3 ms·
I'm sure this was discussed before when this story made the rounds last month, but this continues to make little sense to me: http://code.google.com/speed/webp
by sghael 16y ago
I'm sure this was discussed before when this story made the rounds last month, but this continues to make little sense to me:
http://code.google.com/speed/webp/gallery.html http://code.google.com/speed/webp/gallery.html
WebP might be great, but how can it be a meaningful comparison --
1) take already compressed jpegs from wikipedia and re-compress them using WebP, and then tout the 39% file compression achieved? I can re-compress Jpeg with more Jpeg and also get 39% "more" compression.
2) as with all lossy compression, unless there is some measure of the loss, perceptual or otherwise, what are we really comparing?
- andrewvc 16y agoIf you cut it down, and it's perceptually the same, then it's an obvious win no?
- sghael 16y ago1) How do we know it's perceptually the same? 2) How do we know we couldn't get the same perceptual gains from more JPEG compression?
- wmf 16y agoDid you see their comparison of WebP vs. JPEG at constant quality? http://code.google.com/speed/webp/docs/c_study.html http://code.google.com/speed/webp/docs/c_study.html (Not only is this thread a dupe, but every single comment as well. Nice.)
- ZeroGravitas 16y agoI assumed that Google planned to use it in exactly the way they tested e.g. * thumbnails of jpegs on picasa web galleries * thumbnails of jpegs & pngs on Google image search * thumbnails of videos on Youtube * an Opera Turbo/Mini style proxy server that recompresses jpegs and pngs * a turnkey web-accelerator that recompresses all the images on your site and serves the WebP ones to compatible browsers without you needing to do anything However, I don't think anything like this has actually been announced. Without such announcements the methodology does seem strange.