⚡ electrical engineering

Video Quality Evaluation Methodology and Result of AV2 Compression Performance

This paper outlines the evaluation methodology defined in the AV2 Common Test Conditions and demonstrates that the AV2 (v13.0) standard achieves significant compression efficiency gains over AV1, with BD-rate reductions of 29.81% for PSNR-YUV and 33.79% for VMAF under random access configurations.

Original authors: Zhijun Lei, Vibhoothi Vibhoothi, Dzung Hoang, Yixin Du, Ramzi Khsib

Published 2026-05-18

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Zhijun Lei, Vibhoothi Vibhoothi, Dzung Hoang, Yixin Du, Ramzi Khsib

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you have a massive library of movies, games, and videos. Right now, these files are like heavy, bulky suitcases. They take up a lot of space on your hard drive and are slow to ship across the internet. The Alliance for Open Media (AOMedia) is a group of tech giants (like Google, Apple, Amazon, and Meta) working together to build a new, super-efficient way to pack these suitcases. They call this new packing method AV2.

This paper is essentially the "rulebook" and the "report card" for how they tested this new packing method to make sure it actually works better than the old one (called AV1).

Here is a breakdown of what they did and what they found, using simple analogies:

1. The Goal: Smaller Suitcases, Same Quality

The main job of AV2 is to shrink video files significantly without making the picture look worse. Think of it like folding a shirt. The old way (AV1) folded it neatly, but the new way (AV2) uses a special "origami" technique to make the shirt take up half the space in your suitcase, while still looking perfect when you unfold it.

2. The Testing Ground: The "AV2 Common Test Conditions" (CTC)

To prove AV2 is better, you can't just test it on one movie or one game. You need a fair, standardized test. The authors created a strict rulebook called the CTC.

The Test Subjects: They didn't just use Hollywood movies. They tested:
- Nature: Real-world videos (like a bird flying).
- Synthetic: Computer-generated stuff (like video games and cartoons).
- UGC (User-Generated Content): Videos people post on TikTok or YouTube, which often already look a bit "crunchy" or pixelated.
- Special Formats: High-definition HDR (like a sunset with deep colors) and wide-screen formats.
The Rules: They made sure everyone used the same "measuring tape." They tested the videos at different speeds (like real-time video calls vs. watching a movie later) and different resolutions (from small phone screens to giant 8K TVs).

3. The New Tricks: "Convex Hulls" and "Adaptive Streaming"

One of the coolest new things in this test is how they handle Adaptive Streaming.

The Analogy: Imagine you are sending a video to a friend. If their internet is fast, you send a huge, high-quality suitcase. If their internet is slow, you send a tiny, low-quality suitcase.
The Old Way: You might just guess which size to send.
The AV2 Way (Convex Hull): The test creates a "perfect map" (called a convex hull). It tests the video at every possible size (from 4K down to tiny 360p) and draws a line connecting the best quality for every possible file size. This ensures that no matter how fast or slow the internet is, the video is always packed as efficiently as possible.

4. The Results: The Report Card

After running thousands of tests, they compared AV2 against the old standard (AV1). The results were measured using a metric called BD-rate, which basically asks: "How much smaller can we make the file while keeping the same quality?"

Here is what the numbers say:

The Big Win: On average, AV2 shrinks the file size by about 30% to 34% compared to AV1.
- Translation: If a movie used to take 100 MB to stream, it now only takes about 66 MB, looking exactly the same.
The "Screen Content" Champion: For things like video games, screen sharing, and cartoons, the savings were even bigger—up to 40%. This is huge for people who stream their gameplay or work remotely.
The "Slow Mode" (Real-time): Even when the video has to be sent instantly (like a Zoom call) without waiting to buffer, AV2 still saves about 26% of the space.

5. The Catch: It Takes More Muscle to Pack

There is a trade-off. Because AV2 folds the video so tightly, it takes more computer power to do the folding (encoding).

The Analogy: Packing a suitcase using the old method takes 1 minute. Packing it with the new AV2 method takes about 3 to 4 minutes.
The Good News: The paper notes that while it takes longer to make the file, the unfolding (decoding) on your phone or TV is only about 2 to 4 times slower than before. The authors are currently working on making the "packing" process faster so it doesn't slow down servers too much.

Summary

This paper is the official proof that AV2 is a major upgrade. It uses a rigorous, fair testing method to show that we can now stream high-quality video using significantly less data (about 30% less). This means faster loading times for you, cheaper storage for companies, and the ability to stream 8K and Virtual Reality content without choking the internet. It's a new, super-efficient way to pack our digital world.