This LoRA captures the distinctive avant-garde couture style characterized by dense bead embroidery, colorful crystal embellishments and sculptural fashion design (Inspired by Kevin Germanier). The model learned how to apply complex bead surfaces, bold color blocking and dramatic couture silhouettes to a wide variety of garments.

It works particularly well for high fashion editorial, runway looks and experimental couture designs, but the style can also transfer surprisingly well to casual clothing or unusual concepts.

The training set includes many different types of outfits and couture elements, for example:

• couture dresses covered in dense multicolored bead embroidery

• garments with crystal embellishments and reflective bead textures

• bold geometric color blocking patterns

• sculptural avant-garde silhouettes

• couture jackets and structured tops

• experimental cutout designs

• embellished bodysuits and evening gowns

• dramatic runway couture styling

• extravagant accessories and headpieces

The LoRA focuses on materials, surface textures and design language, not on specific identities or faces.

Because of this, the style can also transfer well to other fashion domains such as:

• casual streetwear

• swimwear

• futuristic fashion concepts

• experimental editorial looks

This is, just like my other clothing LoRAs, not a "single outfit LoRA". Instead, I trained how fabric works, falls, looks and behaves and where the cutouts are places and how much skin is exposed.

Would love if you post your creations here with it.

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

Dataset and training data:

• 105 images

• 10 steps per image

• 10 epochs

• bucks 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.