The latent space has been compared to the Wild West in its lawlessness, but that metaphor disguises the true weirdness of its uncanny valleys. Faceless men struggle to be rendered out of fog, textured spires extend to the horizon, night markets fade in and out of existence as Frank Lloyd Wright’s buildings transform into Pizza Huts. As art is necromanced in the style of dead and living artists, one of the latent space’s valleys fills with the commands of a million sculptors and users. Though this may sound like a great setting for a tabletop role-playing campaign (with genre-mashing, surrealness, and questions about life and humanity), the latent space is becoming a questionable tool in a TTRPG designer or artist’s belt for commercial releases, fan creations, and more. The latent space, put simply, is a metaphorical location full of all of the possible images AI could create, each altered by a different factor or prompt. In this growing world of AI art generation, as the generative tools improve each month, so do the ethical questions about the future of TTRPG art.
AI art generation systems like DALL-E 2, Midjourney, and Disco Diffusion have exploded in popularity in recent months, with the biggest push coming this spring with easy-to-access systems like Craiyon (formerly DALL-E mini) and accounts like Weird Dall-E Mini Generations, showing people just how far AI art has come. Vox’s primer video offers a great explainer, but the main art-generation mechanism is word prompts. You write what you want to see and it spits out art at a freakishly fast rate. Arthur C. Clarke said any sufficiently advanced technology is indistinguishable from magic, and AI art can really get that message across. When game designer Raph D’Amico
Read more on polygon.com