Slides: <a href="https://www.slideshare.net/cloudinarymarketing/imagecon-2019-jon-sneyer" rel="nofollow">https://www.slideshare.net/cloudinarymarketing/imagecon-2019...</a><p>Background: JPEG XL is a combination of Cloudinary's FUIF [1] (successor of FLIF [2]) and Google's Pik [3].<p>Committee Draft (Aug 2019): <a href="https://arxiv.org/abs/1908.03565" rel="nofollow">https://arxiv.org/abs/1908.03565</a><p>Technical Details: <a href="https://www.spiedigitallibrary.org/conference-proceedings-of-spie/11137/111370K/JPEG-XL-next-generation-image-compression-architecture-and-coding-tools/10.1117/12.2529237.full" rel="nofollow">https://www.spiedigitallibrary.org/conference-proceedings-of...</a><p>Features / Goals:<p>- high quality compression (> 60% over JPEG-1)<p>- royalty-free with open source implementations available from the start<p>- versatile: supports alpha transparency, high bit depth (16-bit), lossless compression, animations<p>- progressive decoding / "responsive by design"<p>- legacy-friendly: reversible transcoding of JPEGs with 22% size reduction (demo available [4])<p>Comparisons:<p>- JPEG 2000, JPEG XR: only marginal compression improvements<p>- WebP: limited (8-bit, 4:2:0), no progressive decoding<p>- BPG/HEIF (HEVC): patent-encumbered (not royalty-free), no progressive decoding, complex<p>- AVIF (AV1): no progressive decoding, complex, slow?<p>[1] <a href="https://cloudinary.com/blog/introducing_fuif_responsive_images_by_design" rel="nofollow">https://cloudinary.com/blog/introducing_fuif_responsive_imag...</a><p>[2] <a href="http://flif.info/" rel="nofollow">http://flif.info/</a><p>[3] <a href="https://github.com/google/pik" rel="nofollow">https://github.com/google/pik</a><p>[4] <a href="https://google.github.io/brunsli/" rel="nofollow">https://google.github.io/brunsli/</a>