Google open-sources new JPEG encoder that reduces file sizes by around 35%, and is compatible with existing browsers, photo editing apps, and the JPEG standard
New algorithm is based on human psychovisual system. Images look better, too. — Google has developed and open-sourced …
Context & Ripple Effects
This lands mid-arc in Google's run of open-sourced compression work: two years earlier it shipped Brotli, claiming 20-26% better compression than Zopfli for text, and now it applies the same playbook to images — an encoder built on human psychovisual modeling that shrinks JPEGs by around 35% while staying inside the existing JPEG standard.
The compatibility claim is the strategic point: because the output works in current browsers and photo editing apps unchanged, publishers get smaller files without waiting on a format-adoption war. It also complements the other half of Google's 2017 image push — RAISR, which uses machine learning to make images up to 75% smaller without losing detail — attacking image weight from both the encoding and enhancement sides.
First-order effects
- Web publishers and photo workflows can drop JPEG payloads by roughly 35% today with zero changes to browsers, editing software, or serving infrastructure, since the encoder's output remains standard-compliant JPEG.
Second-order effects
- The improvement raises the bar inside the JPEG ecosystem itself: competing encoders and image-tool vendors must match psychovisual-level optimization or cede the default-compression slot, while Google consolidates its role as the de facto steward of how images get compressed on the web.
Third-order effects
- Google is building a layered open-source stack around legacy standards rather than replacing them — Brotli over text compression, this encoder over JPEG, and later Chrome Lab's Squoosh app bundling those choices into one workflow — shifting image delivery from format wars to algorithmic optimization controlled upstream by toolmakers.
The trend: Google is assembling an open-source compression layer that optimizes within established standards like JPEG rather than pushing new formats, making file-size gains available without ecosystem-wide adoption.