Netflix boss says DVD has two years left | Hacker News Reader

Netflix boss says DVD has two years left

tech.yahoo.com

25 pointsmjfern32 comments

Anyone has numbers on how much bandwidth (a range..) it takes to stream a movie in close-to-DVD quality?

I don't have personal experience with streaming from Netflix (I'm in Canada), so I'm not sure how they do it. But back of the envelope, I would assume that with a modern codec (f.ex. h.264) you could compress the average 2-hour movie to about 1.5 gig without losing too much quality. That's about 750 megs/hour, or about 210k/s. Certainly doable.

But if you want HD that's another story...

5mbps, encoded at 720p, with a decent h264 decoder will far exceed DVD quality with audio and video. Apple TV is a good example of this.

10mbps will approach blu-ray.

I'm curious what the satellite broadcasters are using for their HD streams. I've read anywhere from 8 to 10 mbps for a "good" stream (e.g. HBO), but since they encode variably I bet it cranks way down for other channels (like news channels or comedy central).

The latest x264 can get away with about 800-1000kbps average bitrate for nearly-perfect rips of a DVD (and down to as low as 300-500kbps if you're willing to lose the film grain). Most web services are worse--since they don't use x264 (a large number still use ancient formats like FLV...). There are some exceptions: Vudu, for example, uses x264 and is able to get tolerable quality 1080p in 2.8mbps. But I personally wouldn't bother with that--at 2.8mbps your 1080p certainly won't have any more detail than 720p. Of course, most people don't notice this because most Blu-rays don't have more detail than 720p either.

Note that DVDs themselves are already highly lossy; most DVDs have been lowpassed before encoding, so they have significantly less detail than their resolution would suggest. This is rather unfortunate, as the tradition has stuck with Blu-ray, despite it being completely counterproductive when using a good H.264 encoder.

Broadcast is an even worse situation; most broadcast television is atrocious quality-wise. This is usually because of outdated technology (MPEG-2), highly-lossy transcoding to save bandwidth (Comcast is a huge offender here), or just plain absurd bandwidth restrictions (a major television station in Japan uses 4mbps H.264 for 1080i). Of course, this is all aggravated by the requirement of constant bitrate and short keyframe intervals for broadcast streams.

> Note that DVDs themselves are already highly lossy; most DVDs have been lowpassed before encoding, so they have significantly less detail than their resolution would suggest. This is rather unfortunate, as the tradition has stuck with Blu-ray, despite it being completely counterproductive when using a good H.264 encoder.

Could you explain what you mean by "lowpassed"?

Any idea what keeps the adoption of modern x264 codecs? Are they expensive to license? Is it just inertia?

You just made me wonder how good things would look if they did it right... Do you think it would make a big difference to the untrained eye (you sound like someone in the video business)?

Lowpassing means performing a frequency transform and removing the highest-frequency coefficients. It has a similar effect to taking an image, downscaling it, and then upscaling it again (except without the aliasing that the latter produces). Wikipedia probably has more information.

You can tell the difference yourself easily: get a DVD and Blu-ray version of the same movie and downscale the Blu-ray to DVD resolution. You'll notice the Blu-ray is still a lot sharper and has more detail.

Also, note that x264 is not H.264. H.264 is an international video standard published by the United Nations; x264 is a particular open-source implementation developed by Loren Merritt, me, and a few others. Not all H.264 encoders are very good, in fact, a lot are quite bad. See http://x264dev.multimedia.cx/?p=102 for a comparison I did a bit back and http://x264dev.multimedia.cx/?p=164 for some technical information on why.

The reason that adoption of newer video formats has been slow, at least on the internet, is primarily that there is a huge (server-side) install base of old technology that would take effort and expertise to convert to use new technology. It wouldn't be expensive--in fact, it would save billions of dollars a year in bandwidth--but inertia is a powerful force, as is simple lack of knowledge and awareness. Of course, many sites have already converted over; Youtube and Facebook now both use x264 for most of their videos.

Thank you. Your comments are most elucidating.

And kudos on x264!

One more question; do you know if the movies we see on TV have been encoded from a "master" provided by the studios or something like that, or if it's a re-compression of something like a Bluray or a DVD..?

Usually it's from a master. But the quality loss due to broadcast compression is great enough that the difference between a master and Blu-ray source would hardly be visible.

HDTV is usually better than a DVD though if it's a premium channel from a good provider.

I'd add use of 1080i compared to better alternatives like 720p to that list too.

Netflix themselves recommend 2 Mb/s for "DVD quality". That puts then into the "good home broadband" range.

(I get 1.5 Mb/s on my ADSL at home, but I am in a rural area)

Netflix "DVD quality" is nowhere near true DVD quality. It also only steams stereo audio.

As great as streaming is, there will always be the need for a sense of ownership, or paying for digital content will be dead too. Streaming works because it seems like you are paying for the stream. But I don't think this is the replacement for DVDs, since physical media adds value and sense of ownership to digital content so that people may feel justified towards paying for a digital copy. Streaming is a great way to watch most movies but what about the movie you are a die hard fan? Blu rays are basically DVDs, but they still don't solve the issue of price, instant gratification, and versatility (since it's still hard to find bluray players everywhere).

I think digital downloads may be a possible solution. While it may not be physical, a download might provide sense of ownership, especially if it is not DRMd, and it certainly can provde instant gratification. So it could be the counterpart to streaming for those die hard fans that want their own copy. Maybe supplement it with a bluray disc that comes in the mail and you're set.

i often wish i could stream books. if the kindle allowed me to check out from a digital library and return books, even with a very low monthly fee or very low price per book, that would be sweeeeeet.

This break the economic model for books. It works with libraries because of limited quantity.

This could work if we had a fixed number of digital library books per a zip code.

Can you elaborate? Why do you need to fix the number of digital copies in a geographic area?

When we all went from using Blockbusters (=akin to library) to using Netflix (=akin to what is being proposed upthread), did it break the economic model for DVDs?

The "Northern California Digital Library" (http://califa.lib.overdrive.com) already loans out digital books (read with Adobe Digital Editions software) to patrons of many Northern California library branches. The system enforces limits on how many copies of any book can be checked out and so is just like a library. It's convenient but the selection is, for now, limited. I've found some good books there and really prefer this to going to the library.

And the DRM appears to just work. I don't mind this model for books, music (I'm a Napster-to-Go subscriber), and when the first viable stream-for-rent system for movies arrives I won't mind that either.

If anyone knows a way to take the over on this, please let me know. Even if the claim is that DVD rental revenue will be streaming revenue just for Netflix, and it's not a general industry claim, I'm still taking the over. I can buy claims that actual streams may be higher, because the high utilization early adopters will stream more. But revenue? Overall userbase? General market? I laugh.

No, seriously, I'd like to do a two-year LEAP/long bet on this.

Why? It only took ~4 years for DVD to obliterate VHS, and even less time for DVD sales to exceed those for VHS [1]. Hastings doesn't say if the replacement for DVD will be BluRay or streaming, but I think rapid consumer uptake of new delivery technologies is a reasonable prediction to make.

[1] http://www.screenaustralia.gov.au/gtp/wvprodretail.html

If the studios are smart, it's Blu-ray. If they're not, it's streaming. BUT I'm willing to take the long bet and say DVDs last longer than two years, regardless.

If you are willing to take the other side, let's make a bet.

Reading comprehension for the win - "DVD will only be the "primary delivery format" at the company for the next two years, though he did add that it would stick around in some fashion for the next decade or two."

This link is locking up Safari before it even loads. Can't even get the menu bar to activate. Works fine in Firefox. Anyone else experiencing this?

Works fine in Safari 4.0.3 on OSX 10.6.1.

My old laptop isn't fast enough to play newer video formats, so I need DVDs.

So let's get this straight: consumers of movies are going from low-bitrate DVDs to BluRay and even 3-D TV in the future, and this guy is predicting consumers going for lower quality movies because the delivery mechanism is faster?

Hell if I know, but it sure looks like hardware manufacturers are making different bets looking at the same market. Perhaps NetFlix spins off into a low-end video-at-any-quality delivery service and another part of the market goes 3-D?

I'm constantly surprised by how many people can't tell the difference between HD and standard DVD. I even know some people who don't really care if it's VHS or DVD (well, one girl at least).

Maybe their research shows that a big portion of people are fine with DVD quality, and that the others will just rent the blu-rays via mail while waiting for streaming HD?

That's the thing of it -- obviously there is a market for "video at any speed" and a market for hi-fi/prosumers.

For instance, research has shown that more people actually like hacked-up, low-bandwidth mp3s over higher-quality ones. So perhaps this guy is thinking the video market goes the same direction? That the price for bits will always trend towards zero?

If so, this will be a major shift in the movie business -- as big as MP3 file-sharing was to the music industry.

It's probably better business for Netflix to buy servers and big fat pipes than to buy more gigantic warehouses full of discs.

The cost of handling and mailing discs probably stays fairly constant, while the cost of handling bits goes down (as you said).

Netflix certainly isn't an uninterested observer in this. Things might naturally trend toward more streaming, but Netflix is probably pushing hard too.

I think we're on exactly the same channel. Netflix is betting on a market split between prosumers and casual movie watchers, (where, of course, casual movie-watchers are the biggest share and they pick them up). Hardware vendors (and moviemakers like Cameron) are betting on a trend towards more realism, both in sound and video.

Hard to say where the market goes. For me, I'm always into more and more fidelity with reality. But I think there are a huge chunk of folks who only want bragging rights to have seen the latest flick first.

The problem, of course, is that if NetFlix is right and the market only wants something-right-now, that it might take anything-right-now: that people get stuff for free and NetFlix no longer has any market at all.

Very interesting quote. Not sure if he realizes the implications.

shrug

It's a bet either way and anecdotes are anecdotes, but Netflix has lots of data and personally this is how I behave.

I've dropped TV entirely for Hulu. What I can't get on Hulu I order from iTunes.

If every movie on Netflix were streaming, I'd watch it that way, even at a lower quality.

So, YMMV, maybe you're 100% the opposite and/or you know lots of people who care about 780p, 10455p, and whatever else. But I don't think I'm from Pluto, either.

From a strategic perspective I think he's right in the limit. Bandwidth is going to infinity. If consumers prefer timeliness to quality they'll put up with lower quality and wait for bandwidth to improve.

The end game is the same: high-res everything, instantly.

"this guy is predicting consumers going for lower quality movies because the delivery mechanism is faster?"

Isn't that what happened to music? (more or less)

Yes. Instant streaming (or high-quality, lossy compression) is more important than HD. HD is worthwhile on a subset of material (do you really need the simpsons in 1080p?) while lowering the hurdles for watching any video media is valuable for almost everything (TV, movies, etc.)

According to the intertrons, a total of $750 million was spent on blu-ray discs in 2008 in the US. Netflix's instant streaming service alone brings in at least a 10th of that revenue if not more, and the library is still relatively small. More importantly, streaming is growing much faster than blu-ray adoption, so it's easy to see where these trends are headed.

Reply on news.ycombinator.com