AI Generated Film



KDDI Alpha-U


Geometry Ogilvy Japan

The concept film for KDDI’s αU culture brand offering Web3, Metaverse, Virtual, NFT, etc., was produced with innovative ideas and direction from Geometry Ogilvy Japan, with T&DA and motion design studio New Holland Creative.

Based on four images of future Tokyo drawn by Mayu Yukishita, one of Japan’s leading illustrators, the AI-generated images by Stable Diffusion and Midjourney were visualised using Deforum and RIFE Interpolation. In order to avoid being “just a technical expression” using AI, the complex layers were carefully composited and combined with detailed visual effects to maintain the beautiful static world of Mayu Yukishita’s four pieces of artwork while blending them into a dynamic virtual world. The result is a dynamic virtual world while retaining the beautiful static world of Mayu Yukishita’s four pieces of artwork.

A fusion of creativity, technology, and digital painting. The above making-of video shows how various AI-generated images were combined to create one complete video. The work is a perfect fusion of T&DA’s machine learning pipeline, and motion design partners’ New Holland Creative’s areas of expertise, with the unique expression of AI under control, creating smooth transitions that are easy to see as crafted images. The film is a perfect example of how to control AI’s unique expressions and create smooth transitions that can be viewed as craft. We invite you to see for yourself the various technical approaches that go on behind the scenes of the actual concept movie and the fun of this initiative.