Scrbbl.ai

As AI images continue to make waves thought pop tech, an interesting and fun implementation is due. Thus we present: Scribbl.ai, a game based on Scribbl.io to test your image deciphering skills!

Comment

Inspiration

scribbl.io, Stable Diffusion

What it does

It is game-like scribbl.io, except it has players create an image with a provided prompt that is processed by Stable Diffusion's Image-to-Image processing.

How we built it

We are using: Figma, HTML, CSS, and JavaScript for our frontend, Flask for our backend, MongoDB for the database, and Amazon S3 for storage, but attempted to use RunPOD for cloud computing

Challenges we ran into

Unspecified stable diffusion API We did not check if Digital Ocean Then we ran into AWS issues as we considered creating our own Docker image, but different implementations depend on running stable diffusion without an iPython notebook. Web Scraping via Selenium for a makeshift deployment RunPOD didn't have an API, so we had to try and use SSH or bash commands

Accomplishments that we're proud of

We have a decent back-end to build off of The concept is extensible and works for many different games Can easily add different technologies

What we learned

Understanding the stack and road to deployment more Communication during development (especially between front-end and back-end) could be an improvement We could have had better prioritization Perhaps too ambitious for a Hackathon; could have had more modest base goals and then incrementally add stretch goals

What's next for Scrbbl.ai

We plan to develop the web app to completion Expanding AWS implementation from just S3 Reach out to the community to learn how to leverage stable diffusion APIs Potential monetization in the same vein as other web-based games such as Geoguesser

Built With

Share this project:

Updates