I was involved on some contract work with effectively the same type of technology that Raystream is touting. We didn't hit the kind of compression ratios that Raystream are talking about, but we were able to compress video drastically while still maintaining video quality. That is, the technology is possible, and I've seen it with my own eyes. With that said, the company I was contracting to, could have easily spun the marketing to suggest we were getting 90% compression. To me, it looks like Raystream are making a lot of claims that are half-truths.

Looking at the uncompressed vs Raystream videos, it's difficult to tell if it's doing anything more than h264, x264, or neither. The conclusion being made by the OP may be true, it may not; likely they're exaggerating their marketing to look better than they truly are.

One of the marketing tricks they're using to make their technology look better than it is, is by not showing the videos side by side. Viewing them side by side, it's easy to see that Raystream's video lose quality in their color complexity, image sharpness, and some artifacting shows up if you know what to look for (I can do up some comparison images if anyone is interested.)

When we were testing our compression technology we would run our video through acid tests (difficult video to test the limits of the encoders.) We had a variety of acid tests, and depending upon encoder settings we could tweak each video to look good - however the problem was getting all of the acid test videos to look good at the same settings.

So while Raystream may be able to encode a promotional video that has very low complexity scenes, that doesn't mean this tech is any good at high complexity scenes. In fact, most of the scenes in the promo video our encoders would have had no trouble doing. Things like sports, live concerts, trees, or complex water (the surfer wasn't nearly complex enough, and the original quality of the humming bird wasn't good enough to produce a noticeable difference) were what we used to test our encoders.

I also notice they've got a "Live Streaming" section on their site. On demand content is easy to do, relative to live streaming. For on demand content you have no bounds like cpu, or ram, the biggest barrier is disk io and time to encode. Whereas live streaming is extremely complex. They claim they can do 1.7 Mbs live streaming; we were able to do 1.5 Mbs and less, but at that point the quality suffers significantly, and the video would infrequently go blocky for a frame or two. The only use case is for TVs that no one is going to get close to.

There comes a certain point with this type of technology where you can't stuff a 100 pieces of data, into a container that only holds 10 pieces of data. It doesn't matter how good your compression algorithm is.