Sahillioglu, YusufAydinlilar, MelikeFu, Hongbo and Ghosh, Abhijeet and Kopf, Johannes2018-10-072018-10-072018978-3-03868-074-1https://doi.org/10.2312/pg.20181292https://diglib.eg.org:443/handle/10.2312/pg20181292We present a method that interpolates new shapes between a given pair of source and target shapes. To this end, we utilize a database of related shapes that is used to replace the direct transition from the source to the target by a composition of small transitions. This so-called data-driven interpolation scheme proved useful as long as the database is sufficiently large. We advance this idea one step further by processing the database shapes part by part, which in turn enables realistic interpolations with relatively small databases. We obtain promising preliminary results and point out potential improvements that we intend to address in our future work.10.2312/pg.201812929-11