I've used iStopMotion -- and loved it. Only a customer, not connected with the company in any way.

The Google Code page for Webp makes grand claims that Webp is better than JPEG and JPEG2000 at compressing images, and then points to a study that compares the three methods by having each recompress an image that is already JPEG-compressed. Recompressing a previously-lossily-compressed image is a rather different task than compressed an original image.

It is unclear to me how quality was measured for all the graphs in the study -- was quality measured against the original image? I doubt that -- the images were harvested from Flickr, so the original (pre-JPEG-compression) aren't likely to be available. Instead, the quality was measured against the decompressed images, which have been blurred by the JPEG process.

If one wants to take one's collection of JPEG-compressed images and compress them further, without losing quality, one should decode the Huffman-encoded stream and re-encode using an arithmetic coder. One will save about 10% of the filesize without losing any quality at all. The Q coder is specified in the JPEG standard, so this can be done in a standards-compliant way, though no web browser supports that (which is a problem Webp also has).

I remember that interview: Bill Gates was asserting that people won't pay for bug fixes, but only for new bells and whistles. And he's right! People expect software with no bugs and they expect that the inevitable bugs will be fixed for free. The big problem, of course, is that Microsoft put new bells and whistles at a higher priority than bug fixes since they get paid for the former but do the latter for free.