Electronics and Telecommunications Research Institute (ETRI) said on Sunday it will unveil three next-generation media core technologies, including video coding for AI, at Korea Electronics Show (KES) 2026 to be held at COEX in Seoul from Oct. 13 to 16.

The video coding technology for AI is a compression technology optimised for AI video analysis rather than for people. It efficiently compresses and transmits information needed for AI to detect, classify and track objects. Based on video coding for machines (VCM), ETRI developed a technology that extracts and compresses feature information needed for AI analysis instead of transmitting the entire video. That allows it to maintain video analysis performance while reducing the burden on networks.

ETRI will also unveil an "AI-based photorealistic multi-person 3D human generation and free-viewpoint reproduction technology" that reconstructs the 3D shapes of multiple people with a small number of cameras. It generates 3D shapes from video shot with a limited number of cameras and enables viewing from a desired perspective. ETRI said it can be used for sports broadcasts, performances and extended reality (XR) content.

It will also present a generative AI-based previsualisation video production support technology. The technology automatically converts sketches, storyboards and scripts analysed by AI into editable 3D scenes. It can be used to review scene composition and camera placement in advance of filming for movies, dramas and advertisements.

ETRI plans to use the exhibition to expand technology cooperation with domestic companies and to pursue international standardisation of related technologies.

Lee Tae-jin (이태진), head of ETRI's Media Research Division, said AI is emerging as a major consumer of video, requiring new changes in existing human-centred video compression technologies. He said the AI video coding technology would lay the groundwork for delivering and analysing large-scale video data more efficiently.