This is a bit more of a complex lora so you'll need a patience and the willingness to tweak if things are not working, but they work well 95% of the time.

Works well on base and edit - use a strength of 1.0 plus/minus a bit as you need to adjust. Heavily defaults to nudity, sometimes even with clothing prompts. So just FYI.

Follow the trigger prompt and framework as below for how training was setup. It does work well with natural language descriptions too, if trigger prompt is used a as a trigger word in the image.

• Trigger Prompt

  • "woman tied in a spreadeagle pose"

• Follow up restraints prompt - SEE POSTED EXAMPLES:

  • "wrists and ankles tied with [color] [material] anchored to [(material) posts/(material) frame/bed/table/(material) cross] with [color+material (only if its different than restraints)] "

• Restraint Pose Prompt (trained on specific ones, try others to see what works)

  • standing

  • suspended in air with feet off ground

  • lying on bed/table

  • lying on her belly on a bed/table

• View prompts

  • For bed/table: side view, front view, and top down views should work

  • For standing: mostly front view, but side and rear can also work