Abstract

Subject of study. In this study, we present a novel method for the synthesis of video images using deepfake face swap, a technology that enables the creation of authentic video clips with false or swapped faces without evident traces of manipulation. The creation process of video images of “impossible faces,” such as a chimeric face, whose left and right sides belong to different people, or Thatcherized faces, wherein the eyes and mouth regions are rotated by 180°, is described stepwise using the DeepFaceLab software as an example. Aim of study. This study aimed to present and validate deepfake technology as a method of digitally collaging “impossible face” images. Method. Examples of method implementation were demonstrated using experimental results related to the perception patterns of moving “impossible faces” and their differences. Main results. The phenomena of perception previously recorded under static conditions were reproduced for dynamic models; accordingly, new interpretations can be inferred. Original faces were evaluated positively under static and dynamic conditions, independent of image inversion. Images of “impossible faces” under all conditions were perceived as unattractive, disharmonious, bizarre, and artificial. Expositions with audio speech increased the adequacy of evaluations under conditions of straight orientation. Practical significance. Digital collaging methods can significantly expand the capabilities of researchers in the field of interpersonal perception. Moreover, information technologies can expedite the creation of stimuli models for “impossible face” images, which are necessary for the further investigation of the human psyche in the course of communication.

© 2022 Optica Publishing Group

Dictionaries for image and video-based face recognition [Invited]

Vishal M. Patel, Yi-Chen Chen, Rama Chellappa, and P. Jonathon Phillips
J. Opt. Soc. Am. A 31(5) 1090-1103 (2014)

Passive imaging at 250 GHz for detection of face presentation attacks

Marcin Kowalski
Opt. Express 29(2) 1956-1976 (2021)

Deep learning for the segmentation of preserved photoreceptors on en face optical coherence tomography in two inherited retinal diseases

Acner Camino, Zhuo Wang, Jie Wang, Mark E. Pennesi, Paul Yang, David Huang, Dengwang Li, and Yali Jia
Biomed. Opt. Express 9(7) 3092-3105 (2018)

Adaptive, optical, radial basis function neural network for handwritten digit recognition

Wesley E. Foor and Mark A. Neifeld
Appl. Opt. 34(32) 7545-7555 (1995)

Optical network for real-time face recognition

Hsin-Yu Sidney Li, Yong Qiao, and Demetri Psaltis
Appl. Opt. 32(26) 5026-5035 (1993)