I made too much and need another thread

In case if you are interested...
I released Anisnuff, a finetuned Anima for R-18G scenes.

Download:
https://mega.nz/folder/iB8yRZjJ#61cqDOWrysUmEIluuTO6dg

Sample images:
https://www.pixiv.net/en/artworks/143751523

Setup:
Just copy diffusion_models and lora folders to your ComfyUI.
Use the default VAE and text encoder for Anima.
Drop any of the workflow file and enjoy.

Comment too long, view post No.18332 to see the full comment.

Finally a new version of the GuRO Checkpoint Model is here.
And this time it also possible to download it as a seperate LoRA.


I have successfully trained a guro checkpoint built upon the WAI 14.0 base.
I added 1000 synthetic images to better teach it some concepts. Bleeding of concepts decreased compared to the previous version.
This model was trained on a total dataset of 8,000 images.

A comprehensive list of the tags used during training can be found here:
https://rentry.co/r-18G_illustrious_tags_V2

It is not perfect but much better than the previous iterations. Including, but not limited to:
Hanging.

Comment too long, view post No.13910 to see the full comment.

I'm creating a more broad thread on dead girls lying as if they're just sleeping, specifically ones who are in coffins. AI image models used to be terrible at making these type of images, but with recent advancements, especially with Google's Nano Banana, The results looks much better. This thread isn't limited to images made in Nano Banana, any image and model is welcome in this thread, as long as it's girls and women in repose

Here's an example I've cooked up by giving the model a reference image and a prompt.

Started playing around with moss image yesterday after hearing some hype around it.

I don't think it's a Krea 2 killer but there's certainly some significant advantages.

Pros:
-Completely uncensored out of the box. Nudity, sex, gore, whatever. No need to finangle with uncensor loras, nsfw enhancers, etc.
-Noticeably better seed variety. Krea 2 i've noticed tends to have very little variety between seeds, moss can give you wildly different results with the same prompt.
-Bounding box support - i haven't needed it yet but i can see it coming in handy.
-Small model, so will work well on smaller gpus

Cons:
-sometimes the good seed variety comes with a downside... prompts that work perfectly fine most of the time will sometimes just give you garbage
-higher possibility of weird body anatomy/bad faces (kinda ties into the above)

Comment too long, view post No.30840 to see the full comment.

Im wondering whether anyone managed to get MiniMax to draw a corpse.

No matter what I prompt for, if MiniMax sees a body, it has to breathe. If it has face, it has to blink. If it has mouth, it has to blabber incomprehensibly, but this last issue can be fixed by attaching soundtrack.

Attached video is to demo the issue. Prompt for last segment is:

subject_definitions:
<Subject 1> is a corpse, dead body, blue skirt.

summary:
[reference generation] 

retention_analysis:
<Subject 1> (appears in [Shot 1]): fully_preserved

detailed_description:
Indoor scene, stationary camera

timeline:
Camera pans around dead, headless unmoving body covered in blood, bloody stump of the neck.

overall_soundscape: Silent film, no speech, complete silence.

non_diegetic_music: N/A

edit: seeing preview now, I'd like to point out that character in video is by no means realistic, its 3DCG-like cat-girl. If her head is attached, that is

Some of you will dig this, and some of you wont. Made a musical compilation of some of my favorite AI videos posted here. You guys are so FUCKING AMAZING, I just wanted to give a little something back.

Link: https://mega.nz/file/0twHnIKD#F65B_4_Zqd1Z4cZ26OFw2Pl1Yt5lj2wChoq9VTKeQCo

This is a nearly 7 minute long guro video in a REALISTIC style, preview images are 2d-ified for forum rules. In it, a couple slutty sorority girls make fun of an ugly old woman, only to find out she's a real witch. The witch gives them the worst - and last - night of their lives. First, she dismembers the blonde, sawing her to pieces and butchering her for stew as the asian girl watches helplessly. Then, the witch brutally disembowels and mutilates the asian slut for potion ingredients.

You can find the full video in my mega folder here:
https://mega.nz/folder/YzRg2JjD#qcvfmIAnU-OUqsjRnpsjxg
under the path "long videos/Witch_Last_Laugh/final_video.zip". As with my images thread, the password for every zip in my mega folder is "Bostwick Consistometer" (no quotes).

Enjoy! I kinda slaved away at this one. More videos to come in the future (but probably not many this long)

Always check your meat for tightness.

Controlnet generated from sketchs, thanks for all the guidance from this board

I am usually a daz content creator who posts over on 3dcg but after getting a new AI plug in from renderhub called AIRE I have started to experiment with AI. It uses ComfyUI which I have no idea is if good or bad, I honestly have no idea what I'm doing but I wanted to post some of my experiments and results. If anyone has some tips and tricks, or how to best do prompts I'm all ears. I used the MiniMax H3 template to make some videos, and honestly just tweaked the prompt that loaded with it.

The kind of content I like mainly revolves around strangling and hangings, pretty much all things asphyxiation, not into body destruction, mutilations or the like though stabbing, shooting, and beheading aren't bad.

Just my AI Experiments.

Mei with quatro tits...
This is the old grok v1, before the update that removed the guro in version 1.5.

Please comment and make requests.
Now main goat is minimax H3.

>>31065 from my thread :
Your request >>31084
https://file.garden/aq44TMzo7v--DTmD/Chick_Request.rar (password: gurochick)

Have fun :)

I've been animating scenes from guro comics to good effect in MiniMax. Here's one from glcer.