Techart Conference

Gilwoo Jeong 

■ Abstract

This study explores the feasibility of implementing an AI-based virtual casting system that can be utilized in the previsualization stage of video content production by combining Face Swap and Voice Conversion technologies. In conventional project planning and investment phases, there has been a limitation in effectively visualizing actual casting combinations and directorial approaches. To address this issue, this study constructs a pipeline that synthesizes actors’ faces and voices using artificial intelligence to generate rough teaser videos. Through this approach, it proposes a production environment that enables the prior simulation of various casting combinations and character interpretations. Experimental results confirm the applicability of face and voice synthesis-based previsualization, while also analyzing technical limitations observed in synthesis quality and voice processing. Furthermore, directions for future improvements are suggested. This study demonstrates the potential of extending AI-based synthesis technologies as decision-support tools in the early stages of video content production.

■ Bio

M.F.A. Candidate in Film & Video Production (Directing),

Graduate School of Advanced Imaging Science, Multimedia & Film, Chung-Ang University

B.A. in Broadcasting & Visual Media, Korean National University of Arts (K-Arts), School of Film, TV & Multimedia

- Director of Photography and Editor of the feature documentary Bora Bora (Screened in the Korean Cinema section at the Jeonju International Film Festival)

- Cinematographer of the documentary Children of the Starlight Orchestra (Screened at the Jecheon International Music & Film Festival)

- Director of the documentary Songs of Life Revisited: Eunpyeong Story

- Director of the short film Please Find My Voice (Award-winning work at the SNS 3-Minute Film Festival)

- Director of the short film Talpigak (Screened at the Mise-en-scène Short Film Festival)

- Director of the documentary Driving in the Wind (Screened at the Seoul International Extreme-Short Image & Film Festival)

- Participant in the 2025 AI-CFT Project Program