Fullyconnected-autorefractor | 2020

Video, custom software, neural network (Neural Radiance Fields)

Fullyconnected-autorefractor is a video work that aims to pose questions about the relationship between the outer world and our vision. It includes frames generated with a neural network (Neural Radiance Fields) which is typically used to construct views of new perspectives using a handful of photographs of known perspectives. The frames in this video is a result of asking the neural network to solve a scene that is incongruent in time and space. Through an exercise of this impossible task, Fullyconnected-autorefractor aims to break the consistencies in our vision that we depend on to make sense of the space and encourage an alternative explanation to visual phenomena.