Most commonly accessible AI image and video models aren't good with violence of any sort, even when there aren't any deliberate content filters standing in your way.

The various AI rendering models generally aren’t trained using many violent media examples, or maybe just enough to recognize violence and say "No!" to it. (However, from some of the examples I've seen, Grok knows a bit about physical depravity, accessible to you if content filters don't get in your way.)

While you can eventually get some of the models I've played with to hang a woman, you have to start with an image of a hanging woman in the first place. Otherwise they'll never catch on to the concept of hanging from simply saying something like, "A woman hangs from a noose."

Nor will long, detailed verbal descriptions of what a hanging entails gain you much traction.

Such prompts can get you some wild and crazy results, with botched rope work and bizarre poses and occasional mutant anatomy that could give Hieronymus Bosch nightmares... but only rarely, by wild chance, will the results be at all close to what you'd probably be hoping to get out of a hanging.

So, how to get an image of a hanging woman (or man or futa or Teletubbie, if those are your things) with the look and attire you want? My answer was to make such an image myself, starting with a nude but vanilla AI-generated image of a woman to suite my fancy, a separate image of a noose (easily enough googled), and to then use some careful Photoshop work to put those two things together.

I realize right there this could be a barrier for people hoping to get what they want out of an AI without having to have any personal talent for graphical work... but it's a solution that works for me.

One rendering model I've used to create women (and girls... yes, I'm a perv that way) for my AI experiments is epiCRealism XL. epiCRealism XL is good at making attractive looking people. You can pose these people to some extent, but I think it has a bias toward portraiture over more general posing. It’s not shy about full body nudity for any age.

I hadn’t been concerned so far with creating any male characters, but I gave that a try and, by default it seems, you get all ball sack and little or no penis on these men. If you specify that the man has a large penis at least you get a visible but flaccid penis. epiCRealism XL refused to put a cock on a female body, so you’ll have to produce futa by other means.

epiCRealism XL wasn’t keen on adding expressions of fear or agony to people’s faces either, so those had have to be generated later. There are, of course, plenty of different image renderers to choose from, so I’ll definitely seek out more of them to set up on my new ComfyUI rig for greater flexibility in the future.

Since I can only post one image or video in this first post of a new subject, I’ll attach an example starter image, and other steps in this process, in additional replies below.

Then I google for a picture of a noose (shown in a reply below).

Then (without getting into the details) I put the above two images together in Photoshop to create a starter for use with image-to-video (I2V) processing (combined image shown in a reply below).

Once you’ve got an image like this (at least if you’re going to be using Wan 2.2 and many other video renderers you can run locally) it’s too much to hope for to use a prompt on such an image to change the woman’s facial expression and also get her hauled into the air will her feet dangling for anything more than a moment before your clip ends.

So I used one I2V session simply to produce another image, generating a clip from which to pluck a single good frame, a frame that would become the seed for a better video clip. Here’s the first prompt I used with the initial woman-with-noose image:

A woman hangs from a noose. Her hands are tied together behind her back.

The noose is pulled upward, lifting the woman's entire body until her feet can't reach the floor, causing the woman to dangle freely from the noose that tightens around her neck.

Her face shows wide-eyed terror and panic. Her legs wildly thrash in the air.

Zoom back to show a full body shot with feet that don't touch the floor.

You can use a low frame rate and low resolution to make sure you’re on the right track with your image and your prompt. But then be patient rendering a few tries at a higher resolution, so you get a decently sharp image for the next I2V.

One thing that can help: If you have a negative prompt available, add “motion blur” to the list of things not to do. This makes it more likely any given still frame you select will be a sharp image to use for the next step.

The image I picked from the first temporary-use video is the one included with this post.

I combined that image with this next prompt:

A woman hangs from a noose around her neck, her feet unable to reach the floor. Her hands are tied together behind her back, unable to break free.

She dangles and her entire body swings like a pendulum from her neck. She gasps for air with a look of desperate panic on her face, her mouth staying wide open the whole time. Her legs flail and kick randomly while her body spasms.

The result? A decent, if far from epic, hanging. More work would be needed to get something spectacular, of course, but I’m content for now to show the basic steps I followed. I’m too much of a noob at this point to be sure I’m even giving great advice here, but perhaps it’s enough to help speed up someone else’s learning curve.

The resulting video is in a follow-up reply.