Green Giant
Stream lighter. Watch sharper.
Inspiration: Video streaming uses a huge amount of data, energy, and infrastructure, especially at higher resolutions like 4K. We built Green Giant to explore a simple idea: instead of always streaming the most data-heavy version of a video, can we stream a lower-resolution version and enhance it locally?
Our goal was to reduce bandwidth and energy use without making the viewing experience feel noticeably worse.
What it does: Green Giant is a Chrome extension for YouTube that improves video playback in real time. It lets users stream at a lower source quality while using local visual enhancement to make the video look sharper and clearer.
The extension includes: Real-time WebGL video enhancement An experimental AI upscaling path using ONNX Runtime Web A live estimate of bandwidth and energy savings Direct integration with the YouTube player
The purpose is not just sharper video. The goal is to make streaming more efficient by reducing the amount of data that needs to travel across networks and data centers.
How we built it: We built Green Giant as a Manifest V3 Chrome extension that runs directly on YouTube pages. The extension detects the active video, creates an overlay on top of the player, and processes frames locally in real time.
We used WebGL for the main enhancement pipeline because it allows fast GPU-based image processing. We also experimented with ONNX Runtime Web to test whether AI-assisted upscaling could improve the result.
Our stack included: JavaScript Chrome Extension APIs WebGL Canvas 2D ONNX Runtime Web esbuild
Challenges we ran into: The hardest part was balancing sustainability, quality, and performance. The extension only makes sense if the local enhancement is lightweight enough that it does not cancel out the savings from streaming less data.
We also had to work around YouTube’s dynamic player. Fullscreen mode, theater mode, navigation changes, and quality switches all made overlay alignment more difficult than expected.
Another challenge was getting ONNX Runtime Web working inside a Chrome extension environment, especially around model loading, browser security, and runtime assets.
Accomplishments that we’re proud of: We built a working prototype that connects sustainability with real-time video enhancement. Green Giant can enhance YouTube playback locally, estimate potential savings, and show how lower-bandwidth streaming could still feel high quality to users.
We’re also proud that the project treats sustainability as an engineering problem, not just a message. It asks how software design can reduce waste in everyday digital habits.
What we learned: We learned that sustainable software is often about tradeoffs. Reducing environmental impact does not always mean doing less computation. Sometimes it means moving work to a more efficient place.
We also learned that perceived quality matters more than perfect reconstruction. Users do not need a mathematically identical 4K stream. They need video that still looks good while using fewer resources.
On the technical side, we learned a lot about browser graphics, Chrome extension architecture, real-time rendering, and running AI models in the browser.
What’s next: Next, we want to improve the AI upscaling path, measure real energy and emissions savings more accurately, refine adaptive quality logic, and make the extension more reliable across different video scenarios.
Log in or sign up for Devpost to join the conversation.