Japanese media companies test generative AI for personalized content and production

NTT Docomo has announced the development of a “personalized video automatic composition technology” designed to enhance live streaming experiences. Utilizing Vision Language Models (VLM), the AI can automatically combine close-up shots of a viewer’s favorite performers with wide shots of the overall stage performance. This technology aims to reduce the manual burden on viewers to switch camera angles while ensuring they do not miss key stage moments. Docomo plans to demonstrate this technology alongside its “Acoustic XR” audio tech at the NTT R&D Forum in November 2026.

In a separate initiative, Fuji Television is experimenting with generative AI to streamline program production. The broadcaster is testing a method to visualize episode talks through AI-generated videos, aiming to improve quality while reducing budgets, time, and staffing requirements. A special variety program titled “Hanashi Cinema” will showcase this approach, where directors use video-generation AI to recreate stories told by guests, creating an immersive experience that visualizes the speaker’s narrative in real-time.