Scott Allen

  There is a tendency for perception called pareidolia, in which one’s mind conjures up patterns that one is usually familiar with when one sees a certain object. In Japan, this has been known as “Mitate,” and it has permeated various external arts and cultures, such as the tea ceremony and Japanese gardens.
  When people make a visual inspection, they search for and imagine visual similarities from their memories. I thought that we could extend the viewer’s imagination by creating an impersonal pareidolia device.
  In this work, images and captions are generated in real time for the shadows of everyday items and toys, using img2img in Stable Diffusion and BLIP-2. The viewer creates a shadowgraph, and then presses a button to start the generation. Next, the caption is displayed and the generated image is superimposed on the shadowgraph by projection mapping with a video projector to confirm the correspondence between them.
  With the development of AI, it is expected that the output using AI will become more homogeneous. In this context, this work will function as an entity that draws out the imagination by influencing the cognitive aspects that occur individually, rather than the human output itself.