FLUX v1.0 :
Trained on FLUX dev with 40 photos of Lucy Liu in the 1990s with detailed captions. Tested on FLUX 1.dev (full) and FLUX fp8 and FLUX nf4 ! Use around strength 0.8-1.2. No keywords needed! Distilled CFG around 1-4 and CFG 1.0 (without negative prompt). Clipskip 1. Can be used for example as follows:
Positive : {Artstyle, Character and scene description in usual FLUX fashion}, , <lora:lucyliu_1990s_local_flux_1_standard-000039:1>
Description
FLUX v1.0 :
Trained on FLUX dev with 40 photos of Lucy Liu in the 1990s with detailed captions. Tested on FLUX 1.D (full) and FLUX fp8 and FLUX nf4 ! Use around strength 0.8-1.2. No keywords needed! Distilled CFG around 1-4 and CFG 1.0 (without negative prompt). Clipskip 1. Can be used for example as follows:
Positive : {Artstyle, Character and scene description in usual FLUX fashion}, , <lora:lucyliu_1990s_local_flux_1_standard-000039:1>
FAQ
Comments (7)
Do you train locally?
I captioned it locally
The likeness is amazing! Great work! 👏
Thanks! Did you test it out yourself? If you got good results please post them if you want!
@steffangund Not yet, but will see what I can do! I was just so amazed by the showcase images already!
@redpinkretro Thanks! But its always good to see how people use the resource and it gives me better ideas of how to prompt. All my prompts are relatively simple and uniform.
@steffangund Understandable, I will make a few that hopefully work without triggering any problematic tags. I had screenshots of workflows and landscape pictures without any people trigger "sexual interaction" tags and whatnot 😅