sounds like you want us to be ai artists

>>5545
If you put your thinking cap on, you'll see he wants to make some art himself with ai, he just doesn't know where to start.

The fact you're asking for commands tells me you're very new to this.

For niche things like guro you're gonna need a customized local setup. Don't expect cloud hosting stuff (like civit and others) to facilitate guro related stuff. They vaguely tolerate adjacent content but purge anything explicit.

So I suppose the first question is can you run SD locally?
Current models need a GPU with 8gb VRAM (minimum), but you'll be more comfortable with 12GB and above.

It's also better if you have an NVIDIA GPU since this stuff works on CUDAS. There are solutions in the form of ZLUDA (kind of a CUDA emulation solution for AMD cards), but this is slower.

So first confirm you got usable hardware

most of guro content needs LORA to produce, or you just playing AI gacha and can't get reliable result.

so you need get or train LORAs, even good ckpt to make good quality AI guro pics.

as other have said you really need to run stable diffusion on your own machine. once youve got that set up
you might find this thread useful https://boards.guro.cx/ai/res/5026.html

and this https://rentry.org/R-18G_Stable_Diffusion

thanks for temari and hinata!

what neg prompt do you need to stop multiple heads?

>>5435

use regional prompt plugin, it can get certain result as your expect.

>>5185
thanks a ton for the inpainting advice, it's working pretty well I think, and I'm making some progress learning to do it better.

also working on adding spiked bloody object insertions with previous generations

>>5242
>>5243

looking good! :) How did you make the spikes on the dildo? One by one or did you find a spiked dildo ready to be used somewhere?

>>5219

Someone just uploaded a Lora for straining expressions for Pony and Illu: https://civitai.com/models/1119055?modelVersionId=1257666

>>5252
I used this sketchfab model of a baseball bat with screws in it: https://sketchfab.com/3d-models/baseball-bat-73cfb07cec35484ea28bae804e9d5fd3 i like it cause of the spacing of the spikes, the short spike length, and the spikes being a different color helps. you can find a lot of good stuff from searchs for nail/spike/barbed wire bat/club. but I didn't want it to be brown or wood, so I played around with the color mapping of the image in gimp and made wooden parts black. then I also painted some red splatter on it.

next I've got a few other items from there I haven't tried yet, and will probably be trickier. needles, brazier with fire/hot coals, pliers, a pear of anguish.

These are done with the Lustify SDXL NSFW model and some Loras posted here.

Holy fucking Christ, these are GORGEOUS!

SUPER COOL.

I would subscribe to this magazine.

Great results

Can’t get the noose around lower neck?

>>4478

penis emerging from the mouth, spit emerging from the mouth,
vertical bisection, horizontal bisection.
disembowel,
viscera,
uterus,


clenched teeth, very wide open mouth,
quadruple amputee, amputee, covered stumps,
neck gash, neck slice,
multiple body parts, multiple heads,

Comment too long, view post No.4479 to see the full comment.

>>4479
This needs improvemnt.

For example what is better
single debreast or breast removal?

I need more specific tags. And advice on how ot modify the existing tags.

For example "tongue hanging from the neck stump, "
is too long

>>4479
i think one hand choke and two hands choke can just combine to choke tag

I'd be interested in seeing what tags and prompts can be used to create a greater variety of horrified and pained facial expressions, like:
https://boards.guro.cx/g/koko.php?res=42
https://boards.guro.cx/g/koko.php?res=86353
so far I've seen good response to "wide eyed, constricted pupils"
"open mouth, screaming" "tears"
and occasionally "clenched teeth" though that can also mess up results (the mouth sometimes ends up open, but with clenched teeth superimposed)
having trouble getting good variety though without messing stuff up( been using jinja2 dynamic prompts templates to create variations, but it's easy to end up with a mess)

>>4479
branding iron
burning
torture
fire torture
needles
nails

denailng
painful insertion
spiked insertion
fucked to death
flayed

Comment too long, view post No.4934 to see the full comment.

>>4734
>>4811

Somehow I doubt you'll be accommodated in this thread here. xD ...I'd suggest you repeat your request here: https://boards.guro.cx/fur/res/15075.html

>>4811
Asking around I found out the models can't handle show accurate gore. Yet. Ponyfags are making progress so quick it's driving everybody else mad.
>>4812
Humanized and EqG aren't furry anymore recolored humans "aliens" are.

>>4855

>Humanized and EqG aren't furry anymore recolored humans "aliens" are.

Yeah, my bad. I had no idea what that equestria Girls thing was but judgeing from that sample pic (>>4734) and what a quick google search told me I thought "mlp? that's got to be furry stuff". Still I maintain that this isn't t the best thread here to request that EqG stuff. I mean, just read this threads first post and look at all pics posted here. I bet the OP isn't too happy over these off topic posts... >.< (sry OP!)

>Asking around I found out the models can't handle show accurate gore.

I've experimented a little and got at least a few results you might like. But this should be continued in another thread: https://boards.guro.cx/ai/res/4860.html

>>4865
Good reply. Maybe you're right and it's not the right thread but I figured "what is possible" includes style imitation.

Nice. thanks for sharing!

<lora:singu_broken tits:0.5> <lora:singu_stomach:0.4>
Negative prompt: ahegao, empty eyes
Steps: 20, Sampler: DPM++ 2M SDE, Schedule type: Karras, CFG scale: 5, Seed: 4154010916, Size: 896x1152, Model hash: a881029f19, Model: 7thAnimeXLPonyA_v10, Lora hashes: "singu_broken tits: 9af7338b4b13, singu_stomach: eabf45d40570", Version: f2.0.1v1.10.1-previous-628-gbae1bba8

The model was hit because of censorship. Shame on them. Here's the backup.
https://mega.nz/file/Ua01SJIb#X4V9WWZ8wU1MqsVQbFbHyiZeOdjRzqD5QjHjwzsPxeU

Tags - on display, head display, behead, headless, no head, no arms, no legs, neck cut, neck stump, pole, on dildo. Less working tags - hanged, arm display, leg display, impaled, blood, guro, ryona, leg hook, upside, upside-down. Problem tags - behalf (not work), backbone (work strange), bone (mostly not work).

Reupload
https://civitai.com/models/713333?modelVersionId=798510

Sorry Due to my ignorance, I would love to make these images but I don't know anything about AI, how do I use this? And if you could share more princesses/queens I would really appreciate it.

>>4858
I'm going to assume English isn't your native language based on your post. I only speak English and Japanese so I'll type in English anyways but I've seen lots of resources in Chinese, Japanese, and some others. If you prefer another there should be plenty of resources of those languages, and a good chance at least the basics will be covered in every major language. Basically all beginners guides should cover models, webui and Loras. To be quick and short this is a Lora which modifies a model. You need to get a model and download this Lora.

Also going to cover the basics myself in case you or anyone else wants it.
Since you said you're brand new you have two main options. You can either go the paid route, by using online services hosting stable diffusion models or the local route where you will download it run it yourself on your own machine. The paid route however typically means you cannot generate whatever you want, and will have restrictions. So guro will most likely be blocked. So I do not recommend it.

To run it locally requires a desktop computer with a GPU. This is the best route if you have a PC for it, you need at least 4GB of VRAM ideally 8 or more. If you're phone posting you're out of luck, sorry.

If you go down the local route, note that it isn't going to be a one-click and done process. While I don't think its difficult there is a bit you have to learn to figure out. First you want either Automatic1111 or ComfyUI. These are basically your choice of front-end, how you will interact with your AI models. Have look through both and pick which you like. A1111 I find easier to use, but Comfy seems to be more technical and let you do more and create potentially better results.

A1111: https://github.com/AUTOMATIC1111/stable-diffusion-webui
Comfy: https://github.com/comfyanonymous/ComfyUI


Comment too long, view post No.4859 to see the full comment.

E-hentai would be a good choice. As for the videos, you can upload them on the cloud storage (e.g. Mega, GD) and just share the link.

Love this style, is there more of your art?

It works very well. Thanks for creating this model OP!

>>4631
i think it will be better to add blood and cum

can you please do a PonyXL version for this lora? or send me the training data and i'll do it.

nvm i did it, i used these pics to create a ponyxl, version ;)

>>4314
post link.

Share it with the community!

>>4318

There you go !

https://civitai.com/models/967004/packaged-concept?modelVersionId=1082724

>>4493
Nice. Will test and post results in the following days


Thanks for making it!