Total Selfie: Generating Full-Body Selfies
This addresses a practical issue for users wanting better self-portraits, but it is incremental as it builds on existing diffusion techniques for image generation.
The paper tackles the problem of generating full-body selfies from close-up self-captured photos, which often have limited field of view and distorted facial shapes, by developing a diffusion-based method that uses four input selfies and a background image to produce high-quality, well-composed photos in desired poses.
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.