> That looks very good
> What did you generate them with?

Thanks! 😊

Unfortunately what I used to generate these images won't generate images like this anymore... the website tightented their violence content filters and you can't even get one character to slap another character anymore.

When it WAS working... I used the website joyfun.ai, and on that website I mostly used the JoyFun v3.1 and the Seedance 1.5 Pro text-to-video renderers.

The hanging was really hard to get. It took multiple attempts to get the idea of a hanging woman (or girl) across at all, with wildly randomly different results: all sorts of nonsensical postures and rope use, occasionally combined with grossly distorted anatomy. (I wish I'd saved some of the hilarious and wierd examples to show you.)

Even when I finally got something close, however, the image had the rope wrapped around the girls neck three times, rather than looking like a noose with one loop.

The trick was to capture that almost-correct image, bring it into Photoshop, and remove the extra loops of rope that way.

Once I had the edited image of a hanging girl to use as input for another text-to-video request, I could essentially lock in that starting point, and it was much easier then to concentrate on coaxing the AI to produce a particular scene with the hanging girl.

I'm just a few days away from the arrival of a much more powerful graphics card for my computer. My current card is woefully inadequate for doing this sort of thing on your own computer. Once I get that card, and once I get past the learning curving of setting my own rendering system for Stable Diffusion, I'll be able to continue to do more and better stuff, without as many content restrictions in the way.