AI Dress Try-On Image Inpainting

I am building a virtual try-on feature and need an AI workflow that can both generate and inpaint high-quality images of dresses and skirts onto user photos. The focus is firmly on clothing and fashion items—specifically casual and formal dresses as well as skirts—so the system must understand fabric flow, drape, and alignment with a human silhouette.

Here is what I need done:
• A diffusion-based pipeline (Stable Diffusion, ControlNet, or comparable) that can create new dress and skirt variations from text prompts or reference shots.
• An inpainting module that places those garments convincingly on a provided model photo, matching pose, lighting, and perspective while preserving natural skin tones and background details.
• Output at 2K resolution minimum, delivered as PNGs plus the reproducible code (Python notebook or script) and a short README explaining dependencies, prompt crafting, and how to fine-tune for new styles.

Acceptance criteria
• Seamless boundaries between clothing and body with no color bleed or warping.
• Fabric texture and folds remain realistic at full resolution.
• Separate checkpoints or embeddings for casual and formal wear so I can swap styles quickly.

If you have a proven track record with AI image generation, inpainting, or virtual try-on research, I would love to see a quick sample or demo link when you reply.