If you are using ChatGPT, Gemini, Claude and so on for your prompt generation, you can feed it with <a target="_blank" rel="ugc" href="https://pastes.io/Z5HyYsy9">this</a> paste so the model knows directly how to use the LoRA - example prompts included.This LoRA captures a distinctive modern, sensual power couture style characterized by bold cutouts, sheer fabrics, body-contouring silhouettes and provocative high-fashion design (Inspired by LaQuan Smith). The model learned how to combine transparent and opaque materials, construct sculptural cutouts, create dramatic body-framing shapes and apply luxurious textures to a wide variety of garments.It works particularly well for high fashion editorial, luxury nightlife looks, runway fashion and sensual eveningwear, but the design language can also transfer surprisingly well to casual clothing, tailored outfits and futuristic fashion concepts.The training set includes many different types of outfits and couture elements, for example:<ul><li>body-hugging mini, midi and floor-length dresses with dramatic cutouts</li><li>sheer mesh catsuits, bodysuits and transparent evening gowns</li><li>strategically placed opaque panels combined with transparent fabrics</li><li>glossy patent leather-like fabrics and fitted jumpsuits</li><li>luxurious velvet and satin with draped, ruched or sculpted construction</li><li>dramatic lace-up designs, plunging necklines and high slits</li><li>embellished net fabrics, metallic studs and sparkling details</li><li>sculptural chrome bust plates and reflective metallic elements</li><li>chainmail-inspired tops, cascading crystal strands and metallic chain fringe</li><li>sharply tailored blazers, power suits and structured jackets</li><li>asymmetrical designs, abstract glossy panels and unconventional cutout shapes</li><li>bold monochrome colors, including black, red, pink, orange, white and rich jewel tones</li></ul>The LoRA focuses on materials, fabric behavior, transparency, cutout construction and body-framing design language, not on specific identities or faces.Because of this, the style can also transfer well to other fashion domains such as:<ul><li>provocative clubwear and nightlife fashion</li><li>casual streetwear and luxury urban fashion</li><li>swimwear-inspired couture and bodysuits</li><li>futuristic and metallic fashion concepts</li><li>experimental editorial and runway looks</li><li>glamorous eveningwear and red-carpet fashion</li></ul>This is, just like my other clothing LoRAs, not a "single outfit LoRA". Instead, I trained how different materials look, fall, stretch and behave, how transparent and opaque fabrics interact, where cutouts are placed and how the clothing follows and reveals the body.A small tip for sheer and transparent outfits:The LoRA can produce very revealing mesh designs, sometimes with visible nipples underneath the transparent fabric (good looking nipples are not trained!). If you prefer strategic coverage, explicitly describe opaque black bust panels, lined bust areas or sculpted opaque chest panels covering the breasts.For example:<code>sheer black mesh catsuit with sculpted opaque black bust panels, transparent sleeves and legs, and body-framing panel construction</code>This worked extremely reliably during my testing and allows you to keep the transparent couture aesthetic while controlling exactly which areas are covered.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:<ul><li>85 images</li><li>8 steps per image</li><li>10 epochs</li><li>bucks with 1152px max size, no auto scaling</li><li>cosine scheduler with 0.2 warmup and 0.8 decay</li><li>Captioning was done with OpenAI</li><li>alpha == dim: 32</li></ul>If you use it with other LoRAs, for example a character LoRA, try to lower the weight — I had good results at around 0.4.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.

Files with SHA256: 62c1884b9858e01f31e88d3408f45161e50d68720ea7466fa2f6bdf964a6ba8e