Approach Paper & Methodology (aw + pw + awin + pwin + s1 Fac) to FFac GANs [145–151] DM, CycleGAN [152] (aw + pw + awin + pwin + s1 Fac) to RFac pix2pix [153] FFac to FFac CycleGAN [154]; StyleGAN2 [155] and component elements (s1 Fac), represented as (aw + (pw) + awin + (pwin) + s1 Fac) to FFac [145–152]. Furthermore, complete facade and roof images for all four directions of a building can be generated style-similar residential facade from semantic labeling according to the user-provided example. [152] Wenyuan Sun, Ping Zhou, Yangang Wang, Zongpu Yu, Jing Jin, and Guangquan Zhou. 3d face parsing via