How a decade of open source experiments led to JPEG XL, the next-generation image standard
By
@GoogleOSS
A five-star bake. Worth schmearing, sharing, saving.
Summary
This article chronicles the decade-long open source journey behind JPEG XL (JXL), a next-generation image coding standard. It details how experiments in psychovisual modeling and compression—including projects like Guetzli, Butteraugli, and Pik—led to breakthroughs in perceptual image quality. The authors explain how JPEG XL addresses modern display needs like HDR and Wide Color Gamut while offering features such as lossless recompression of legacy JPEGs, efficient encoding/decoding, and royalty-free licensing. The piece positions JXL as the culmination of collaborative open source research that balances visual fidelity with bandwidth efficiency.
Key quotes
· 3 pulledThe internet runs on images. Since the early days of the web, there has been a relentless tension between visual fidelity and bandwidth.
The road to JPEG XL (JXL) wasn't a straight line. It was a decade-long exploration through open source experiments that reshaped our understanding of how humans perceive images.
JPEG XL is designed to be the universal image format that can replace both JPEG and its various successors, offering backward compatibility while pushing forward the boundaries of compression efficiency.
You might also wanna read

Google's Nano Banana Pro AI Generates Realistic Smartphone-Style Photos
Google's Nano Banana Pro AI image generation model produces remarkably realistic images that mimic the look and feel of smartphone photos, i
Reve 2.0: A layout-first AI model for precise 4K image generation and editing
Reve 2.0 is a 4K image generation and editing model that uses a layout-first architecture to solve the precision problem in prompt-based ima
LoomVideo: A 5B-Parameter Unified Model for Efficient Video Generation and Editing
LoomVideo is a new 5-billion parameter unified architecture for video generation and editing that addresses computational bottlenecks in exi

Google's Gemini Omni Flash model makes AI video generation from real images surprisingly easy
The article explores the author's experiment recreating a Google Gemini ad featuring a deer named Buddy, reflecting on how generative AI vid
AI Platform for Advanced Image and Video Enhancement and Analysis Used in Legal System
The article describes an AI-powered platform for image and video enhancement and analysis that goes beyond current market capabilities. The
