This LoRA is experimental and I am seeking feedback.

This LoRA is trained to use ControlNet (Depth).

The effect of not using ControlNet is undefined.

The purpose of this LoRA is to bind pretty girls beautifully.

(Assumes the Anime model).

This LoRA is used in the following steps

Step1.

Prepare a base image.

Step2.

Using any paint tool, draw the position where the rope is to be drawn.

It is recommended to use high contrast color as much as possible.

Step3.

Convert the image of Step 2 to Depth Mapping and generate the image.

At this time, the CheckPoint and the prompt used in Step 1 should be used, and the prompt should be

<lora:kinbaku:1>kinbaku,

is added to the prompt.

At this time, optionally add

shibari,shibari over clothes, crotch rope, and

etc. can also be added optionally.

Adjust ControlNet's Control Weight value around 0.9 - 1.3.

Step4.

The final step is complete.

If the rope is not drawn well, you may need to adjust the image in Step 2.

This LoRA may still not succeed because of insufficient data.

Please.

I want the Depth Mapping image you created in Step3.

Thanks