> There's been more recent updates.
> https://blog.mozilla.org/research/2014/07/15/mozilla-advances-jpeg-encoding-with-mozjpeg-2-0/ > > The test method, data, etc are public and the results can be verified or > rerun with updated encoders. > As you can see the conclusion is very different from your anecdotal results. It's often anecdotal if the results don't agree with you. Further, there's something very peculiar about the similarity of curves between WebP and the two JPEG encoders at higher compression ratios. Anyone can easily verify that WebP performs significantly perceptually better then JPEG at higher compression (quality/bits-per-pixel), and much more so the larger the image size. It seems to not be stated what compression options were used for WebP. Perceptual may not be quite the same as SSIM but they should at least correlate, right? However, the web is awash in over-compressed JPEGs. We're not really interested in WebP for archival quality use cases. So the 5% gain in compression/quality of mozjpeg 2.0 is great, but claiming 3.0 will be competitive/superior WebP, while retaining full compatibility with original decades old spec is intuitively problematic for a number of reasons. It's like claiming one has developed a new encoder for the similarly-ancient Zip format that achieves 7Zip like performance with no changes to the decompressor algorithm. You see, like WebP, 7Zip includes all the compression techniques that is in the older competitor offers but with many more tunings/options (larger dictionaries, newer compression algorithms, again like WebP). Like WebP's variable compression blocks, 7Zip offers filetype-sorting/solid block compression, hugely contributing to better compression ratios alone that get better vs Zip with the number of small files (as WebP does with larger photographic images). I've yet to read one compelling argument against WebP on technical grounds because there isn't one. Even if it's encoding efficiency were similar to baseline JPEG, it's other features (lossy animation, alpha) would be enough to deserve consideration as next-gen image format for the web. Paul _______________________________________________ dev-media mailing list [email protected] https://lists.mozilla.org/listinfo/dev-media
