A nude/NSFW capable model that focuses on images with a female primary subject.
It is heavily biased towards nude images.
Recommended Settings
Sampler is dpmpp_2m (DPM++ 2M) and the scheduler is beta
No VAE or CLIP is baked in. Use separate sources for those.
Each sample/preview image contains the used workflow. Here's a quick article with simpler more beginner-friendly workflow. This is a recommended starting point.
FP8 Download
For some reason, when you upload multiple formats (BF16 + FP8) on one model, which they now support, they hide the FP8 under the BF16 model group in the side menu...
Early Access is enabled to support the development of new versions and models.
Description
FAQ
Comments (37)
Which image do you reckon should be the front cover? Vote for the best image with thumbs up and I'll make it the cover :)
the thumbs up are wrong. your cover is the milk drinker, clearly.
@Narz That's what I was thinking was gonna be the cover as well. Most impressive/interesting image. But I guess not the most appealing in general?
@6tZ i'm of the cynical opinion that people are mesmerized by virtual girlfriends and can't see much beyond symmetrical semireal Pony-bimbos.. so i'm admittedly biased lol
5000 Buzz?
I mean, that's just the minimum. There's nothing stopping you from increasing it.
Thank you for your patronage!
In case you would rather wait, just click the 2.3 version. It's almost as good!
It's the GGUF version but the extension is safetensors. Does that need to change for ComfyUI? GREAT WORK BTW
Make sure you are on the correct model card version at the top row.
Which exact version are you getting the incorrect format for?
Should you gen using this checkpoint or should you gen using original Flux-8 first but use this checkpoint?
This is a separately trained model, so you use either this, or the original Flux-8.
So if you want nude images, you use this model.
If you want normal flux, you use the original model.
@6tZ I have both normal and nude, so my choice should be this or that?
@Arika07 I don't understand what you mean.
Based on what you want to generate, you switch models.
Get both, use both for what they are good at :)
@6tZ Sorry I come back for another comment. Is Penis not possible to generate at all in this checkpoint? ^^; I try so many Penis keyword, every boys turn to have pussy only!
@Arika07 Sorry about that! I'm trying to get male and female in at the same time, but you mostly just get mussies. If you want male, I suggest my Pulsar model:
https://civitai.com/models/884686/pulsar-male-gay-nsfw-base-model-flux
@6tZ Hi. It took a lot of time to train my Flux lora (character, my own) data. So if i train on regular fp8 Flux Lora, will it work on both of your male checkpoint and female checkpoint NSFW? Or do i need to train Lora from one of your checkpoint to make it work. Each separately trained?
@Arika07 I'm sorry but I don't have an answer. I haven't tried this.
I would love to know the answer though! So I would really appreciate to hear back from you if you end up making a version of it.
What I know, is that I have always used the Flux base model for my trainings.
I have tried training with this model and others as the training checkpoint, but it doesn't make it better from what I have seen.
Both seem to work fine.
@6tZ Because it took really long time to train, okay, I will start doing it tonight then, with the base model first and see if it can use on both checkpoint. Last question. Is there any checkpoint for Flux that is possible to have hetero sex yet? I means both male and female in the same image. Because your checkpoint would only allow me to choose male or female in separate checkpoint, so having them on sex is not yet possible ^^; Is there no any possible way for that scene now?
@Arika07 As far as I'm aware, no. There is none.
I'm trying with this one, but it's not working. It's doing males and females at the same time, but no separate genders.
It's trained and has several trained LoRAs of this subject merged in, but it's not picking up on it at all.
I think it needs very special multi-concept training into the checkpoint itself, which is something I cannot train locally. So I need to run training for that on the cloud, and even if I get that training set up, there's no guarantee it will work.
I don't use a GUI (no Forge or Comfy or ANYTHING) and run FLUX programatically on CPU alone using Python scripts directly from a command-line.
I have NO IDEA how to even refer to these in a python script so cannot use them. I wish I could. When I figure it out, I'll write it up in an article to help everyone.
Until that happy day, I'd LOVE to try one of these, likely the 8bit ones, but have no idea how to use them.
But they look great so... great work. Thank You.
You are crazy. How long do your gens take
@gub An average of 7 to 8 minutes each when running three additional LoRAs. NOT as mad as you'd think. Whole system has no GPU, no CUDA and only CPU. My friend got two rack servers out of a skip (dumpster) behind an industrial unit and gave me one to play wih. Check out my article;; https://civitai.com/articles/8029/flux1-schnell-flux1-dev-on-an-elderly-computer
That's cool! I have no idea how to convert to nf4 format, or whatever it is called, to run on potatoes. If I figure it out I'll be sure to post a potato-model for you @Davros666 !
@6tZ fp32
@Davros666 bro...get yourself a GPU. That is inhumane. You don't deserve this
@gub I enjoy pain, confusion and anguish. It's why I live in the UK.
@Davros666 that's amazing, working out how to use "e-waste" to run state of the art AI models.
Another great NSFW checkpoint except... again, the obsession with shaved ladies. Why? Why can't we just have natural people as they are, simply?
Bro go download a Lora to make your women hairy. Most people don't want to see that
My guess is that it's harder to achieve good quality dataset for training if you want what you want. That's the world we live in.
Every woman I've been with has, of their own accord, shaved their pussy. I've had to specifically ask them to keep it a bit more natural, and most of them were against the idea. I think shaved is, in this day and age, more "normal" than unshaved.
@grossExaggeration There are 4 billion woman on this planet. You've asked every woman you've been with. Would you agree this is the definition of anecdotal? username checks out.
See my works, dude) There are a lot of natural&unshaved women!
@sparkleboy599 I'm cooking something for you right now. I hope you will like it :)
@6tZ Looking forward to it!!!
@sparkleboy599 https://civitai.com/models/1324291/retro-vintage-realistic-ilxl-illustrious-xl-nsfw-sfw
Only Illustrious done so far. The Flux training didn't take it seems. Need to retry or train longer. Very weird honestly.
@6tZ let me know when a flux version is available. Also, do you train your model with masks over faces, so that it won't affect loras?
@sparkleboy599 I do not train with masks over faces, no. From my light testing it's not really needed.