Originally Posted by Scott Wilkinson
Exactly correct. Their goal is to preserve a given quality level at a lower bitrate.
Their goal is to further compress already compressed content as much as possible without 'noticeably' lowering quality. 'Noticeably' as according to their quality metric. It's a rather idiotic premise since it requires the content to already be lossy encoded at an arbitrary bitrate.
If we take a specific, unchanging source - for example the latest X-Men movie: if you feed Beamr video an uncompressed deliverable, it would give you an encode that is 50Mbps+. Because that's the only way that you can achieve something almost indistinguishable from source....very high bitrate.
If you feed it the movie - this time already compressed to 8Mbps H.264 - it will give you maybe a 6Mbps output, because it's quality metric feels that it is 'about the same'.
If you feed it the same movie, this time compressed to 3Mbps H.264, it will give you an output of maybe 2Mbps.
You can surely see the problem. It's the same movie, yet Beamr has given us 3 widely different bitrates. It's not an absolute metric.
It doesn't suggest to us what bitrate we should use to deliver good quality content. For content providers it's completely useless since it requires an input that has already been encoded at an arbitrary bitrate
Beamr won't tell you what bitrate to deliver your content at. You feed it a video and it simply re-encodes it at a lower bitrate and says "hey, this will look approximately the same". Anyone with a mathematical mind should see the problem here. Small differences are inherently indistinguishable. It's a parlour trick. You can achieve the same results by randomly choosing a % between 5 and 20 and re-encoding a clip with that percent less bitrate. Guess what - most people will not tell the difference. Whoop de do. You save money on bandwidth. Send your cheque to me, thanks.
Unfortunately, I think what their technology does, and probably the entirely of what i've written, will go right over the heads of most readers. The comment by fookoo_2010 should be proof of that. I'll say it again, Beamr have made no contribution to encoding efficiency. Whatever quality their software can achieve at a given bitrate... was already acheiveable at that bitrate with the existing encoders - the same encoding libraries that they use in their software without giving credit to the original developers.