Digital Face Beautification

Tommer Leyvand, Daniel Cohen-Or, Gideon Dror and Dani Lischinski

ACM SIGGRAPH 2006 Technical Sketches


A few results demonstrating our digital face beautification technique on both women and men:
Input images on the top row and their corresponding results on the bottom row.

* Input images courtesy of © Akira Gomi and © AR face database (bottom right image is me)


SIGGRAPH 2008 paper page with additional details on this work (here)

This sketch presents a novel method for digital face beautification: given a frontal photograph of a face (a portrait), our method automatically increases the predicted attractiveness rating of the face. The main challenge is to achieve this goal while introducing only minute, subtle modifications to the original image, such that the resulting “beautified” face maintains a strong, unmistakable similarity to the original, as demonstrated by the pair of faces shown in Figure 1. The effectiveness of the proposed method was experimentally
validated by a group of test subjects who consistently rated the modified faces as more attractive than the original ones. Professional photographers have been retouching and deblemishing their subjects ever since the invention of photography. It may be safely assumed that any model that we encounter on a magazine cover today has been digitally manipulated by a skilled, talented retouching artist. Since the human face is arguably the most frequently photographed object on earth, a tool such as ours would be
a useful and welcome addition to the ever-growing arsenal of image enhancement and retouching tools available in today’s digital image editing packages. The potential of such a tool for motion picture special effects and advertising is also quite obvious.

SIGGRAPH 2006 Sketch PDF (2.3MB)

Application Demo DivX Video* (~15MB) (free DivX codec available here)

Beautification application screenshots:


* Thanks to David Logan for the video voice over

Back to my research page