In my testing it works well with T2V and I2V both. Use on high noise only, it was trained on images so all motion comes from default "understanding" within the Wan 2.2 checkpoint. I haven't had to use it with low noise.

Strength of 1 to 1.6 works well - crank up or down as needed; typically i use around 1.

Trigger phrase: woman has [color] ring gag in mouth fastened with [color] leather straps.

Notes:

• The act of applying the gag itself seems to not work well on T2V. Usually all T2V just produces the woman with a ring gag in her mouth already. The act of applying the gag works better on I2V.

• T2V seems to default more towards silver/black ring gags and black leather straps. I2V seems to be able to be more flexible on the color of the ring gag and straps.