GaussiScape | Devpost

Inspiration

Transforming 2D images to 3D affects fields that are worth hundreds of BILLIONS of dollars. 3D environments are significant due to several applications and here are some reasons why this conversion is crucial: Enhanced Visualization and Immersion: Depth Perception: 3D models provide a clearer understanding of depth and spatial relationships between objects, something that's not easily captured in 2D. Interactivity: Users can view 3D models from any angle, providing a more comprehensive perspective and interactive experience. Medical Imaging: Diagnosis: 3D reconstruction from 2D medical scans, like MRI or CT, allows doctors to view and analyze anomalies in the body more precisely. Surgery Planning: Surgeons can plan interventions better by visualizing the exact morphology and position of organs, tumors, or vessels in 3D. Entertainment and Media: Video Games: Modern games rely on 3D environments and characters for realistic and immersive experiences. Movies: 3D modeling and animation have become staples in film production, especially in CGI-intensive movies. However converting 2D to 3D is extremely difficult. There is a steep learning curve, a powerful amount of software needed, and a high cost involved. In this project, we seeked to use the emerging use Gaussian splatting technology to make this process much easier.

What it does

Leveraging the breakthrough with Gaussian splatting technology, GaussiScape is a web tool that transforms simple 2D imagery into immersive 3D landscapes. With applications spanning virtual/augmented reality, education, and medical imaging we're not just creating a tool—we're pioneering a 3D digital revolution.

GaussiScape takes a collection of images of objects or environments and converts them into a virtual 3D environment, maintaining visuals, shape, and quality. Say goodbye to extreme learning curves, insane costs, and ridiculous softwares.

How we built it

Frontend: React, JavaScript, Vanilla HTML/CSS

Backend: Flask, Python, Gaussian Splatting, Embeddings, Node.js

Challenges we ran into

With Gaussian Splatting being such a new technology, we ran into many errors while building our model. Documentation was unclear, not many people had built stuff before on it so there was a lot of trial and error

Accomplishments that we're proud of

Making a very impressive UI and finishing the project even though we were working with a new technology in unfamiliar territory

What we learned

We learned to be alot more patient with the process and understand that some functionalities didn’t work with the model simply because it was so new so, we could always find a fix if we worked towards it later down the line

What's next for GaussiScape

We want to implement these 3D environments into specific fields. Bettering environments in the metaverse, making VR more real, the options are limitless.