Realsynth.

Environment Artist
Realsynth, 2017
Synthetic annotated image data

I worked together with Realsynth for three months to create this environment to be used with AI technologies for railway. This is the first project I worked on that wasn't made to be looked at humans but by computer software.

Regular image data for AI learning purposes requires a lot of manual work. With synthetic image data created like this in a game engine you can get around a lot of that work by automating the process. It also makes it possible to simulate different weather conditions such as fog and rain and other scenarios that would be either to coastly or dangerous to perform in real life.

I was the only artist/tech artist on the project and my role was to create the environment around the railway tracks from reference video. The trees and grass are from asset packs, the rest of the assets were modeled by me. Because of the nature of the project a picture perfect look wasn't the goal but more the accuracy and placement.

Signal Blueprint

To be able to make all the needed variations for the signals I set up a blueprint with options to be able to do different configurations. It's possible for example to choose what lamp layout the signal should have, what lamp should be on or off, blinking and so on.