cmpt494/495 capstone project
Published:
- Supervisor: Ke Li (keli@sfu.ca)
- Expected Group Size: 4 (currently enrolled 2)
Brief Explanation of Subject Matter
3D assets take hours or days for artists to create manually. Machine learning (ML) enables 3D assets to be created automatically from real-world images captured from multiple angles. However, unlike with 3D assets created by artists, it is difficult for users to tweak and deform the 3D assets created automatically. The goal of this project is to build an interactive viewer and editor of automatically created 3D assets, to simplify and speed up the editing process. It will build on an existing software application for viewing 3D assets that does not have editing capability, whose code will be provided.
Expected Outcomes
- Designing and implementing user interface and software architecture for editing 3D assets
- Integrating user interface with the back-end machine learning model used to render the 3D assets
- Integrating with various 3D graphics frameworks to enable rendering across devices
- Optimizing the machine learning model to increase the rendering speed
Tasks Involved for The Students
Breadth: Students will learn how to design good user interfaces, integrate user-facing applications with machine learning models, and optimize machine learning models.
- Depth: Students will learn how to use machine learning frameworks, manage concurrency and integrate with open-source 3D graphics frameworks and visualization libraries.