Apple’s PICO cuts image files by one-third at equal quality
Apple has announced PICO, a new machine learning-based image compression codec designed to achieve higher perceptual quality with smaller file sizes than existing codecs like AV1, VVC, and JPEG-AI. PICO can reduce image data size by up to one-third for the same quality and offers a 20-40% bitrate reduction compared to other leading machine learning codecs, while also being optimized for practical on-device processing and inter-device compatibility.
Key Takeaways
- PICO stands for Perceptual Image Codec and uses a neural network instead of only traditional hand-designed compression steps.
- Apple says PICO can match image quality with up to one-third less data than AV1, AV2, VVC, ECM, and JPEG-AI.
- Compared with other leading machine learning codecs, PICO is said to cut bitrate by 20% to 40%.
- On an iPhone 17 Pro Max, Apple says PICO can encode a 12-megapixel image in 230 milliseconds and decode it in 150 milliseconds.
- Apple tested PICO with 74,925 paired comparisons from 610 evaluators across CLIC 2020 Test, Kodak, and DIV2K datasets.
Why It Matters
PICO gives Apple a codec that pairs smaller files with perceptual quality targets, not just pixel matching, which matters for image-heavy workflows where storage and bandwidth are tight. The broader signal is that Apple is pushing machine-learning compression toward practical deployment, with explicit attention to rate control and inter-device compatibility rather than lab-only quality gains. What to watch next is whether Apple extends PICO beyond research by showing support for simple composite images like cartoons, where the paper says it may need a higher bitrate than conventional codecs.
Read full article at gigazine.net
Get this in your inbox → Subscribe
Enjoy our coverage?
Add StreamingMeme as a preferred source on Google to see more of our streaming news at the top of your Search results.
Add as preferred source