Welcome to The Fux!
The goal of this model is to unlock full NSFW capabilities in Flux, one step at a time, starting with a photorealistic female focus. Enjoy the show!
Recommended Settings
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...
Description
FAQ
Comments (16)
Can you release a nunchaku version of the latest one for the Poor GPU people?
Thank you for your efforts!
Q4_K_S GGUF uploaded! Q8 coming next.
6tZ thank you for your efforts but I have noticed that GGUF is slower + gives similar quality (or sometimes worse) when compared with svdq / nunchaku ones. Not sure if it is a general thing or just on my device / workflow.
blacksheepextension870 Isn't nunchaku a node? Not sure what svdq is. But yes, GGUF is slower than normal and worse quality. This is how it works. The quantization worsens the result and makes it slower when using.
Hi, relative noob here, will this work with character loras that work with flux1d? If not, any nsfw checkpoint that is best?
Can I ask how you are able to use the model with T5xxl_fp16.safetensors and clip_l.safetensors in Forge? I cant find any UI to add the clips etc. I sense I am misunderstanding something here.
when trying to run a generation with the checkpoint I get a "AssertionError: You do not have CLIP state dict!" error... I do use ComfyUI also, but I use Forge for certain things, and I would love to get this working there. Cheers
(sorry for duplicate comment. Posted it in the wrong place first time)
You need to have an up-to-date Forge, and these selections are in the top bar, to the right of your model /checkpoint selection.
The area is called "VAE / Text Encoder".
I tried this now. Both safetensors are added in the VAE / Text Encoder, but I still get this error. Any ideas?
@jay_rich @6tZ im getting the same sort of error for Forge.
, line 63, in load_huggingface_component assert isinstance(state_dict, dict) and len(state_dict) > 16, 'You do not have CLIP state dict!'
AssertionError: You do not have CLIP state dict!
I dont have a clip option at the top. I have
UI type set to flux:
Checkpoint, Vae/Text encoder, Diffusion in Low Bits, Swap method, Swap Location, GPU weight. No Clip though
@Megasherru this is the correct one: Vae/Text encoder. I understand it as you're supposed add both of the safetensors 6tZ mentions above. Both the clip_l and theT5xxl_fp16.
I did that, but it still errors :/
@jay_rich I just tested in Forge myself, and it works here.
I loaded the following models in the VAE/Text-Encoder area:
flux_vae.safetensors (also known as ae.safetensors, or ae.sft sometimes
clip_l.safetensors
t5xxl_fp16.safetensors
Sampling: DPM++ 2M
Scheduler: Simple
Steps: 24
Distilled CFG Scale: 1
CFG Scale: 4
Here's a link to the sample image generated like this with Forge.
https://civitai.com/posts/19274850
I would strongly suggest looking at ComfyUI regardless, it's just so much more stable in general.
@6tZ Actually it is just me who is being stupid! I totally missed adding ae.safetensors too! It's working now. Thanks for helping!
@jay_rich Perfect! Let's see if @Megasherru was having the same problem :)
@6tZ I also fixed my problem, just had a Skill Issue xD
How you create prompts?
What do you mean?
Are you talking about my methods, or how you enter a prompt in a generator?