What we train with (b) What we train for (c) What we can do without explicit training Figure 1. (a) We train with our scale-consistent positional encodings and modified generator architecture that enables (b) synthesis of arbitrary scales and multi-scale consistency. (c) We showcase our results for novel configurations not encountered during training such as generation at never-seen-before scales, extrapolation and editing using spatial transformations such warping and stretching.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.