Another Easy Re-Gendering

Today’s original image was another Civitai found image, created by noosprog80:

Girl in VR Glasses

Somewhat unusual for me, this is not a Flux or ZImage image, but instead Juggernaut XL, which I think comes from the wild west of Stable Diffusion checkpoints / fine-tunes 1. Luckily, it was an “english language” prompt for the most part, and I didn’t have to delve into the wild world of Danbooru prompts. (The entire idea of Danbooru prompts for generative AI seems like someone is missing the point of both Danbooru and generative AI. But it’s the early days of AI, and I think we should expect all sorts of both patently crazy things as well as things that will only seem odd in retrospect.)

The very first prompt worked on the obvious things: the gendered pronouns, and the very specific details around clothing. Clothing ended up being “He is dressed in a tight-fitting black leather harness and a black leather jockstrap. The harness shows off his broad, muscular chest.” This was the image:

Man in VR Glasses, V1

I did two more images, this time letting the seed float. I ended up liking both of the images better than the original; it felt like the men were more “open” to the viewer. Here’s one of them:

Man in VR Glasses, V1

I continued to explore a little, switching it to a landscape format, and seeing about bondage elements. However, I didn’t feel like I had a story line to pursue, and it was beyond the scope for “simple” re-gendering.

  1. At some point, I want to crawl all of Civitai, and build a phylogenic map of all the different models, finetunes, and checkpoints that are out there. ↩