An advanced action-focused LoRA specifically trained to capture the realistic process of putting on, pulling up, and adjusting thigh-high stockings (hold-ups). Unlike static clothing models, this LoRA perfectly simulates the physical interaction between hands and nylon fabric, including stretching over the feet, smoothing out wrinkles, and firmly pressing the elastic lace top bands against the thighs. Optimized for high variability, clean anatomy, and natural skin gradients under tension. Works seamlessly with Z Image Turbo (ZIT) and Flux models.
Description
Major upgrade over V1. Completely rebuilt from scratch using an advanced 4-layer captioning system (Dataset V2). This version focuses heavily on active clothing mechanics, fixing previous anatomy bugs, and delivering vastly superior interaction between hands and the nylon fabric. Expect much more realistic stretching, highly detailed lace top bands, and perfect skin contours during the action of pulling up and smoothing the stockings.