Hi! I want to know how to output the correspondence map with deep image analogy, similar to the figures described in the paper. Thank you!