
AI demo film for AI Festa
KT
Events and expos · Storyboard to final cut in 6 days
The project
A demo film for AI Festa showing KT's multimodal AI, "Mi:dm K Multimodal", reading the audio, images and text of drama scenes together and answering questions about them. The event was close, so we delivered everything from storyboard to final cut within 6 days.
Brief
Visitors had to grasp, in a few seconds, that the AI understands the dialogue, picture and subtitles of a video at once. We designed demo UI screens scene by scene and showed it as a conversation of questions and answers.
Structure
- Scene 1: the AI explains the line "He crossed the Jordan River", said by a character reading an obituary, as a Korean idiom for death
- Scene 2: from the dialogue and the picture, it works out the situation of a character in a hospital gown and what caused it
- Scene 3: it identifies a gold object on screen as a watch and explains the incident tied to it
Revisions
At the mid-point review, the direction was to bring out more strongly the idea that three kinds of information are analysed as one. So we added a shot where the screen pulls back and the VOICE, IMAGE and TEXT icons circle and join together.
Delivered
One 73-second horizontal film, built on the on-screen conversation and the drama audio with no voice-over.







