This page was machine-translated and may differ from the original. View original
SKT and KAIST's AI Video Synthesis Technology Accepted as ECCV 2026 Paper
'InsertAnywhere' based on 4D scene understanding automatically implements even shadows and reflections
SK Telecom (SKT) announced on the 30th that a paper on 'InsertAnywhere,' an AI image synthesis technology jointly developed with Professor Jae-Geol Ju's research team at KAIST, has been accepted for ECCV 2026 (European Conference on Computer Vision), which will be held in Malmö, Sweden, this coming September.
ECCV is classified as one of the world's top three computer vision conferences, along with CVPR (Computer Vision and Pattern Recognition Conference) and ICCV (International Conference on Computer Vision).
InsertAnywhere is a technology that adds objects to existing video that were not present at the time of filming.
The existing method had problems such as the position and size of inserted objects becoming unstable during camera movement or subject overlap, or objects that should be hidden remaining on the screen.
The unnaturalness of shadow and reflection expressions was also pointed out as a limitation.
To resolve this, '4D scene understanding technology' was applied, which adds a time axis to the 3D spatial structure.
If the user specifies the insertion location just once in a particular scene, the AI automatically expands that information to the entire video.
SKT explained that it stably maintains the shape and position even when the camera moves or objects are obscured, and automatically generates shadow, reflection, and lighting effects around inserted objects to produce natural results.
SKT has already commercialized InsertAnywhere by applying it to its virtual product placement (VPPL) solution, 'AdFlux'.
AdFlux includes △Go All the Way Solo Tour △Our It has been used in broadcast programs such as "Meet Now," "Laughing," and "Call Anytime," and patent applications for related technology have also been completed.
Yang Seung-ji, Head of Enterprise Technology Support at SKT, stated, “Building on the adoption of ECCV 2026, we will continue to advance high-quality AI video synthesis technology that can be practically utilized in the advertising and media fields.”
본 기사에 대한 정정·반론·추후보도 청구는 보도 청구 안내를, 그간 게재된 보도문은 정정·반론보도 모아보기를 참고해 주세요.














