The sweet spot is between 0.50-0.30. Anything higher will most likely end up looking like the training data, which was a render I made in blender.
The hogtie LoRAs on this site are good but few of them have that head to ankles tether pull that can make a hogtie so much better. So, this fixes that.

