New site with codec comparisons going back to 2001

[RadicalEd]

I just finished my new website, and included some of the bigger codec comparisons (video and audio) that I've done over the years going all the way back to 2001. You can really get a feel for the progression of technology. There are screenshot comparisons done in JavaScript, PSNR and SSIM charts and graphs, as well as a written analysis. The 2001 comparison even has full perceptual ranking data from a group of viewers. Anyway, check it out (2008 coming soon): http://mike.command-q.org/Codec.html

The site itself is my online portfolio, feel free to look around if you want to.

[Avenger007]

Nice site. :)
Where are the samples and what are the various encoder settings? :confused:
And what about live action comparisons instead of only anime?

[RadicalEd]

what are the various encoder settings? :confused:
Unfortunately, outside of the bitrate, I didn't keep this information. However, I usually strive to use the best quality settings possible.
Where are the samples
You mean video files? I didn't put any of those up, sorry. I figured the screenshots would be good enough.

[RadicalEd]

And what about live action comparisons instead of only anime?
The 2001 comparison includes live action, as will the coming 2008 comparison. I mostly used that particular anime (2004 and 2007) because the source is incredibly clean and brings out the flaws of the encoders well.

[zambelli]

Unfortunately, outside of the bitrate, I didn't keep this information. However, I usually strive to use the best quality settings possible.
Best according to whom?

A codec comparison that doesn't include encoding settings is worthless, in my opinion. No offense. But it's like comparing cars and only keeping track of the number of doors on each model.

I would also argue that saying a thing or two about decoder configuration (i.e. which decoder? post-proc or not?) would've been useful.

[Dark Shikari]

A codec comparison that doesn't include encoding settings is worthless, in my opinion.Agreed. Such things lead to non-replicable tests, which leads to cases like the recent VP8 marketing hilarity, where they compare to x264 (where x264 looks much worse)... and then when everyone tries to replicate their test, x264 comes out much ahead. :rolleyes: