finally a dataset that could actually help us tackle the value alignment problem in generative art models. maybe we'll get some decent, non-weird human depictions out of this. https://www.reddit.com/user/hafftka