Gynecology Chair!?

Yes indeed.

A beautiful {Russian|French|Swedish|Swiss|African|Latina|Austrian|Dutch|English|American|Siberian|Chinese|Ukrainian|Thai|Czech|Japanese|Indian|Caucasian|Spanish|Portuguese} woman is sitting in a gynecology chair, ankles bound, wrists bound, legs in stirrups,

She has {very long|long|short} {blonde|brown|dirty-blonde|black|auburn} hair and is wearing a bikini,
{
A doctor is {putting a blindfold over her eyes|{sitting|standing} near her|securing her wrist to a support}||
}

indoor lighting, doctor's office, {high-angle|three-quarter|front} view

Training Notes

WAN2.2 is a lot easier to train. I was really concerned about the time required to train two models and test two models but I am having lots of luck with shorter clips and the resulting models seem so much stabler than what I was used to with HunyuanVideo that epoch selection is taking me way less time.

I was exploring a single LORA file with a fellow user but it did come with slightly less quality, so two models for full quality it is!

My captioning style matches my wildcard prompts pretty closely in case you are curious.

A note about LTX2.3, as I've been getting requests for it. Everytime I read up on whether or not it's worth trying that out, I keep seeing that WAN2.2 is better, so... not planning on doing that anytime soon.

Disclaimer

Ask your doctor!