To stream everywhere, Netflix encodes each movie 120 times
gigaom.com
gigaom.com
I think it only makes sense to lazily-transcode videos on demand (and cache them) when you have millions of videos (eg. Youtube). Netflix and Hulu have much smaller catalogs, so they can afford to transcode ahead of time, and it simplifies their processes.
As for the actual encoding, nearly all the computation is very specific to the output resolution/quality. Luckily encoding multiple videos can be spread over many machines, and encoders can take advantage of multiple cores.
Apparently they are quite comfortable encoding as many as 120 instantiations of their videos.
However they could still lazily trans code on demand (and cache the result).
Youtube offer their videos in vp8 along with h264.
There is more to it than just the number of videos (or storage requirements). You can get dedicated hardware h264 encoders easily (and relatively cheaply), not to mention the decoders in mobile devices. There's also quality to consider, effective quality per MB (important when targeting mobile devices), and all the things that fall into that (motion handling, prediction, etc).
I don't claim to be an expert, but from what I've read Ogg & WebM are fighting an uphill battle on the technical side of things. Part of this is no doubt due to h264 patents, but it's a shame the free solution isn't also the technically superior one.
We don't actually use the GPU instances yet, but I think I heard we're moving over to them (need to check on that).
> Simply put, content owners who can deliver quickly and without error are getting more licensing revenue from Netflix.
Really? Can Netflix afford to be picky with content owners? It seems their main problem is the limited size of their online catalog...? They should take anything they can get!
I just finished watching "Elite Squad 2"; it's such a great movie (really, it is: see http://www.imdb.com/title/tt1555149/) that I wanted to watch the first one, but it's "only available on DVD".
Why, why, why?
We have $X to spend. We have $Y left at the end of the year ($Y < $X). We go though our system and say "who can deliver us the quickest, most error free, content? Ok, give them $Y for whatever they have that our customers might enjoy.
The reason you only see Elite Squad 2 is the same reason that you only see a lot of stuff -- because the price they wanted for the other one was too much.
(I'm not sure that the technical quality of transmitted files is the best criterium to select content, though, even when it's the end of the year and we're talking about "leftover money").
Let me put it to you this way: If it ain't on Netflix, I ain't watching it. Now, I don't consider myself typical; usually I'm willing to do research, dig into things, I know multiple ways to "access" content, etc, etc. But at the end of the day, I'm tired (and lazy), and when it comes down to it, I'm browsing Netflix, which is full of stuff, so I almost never notice if something is missing. In this respect, I'm willing to bet I'm typical.
Content owners: either get on the bandwagon (and that means distributing to everyone, not just Netflix, who doesn't require exclusive deals, I might add), and make your stuff available easily, else you will be left behind and rightfully forgotten.
I do also have Amazon Prime, but that has enough crossover with Netflix that I mostly use the shipping benefits. If I'm looking for something, I will check CanIStream.it (aka, Don'tMakeMeFuckingPirate.It), and if it's not available on Netflix/Amazon, I'll wait. Sooner or later it will be.
Don't blame Netflix for it. Blame the arduous contracts which must be negotiated in order for Netflix to carry the content it does.
So Silverlight not beign supported on Linux is not the real issue here.
But then you're forgetting: poverty hasn't quite abandoned us yet, has it?
Even with all of the modern technology and what not -- people are still dying of hunger.
We first need to start with these base human needs.
Elevate the standard of living of the people on the other side of digital (and economic) divide, and bring them up to standard of an average first-world-country person.