This is a moving image production of a single story that was developed by layering and connecting automatically generated images based on keywords. Each of the story’s nine stages – from a beginning to an end (from birth to death) – is represented through the keywords of “ground,” “moon,” “water,” “green,” sky,” “rain,” “light,” “death”, and “sun”. Each of the nine original images by particles of pointillism was assigned as one of the nine stories that make up the project. The context for the keywords assigned to each image evolved as the work proceeded from the starting point (src) “ground” to the next stage, “moon”, and then ultimately to the ending point (dst) “sun”. At each stage of the process, the changing keywords induced shifts in the contour and hue within the abstract image, thus facilitating its evolution.
VQGAN in machine learning technology can set two keywords and automatically generate colors and shapes that evoke images from the related searches. By setting a quantitative influence level for each keyword, it is possible to mass-produce images that reflect the set level of influence. Considering this technology as a device for mass-producing abstract visual images from keyword-extracted linguistic images, the motivation for our work was to consider what would happen if we used the process to produce complex word trajectories such as poems and novels. The complex language structures of literary works have long been known to induce vivid images in the minds of their readers, but this technology automatically mass-produces abstract visual images as such works are read. In this project, which generated moving images showing a progression from a beginning to an end (from birth to death), the development of the generated images from one keyword to the next would not have been possible without the assistance of the generative algorithm. This work evokes images for a story based on the essence of nine keywords in order to create a unique visual experience.
