This Lora helps produce a girl with her arms bound behind her back, mainly with rope. Models are already alright with this, especially any bondage mixes, but this was made to help with consistency and test the effect of only using illustrations to train bondage. Plus, it helps if you don't want to use a bondage mix either. Generally hands are just obscured while using this, it definitely doesn't help them when they do show. Although I did tag what I saw, I didn't end up including enough "from behind" images in the training set for this to perform well for that.
Trigger Words:
bound arms, arms behind backtechnically only "bound arms" is the trigger, but adding "arms behind back" seems to improve its performance.
Note: one issue I've found is a tendency to try and add a gag regardless of prompt, so add
gag, gaggedto the negative prompt to counteract that.
Description
Trained on animefull-final-pruned using 564 images with 2 repeats, batches of 3. Used DAdaptAdam
FAQ
Comments (8)
i'd love a version of this that is rear view!
Another user has got the covered actually! https://civitai.com/models/96229/bound-shibari-from-behind
very good lora, hope could do it on 2 girls in the same picture
In theory it could, though I think I did only train it on single subject images. I did the same with my tape gag one and that worked for multiple subjects when it was the only active Lora. Worst case you ought to be able to make it work with latent couple.
yes, i did 2 girls with no other loras, when working with other characters loras only could generate one character, but that's alright, I can redo characters face with inpaint
@eric18zhao Yeah I've just never had any real luck getting multiple character Loras to work at once personally
i have add one of my ai work, but don't know why the prompts were unable to be added
It seems this lora tend to be cartoonish even prompt requires to be realistic. Also can use more gestures.