Purpose

Limited to one specific pose:

• Female standing, facing forward, with knees and toes turned inwards, AKA a knock-knees and pigeon-toe.

• This isn't a good lora. I'll make a v1.0 some day. But it does this one pose, so I'm sharing it.

Gallery images:

• left side = lora, right side = no lora

• Workflows are embedded for the Krea gallery

Krea2 version

Usage:

• Use strength is 0.6 to 0.8. There's no trigger.

• Even though it wasn't captioned, adding "her knees are pressed together and feet touching" to your prompt will enhance the pose.

• Don't use "shy pose" in your prompt unless you want the hands near the face, and face turned away. That effect isn't caused by this lora, but from the Krea2 base.

Limitations, sadly:

• No males

• No Side-profile

• No Sitting or lying down poses

• Causes slight de-aging

Training method:

• I used LLM captions and 20 images: 10 closeups of legs+feet, and 10 full body with all faces obscured.

• I used 1,200 steps, starting with super high learning rate and sharply descending, then trail and error block tuning.

• Normal learning rate with more steps would have been better!

ZiT version

Usage:

• Start your prompt with the full phrase, "A woman is standing in a shy pose with her toes pointing together."

• Use strength 1.3.

Limitations, sadly:

• Only one person

• No males

• No side profile

• No Sitting or lying down poses

• Impacts faces and other aspects of the image

Training method:

• Trained for $2 on a service that had no settings

• I used 20 images: legs and feet only (I learned later this is a bad technique)