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 (15)
This model cannot generate male images with male characteristics. When combined with Lora related to male features, it will reduce the image quality and make the generated images very blurry!
So first you complain, then you ask for help, then you remove your failed setup so I have a harder time helping you?
Good job Mr_Fei!
https://civitai.com/posts/25976197
This post shows the usage without and with a penis-LoRA. It can do penises with both, and LoRAs work fine.
@6tZ Wait a moment. The website has blocked me. I'll re-upload it.
@6tZ The workflow has been placed under version 5.0!
I am using version 5.0 which works great, except when i want higher resolution image settings. I am using it with a LoRa which has conflicting sampler settings. Can you explain the difference between the new 5.1 vs the 5.0? Thank you!
5.1 adds a trained realism images for better skin detail, and also some style LoRAs like:
https://civitai.com/models/1727659
nice
I thought it was funny that someone gave you a thumbsdown on that comment. And then I saw it was you.
I have a 4090 with 32GB DDR5 RAM. Taking ages to generate one image. Tried using both your workflow and the linked simple ComfyUI workflow. Any suggestions or is it expected to take several minutes?
I didn't know they made 4090s with 32gb RAM.
I can get an image in 10-30 seconds with the 5090 at least. Not sure about a 4090, but should be at most 30-60 seconds.
It sounds really weird, perhaps, like, pytorch isn't correctly running, or you're not utilizing the GPU fully. Does it work well with other models, non-flux?
@6tZ Oh sry I should've clarified. 4090 is still 24GB lol. 32GB RAM is my system memory not VRAM.
It works fine with other non-flux models, once models load in it takes about 10-15 seconds to generate.
On other flux models, it takes about 30 seconds tops.
Edit: The only other flux based model I tried is a unet model so idk if that changes anything.
Changed sampler to dpmpp_2m_sde from dpmpp_2m_sde_gpu and generations are much faster now. Taking about 25 seconds each.
@6tZ mine is 1 min for a picture with flux in forge with 5090. how you do you do this ?
@supup Easy, drop forge > Switch to ComfyUI :P