Flux-Moja

A checkpoint trained on photographs of women in wet clothing.

Hello! We have had some success with this method, which is to train a checkpoint instead of a LoRA. Others have had great success with LoRA, but the checkpoint matched our style a little better.

The latest version, V6, is trained on base Flux Krea: PLEASE OBSERVE THE LICENSING LIMITATIONS OF THIS MODEL.

Hints for Use/Prompting

We have seen decent results come out of this model in a variety of prompt styles:

• telegraphic speech: large format sharpness. toothy film grain. cross-processed color. glistening silk wrap top in gun-metal gray.

• objective and analytical: The lighting, achieved with a single, harsh key light and a ring flash, continues to create dramatic contrast, emphasizing the intensified hues and the glistening water droplets.

• pseudo-intellectual art bullshit: the paragon of hi-fidelity pleonasms: the woman trying to live her best life. rappelling across the chasms of surreality, in and away from the blinkered philistine pig-ignorance of bourgeois america.

Recommended Parameters

Samplers: DPM++ 2M, Euler a, Euler, IPNDM

Schedulers: SGM Uniform, Simple, Beta

Steps: 22-26

CFG: 2-4.5. I have seen good results come out of every value in this range, but it all depends on the aesthetic and the prompt.