Volumetric capture system on set to 3D scan the actors with an array of specialised cameras, reconstructing them using software to create hyper-realistic, moving, 3D models of the talent and their performance. These were then placed into the miniature CG environments, allowing for easy repositioning adjustment. Most importantly, they were lit using the same CG light sources as the rest of the miniature elements, allowing for smooth camera moves around the UMass ecosystems.  

One of the most complex scenes to craft was the train, needing to balance out the freedom of movement in the wides, and the small-scale nuance required for the close ups of the laptop. For better control, the team built a train set on a stage, hanging a long LED screen outside the train car window which ran an Unreal Engine environment. 

The passing landscape was built from scratch, developed as an optimised and fully controllable procedural system of spawning city levels in real-time.

The team was thus able to create a realistic feeling of motion that was easy to manage on set: they could change the speed of the train, the sun position and weather, lensing or camera angles, or even the motion blur of the outside. They could also ensure that the scene was accurately lit by the moving scenery, generating moving lights and shadows inside the train car, as well as realistic reflections and refraction in the windows and other surfaces.