This LoRA was trained on different undressing images with different materials and garments, including coats, jackets, bras, dresses, tops, bodysuits, strings and pants and probably more that I didn't mention. The goal was to teach Flux how undressing works, where the hands are placed, what happens with the fabric and what gets exposed. The captions described the placement of hands, how and where they grab the fabric and in which direction the garment is being removed, making it possible to control the motion pretty well. The dataset was mainly NSFW, but SFW is possible - you just have to tweak your prompt and mention that specific parts are covered. Avoid the word "revealing" and "exposing" in these cases. Check the showcase to get an idea how to prompt.

Would love if you post your creations here with it.

The LoRA was trained on the default flux1-dev model for maximum compatibility.

Important: this is not an anatomy NSFW LoRA, load a NSFW LoRA or model for anatomy!

Usage: Always use the trigger word undress_dynamic, then add the matching trigger word of your desired action (undress_pull_up, undress_pull_down, undress_unzip, undress_open_front, undress_slide_off, undress_pull_strap, undress_lift). You can also mix if it makes sense, no responsibility taken if you use conflicting combinations like pull up and pull down ;)

Dataset and training data:

• 124 images

• 10 steps per image

• 10 epochs

• buckets with 1152px max size, no auto scaling

• cosine scheduler with 0.2 warmup and 0.8 decay

• Captioning was done with OpenAI

• alpha == dim: 32

If you use it with other loras, for example a character lora, try to lower the weight - I had good results at around 0.6.

As always: I did not polish any of the images in the showcase, they all came out like this. I did 2 images per prompt and picked the better looking one.