Talking about Guro, the Dakimakura cover is meowhat I need you to share with any methods to kill your favorite characters.

I have a recommended LoRA for Illustrious at CivitAI:
https://civitai.com/models/1795249/ayaswanilv2

For every user of the apps you're using, including PixAI, you need any Guro LoRA to generate your Dakimakura cover and make sure you share some of your prompts and negative prompts.

For every fan of Guro, this page is about any prompts you want to kill the Powerpuff Girls with AI, including decapitation, hanging, garrote, strangulation, gun fire, headshot, sagittal or transverse plane cutting and so on. Only three sisters of the Utonium family, Blossom, Bubbles and Buttercup, can be added whichever one you want, and so can their Bleedman-verse counterparts.


For example, here's the prompt list of one of my favorite beheaded Blossom(PPGD) art in her Megaville Elementary School uniform with the Absolute Territory generated at PixAI from my original one at my Pixiv:

1girl, solo, [[[blossom (ppg), young girl, long hair, blunt bangs, orange hair, school uniform, megaville_elementary, ((white short sleeved puffy blouse)), ((pink sleeveless vest over blouse)), ((tucked blouse)), ((large red bowtie)), short skirt, pink uniform skirt, ((plain skirt)), white panties, ((white thighhighs)), ((no shoes)), zettai ryouiki]]], yukihara, (((no head, ero guro, guro, lifeless, corpse, death, death face, a lot of blood, blood on neck, headless girl, blood on face, neck spray, detached head, beheading, amputated neck, beheaded girl, tears, decapitation, disembodied head, necrophilia, empty eyes, bags under eyes, nsfw, peeing self, clouding eyes, half-closed))), anime screencap, anime coloring, lying on back, from above, knees together, feet apart, ((emotionless, expressionless)), full body, perfect anatomy, perfect body, perfect hands, separate views, colorful shaders, colorful outlines, perfect anatomy, perfect body, perfect hands, uncensored, Hands with 5 fingers of different lengths, on (right hand) and (left hand), nail plate on 5 fingers (right and left)

Quality: score_9, score_8_up, score_7_up, score_6_up, score_5_up, score_4_up, best quality, masterpiece, 4k, high quality, best quality, perfect lighting, detailed, perfect body, perfect anatomy, perfect hands, very aesthetic, anime BREAK, source_anime


So show me your own AI generated Guro arts with the Powerpuff Girls and leave your prompts within.

Ultimately I want this scene as something that would happen after the woman has been nailed to the cross, the ropes cut away, and she been raised up to hang from the nails.

But, as hard as it was to get this reasonably close to right, I'm pretty sure a good take on hammering big spikes through hands and feet is going to take even more work.

The tough part was convincing Wan 2.2 to make the sword go through those massive breasts. No matter what wording I tried the model had a hard time at deciding to do anything more than have the spear tip poke into her nipples or slash its way over the skin on top... when it wasn't otherwise doing completely bizarre shit way off from what my prompt asked for.


I tried to use "Time to Motion", but that didn't help.

The big breakthrough was finding a Wan 2.2 workflow that let me provide both a start and an end image. That meant I had to use some I2I and Photoshop to create an end frame image, but at least once I had that I finally made the tit skewering action something the model could grasp.

This was the prompt that finally got the job done:


Comment too long, view post No.15958 to see the full comment.

We need a version of this where we can see her face!

If semi-realistic video with a young character, snuff, and borderline sexual activity is an issue for you... don't click on this link.

(Borderline for being sensual but very softcore in one case, borderline because of non-living status in another.)

Also be warned that, since this video is longer than the usual 5-10 second clips posted here, the video might stutter and freeze up if you try to play it directly in your browser without fully downloading it first.

Download: https://files.catbox.moe/0kd3tq.mp4

The original characters in this video are not real people, they were text-to-image creations from epiCRealism XL. This, and some Photoshop work on the epiCRealism XL output, was the starting point for images used to seed image-to-video generation.

Numerous short Wan 2.2 clips were then generated. I tried to go back to the epiCRealism material when I could to start each clip, to try to avoid character drift, but many clips started with a frame from another clip.

Some clips were generated with no intention of being used for anything other than extracting a single frame with from the clip, with the right posing and positioning, to function as a good starter image for a different clip.

Comment too long, view post No.15483 to see the full comment.

I really do need a proper anatomical model. Current AI can only produce bloody cross-sections or fundamentally flawed anatomical structures, and the texture of grilled meat is also generated quite randomly.

This video is obviously NOT what one wants when depicting a nail being hammered through the palm* of someone's hands. Obviously the crucifier should not be able to casually push a nail into the palm of the victim's hand as if it were made of putty, then take her hand away before using a hammer on the nail.

I tried to make this very clear with my prompt:

Pam is a naked woman tied down at her wrists and ankles to a wooden cross on the ground. Pam CANNOT MOVE HER HANDS. PAM CANNOT MOVE HER FEET. She can only squirm helplessly.

Mary is the woman holding the nail.

Pam's arm repeatedly tugs against the ropes, but her struggle against the ropes is futile.

Treat Pam's hand and the palm of her hand as if they are as tough as wood. Mary CANNOT simply push the nail through Pam's hand.

While keeping a hold on the nail Mary hammers the nail through the palm of Pam's hand and into the wood below.

Comment too long, view post No.15979 to see the full comment.

the wan2.2 model performs well on the 'nipple fuck' tag, even though its resolution is only 640x432.(nipple insertion,nipple fuck:1.6),The girl remained completely still, while two men thrust their penises continuously into their nipples. Finally ejaculate through the passage of the nipple.The first paragraph of the video did not perform very well due to my failure to debug the prompt words properly, but after modification, the model's understanding of nipple fuck in the following two videos has improved significantly.

I'd like to start a running discussion on AI image and video tech, and how to set yourself up to run your own text-to-image (T2I) and text-to-video (T2V) on your own computer.

I'm a noob at this myself. I've just managed to get SwarmUI up and running with a newly-arrived video card (NVIDIA RTX 3090-based card with 24GB), after an unsuccesful attempt a week ago with my old video card, a card that proved to be utterly inadequate to the task.

I've managed so far to generate a couple of boring images of houses and a couple of low-quality videos of cars driving down a street. Whoop tee doo. Enough to show my system basically works, way short of pleasing results.

I found the process of getting Wan 2.2 (the only video model I've installed so far) working very frustrating. I'm fairly sure it's only half working given the quality of my results so far.

I'd like this thread to be a place where noobs like me can share what we learn as we go along, and where those with more experience can hopefully jump in with advice.

I used ChatGPT to help get as far as I have, but it was often as misleading as it was helpful. The tutorials I've found online all assume too much previous experience for their advice to be easily followed.

huggingface.co is apparently the place to go for downloading models, LoRAs, and all the other crap I don't know much about yet. At this point I find huggingface.co more overwhelming than helpful, but I hope to get beyond that point in time.

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.

Comment too long, view post No.15552 to see the full comment.

Fuck yeah, I love AI.

This is a repeat of a previous thread, but with the files elsewhere-hosted because of being a little too realistic (though not realistic enough to fool anyone the action could possible be real).

Full download: https://www.afilehost.com/vPbe3Pl23lX6

Due to most artist refused to draw R-18G umamusume content, let's make a thread for uma only

I couldn’t find any uterus removal concept on this board, but I saw someone had generated it on pixiv, so I decided to create one myself and train on those images. The generated output looks ok-ish to me, but with some in-painting touch-ups, it could be even better.

-------
Discovered this board with someone who created the transverse and saggital cutting, so I would like to thank him by sharing my creation to the community, and if you appreciate my work, please (too) support/donate to the guro artist you like & loved instead.

-------
Review tags and details in the archive
password: w51=WdL'
https://mega.nz/file/201EHayC#LoXkPpbUQ-kZmWaegpwA5ZWJaKrl-msZwJJ5WR76PsE