The model is fp8 and based on Flux.1 D. It comes without VAE, Clip and textencoder to save disc space. So be sure to load those files in your workflow. You can find example workflows embedded in my posted images.

Suggest parameters for image generation:

Sampler: Euler

Scheduler: Simple

Steps: 20

Distilled CFG: 3.5

Other sampler/scheduler combination with great results:

EULER / beta with 25 steps

DDIM / ddim_uniform with 25 - 38 steps and CFG 2.7

UNI_PC_BH2 / beta with 30 steps and CFG 3.5

NSFW images are possible but thats not the area were this model wants to shine.

For CLIP I recommend using the new improved version from here:

https://huggingface.co/zer0int/CLIP-GmP-ViT-L-14/blob/main/ViT-L-14-TEXT-detail-improved-hiT-GmP-TE-only-HF.safetensors

Have fun - drop your images here and leave a review if you like.

Thanks for trying and sharing your experiences.