Actually, Apple uses Sorenson's codec to encode the movie trailers.

So could this Sorenson be used to produce the same video quality as seen on the apple site with their steaming video trailers? Is it possible?

If you have a a couple of million laying around and shoot things the way videos and movies are shot, sure.

It has more to do with the original footage quality than compression.

There is absolutely nothing that is inherently better about Quicktime compression. The clips on the Apple site start with pristine source material and are encoded by compression specialists who make adjustments on a scene-by-scene basis. Personally, I think that WMV, DivX or VP6 Flash can look just as good, if not better, than Quicktime.

John

Here is a video I filmed in HD and encoded to DivX HD. It streams from Stage6's server. Take a look at the quality. Also, here is the same file at a lower resolution still in DivX, it has great quality encoding as well.

j razz

First of all, we are WAY down the learning curve on compression technology, and while each technology has its own advantages, there are not going to be huge gains in quality as the years go by. That's just doesn't happen after thousands of engineers have spent the past twenty years perfecting DCT and wavelet compression.

Also, while the high-end encoders have all sorts of manual tools that professionals can use, these help avoid problems at specific boundary conditions, such as transitions and dissolves. They do virtually nothing to help improve the general look of most of the video. In other words, there is no "magic."

There are a few reasons why the QT stuff at the Apple site looks better. One has been mentioned, namely that they are working with very high quality sources. The second is more important: 24p. When going from 60i SD or HD material to web streaming, the footage has 60 temporal events every second, compared to just 24. It is interlaced, and yet is going to be played back on a progressive device (your computer monitor) and therefore some sort of deinterlacing has to take place. The big issue, however, is the 24 vs. 60 temporal events.

If you have any true 24p footage lying around, encode that, using the appropriate progressive presets for your favorite streaming codec, and see if it doesn't look pretty darned good. Oh yes, don't forget to pre-process with a signal processor that removes all residual "snow" and other random noise. This gets back to the comments about having a "pristine" source. Random noise absolutely kills low bitrate encoding because the small number of bits available for the encode get "wasted" trying to chase around all the random noise dots in each frame.

Is it possible?

Yes.

But remember, the finished image isn't going to look any better than what you started with.