Analuk is my most popular model, already made for SD1.5 and SDXL, here is her SD3 version.

Image generation info:
Weight 1 is fine, you even can raise it a bit.
I used the default ComfyUI workflow coming with SD3 medium (just added the Lora loader): based SD3 model, 28 steps, cfg4, euler, random seed.

Prompts stay simple: photograh of a lady wearing a floral dress, creation by <artist name> of a lady, ...

Negative: obese, featureless but you can try without any.