Total Selfie: Generating Full-Body Selfies

TL;DR

This paper introduces a diffusion-based method to generate full-body selfies from close-up selfies, enabling realistic, well-composed images with desired poses and backgrounds.

Contribution

It presents a novel diffusion approach that synthesizes full-body selfies from multiple input selfies, background, and target pose, addressing perspective distortion issues.

Findings

Produces high-quality, realistic full-body selfies

Successfully incorporates desired poses and backgrounds

Addresses perspective distortion in self-captured photos

Abstract

We present a method to generate full-body selfies from photographs originally taken at arms length. Because self-captured photos are typically taken close up, they have limited field of view and exaggerated perspective that distorts facial shapes. We instead seek to generate the photo some one else would take of you from a few feet away. Our approach takes as input four selfies of your face and body, a background image, and generates a full-body selfie in a desired target pose. We introduce a novel diffusion-based approach to combine all of this information into high-quality, well-composed photos of you with the desired pose and background.