AI photobooth

Our initial test picture that we wanted to manipulate
A fun Bob Ross inspired AI generate image
Something a little more anime-inspired
The team proud of their accomplishments
One of our team members is now looking away from the camera
We turned into a crowd!
This one is especially apocalyptic

Inspiration

A passion project by one of our team members and a long-running idea, inspired by recent developments in AI image generation was allowed to come to fruition in a fun and time-crunching manner through the 2023 HackUTA event.

What it does

A fun and intuitive program that allows users to take pictures and see their portraits transform through AI, the AI photobooth enables inspiration and encourages fun through a simple application that takes a photo, processes it, and returns it to the user via a QR code in order to retrieve their works of art.

How we built it

Using a Raspberry Pi, ArduCam Hawkeye camera module, and a spare monitor, we were able to remotely develop using Github services in order to program, troubleshoot, and deploy our application in a timely manner.

Challenges we ran into

We couldn't figure out why our locally hosted images would not upload to Replicate's servers, we instead decided to use our already deployed Google Cloud server to serve the images to the API. Learning the Git workflow was a hurdle we had to go through. Raspberry Pi dependencies were tedious, but our team members managed to pull through and deploy our program.

Accomplishments that we're proud of

An impressive integration of the Replicate API as well as the Google Cloud program

What we learned

We learned how to implement Replicate's APIs to remotely run stable diffusion for both prompt generation as well as image generation. We utilized Google's cloud services in order to both host the images as well as serve them to our users via QR codes.