Paper dreams : real-time collaboration with machine learning for artistic expression

Abstract

Collaboration between human and machine can lead to the novel generation and combination of ideas. Paper Dreams explores how the real-time generation of ideas and visuals based off of user sketch, text, and voice input can encourage additional creativity, under the context of visual story development. We discuss the definition of and importance of creativity, and outline past tools for artistic expression and machine learning based art. We then explore how Paper Dreams' system of processing user sketches and generation of color and classes is unique and impactful. Emerging from past work done on Paper Dreams, we discuss how an initial user study motivated additional technical and design changes to support easier user manipulation and data collection with the system. We then detail the design and development of these technical changes, and the decisions and challenges surrounding these steps. Finally, we demonstrate how these system changes contribute towards future work for Paper Dreams.