Autocomplete Hand-drawn Animations

摘要

Hand-drawn animation is a major art form and communicationrnmedium, but can be challenging to produce. We present a systemrnto help people create frame-by-frame animations through manualrnsketches. We design our interface to be minimalistic: it containsrnonly a canvas and a few controls. When users draw on the canvas,rnour system silently analyzes all past sketches and predicts whatrnmight be drawn in the future across spatial locations and temporalrnframes. The interface also offers suggestions to beautify existingrndrawings. Our system can reduce manual workload and improvernoutput quality without compromising natural drawing flow and control:rnusers can accept, ignore, or modify such predictions visualizedrnon the canvas by simple gestures. Our key idea is to extend thernlocal similarity method in [Xing et al. 2014], which handles onlyrnlow-level spatial repetitions such as hatches within a single frame,rnto a global similarity that can capture high-level structures acrossrnmultiple frames such as dynamic objects. We evaluate our systemrnthrough a preliminary user study and confirm that it can enhancernboth users’ objective performance and subjective satisfaction.