Facial reshaping operator for controllable face beautification

Posting attractive facial photos is part of everyday life in the social media era. Motivated by demand, we propose a lightweight method to automatically and efficiently beautify shapes both portrait non-portrait faces photos, while allowing users customize beautification individual features. Previous methods focus on mostly frontal neutral faces, without incorporating user controllability process. To address these restrictions, Facial Reshaping Operator representation, which affine-invariant, captures pairwise geometric configuration landmarks, allows for efficient face with user-specified weights parts. We also an unsupervised operator space where input iteratively pulled towards local nearby density mode improved attractiveness. Our distinguishes itself from commercial tools that it mildly enhances altering makeups or complexions, complements lack fine-grained control attractiveness users. The experimental results show our improves shape large range poses expressions, demonstrating potential applicability seen such as Facebook Instagram everyday.