Performaxx | Devpost

Base template before making any inputs

Inspiration

We were inspired by the rise of the performative aesthetic, the intersection of identity, art, and irony in online culture. From matcha and tote bags to Clairo-core playlists, we wanted to capture that vibe with AI. The goal was to create a tool that helps users improve their performative energy through creativity, not judgment.

What it does

Performaxx analyzes an uploaded image and gives it a performative score from 0–100, where 100 represents peak aesthetic performativity. It also:

Modifies the given image to improve performative aesthetics and fully maximizing performative prowess.

Includes an AI chatbox restricted to performative assistance, powered by Gemini

Generates a personalized shopping list with missing aesthetic items and prices

Plays a matching random performative song

How we built it

Using various python libraries, we created a gui where you can upload an image. This image is then processed and analyzed by a synced Gemini API, which returns a performative score and a version of the image in which the person has a maximum performative score.

We then randomly selected a song that matched the aesthetic of the image uploaded, and played it by calling Deezer API.

Then, by hardcoding various performative items the AI might find missing, we constructed a shopping list with prices that the user is missing.

Finally we created a wrapper which connects to Gemini and inputs performative related queries to be answered.

Challenges we ran into

Wanting to use Spotify, we built the entire project focused around the Spotify API, having past experience with it. Unfortunately, we found out that it was recently discontinued, and had to switch to Deezer last second, causing a lot of panic and stress for us.

Because of the way that Pygame works (the library we used to create the gui), it only renders things frame by frame, which meant that when we played the song, it waited for the entire thing to finish before moving on and generating the next frame, which was our newly generated image. This was a big issue, since a 30s+ wait time is terrible when waiting.

Accomplishments that we're proud of

Building a fully functional image-based scoring system in under 12 hours

Learning how to wrap an AI to use as a chat bot in our program

Using AI to generate more performative versions of pictures

Creating a costume good enough to win the costume award.

What we learned

We learned the best way to utilize AI to build our program, as well as integrating it into key components of our project. Things like the chatbox, image uploader, image generator, etc. where all very new and foreign concepts to all of us, meaning we had to learn how to create and utilize them efficiently within a 7 hour hackathon.