MEET ME IN DIFFUSION

Meet Me in Diffusion is a critical video art piece that aims to reveal the cultural symptoms in the current AI image generation field. Despite tools like Stable Diffusion providing users with diverse operation methods, their ecosystem has fostered a homogenized aesthetic tendency, particularly an obsession with anime female character images. This work, in the form of video, employs AI image generation technology to discuss the complex entanglement between the operability of technology and the singularity of generated content.

The video begins with a parameterized modeled Earth. Through Stable Diffusion's image-to-image function, the Earth's texture is redrawn into anime girl images that flood the CivitAI website. Following this, the artist utilizes Controlnet's OpenPose function to match constellation maps with human skeletal points, generating "Constellation Beauties." The scene then switches to a boxing match. The artist uses the AI tool Cutie to extract the image of one of the boxers and employs ProPAInter to erase him from the original video. The extracted boxer's image is then redrawn by Stable Diffusion, presenting a virtual image that differs in every frame. By placing the virtual image back into the original video, a fight between the real boxer's image and the virtual image is formed, creating a strong contrast between the virtual and the real.

This work aims to showcase the artist's ability to apply AI technology and his insight into technological and cultural phenomena. By creatively appropriating and transforming AI tools, the artist reveals the complex interaction between technology and culture, providing the audience with a perspective to reflect on the AI image generation ecosystem. Meet Me in Diffusion presents a non-AIGC path of AI art creation, demonstrating a possibility of the fusion of technology and art.