IBM x Ferrari
I worked with Field to help visualise, prototype and create a Data-Visualisation in Unreal for a real-time racing simulator in Unreal 5. For more information visit: https://field.io/work/ibm-ferrari. Credits at the bottom of the page
This is beyond a driving simulator, it’s a demonstration of speed + intelligence. When fans sit at the wheel, they grasp what we discovered with IBM + Ferrari — that dynamic intelligent systems can turn raw telemetry into rapid feedback + real-time understanding.
R&D
There was a huge amount of R&D in the project and this is just a slice of everything I contributed, but of course there were so many talented people on the team that made this project come to fruition.
During the R&D stage I used Reinforcement Learning to train some AI car to navigate the race track while we were waiting for some live data. This meant we could develop and test ideas without having to wait for the data to be recorded.
Testing out camera ideas with just the AI cars.
One concept that popped up was having multiple trails of past recordings display at the same time, so I used the AI cars to experiment given that will perform differently every run.
As with most Field projects, data visualisation is really important part of the process. We were looking into ways how we could visualise the car data in Unreal and this was a small exploration at doing real-time smoke that would be generated from the car braking/moving.
Transparent materials became an strong idea at one stage of the project, so I was looking into developing some materials that would in the brake data and be visualised on a transparent surface on top of the material.
A paint stroke aesthetic became a big part of the discussion for the design of the environment. I development a system and integrated a look into Unreal where we could blend between paintstrokes on side of the car and also top down a
Using Houdini, I created a tool that would generate the strokes around the sides of the track and larger ones top down on the surface of the track for higher wider shots. This meant it was easy to art direct and make changes across the whole track directly in Unreal. The setup smoothed out the strokes around the edges, where sometimes you would get weird intersections if the turn is too tight.
The paint stroke material had it’s own functionality where I could control the speed of the movement of the paintstrokes, this was linked to the car speed, so when the car was moving we could exaggerate the feeling of the speed by speeding up the paint strokes.
Dorian Gourg were working with C4D to do the style frames and we were curious how we could translate the designs across to Unreal. I used Datasmith to bring across the Redshift materials and the we got about 90% of the look in Unreal without much further work.
The trails were a key feature of the project and we were experimenting in Unreal how they could be visualised while also being influenced by the race data. In this test I’m generating some trails from the car and using the brake data in the colour
I also looked at developing some trails for the top down view that had a 3D trail element that we could play with different angles
Not the final version of the project, but some lookdev I was doing for the car. At this point the wheels are still not reacting to the ground properly, but we had a driver rigged who would steer. Model and Textures by Elliot Brett.
Prototypes
As in every R&D stage there is a lot of development and concepts that don’t make it to the final product. Here are just some rough prototypes I built to visualise and test the simulator early on in the development stage.
This a flow diagram and some system thinking of how the whole Unreal side of the project would work - how the race data flows into Unreal and influences different components such as the materials and camera systems.
In the early stages of the project, we were playing with the idea that visualisation would all be on one screen. I was tasked with mocking up prototypes so we could creatively review ideas in situ. These prototypes were less technically focused and more about creative beats.
At one point we played with the idea of having a cinematic intro and some sliding transitions. This prototype was created to visualise the whole experience during development stage so that we could string together a lot of ideas.
This prototype focused on cameras and cinematic style. Working with creative director Vincent Venema and 3D designer Sam Care, we collaborated ideas on how we could bring together a cinematic vision.
I was responsible for the camera system which was a single camera blueprint that could cover different scenarios such as static tracking, orbiting speed based on car speed, dynamic zoom based on car speed, etc.
We went through quite a few iterations, including go pro-like action cams that were attached to the car, cinematic drone cameras but we had to reduce them to fit the specs of the project.
Created by
FIELD.IO
Client
IBM
Executive Creative Direction
Xander Marritt
Creative Direction
Vincent Venema
Technical Direction
Jonas Otto
Production
Kerem Demirayak
Design
Jason Wiley
Audrey Ma
Max Palmer
Development
James Lee
Casper Wortmann
Ezekiel Aquino
Sargey Rose
Shadi El Hajj
Lukas Schmeck
Marc Winklhofer
3D Design
Julien Bauzin
Sam Care
Dorian Gourg
Elliot Brett
Brendan Mannion