How To Reduce Image Size With WebP Automagically • Hacker News Board

Re: How To Reduce Image Size With WebP Automagically

#21

Post by notatoad » Sat, Apr 13, 2013, 2:12 AM UTC

eridius wrote: post #3

> But with CDN Connect’s automatic image re-encoding, this 52kb PNG shrunk to a 11kb WebP image. That’s a 78% size reduction! Just imagine this image was referenced on every page of your site, and you had handfuls of similar images to download. The potential bandwidth savings and performance improvements are enormous. What utter garbage. Does CDN Connect want us to believe browsers don't cache images? Sure, that's a…

depends what your client is. on a desktop, yeah it's going to download once. on your phone, where it really matters, that's going to end up being a saving just about every time you visit the site.

Re: How To Reduce Image Size With WebP Automagically

#22

Post by snotrockets » Sat, Apr 13, 2013, 11:03 AM UTC

riffraff wrote: post #15

FWIW, pngquant compresses the png to 18kb without any visible (to me) artifact. If you are not using pngq/pngquant and the likes to compress alpha png I advise you to try them out.

I just tested pngquant vs webp on a corpora of 1000 png images.

WebP beat pngquant by a factor of 2.36 (I hope to release the full report and data set soon.)

Re: How To Reduce Image Size With WebP Automagically

#23

Post by artificialidiot » Sun, Apr 14, 2013, 8:44 AM UTC

derefr wrote: post #19

celoyd wrote: post #12

Earlier quoted context omitted.

In URLs, and especially user-visible URLs, file suffixes at all should be considered an only occasionally necessary evil. They’re ugly, redundant, brittle, and generally a headache. TBL wrote about this in 1998: http://www.w3.org/Provider/Style/URI.html#remove

Though probably do transmit a "Content-Disposition: inline; filename=foo.jpg" header. Otherwise, the OSes that infer media type from extension will get confused when the user goes to save the file and it appears extensionless. (That might be called a browser bug, actually; it'd sure be nice if browsers on these OSes could take the observed media type and tack on an extension to extensionless files on Save As to ensur…

it is neither the OS' nor browsers responsibility to launch correct application for a file. There are launchers that examine file headers (no need to dup it on filesystem) and keep track of your application preferences for a given mime type. They may be called file manager, shell or something similar and responsibility lies there. 8.3 file naming is a historical baggage and nothing more.