147,483,646

Using natural language, AWS explored prompts focused on the cloud infrastructure used for F1 data, artistic representations of the Circuit Gilles Villeneuve, a combination of all historic GRAND PRIX DU CANADA trophies, and a trophy that embodied the millions of data points being transmitted from the car in real-time. Prompts like: “Design a trophy representing billions of data points” or “Design a trophy inspired by the flow of data from the car to the cloud” were turned into realistic, studio-quality images that could be edited and refined.

Prompt: Design an F1 trophy representing billions of data points transmitted from an F1 race car. Trophy materials should be silver. Prompt: Design a trophy inspired by the flow of data from a race car on the track to the cloud. Trophy materials should be silver.

If you’re just getting started with Amazon Titan, this video demo provides a good introduction

The designs were incredible, but had limitations around manufacturing feasibility, weight, and concerns around structural integrity. AWS explored alternative prompts and proprietary data to produce images consistent with AWS branding and to meet trophy design requirements set by F1.

Inspiration from the F1 next-gen car project

The most promising designs came from inputs inspired by the next-gen F1 race car project, specifically prompts focused on the use of computational fluid dynamics (CFD) and AWS High Performance Computing (HPC) to study the flow of fluids (i.e., the air around the F1 car) and test the aerodynamics of cars while racing.

Computational Fluid Dynamics (CFD) visualization used as inspiration for the final trophy design

To watch a full webinar on CFD for Motorsport, click here.

In an effort to generate closer wheel-to-wheel action on the track, F1 aerodynamicists focused on designing a car that can produce a smaller wake, while maintaining the required degree of downforce, reduce the adverse effects of driving through another car’s wake, and ensure peak speeds. The design changes included wheel wake control devices, a simplified front wing that diverts airflow off the front wheels, a more sculpted rear wing to effectively draw air in from the sides and lift it above the car following behind. Amazon Science dove deeper into the project in this blog.

The initial prompts that focused on this project started with “Design a trophy inspired by the use of Computational Fluid Dynamics and AWS High Power Computing”. Then the prompts were honed in on the wheel wake example, such as “Design a trophy mirroring an F1 car’s front wing”. Additional language around materials, trophy base, and background were added. Those designs were close, but the team continued to experiment until they found the right prompt.

AWS landed on the following prompt: “A sleek, aerodynamic silver trophy on a square marble base, its twisted airflow form resembling the intricate aerodynamic patterns around a Formula 1 race car. The trophy’s design incorporates swirls and vortices, symbolizing the complex computational fluid dynamics simulations run on AWS HPC used in F1 car development. The background is a muted grey, with subtle motion blur effects to convey a sense of speed.”

To create multiple images for ideation, the seed value—which determines the initial noise setting—was increased generating multiple distinct images for each numerical value. Amazon Titan Image supports a vast number of seeds, precisely 2^64 or approximately 18.4 quintillion unique seed values. By changing the seed value, AWS created hundreds of variations to review in less than 10 seconds—providing a wide array of options to choose from.

Trophy design variations based on single prompt with an increase in seeds

After internal reviews, the final design was selected for its stunning and unique aerodynamic design and signature airflow twist.

Generative AI-designed trophy using Amazon Titan Image Generator on Amazon Bedrock

Generative AI-designed trophy using Amazon Titan Image Generator on Amazon Bedrock

From generative AI to reality

The 2D trophy design image from Amazon Titan Image Generator was sent to a specialist 3D computer-aided design (CAD) designer. AWS and the CAD designer worked together to add subtle but intricate details to celebrate the race’s home country—like a maple leaf and waves to represent the St. Lawrence River. That resulted in a 3D render used to 3D print a full-size resin form. A traditional silversmith took the full-size resin form and over the course of 72 hours—and using a process known as electroforming—grew sterling silver on the resin form in a bath of cyanide. This process allowed for the silversmith to create the unique design accurately and the use of electroforming allowed for all four trophies: 1st Place Driver, 2nd Place Driver, 3rd Place Driver, and Winning Constructor’s Trophy to be identical versions of a single design. Then the silversmith finessed the shape of the trophy by hand, a process that dates back centuries, to form each individual detail. The result captures the rich and elegant tradition of F1 while using generative AI to accelerate innovation.

Image of silversmithing process during the trophy manufacturing process.

Design your own F1 trophy with generative AI

Our team had so much fun during this process that we want to give you the opportunity to use generative AI to design your own F1 trophy. With AWS PartyRock, a generative AI playground, you can easily experiment, create, and ideate on designs for your own F1 trophy.

Visit the PartyRock Trophy Sweepstakes and start designing today. Not only will you get to experiment with prompt engineering and learn about generative AI, but you can even enter to win a paid, all-access VIP trip to a 2025 FORMULA 1 GRAND PRIX. Entries are due by October 31, 2024, and the selected winner will be announced at AWS re:Invent 2024. Terms and conditions do apply.

So, what are you waiting for? Unleash your creativity, explore the power of generative AI, and design your own F1 trophy.

Doug Buser

Doug Buser

Doug Buser is a Senior Marketing Manager for AWS Media, Sports, and Entertainment. He works with AWS Sports customers and partners to share their data, ML/AI, IoT, and other workload journeys. Prior to joining Amazon, he led marketing and strategic communications for companies in the energy, travel, sports, and technology sectors.

David Green

David Green

David Green is a Principal, Product SA at Amazon Web Services. David works closely with customers and AWS product and service teams to help deliver easy to use services and features based on customer input and feedback.