How to use:
Start with a weight of 0.7 or so. Use the trigger word "handtie" near the front. However, this is where it gets tricky - if the person's arms aren't together, then you won't see them their arms bound, so I recommend you add something like "hands bound together with rope" somewhere in the prompt as well. Or use one of my sample images in a ControlNet on top of this LoRA.
Although they're not the point of this LoRA, you can also add lingerie and blindfolds to your images in the prompt since a lot of training images contained those and were labeled accordingly.
Description
FAQ
Comments (3)
What checkpoint did you train this on / use for your image generation? Having some trouble, so keen to know how you got it work.
Good question and I'm glad you asked. Technically, I fine tuned it against SD 1.5 but what I did was I fine tuned a LoRA of Objective 2.0 that I extracted (using SD 1.5 as the base) on my own dataset. Then I applied that LoRA to Objective 2.0 to generate the images.
Your models are great in general, but I'm having trouble with this one. I think I'm doing something wrong (others got great results), all my pics are coming out messed up. I posted a few examples of my results here (now: 2/8/2024, 11PM EST), do you have an idea what I'm missing?
And is there a specific term for my problem (mixed kaleidoscope of colors in pic)? I've been calling it "noisy", and i want people to know what I mean.
Thanks a lot, I love your work!
Details
Available On (1 platform)
Same model published on other platforms. May have additional downloads or version variants.