Short Film to Premiere at 43rd BISFF and BISFF in Paris

The Electronics and Telecommunications Research Institute (ETRI) has developed a short film titled Cinema Genesis, which utilizes artificial intelligence to transform archival 2D film footage into 3D stereoscopic video.

The film, which has a duration of approximately four minutes, is scheduled to be screened at the 43rd Busan International Short Film Festival (BISFF) in South Korea. Following its premiere, the work will be showcased at BISFF in Paris, an extension of the festival held in Paris, France.

AI Technology and Restoration Process

ETRI applied two primary core technologies to achieve the 2D-to-3D conversion. The first is depth estimation technology, which allows an AI to autonomously calculate the distance of objects within a flat image to create a perceived sense of depth. The second technology automatically converts the 2D video into a 3D format based on those estimations.

To address the quality of the original archival materials, ETRI utilized Generative Adversarial Network (GAN) technology. This process involves two AI systems competing against one another to improve the final output, which allows the AI to restore and colorize photos and videos that were previously damaged or blurry.

Subject Matter of Cinema Genesis

The film focuses on the historical figures who contributed to the evolution of filmmaking. By reconstructing materials from early cinema history, the project highlights the work of several pioneers:

• Eadweard Muybridge, who demonstrated the possibility of moving images through successive photographs of a running horse.
• Étienne-Jules Marey, who researched methods for recording the movements of humans and animals on camera.
• Thomas Edison, the inventor of the projector used to show videos by running film.
• The Lumière brothers, who organized the first public film screening in 1895.
• Georges Méliès, a pioneer of special effects and screen compositing.

These historical figures’ old photographs and video materials were restored and transformed into the final 3D stereoscopic video format through the AI-driven process developed by ETRI.