At first glance, the term appears to be a random assemblage of words: a common placeholder name ("Jane Doe") attached to a cryptic suffix ("BlobCG"). However, a deeper dive reveals a fascinating intersection of 3D rendering, privacy advocacy, and generative AI art.
The answer lies in the . As of late 2023 and 2024, a massive ethical debate has surrounded AI image and video generators (Stable Diffusion, Midjourney, Runway Gen-2). These models are trained on billions of images, many of which include real human faces—without consent. jane doe blobcg
In response, a coalition of open-source developers and privacy activists began creating "Poisoned" or "Anonymized" datasets. is the codename for a specific dataset containing 10,000+ renders of a generic, blob-based human figure. The goal is to train AI models to understand human anatomy and movement without ever seeing a real photograph of a person. At first glance, the term appears to be
