Genforall

Inspiration

Our motivation comes from the ability of generative AI to create personalized and distinctive ideas in fashion and home design. By utilizing generative models' creativity, we hope to deliver design solutions that are specifically catered to individual preference

What it does

Genforall uses stable diffusions model to generate layouts, and design variations of the input image either using prompts and depth maps.

How we built it

it's built using stable diffusion and conditional GAN

Challenges we ran into

one of the main challenges I faced is computational and memory usage. Fitting multiple models and generating variants took lot of memory usage

Accomplishments that we're proud of

it generalizes across furniture design, layout generation, wardrobe personalisation,etc.

What we learned

diffusion models

What's next for Genforall

I need to train it on monocular depth estimation using conditional GAN for pix2pix purpose so that it can create inpaints and generate variations for borders andfree-handd sketches. It's an ongoing process and i currently used the validation model for training and hope it works well.

Built With

• stablediffusion
• streamlit
• torch