DreamScape

Problem ID
Demo screen 1
Demo screen 2

Inspiration 🪄

We were inspired to build a project using Stable Diffusion technologies given the impressive outputs of the existing models 🤯. This technology pairs especially well with creative storytelling; as such, we considered relevant pain points that this technology could help address. We deployed our own Stable Diffusion Webui (https://github.com/AUTOMATIC1111/stable-diffusion-webui) that uses FastAPI, we narrowed in on the concept of dream journalling and thought through how we could apply Stable Diffusion to enhance the process.

What it does 🤔

DreamScape is an immersive dream journalling platform 📝. Simply describe your dream, and illustrative visualizations will be instantly generated (powered by Stable Diffusion AI 🧑‍🎨 ). The platform offers features to store your past dreams and share your dreams with your friends. Our goal is to help users capture the magic of their dreams and imagination 💭!

How we built it 🛠️

Framework: Next.js, Typescript Database: MongoDB Authentication: NextAuth Technologies: Stable Diffusion (sent prompts along with 20 training steps, CFG_scale of 7, sent previous image's seed to produce images similar to the previous one generated to maintain consistency across images) Styling: Flowbite, Tailwind CSS

Challenges we ran into 🫠

The first challenge that we ran into was idea generation. It took nearly 3 hours of research and discussion to finalize our idea, which was much longer than we had anticipated for this step of the project. We found ourselves trying to come up with solutions prior to identifying existing problems and pain points to solve 🔍, which resulted in some inefficiencies during discussion. Moving forward, it would be very helpful to use design thinking 💡 to approach the ideation process and to think about potential ideas prior to the start of the hackathon. A second major challenge that we faced was working with Next.js 14... it had weird API routing compared to old technique.

Accomplishments that we're proud of 😎

We are proud of working with new technologies such as Stable Diffusion. Given the recency of technological developments in the AI space, we believe that there is huge potential in the application of such technologies. We hope that our project is able to make an impact by demonstrating creativity and innovation 🤓.

What we learned 🥹

We learned how to work with Typescript and a NoSQL database (MongoDB).

What's next for DreamScape 🗿

Next steps for this project would be to polish the features and UI of the platform to ensure a seamless and delightful user experience 😊. If we had more time, it would have been great to finish building out the platform's social features, allowing users to interact with the dreamboards of their friends and share their own. From a business model perspective, it would be meaningful to think of ways to incentivize consistent journal entries for users. To improve the accuracy of generated images, we could potentially allow users to customize some Stable Diffusion call parameters and enhance user inputted descriptions using NLP. Additionally, it would interesting to try training a custom Stable Diffusion model to generate dream-like illustrations ☁️. Alternatively, we could consider offering users the ability to customize the art style of their dream visualizations -- this could be quickly implemented by including a field where users can pick from a selection of style LoRAs.