Neural Segmentation Based Apps

Came in to help develop some effects for a set of very ambitious neural segmentation based apps, in which the effects would alter video or audio of different segments of the clips, as well as creating editor panels inside Unity to help designers do their job in a more comfortable way.

The main effects I worked on were relating to what was called Imposter Effect, in which you’d record your own video and audio, and it will appear on top of the face of someone in the video on playback, or as part of an angel or devil figure flying over the shoulders of the target.

It was a very dynamic environment in which every two weeks we had widely different tasks and aims, and we were expected to have capacity to diagnose problems fast and determine our own task-solving schedules, leading to collaborating hand in hand with the Unity AR Team themselves on preparation for their Unite 2018 Conference.

NOTE: I had no role in the neural network segmentation part itself, I was only utilising the results of it.