(20 Mar 2026, 15:32 )Like Ra Wrote: Any examples?
Grok Imagine is really, really good at spells. The results look like they're from a movie. Upload the image of yourself / your representation, plus the person you want to be casting the spell. If you're using an A.I. generated image for the spell caster, it actually makes it more photorealistic to fit with yours.
I've been using LLMs to write the prompts for me. For instance I will ask it to write 10 historically-evidenced spells where in the foreground a woman is casting a spell to make a man impotent, or locked in chastity, or bound to her, etc, while in the background or to the side he looks on (usually spellbound / befuddled / concerned, etc).
Grok 4.20 is pretty good at it, but not the best. I like the Chinese models such as Z.ai's GLM-5, DeepSeek, Kimi K2.5, and others you can find on OpenRouter.
I'm using this format for the prompts:
Title/Summary
Visual
Action
Audio
It can follow detailed requests for sound effects and speech. Music is slightly behind but still worth playing around with.
A paid SuperGrok account is pretty much essential as apparently free generations are very limited now. Also, people using it to remove clothes / generate conventional sex acts, have been complaining non-stop for months that it won't let them. So it goes without saying, don't expect to be able to do that.
But for fetish, femdom, hypnosis, fantasy, role-play and suggestive themes, it's mind-blowing what the Grok's new multiple image to video model can do. It literally feels like it allows you to think in erotic video, in a way that's never been possible before.
