Virtual Try-On LoRA

Inspiration

We saw the power of new AI models and had a simple, practical idea: use them to solve a real-world problem. The uncertainty of online clothes shopping was the perfect target. We trained a specific LoRA to master the single task of realistically putting digital clothes onto real people.

What it does

Virtual Try on lora put clothes on a person/model. Firstly, we have to combine model photos and clothes. Then the try on lora that we trained effectively place the clothes realistically on the person.

How we built it

We used FLUX.1 Kontext [dev] as base model We trained a specialized LoRA. For lora training, we made a special dataset for this. Also, We built a interactive Gradio app and hosted it on Hugging Face Spaces so anyone can try it instantly.

Challenges we ran into

Creating perfect dataset was challenging. We need to create and put clothing on images dataset manually as I didn't found this type specific dataset for my need. Creating Gradio web app was also challenging in this short time. But at last, the results went well.

Categories that aligns with the project

Best LoRA (I trained a custom lora) Best Creative Vision(Made an interactive web app to use it) Best local use case (Provided clear instruction in model page to run the gradio app locally)

Accomplishments that we're proud of

It actually works! We have created a virtual try on LoRA by which we can try on any garments on a model , not just another style transfer. We are most proud of building a fully functional, public demo that people can use and have fun with right now.

What we learned

This project was a deep dive into the creative potential of the FLUX.1 Kontext [dev] model. We learned that fine-tuning a custom LoRA can unlock entirely new abilities in a model.

What's next for Virtual Try-On

I am going to make a AI image generator and photo editor SaaS. So, we can add very good features on my SaaS and explore more possibility.

Built With

• flux-kontext-dev
• gradio
• lora-training
• python