Netflix To Re-Encode Entire 1 Petabyte Video Catalogue In 2016 To Save Bandwidth (variety.com)

An anonymous reader writes: Netflix has spent four years developing a new and more efficient video-encoding process that can shave off 20% in terms of space and bandwidth without reducing the quality of streamed video. With streaming video accounting for 70% of broadband use, the saving is much-needed, although the advent of 4K streaming, higher frame rates and HDR are likely to account for it all soon after. Netflix video algorithms manager Anne Aaron explained to Variety that certain types of video benefit little from the one-size-fits-all compression approach that Netflix has been using until now: "You shouldn't allocate the same amount of bits for My Little Pony as for The Avengers."

Comments

My little pony

[Eunuchswear]

"You shouldn't allocate the same amount of bits for My Little Pony as for The Avengers."

So they're dropping the resolution for The Avengers?

Re:My little pony

[radiumsoup]

damn, mod points just ran out. Best first post in a long, long time :D

Re:My little pony

[Anonymous Coward]

"You shouldn't allocate the same amount of bits for My Little Pony as for The Avengers."

So they're dropping the resolution for The Avengers?

+1. Just compare the fan base.

Re:My little pony

[Anonymous Coward]

Absolutely. I need to see every little pony in all its unpixelated, 4K glory.

Wait... that didn't come out right.

Re:My little pony

[Eunuchswear]

You couldn't believe how hard it was to type that fast without speeling erreurs.

Re:My little pony

[Tablizer]

Good! You don't want angry 4-year-olds, trust me.

Re:My little pony

[Anonymous Coward]

4-year-olds? I love that show, dude. Don't pixelate the pony, bro!

Compression ruins the glistening of the glitter and the graceful flowing of manes during brisk gallops.
     

Re:My little pony

[Anonymous Coward]

You think that 4 y/os make up the majority of MLP viewers? Cute.

I remember the twitter shitstorm that got stirred up because MLP became recommended if you watched Doctor Who, Supernatural, and several other similar genre shows. Seems @CopingWithPony hasn't updated in almost 3 years though. Guess whoever it was got bored.

Re:My little pony

[Kagato]

On a serious note, animated content is much harder for the 8-bit encoding. It's the hard edges with high contrast cell shading. You get a lot more compression artifacts than a typical movie. You can resolve this by using 10-bit encoding, but there's a lot of Netflix devices with embedding video codecs. They really can't change, and almost none of the chipsets out there support 10-bit decoding. So that leaves option two, which is to increase the bitrate.

Re:My little pony

[malditaenvidia]

I love that show

Why? What's wrong with you?

Re:My little pony

[PRMan]

This isn't true and the article goes into great depth about how much easier it is than Avengers.

Re:My little pony

[Anonymous Coward]

I guess you haven't heard of the "Bronies" sub-culture. Where adults pretend to actually like My Little Ponies ultimately to be able to come together and live out some slightly less weird fantasies of the adult variety.

Re:My little pony

[subk]

animated content is much harder for the 8-bit encoding. It's the hard edges with high contrast cell shading.

Nonsense. Animation is optimized for compression.. It's already dithered to a limited color set! Whole swathes of pixels get the same value. Sure, you might be able to SEE the compression.. But it is a breeze for the hardware, and bitrates can be dramatically reduced compared to the baseband 4-2-2 video.

Re:My little pony

[marcansoft]

That animated content benefits from 10-bit encoding is true. That has less to do with hard edges and more to do with banding artifacts on flat shaded areas - TFA actually goes into that, mentioning soft focus and fog as producing hard-to-encode gradients, the same kind of gradients present in many kinds of animation and which would benefit from using 10-bit mode. Hard edges do tend to be hard to encode with typical video codecs too (but 10-bit probably won't help you there).

However, My Little Pony isn't a particularly good example, because it's full of completely flat areas that are trivial to encode. It might take a higher quality setting than you might expect to look crisp, but at the end of the day, you're going to be spending fewer bits per frame on it than on The Avengers. Animation has its own set of encoding tradeoffs/challenges (which is why good encoders have presets tuned for animation).

Re:My little pony

[Anonymous Coward]

Disregarding the incorrectness of your central point, there's a third option that you leave out:

Encode at 10bpp for the devices (like every PC, ever) that do support 10bpp decode, and at 8bpp for the devices that do not. This doubles your storage requirements, but I expect that the data transmission costs would make up for the additional storage expense.

Re:My little pony

[Kagato]

I suggest you google "Why 10-bit anime".

Re:My little pony

[Falos]

I can't tell if she's saying the streams have different priorities or if they should be compressed with different codecs. Live action and animation act rather differently on screens. My understanding was jpgs are better suited to photographs and pngs for flatter images (ie website screencap, logos), so I imagine there's also best-choice temporal/spatial tech for video. And it shouldn't be too hard to write an algorithm to classify the strea- wait, they would already be tagged.

Or maybe not, and she's bashing one of the titles. And while everyone else is laughing with OP, I'm genuinely uncertain which. TV is a big source of culture sync, and I don't supplement with enough other sources (eg current literature, regular community events, even advertising) to keep in the loop.

Re:My little pony

[auld_wyrm]

I'd really rather not, thank you.

Re:My little pony

[r1348]

If you don't already know who "bronies" are, don't look for them.

Re:My little pony

[Tablizer]

Example

Re:My little pony

[Eunuchswear]

Ho ho ho, insulted by uid 666. I think I'll go and cry.

Re: My little pony

[Anonymous Coward]

The new (as in last 10 years or whatever) show is actuallly really funny, and littered with pop culture references obviously aimed at adults. Once you have children, that kind of thing will be highly valued when it comes to sutting down with them and watching a cartoon.

Re:My little pony

[Mr D from 63]

I can't tell if she's saying the streams have different priorities or if they should be compressed with different codecs.

It seems both. I'd say the best way to save bandwidth is to reduce the popular titles first. All the latest compression technologies can't reduce the existing bandwidth used for watching Vanilla Sky.

Re:My little pony

[Anonymous Coward]

I recode a lot of video (don't ask), and in my experience, animation is a lot easier on the resulting files for the same quality. The absolute worst you can do if you want small high quality files is "film grain", whether it is from a bad source or artificially added for artistic reasons. Second worst is a badly compressed source with lots of artifacts. Then there's video with lots of small objects moving across a detailed background. The hard contrasts at the edges of cell shaded videos are only problematic if you don't have a good quality source, i.e. if your source already has lots of compression artifacts.

Re:My little pony

[Anonymous Coward]

Yes, we have, and they are creepy as hell.

Re:My little pony

[Anonymous Coward]

If they drop so much as one pixel from my 4K stream of My Little Pony, I'm gone!

Re:My little pony

[sexconker]

I suggest no one do that, because it's fucking horse shit. I know the fucking document you're referring to, and it's wrong, wrong, wrong.
There is no reason to use 10-bit encoding over 8-bit encoding unless you're using a shitty fucking encoder that introduces rounding errors at every step. Decent encoders work in a higher precision before shitting out the encoded result.

Re:My little pony

[Anonymous Coward]

I suggest you google "Why 10-bit anime".

Because you're a fucking autist obsessed with your bootleg cartoons from Japan?

Re:My little pony

[Anonymous Coward]

I doubt an autist would suggest a website using the word "animes"

Re:My little pony

[Anonymous Coward]

So you know more than the professional who is actually doing this project? Typical.

Re:My little pony

[Anonymous Coward]

That's great, nobody used that word but you.

Re:My little pony

[Krishnoid]

Isn't this exactly what Flash (and similar vector formats) are good at reproducing, at any resolution?

Re:My little pony

[Chris6502]

Avengers are ponies: https://s-media-cache-ak0.pini...

Re:My little pony

[DanJ_UK]

Better example

Re:My little pony

[Anonymous Coward]

Oh boy, you are so wrong about that. The truth is, cel animation content is far easier to encode. If you're seeing artifacts, you're doing it wrong.

To do it right: In HandBrake, go to the Advanced tab of the H.264 encoding settings and set the Adaptive Quantization and Psychovisual Rate Distortion sliders both to the far left, e.g., aq-strength=0.0:psy-rd=0.0,0.00.

Boom— beautiful, non-artifacty cel animation.

Re:My little pony

[Anonymous Coward]

p.s. I reencoded all of the Futurama DVDs this way, using HandBrake, and they came out looking much better than the original MPEG video from the DVDs — and were much, much smaller.

Re:My little pony

[Rob Lister]

Anne Aaron is the project manager.

She doesn't seem particularly misogynistic.

Re:My little pony

[TechyImmigrant]

I'd really rather not, thank you.

The bronies will be furious.

Re:My little pony

[justthinkit]

However, My Little Pony isn't a particularly good example, because it's full of completely flat areas that are trivial to encode. It might take a higher quality setting than you might expect to look crisp, but at the end of the day, you're going to be spending fewer bits per frame on it than on The Avengers.

Ah, this is why it is a perfect example. "Easy to encode" equals "will save a lot of space" in the new format.

Re:My little pony

[Anonymous Coward]

What you said didn't come out right either. :D

Re:My little pony

[Shirley Marquez]

I appreciate the humor. But we know that in reality, it means fewer bits for My Little Pony. The images don't change as quickly and the relatively simple line drawings that it uses make it easier to encode; you can get the same level of quality with a significantly lower bit rate.

Netflix could save even more bits by reencoding with H.265. (My understanding is that they already use H.265 for 4K streams.) But that would have unfortunate effects on older devices that lack hardware H.265 decoding. It would mean choppy playback and/or decreased battery life on older computers and mobile devices, so it should either be opt-in or decided by some sort of automatic detection of client capabilities.

Re:My little pony

[CrankyFool]

Nope. Anne Aaron is not the project manager. She's the engineering manager.

You can find her career history at https://www.linkedin.com/in/an... -- after getting her PhD in EE from Stanford (thesis was about video encoding) she was a software engineer for 14 years until she was promoted in September/2014 to manager.

(Full disclaimer: She's a coworker of mine)

Re:My little pony

[RockDoctor]

Pretty good username too. Props!

Another year, another video codec...

[Viol8]

Shaving 20% off seems pretty optimistic to me. Unless they've suddenly discovered some whole new realm of compression mathematics I'd be surprised if thats anything more than a peak compression in some rare edge cases.

Re:Another year, another video codec...

[hawguy]

Shaving 20% off seems pretty optimistic to me. Unless they've suddenly discovered some whole new realm of compression mathematics I'd be surprised if thats anything more than a peak compression in some rare edge cases.

Sounds more like as a part of re-compression, they are going to drop the bitrate (and video quality?) for videos that don't "need" it:

certain types of video benefit little from the one-size-fits-all compression approach that Netflix has been using until now: "You shouldn't allocate the same amount of bits for My Little Pony as for The Avengers."

Re:Another year, another video codec...

[c0d3g33k]

I suspect that by the time of this announcement they have already done the testing, so have a good idea of how much they can optimize. From the article, it's more about optimizing compression parameters to fit the source material rather than taking a one-size-fits-all approach.

Re:Another year, another video codec...

[Anonymous Coward]

Could it be that the new larger total volume allows for the existance of larger patterns (or more frequent same sized ones) that then allow for a better compression rate when the entire library taken as a whole?

Re:Another year, another video codec...

[magarity]

Sounds more like as a part of re-compression, they are going to drop the bitrate (and video quality?) for videos that don't "need" it:

Is it the bitrate that's being changed? I took the part about Pony vs Avengers to mean animation vs live action are better compressed by completely different algorythms.

Re:Another year, another video codec...

[Anonymous Coward]

The article is actually in English, you know.

"The new system will encode from the raw source material more intelligently, considering whether or not the material itself can really benefit from higher bit-rates, or whether identical quality can be maintained with less space and bandwidth."

Re:Another year, another video codec...

[Anonymous Coward]

Not really, just a "tech journalist" completely misinterpreting how VBR encoding works.

Re:Another year, another video codec...

[adolf]

The Avengers is not animated?

Re:Another year, another video codec...

[JackieBrown]

I am in the process of moving my fiancee's dvd/bluray collection to my server and putting her physical copies in storage. Using Handbrake, switching from x264 to x265 saves me at lease 10 % on dvd sources and closer to 30+% on the bluray sources.

Re:Another year, another video codec...

[Junta]

They suggest that they are going to have an automated process to transcode content a few ways and use some automated quality check to decide what meets the threshold. Or they are trying to make changing from CBR to VBR sound more impressive than it is.

Re:Another year, another video codec...

[Anonymous Coward]

switching from x264 to x265 saves me at lease 10 %

Use more aggressive setting and you can save 50% or more. You will not see the difference unless you are one of those idiots who buys 400 USD network and audio cables ;)

Re:Another year, another video codec...

[PRMan]

They have. Mathematically, My Little Pony doesn't need 5.8 Mbps for 1080p. It can be done at about 1.5 Mbps because of the simple color palette.

Re:Another year, another video codec...

[bagofbeans]

I remember ripping my CD collection to ogg, only have to do it again years later to flac when space got cheaper. The ogg was fine, but not a good source for re-encoding to another format such as mp3.

If I was going to rip movies, I'd keep the original streams. You'll never spare the time again to re-rip, even if you you think now that you will.

Re:Another year, another video codec...

[PRMan]

Pretty much...

Re:Another year, another video codec...

[PRMan]

But Netflix can't do this since most of the hardware consuming their movies has an x264 chip built in.

Re:Another year, another video codec...

[known_coward_69]

my guess is that cartoons, especially older cartoons have large parts of the frame with the same color which can be compressed more easily. or they will simply stream cartoons at lower bitrates

Re:Another year, another video codec...

[Wrath0fb0b]

And now you have x265 that's not hardware accelerated on anything but the most modern GPUs (and even then, only partially -- certainly not suitable for any set-top-box, tablet or mobile phone. Heck, even a laptop that has partial GPU supporting (or none) will burn through tons of battery watching it on a flight with no power plugs.

Netflix has to support all those platforms (and probably worse ones) -- and then you enter the idea of having multiple copies for every asset :-(

Re:Another year, another video codec...

[jedidiah]

They might also be able to employ different and better options on the transcoder. When I first started transcoding my own stuff I optimized for quality and total processing time. The files were meant for HTPC use and mobile video devices weren't terribly common yet. What I ended up with was something that early iDevices couldn't even handle (although Archos could).

They may simply be adjusting for more modern and more capable devices.

Re:Another year, another video codec...

[jedidiah]

A 10% storage increase is not nearly enough to justify x265. It's a bleeding edge codec with little GPU hardware support and ungodly CPU decoding requirements.

Re-compressing BD kind of defeats the whole point of bothering with BD. Plus you magnify the aforementioned decode support issues.

Re:Another year, another video codec...

[JackieBrown]

I'm not using very aggressive encoding settings and my google player box has been able to play all of the movies I ripped so far (I know it's not hardware accelerated but as long as it plays without stuttering, it makes no difference to me.)

Guess I'll double check my amazon fire (not the 4k model) and make sure it's able to play those too.

Re:Another year, another video codec...

[JackieBrown]

10 % for dvds and 30 % for blurays. And that was compared to the x264 rip not the source.

Now that you mention it though, I might have to debate the sd rips going to x265. Not sure if it's worth the extra encoding time on those. It is easy to see the plus, though, for the HD stuff. The storage savings add up fast.

As far as just putting the original on my server, i thought about it but 20-30 GB a movie would add up pretty fast (and that is with removing most of the sound tracks and all the extras).

Re:Another year, another video codec...

[jensend]

On the contrary, I'm surprised they're only saving 20%. Look at the difference between VP9 or X265 and the VC-1 encoder they used exclusively for their first few years. And given the costs they incur using all this bandwidth they could definitely throw a good number of top quality engineers at figuring out a rate control algorithm that's more suited to their rather unusual use case.

Re:Another year, another video codec...

[Anonymous Coward]

They're using Pied Piper's middle out compression.

Re:Another year, another video codec...

[powerlord]

I think its more a matter of (potentially) saving 20% versus what they are using now.

Didn't read the whole blurb, but they are probably going full-in on HEVC ( https://en.wikipedia.org/wiki/... )

The Wikipedia article mentions 50-60% savings using HEVC over H.264 depending on resolution, but chances are these are optimum subjective results. My Little Pony (and other animated shows) would probably encode much tighter compared to media that is action/visually dynamic.

Netflix may be looking at that 20% as a realistic figure for their library as a whole, baring in mind realistic encoding savings and the size of their catalog in various categories.

From 2 years ago which seems to forecast where they are now: http://www.streamingmedia.com/...

Re:Another year, another video codec...

[Spy Handler]

20% savings is believable. Pirates easily achieve that much without sacrificing video quality. Of course pirates do it by hand, eyeballing each scene and adjusting the bitrate and compression, so it's a lot of labor. Also it's almost an art form. It's incredible how much video quality a good pirate encoder can fit inside 700mb (to fit on a CD-R... this used to be the standard back in the day)

Of course Netflix can't hire teams of pirates to encode their billion hours of video, so they'd have to automate it. But thinking that current video codecs have reached the limit of efficiency or close to it is absolutely not true.

Re:Another year, another video codec...

[danbob999]

My smart TV can play HEVC content just fine (using the web browser). I doubt it could do it without any hardware acceleration.
Most smart TVs use underpowered smartphone chips.

Re:Another year, another video codec...

[Anonymous Coward]

This so much this.

I have a fairly sizable CD collection. There is no going back to re-rip them. The DVDs and bluray? I just did 1:1 from the start menus/special features/whatever. Space is starting to get a bit lean though. May need to get another 10-15TB. That should last me a few years before I have to buy again.

Re:Another year, another video codec...

[Kjella]

I am in the process of moving my fiancee's dvd/bluray collection to my server and putting her physical copies in storage. Using Handbrake, switching from x264 to x265 saves me at lease 10 % on dvd sources and closer to 30+% on the bluray sources.

And in dollars? You can get an 8TB drive perfect for the task for $260, it's only the random write speeds that are slow. A remux (best quality you can get) is ususally around 25 GB / BluRay or 320 on a drive... less than $1 each. Seems like a fairly cheap investment compared to buying all the discs...

Re:Another year, another video codec...

[JackieBrown]

I'll look at that. In real world, do you notice much difference between a NAS harddrive and a desktop harddrive in a home server?

Re:Another year, another video codec...

[Anonymous Coward]

I have status cables, where do I look to see the difference?

Re:Another year, another video codec...

[hankwang]

Can Handbrake handle subtitles and dvd chapter/title formats well? A while ago I considered ripping my dvd collection. Tried several Linux rippers and gave up after the third dvd.

Converting subtitles to text by OCR'ing low-res subtitle bitmaps Wasa dl5a5 fer... Online subtitle rips were out of sync due to deleted scenes or 24/25 fps mismatches. Figuring out which chapters/titles correspond to the main feature film or episodes and which ones were trailers and copyright notices was also a pain. I also recall annoyances with deinterlacing and 4:3 versus 16:9 formats. Ripping an audio cd is infinitely more easy.

I decided to stick with just finding the dvd if I wanted to rewatch a movie and sit through the silly unskippable parts.

Re:Another year, another video codec...

[JackieBrown]

Can Handbrake handle subtitles and dvd chapter/title formats well?

It does this well. I copy that stuff over too. No manual conversion needed

Re:Another year, another video codec...

[93 Escort Wagon]

I wonder if it's as simple as moving from encoding at a constant bitrate to using a variable bitrate with a quality floor - e.g. Handbrake/x.264's "CR" option.

Re:Another year, another video codec...

[lgw]

I'd keep the original streams. You'll never spare the time again to re-rip, even if you you think now that you will.

Original storage for BluRay is nuts - you can get 10x compression with h.264 with results indistinguishable, screenshot-to-screenshot, by the eye.

For BluRay, compression with those good settings is slower than ripping, on reasonably good hardware. With h.256 you get better results, compression-wise, but it takes even longer.

At some point I may start re-compressing my h.264-encoded BR library using h.265. I doubt I'll bother to re-rip, unless experiments show me I can actually see the difference.

Re:Another year, another video codec...

[SuiteSisterMary]

Netflix would be best served by having multiple formats encoded and having players report capability, and sending the appropriate format.

Re:Another year, another video codec...

[Anonymous Coward]

x264 is an open source software codec for h264 video. Nobody has a x264 chip build in.

Re:Another year, another video codec...

[Anonymous Coward]

HEVC Advance changed that. No way are they going to pay 0.5% of their revenue for licensing.

Re:Another year, another video codec...

[Anonymous Coward]

Original storage for BluRay is nuts - you can get 10x compression with h.264 with results indistinguishable, screenshot-to-screenshot, by the eye.

No.

Re:Another year, another video codec...

[ceoyoyo]

If they're going to save space then they don't have much choice but to drop the bit rate.

Re:Another year, another video codec...

[omnichad]

Re-compressing BD kind of defeats the whole point of bothering with BD. Plus you magnify the aforementioned decode support issues.

I disagree there. They have 50GB to fill and they're going to use as much as they can and likely CBR. If you use Handbrake with CRF at 18 or so, you're not going to see a difference, and you're going to save a bit on hardware if you have a large collection to rip.

But when ripping TV content where several episodes are crammed onto one disc, compare the output to the original. You may have made a larger file.

As for DTS-MA, you could probably extract DTS core and still have way better than DVD audio. For no explainable reason, I preserve the full DTS-MA.

Re:Another year, another video codec...

[ceoyoyo]

That's the great thing about pirates, you don't have to hire them. Netflix just needs to hire someone to search torrent sites and download their catalog.

Re:Another year, another video codec...

[omnichad]

It does subtitles really well. Though I keep it in an MKV container and preserve the subtitles form the disc in their original encoding.

It does add chapter markers, but since the disc doesn't have labels for those chapters, it may or may not be that useful to you.

Don't de-interlace your video when ripping. You're effectively cutting your resolution in half. Better to play back on 60fps capable hardware and have your playback de-interlace at 60fps. It will display each half consecutively, interpolating the missing lines. See yadif 2x in VLC, for an example.

Re:Another year, another video codec...

[lgw]

No.

Sorry, I'm comparing the compressed main title to the raw DVD image, which is filled with all sorts of other stuff. Even then, 10x is rather best-case, and assumes you're only keeping the DTS-quality sound, not the "goes with my $1000 patch cords" raw audio tracks.

Re:Another year, another video codec...

[CelticWhisper]

You shouldn't. "NAS" drives, at least as they come from WD, are just Greens with TLER so they don't drop out of RAID. If you're not RAIDing, or if you're doing software RAID with MDADM or ZFS, shouldn't matter what you use. Even with hardware RAID, it really only matters if you're doing parity RAID. 1 or 10 won't care.

Re:Another year, another video codec...

[Anonymous Coward]

By don't need it you mean that they have discovered that kids don't care about definition so all kids content will be re-encoded with minecraft style blockyness.

Re:Another year, another video codec...

[jedidiah]

You have no idea what you're talking about. You clearly have no clue how large the image files on a BD are and that's just the basic "yeah I've actually done this before" kind of info.

Re:Another year, another video codec...

[jedidiah]

??? Converting subtitles? Why shoot yourself in the foot like that and over complicate the problem? Just extract the image caption you want and remux it into the MKV. Not hard even with the command line tools.

The logic of sorting out main features for a film and episodes on a TV DVD is actually pretty trivial. Just look at the table of contents and see what the run lengths are.

If shiny happy GUI tools can't manage that, it's really embarassing.

Re:Another year, another video codec...

[omnichad]

I've ripped a good 100 of them. The video/audio average around 39GB-45GB for the main feature with only English audio. No, I haven't mastered a Blu-Ray personally, but that's only because the software costs so much.

If you're going to disagree, at least offer a counterpoint.

Re:Another year, another video codec...

[Anonymous Coward]

Agreed. If you transcode H.264 to H.265, you should be seeing file sizes that are half the size at no noticeable loss of quality.

Re:Another year, another video codec...

[Anonymous Coward]

Can Handbrake handle subtitles and dvd chapter/title formats well?

ok and why do you care about those?

Re:Another year, another video codec...

[retchdog]

Eh, even if that were all, it's still not entirely trivial to automatically tune the bitrate for each of 1 petabyte of video files without a lot of human intervention and/or wasted CPU time. At that scale, a minor improvement can still save a lot of money.

Re:Another year, another video codec...

[retchdog]

No. Seeing significant gains in compression ratio from that level of context would be a breakthrough in AI. At any rate, we're nowhere near that point yet, at least in practice. I wouldn't be surprised if there were some wonky deep neural network paper on this, but don't hold your breath for it to become usable.

Re:Another year, another video codec...

[retchdog]

Should we interpret this as a cry for help?

Re:Another year, another video codec...

[greenfruitsalad]

agreed, you should be seeing 20x the electricity bill and have your CPU unavailable for most of 2016.

Re:Another year, another video codec...

[Shirley Marquez]

People doing disc encoding generally plan to use all or most of the capacity of the disc; they want to make the disc look as good as possible given the technical limits of the media. Putting more bits on the disc (so long as you don't have to step up to a higher capacity version) adds nothing to the manufacturing cost. So a major release of a new film on Blu-Ray will be on a BD50 dual-layer disk and use 40GB or more unless the film is very short. Older catalog titles that originally release at $10 or less might be done on a single layer BD25 disc to lower the manufacturing cost. DVD mastering is similar; major films will fill a DVD9 while catalog releases may be on a DVD5. Some early DVDs were DVD10 flip-disc releases with anamorphic widescreen on one side and 4:3 fullscreen on the other, but I haven't seen a new release done that way for years.

When Ultra Blu-Ray comes out next year I expect to see a similar differentiation. New flagship releases will be BDXL100 discs (three 33GB layers) but older titles might be done on BD50; those can be manufactured on existing Blu-Ray pressing lines rather than requiring new manufacturing equipment. I don't expect to see BDXL66 (two 33GB layers) get used often; the makers will either spring for the full 100GB or compress down a bit more to get to 50GB.

Re:Another year, another video codec...

[Shirley Marquez]

Full featured Blu-Ray mastering software is still expensive. But if you can live with more limited capabilities it doesn't have to be. Sony's Movie Studio 13 Suite lists for $140 and B&H is currently selling it for $90; it can create Blu-Ray discs with basic menus. You will also need a Blu-Ray drive for your computer, which starts at around $50 for an internal SATA drive or $75 for an external USB drive. Perhaps the most important limitation is that it can't do surround sound; for that you'll have to step up to a more complete (and expensive) solution.

Of course, for some users that's still a lot of money. But it's a lot less expensive than what you would have needed a couple of years ago.

Re:Another year, another video codec...

[Shirley Marquez]

They are talking about improving the process for encoding with their current codec. Going to H.265 might happen in the future but right now it would only work well for a small percentage of playback devices, so they could do it as an option but would still have to offer the existing encoder as well.

Re:Another year, another video codec...

[omnichad]

Looks like Sony sells just the authoring software standalone for $39.95. Same one that's in Movie Studio Suite. But having played with Sonic Scenarist, and worked with DVD Studio Pro, I don't know if I could handle using something at the iDVD level with basic themes and graphics imports. If Apple had backed Blu-Ray instead of HD-DVD, DVD Studio Pro would be useful for Blu-Ray. It has full HD-DVD authoring.

That leaves me with DVD Architect Pro from the Sony side. Not too bad at $199. But I've been out of the video production game for a while. No idea if I'll ever need it. I do have a Blu-Ray burner, though. I've only used it for ripping my movie collection.

Re:Another year, another video codec...

[Anonymous Coward]

There are adaptive bit rate compression algorithms that look at the characteristics of small areas of the image (flat, low gradient, sharp edges, noisy like hair/grass) and assign different numbers of bits to encode it. That is what they are likely talking about. I did my master's thesis on one such algorithm

Re:Another year, another video codec...

[Shirley Marquez]

You can download trial versions to see if they do enough to keep you happy.

Unlike iMovie and the like, the Sony Studio applications use the same interface as the Pro versions and just take out some features. They also bundle fewer assets and plugins. In DVD Architect Studio, the most notable limitation (other than the track count) is that it can only do stereo PCM audio; no Dolby Digital (that encoder would add to the cost of the program) and no multichannel PCM.

Both the Studio and Pro versions of DVD Architect have limited video encoding options: CBR only and the only progressive frame rate available is 24p. Sony Vegas has more options; if you're working entirely in the Sony ecosystem the expectation is that you will encode the video there and then bring the encoded video tracks into DVD Architect. I believe that will also work with video from non-Sony applications but I've never tried that. Similarly, you can import encoded audio tracks.

Your level of experience is well beyond the "give simple discs to my friends" level, so I suspect you would find the limitations of the Studio version to be too restrictive. But as you point out, $200 for the full version isn't too bad and better prices are available: B&H sells it for $155.

Re:Another year, another video codec...

[omnichad]

I usually encode my own video, so it's possible even the pro version doesn't let you take in your pre-encoded video and dolby digital audio and let it go through unchanged. I might try sometime.

Re:Another year, another video codec...

[hankwang]

DVD subtitles are low-res, blocky bitmaps that don't scale nicely. Tghat's why I'd like to have them in ascii or utf-8 format. Sorting out and labeling the tracks of a dvd with a few 20-minute episodes of a series, trailers, and special features is not mentally challenging, but it still takes time and the balance of that time versus sitting through the unskippable parts for the two times that I might re-watch that dvd in the future tips in favor of not ripping.

Re:Another year, another video codec...

[powerlord]

How many of the existing devices are programable?

At this point, most devices are capable of downloading updates to playback apps, and the fact that they are connected to broadband for streaming video handles the required internet connection.

I've seen points in the past where playback mandated an app update, so I wouldn't discount that as a possibility.

Re:Another year, another video codec...

[Shirley Marquez]

Devices are programmable. But devices (as opposed to computers) often have relatively wimpy CPUs and depend on specialized decoding hardware for acceptable performance. Switching from a codec that has hardware support to one that does not will not work well on such devices. That includes most set-top boxes and smart TVs.

Mobile has the same problem on older devices but a different one on new devices.The CPU might have enough computing power to handle the new codec but doing the decoding in the CPU rather than the video hardware is likely to shorten battery life considerably.

So I think my basic point stands. Switching to H.265 is not viable for the majority of the installed base right now, but if Netflix is willing to do dual encoding they could get some benefit by using it for the users that can handle it. That will change with time as more of the installed base gets H.265 hardware decoding. But perhaps by then the hot topic will be whether to switch to Daala: https://xiph.org/daala/

Meanwhile, if they have found a way to improve the efficiency of their H.264 encoding that's a win all around. Lower bandwidth bills for Netflix, less congestion on the internet pipes for everybody, and everybody's systems still work. And they plan to do the re-encoding by using their existing hardware during off-peak hours so the only expense is a bit of extra electricity, because modern computers consume more power when they're actually doing something than when they are idle.

"without reducing the quality of streamed video"

[Anonymous Coward]

Yeah, right.

Neat... but why?

[Kenja]

So they should be able to do this without effecting quality due to better compression algorithms being available now, but what problem is this trying to address? Other then keeping that one intern they don't like locked in a small room running HandBreak 24/7.

Re:Neat... but why?

[medv4380]

Look at the comparison. I'm willing to bet a large amount of their audience happen to be watching cartoon, or anime. When your color pallet is reduced you don't need to save the entire thing in 32bit color to get the same exact quality. Some will be shaved, and others wont be shaved that much.

Re:Neat... but why?

[Anonymous Coward]

Because that would require updating their app on dozens of platforms they haven't touched in years. BluRay players, smart TVs, and old game consoles that still work but wouldn't be able to make sense of a new format

Re:Neat... but why?

[c0d3g33k]

What problem is this trying to address?

Saving on bandwidth costs?
Providing a better streaming experience for customers on poor or throttled connections?
Storage space savings?
Getting the satisfaction of doing something better because why not?

Re:Neat... but why?

[loony]

Three words: Comcast data cap...

Peter.

Re:Neat... but why?

[Junta]

Actually I doubt they'd reduce the color precision from 24 bit (it's not 32 bit).

In fact, content with lots of synthetic content can sometimes be smaller by being 30 bit instead of 24 bit (think how often synthetic content puts in gradients, with 24 bit those gradients are more dithered than 30 bit, and the compression algorithms struggle a bit more with what appears to be 'noisy' content from dithering compared to less noisy undithered content).

Of course this is using general purpose algorithms that are used for both animated and photorealistic alike. There may be some gains in theory from a codec focused exclusively on animated content, but in practice no one seems to think it's worth the trouble to pursue.

Re:Neat... but why?

[Anonymous Coward]

Yea, I thought this was pretty obvious. Everyone knows ISPs are already starting to screw their customers with data caps. Some customers might mull cancelling Netflix if it became a serious enough problem financially or they might just go back to cable because it's all you can watch. A 20% reduction would be spectacular. Hopefully they can find more ways to cut their data usage.

Now if Steam could get game companies to find ways to shave data too...

Re:Neat... but why?

[c0d3g33k]

What problem is this trying to address?

Saving on bandwidth costs?
Providing a better streaming experience for customers on poor or throttled connections?
Storage space savings?

So nothing that's REALLY important then?

So what would something REALLY important be?

Re:Neat... but why?

[Junta]

Maybe better implementations of existing algorithms, but they have to tread carefully about new algorithms.

I assume today they use H264 and the re-encode would just be newer implementation of H264 encode/different settings than they used before.

They could get more from jumping to HEVC, but netflix is on crap tons of smart TVs and such.

Of course that's not to say they transcode to HEVC and client advertises whether it's H264 or HEVC and netflix just keeps both H264 and HEVC live on their CDN, if capacity is no big deal.

It's obvious it saves bandwidth, and if Netflix were having someone manually babysit instances of a transcoder that would be ridiculous. Every transcoder is automation friendly and assuming they've done it all right it's a matter of kicking off a massive parallel job to chew on the library using their idle capacity. It's a common thing to do when you have low priority workload and gobs of servers. Now if they are having people hand review the result of the transcode for video quality issues that would be expensive, but I doubt they would. They would cherry pick a few representative scenes of content and verify their batch job looks good and fire away. Then they would rely upon user issue reports to catch anything they may have missed.

Re:Neat... but why?

[Applehu Akbar]

I'm struggling with a data cap too, so I'm saving Netflix bandwidth by getting all my content on DVDs.

Re:Neat... but why?

[known_coward_69]

saving bandwidth and saving them money from buying new CDN servers. they are probably at the point where a lot of their servers are running at capacity and it's cheaper to lower the bitrate than buy new servers, routers and pay for the installation and everything that goes along with it

Re:Neat... but why?

[pnutjam]

Pied Piper strikes again? Hulu just brought piss to a shit fight.

Re:Neat... but why?

[Anonymous Coward]

What problem is this trying to address?

Saving on bandwidth costs?
Providing a better streaming experience for customers on poor or throttled connections?
Storage space savings?
Getting the satisfaction of doing something better because why not?

Showing up that Gavin Belson.

Re:Neat... but why?

[Minwee]

What problem is this trying to address?

Saving on bandwidth costs? Providing a better streaming experience for customers on poor or throttled connections? Storage space savings?

So nothing that's REALLY important then?

So what would something REALLY important be?

Increased fidelity for rendering skin tones?

Re:Neat... but why?

[cdrudge]

Improved sarcasm enhancement

Re:Neat... but why?

[sexconker]

Horse shit.
There is more "noise" (signal) in a 10-bpc source than an 8-bpc source OR an 8-bpc conversion of a 10-bpc source.
You just can't see it as easily with your eyes.

Re:Neat... but why?

[Anonymous Coward]

I thought that's exactly what Netflix did: send DVDs through mail.

Re:Neat... but why?

[Mariner28]

So can someone tell me how this doesn't violate Net Neutrality rules? Video streamed from Comcast's own source properties doesn't count towards data caps, yet watching the same movie from Netflix does? And Netflix has their own caching servers installed directly inside Comcast's distribution network? WTF?

Re:Neat... but why?

[c0d3g33k]

Improved sarcasm enhancement

You're spot on there - your sarcasm enhancement wasn't effective at all. The sarcasm totally came across as a hastily submitted sentence fragment that required no thought at all apart from an overestimation of your own cleverness and talent at subtle humor.

Re:Neat... but why?

[omnichad]

Haven't been to Dollar General lately, have you? They've been selling off the majority of their DVDs and not replacing damaged ones. I have movies that have been on my queue for years that are now (permanently) unavailable. I had a big backlog and now I don't really have a way to watch them - they're certainly not available for streaming. And they already put rental stores out of business. I would literally have to buy some of these movies to watch them.

Re:Neat... but why?

[omnichad]

having people hand review the result of the transcode for video quality issues

I can assure you they don't do that. I watched through Buffy the Vampire Slayer on Netflix recently (for the first time). There are bad spots on the Beta master tapes they were transferring from and the picture actually goes blue for brief moments (as you'd see on some VCRs when there's no usable signal).

Re:Neat... but why?

[Anonymous Coward]

For poto realtistic, yes.

For animated content, certain patterns are common that are more compression friendly at un-'dithered' 10 bit per color than dithered 8 bit per color. True, if both are dithered or neither, 8 per color will be smaller, but banding at 8 bits per color is more pronounced without dithering it. The affinity for simpler colors and gradients in synthetic content just don't play well with usual assumptions about 24 bit color.

Re:Neat... but why?

[Anonymous Coward]

A few ideas why, without looking at their annual report:
- Is bandwidth their primary cost, or licensing?
- Do they pay the same licensing costs for My Little Pony or Avengers? Who complains first when video quality is bad?
- When did they last establish bandwidth agreements, and are they up for renewal?
- Does lower bandwidth allow growth through entry into markets that have less bandwidth? /BB/

Re:Neat... but why?

[Anonymous Coward]

So they should be able to do this without affecting quality due to better

FTFY

Re:Neat... but why?

[Anonymous Coward]

So can someone tell me how this doesn't violate Net Neutrality rules? Video streamed from Comcast's own source properties doesn't count towards data caps, yet watching the same movie from Netflix does? And Netflix has their own caching servers installed directly inside Comcast's distribution network? WTF?

Wish I could mod this "-1, Off Topic"

Re:Neat... but why?

[Shirley Marquez]

The problem they are addressing is the cost of bandwidth. When you're sending out bits on the scale of Netflix the bills add up. Saving 20% of that is well worth the effort.

1 Petabyte?

[Anonymous Coward]

Wow, 1 Petabyte... pretty soon they might surpass the size of the first volume of my porn collection.

90's sat tech

[silas_moeckel]

Back in the MPEG2 Sat days they regularly used different bit rates depending on content talking heads very little compared to full out for sports and action movies. An actual knowledgeable encoding tech can do wonders, higher quality source material can also do wonders.

Re:90's sat tech

[fastest fascist]

Netflix probably aren't too keen on the idea of paying people to puzzle over what compression would best suit each and every item in their 1-Petabyte video library.

Re:90's sat tech

[silas_moeckel]

It could make sense for highly viewed content. While the entire library is that big the highly viewed content not so much. Can guestimate from the their appliance has 288 raw TB or so that should be enough to fit all the high demand content.

Re:90's sat tech

[c0d3g33k]

Netflix probably aren't too keen on the idea of paying people to puzzle over what compression would best suit each and every item in their 1-Petabyte video library.

The summary says they spent four years developing the new approach. I suspect that paying people to puzzle over (in layman's terms: do research) how to improve the encoding across their Petabyte video library was exactly what they did.

Re:90's sat tech

[SScorpio]

They still do this today. Cable companies even mess with the bitrates for different channels. An HD antenna can usually beat the re-compressed version you get over cable.

Re:90's sat tech

[SydShamino]

Netflix probably aren't too keen on the idea of paying people to puzzle over what compression would best suit each and every item in their 1-Petabyte video library.

I think they would just encode each item with every compression algorithm they support, then spot check with humans to find the lowest-bandwidth option deemed "acceptable" for a given viewing device and style of video. It need not be the same algorithm for each device; an animated TV show going to a Wii or a cell phone might allow for a lower-bandwidth, lower-quality option than streaming to a Roku or Apple TV, for example, and both could be lower bandwidth than a Hollywood movie. Their cost isn't in storing multiple copies of each video, it's in streaming those videos, so anything they can do to lower the bandwidth is good for their bottom line.

Re:90's sat tech

[SydShamino]

288 TB should be enough for anybody

Re:90's sat tech

[omnichad]

An HD antenna can usually beat

Will always beat. It's re-compressed from what the cable company gets over antenna. You can't bring picture quality back by re-encoding.

and the Star Wars Prequels?

[Anonymous Coward]

Presumably the three of those are compressed down to zero bits?

Re:and the Star Wars Prequels?

[Eunuchswear]

Whereas the original Star Wars is compressed down to one bit -- Solo fires first yes/no.

Re:and the Star Wars Prequels?

[Higaran]

They exist, get over it, you might decide you actually enjoy them depending on how much lens flare they put into this next one.

Re:and the Star Wars Prequels?

[tepples]

You still need II and III for Machete order (IV V II III VI). This order makes the most narrative sense by skipping the toy commercial that is The Phantom Menace and treating Attack of the Clones and Revenge of the Sith as an extended flashback.

Multipass

[bjb_admin]

Almost sounds to me like they have switched to multiple pass encoding, rather than a fixed quality/bandwidth setting.

"The new system will encode from the raw source material more intelligently, considering whether or not the material itself can really benefit from higher bit-rates, or whether identical quality can be maintained with less space and bandwidth."

Re:Multipass

[Anonymous Coward]

I came here to say this. A lot of naysayers here saying it's impossible. It's completely possible, even with the same codec, if you just use it properly. Think the size of a VBR mp3 vs. a CBR mp3. Huge space savings and identical quality. And yet, the same algorithm.

Re:Multipass

[Anonymous Coward]

VBR mp3 and CBR mp3 are not even remotely similar.

VBR adjusts the rate for silent parts of the song, CBR does not.

Re:Multipass

[Anonymous Coward]

You got it backwards. Fixed quality is what they have switched to. Multiple pass encoding is used to find the maximum quality setting that will fit a predetermined file size budget.

Re:Multipass

[Anonymous Coward]

WRONG. VBR adjusts the rate to match up with the highest detected frequency and accurately sample only as far as needed for that slice of time.

Re:Multipass

[sexconker]

That's a good WRONG. Love it.

Repetition

[Impy the Impiuos Imp]

They can save about 500% of my bandwidth by just letting me perma-download Family Guy, American Dad, and Buffy, which I keep watching over and over and over again.

Re:Repetition

[OzPeter]

They can save about 500% of my bandwidth by just letting me perma-download Family Guy, American Dad, and Buffy, which I keep watching over and over and over again.

I would love the ability to preload a fixed number of shows to a device even if it was for a time limited period.

Re:Repetition

[c0d3g33k]

They can save about 500% of my bandwidth by just letting me perma-download Family Guy, American Dad, and Buffy, which I keep watching over and over and over again.

Look into something called "Boxed Sets".

Re:Repetition

[OzPeter]

Look into something called "Boxed Sets".

So then .. looks like I'll have to buy the white album .. again.

Re:Repetition

[jedidiah]

If you really care about what you're watching and you're planning on watching it again, it really makes much more sense to just get the boxed set. Rip or not but you will still have it after it disappears from Netflix.

Plus you won't have to worry about the quality of the stream you're getting from Netflix or any other shenanigans they might pull with the original content.

You don't have to have the entire run of 200 series on your media server. So the HTPC option doesn't need to be too complicated. '-)

Re:Repetition

[93 Escort Wagon]

If you really care about what you're watching and you're planning on watching it again, it really makes much more sense to just get the boxed set.

This is basically what we do. Netflix (DVD or streaming) is how we watch a movie we want to see the first time. If it's one we really love and are likely to want to watch many more times in the future - The Fifth Element comes immediately to mind - then we buy it, I rip it, and it goes on an old MacBook Pro that acts as our in-house video server.

Thing is, I bet we watch 50 movies for every one we purchase. It's just not that often that I want to see a movie again, having seen it once - even if we enjoyed it. This is even more true with TV series - once you know whodunnit in any particular episode of Foyle's War, there's not really a point in watching it again.

There are exceptions to this rule, of course. I have to own every James Bond movie... even the sucky ones. So I'll probably just buy Spectre as soon as it's available.

Re:Repetition

[quenda]

it really makes much more sense to just get the boxed set. Rip or not but you will still have it after it disappears from Netflix.

Too much effort to rip all that physical media, especially when other people have already done it for you. Nobody rips CDs any more, you just download it from a paid service or torrents etc.

    Netflix could save the world obscene amounts of bandwidth (billions of dollars in infrastructure upgrades?) if they'd just allow clients to cache programs. If only it could off-peak download the next episode in a series I'm watching, or movies added to "My List", the world would be a better place. I really believe Netflix could prosper with no DRM at all, and an open API, but the studios are blocking it. Just as they did with music, but the industry did not collapse when Apple removed DRM.
Netflix adds real value, that people are happy to pay for.

Re:Repetition

[tepples]

Netflix could save the world obscene amounts of bandwidth (billions of dollars in infrastructure upgrades?) if they'd just allow clients to cache programs.

Columbia, Disney, Fox, Paramount, Universal, and Warner Bros. could save the world obscene amounts of bandwidth (billions of dollars in infrastructure upgrades?) if they'd just allow Netflix to cache programs.

class action suit

[sunking2]

As a My Little Pony enthusiast who pays the same per month as everyone else I demand the same quality as the Avengers.

Re:class action suit

[Anonymous Coward]

Maybe you should invest some time in a shrink.

Re:class action suit

[Kjella]

Are you sure you haven't agreed to a binding arbitration agreement? >:-)

Re:class action suit

[Tablizer]

Pony haters will hate.

Re:class action suit

[Anonymous Coward]

How do you think I got my ponies so little?

Re:class action suit

[Anonymous Coward]

Only on broken adult men who like them. If you're an 8yo girl, that's fine.

Re:class action suit

[Anonymous Coward]

Haters gonna hate, ponies gonna pwn.

Re:class action suit

[Anonymous Coward]

Wish granted. They will be lowering the quality of My Little Pony to match Avengers.

Re:class action suit

[argStyopa]

If you're a fan, why would you want to dumb-down My Little Pony?

Re:class action suit

[Ksevio]

But the new algorithm means you'll be getting BETTER quality for your cartoon using LESS bandwidth than the noise-filled action movie.

Wasn't this the entire plot....

[tekrat]

Wasn't this the entire plot to that HBO series "Silicon Valley", where a group of geeks form a company to re-encode petabytes of porn?

I never really got to see the whole series, but it came on after Game of Thrones, so I saw the occasional episode.....

Re:Wasn't this the entire plot....

[godrik]

I think you are confused. Game of Thrones is the porn they were rencoding! :)

The Open Edge Content Delivery Network, perhaps?

[fredan]

TOECDN would have fixed this shit already. We don't want less quality, we want more!

What we need is the death of "bandwidth caps"

[nerdyalien]

I watch YouTube a lot, on average about 2-3 hours a day. As of late, I live in a country where there is a bandwidth cap of 40 GB/month. And I have no option but to YouTube at 144p to avoid extra bandwidth charges.

I applaud all efforts by tech companies to reduce bandwidth usage (and not to forget, making inter-webs more exciting). Then again, none of those efforts matter, if bandwidth caps are forcing consumers to use internet like back in 90s.

Re:What we need is the death of "bandwidth caps"

[Anonymous Coward]

You could move to Kansas City.

Re:What we need is the death of "bandwidth caps"

[c0d3g33k]

Or find another hobby.

Re:What we need is the death of "bandwidth caps"

[Anonymous Coward]

>I watch YouTube a lot, on average about 2-3 hours a day.

How about getting a life!?

Re:What we need is the death of "bandwidth caps"

[Anonymous Coward]

a bandwidth cap of 40 GB/month.

That's a data cap.

Bandwidth cap would be your maximum speed.

Re:What we need is the death of "bandwidth caps"

[Anonymous Coward]

In much of the US, we don't do bandwidth caps, but instead the local phone companies and cable companies limit the bandwidth. I'd rather have a cap so I could burst to faster speeds. At my condo in downtown Seattle, I have 160 kbps DSL. I'm lucky because most of my building can't even get that. They're stuck on ISDN with per minute charges. Comcast has tried to offer service to the block, but the city hates the Internet and will not allow them permission to install pedestals. Satellite isn't an option because of the tall buildings to the south. I wish people in cities in the US had connections fast enough to use Netflix. Personally, I've never been in a home with a connection fast enough to stream video, so I know that 70% number that keeps getting thrown around is BS.

Re:What we need is the death of "bandwidth caps"

[CastrTroy]

How else do you expect internet providers to limit people from using up all the available bandwidth? Even with a relatively slow connection of 5 Mbit/s, you could pull down 1.6 Terabytes every month if you left your connection running 24/7. And a 5 Mbit connection is probably slower than most people want in their homes. Because they want to be able to download a small number of things relatively quickly.

I'd much rather have a 100 Mbit connection that was limited to a certain number of gigabytes a month than have a 1 Mbit connection to ensure that I wasn't able to download more than my fair share in a month even if I maxed it out every second of the month.

I'll admit that 40 GB per month is pretty low, but there has to be limits on how much people will consume. Especially with more and more content being made available over the internet. Back when cable internet first came out, most ISPs didn't have bandwidth caps, simply because there wasn't enough content out there for most people to use an appreciable amount of bandwidth. Now with providers coming online with 4K video at 60 FPS, it starts to become a real concern and even the average person can use 500 GB per month without even trying. Having limits makes people think about how much of the limited resource they are using and budget accordingly.

Re:What we need is the death of "bandwidth caps"

[Anonymous Coward]

You could always move out to the countryside where we have no infrastructure (so you have to maintain a car), everything costs twice as much (food, fuel, gasoline, etc), you have to drive 40 miles to the grocery store, AND have crappy bandwidth.

Re:What we need is the death of "bandwidth caps"

[Anonymous Coward]

Wow, I'm always reading how Seattle's Internet sucks and that's even if they can get Internet at all to homes. Is it really that bad there? Makes me ecstatic about my 15Mbps download connection. I always felt my 15Mbps down (sometimes bursting to 18Mbps), 1Mbps up was pathetically slow, but I guess I have it good.

Re:What we need is the death of "bandwidth caps"

[Anonymous Coward]

As someone who just used 1.2TB this month, I feel absolutely horrified with what you have to live with.

Re:What we need is the death of "bandwidth caps"

[Anonymous Coward]

Except the "limited" resource is less like a pool of stuff that runs out and more like lanes on a highway.

It really doesn't matter how much everyone individually is consuming.
What matters is how much is being consumed at the same time.

I could understand if, at peak hours, ISPs had no choice but to employ QoS and throttle the top speed of everyone's connection.
Granted, that's a sign that their infrastructure needs improvement.
But still, "bandwidth caps" is attacking the problem from the wrong angle.

Re:What we need is the death of "bandwidth caps"

[Anonymous Coward]

That's complete bullshit. Network capacity isn't an amount, it's a rate. If it is a capacity problem they can simply cap the speeds during commercial hours. Contrary to popular belief, people at home are not the biggest consumers of network bandwidth.

They just want to milk more money out of the costumers, like the cable companies.

Re: What we need is the death of "bandwidth caps"

[Anonymous Coward]

The city is hardcore anti-Internet. They won't allow the new building where I live to get cable TV, because then they would have no way to block Comcast from selling Internet access.

Re:What we need is the death of "bandwidth caps"

[Anonymous Coward]

GB limited bandwidth caps only serve to show the ISP is third tier at best. They obviously don't buy internet bandwidth in any normal fashion.

Why? A 40 GB cap just serves to tell people "Only use it for an hour or two at peak times when it is the most congested!". Now you need really fat pipes that sit unused most of the day.

A proper method to limit usage that will actually save the ISP money would be to figure out your 95th percentile peak hours (You know, the same way you're billed by the first tier providers you're supposed to be buying bandwidth from...) and cap usage during that time. Let's say those hours are 7 - 9 pm. You get 40 GB total usage during that time, and if you step outside of your cap you're either billed extra or your internet turns off from 7-9.

Outside of those peak hours, your usage is unlimited. You want to decrease your peak hours and increase your non-peak hours to balance the network and make the best use of the bandwidth you have.

Now let's think of the "abuser", the one who does 1.6 TB a month at 5 mbits (I've been there!). If your peak hours are 7-9 pm, and at all other times your traffic is low enough that another 5 mbit doesn't max our your pipes, that abuser is only causing trouble for 3 hours a day. The rest of the data is inconsequential to running your business. By pricing your services properly, that abuser now becomes a good customer. They adjust their internet usage in just a small way (ending downloads for 3 hours a day, reducing their consumption to 1.4 TB) and everyone else wins (the internet goes 5 mbits faster during those congestion times); and you get to keep them as a paying customer that is now a profitable customer.

Why ISPs keep doing this dumb thing of telling those customers to leave, I don't know. That heavy user is typically the one family and friends ask which ISP to use. You think they're going to recommend one that worked to kick them off the network? Especially when a mutually beneficial solution was easy to implement?

And yes, if you are Netflix, you can work your application around peak hours. There is no reason why Netflix couldn't precache content it knows you are interested in during the least used hours of the day. Now you magically have 0 mbits usage from 7-9 pm. It's the ISPs own fault for having an idiotic business model, especially since they resell service from top tier ISPs that are willing to explain the proper business model to them. For free. Ugh...

Re:What we need is the death of "bandwidth caps"

[Anonymous Coward]

I don't think we need to get rid of bandwidth caps but we need to make them reasonable. They're too low on consumer connections for the speeds sold now.

It should not be possible to use them in a day.

Re:What we need is the death of "bandwidth caps"

[Anonymous Coward]

a less rude way to say it is that perhaps GP should consider finding a new hobby. one with a tangible outcome or that involves interacting with other people. (not good at it? then you need more practice)

yeah so upon rereading I'm going to stop looking at /. now and go enjoy the day.

Re:What we need is the death of "bandwidth caps"

[93 Escort Wagon]

I hear everything's up to date in Kansas City - they've gone about as far as they can go.

Re:What we need is the death of "bandwidth caps"

[esperto]

As someone that uses 7 to 8 TB a month on average, I just want to brag.

Re:What we need is the death of "bandwidth caps"

[Anonymous Coward]

Protip: Go to your public library, install the "Video DownloadHelper" add-on in Firefox, and download from Youtube in glorious 1080p to your hearts content. Then you can go home and watch the videos. That's what I do (fuck off comcast).

Re:What we need is the death of "bandwidth caps"

[Anonymous Coward]

If you want it this way, make it a simple one click for admins to do. Also, shutting off Internet completely is too traumatic. Drop it to kbit/s speeds, sure. Kill it completely = arggggggh.

Re:What we need is the death of "bandwidth caps"

[Anonymous Coward]

What is an "inter-web"?

I never heard of that technology. Is it something unique to your region of the world?

Seriously! Inquiring minds want to know "what is an inter-web"?

Re:What we need is the death of "bandwidth caps"

[Anonymous Coward]

Welcome to Australia.

Re:What we need is the death of "bandwidth caps"

[Anonymous Coward]

1.6TB a month is nothing ...

But i guess i'm just weird as we do sell services at 10-15€ a month with 100mbps unmetered connectivity or at times even "1Gbps" ... we have customers paying 15€ a month and doing 40TB+ a month.

The only explanation for bandwidth caps of gigabytes (not terabytes) is something is wrong in business model OR lack of competition in the area

Current market rate for transit is around 0.20€ per Mbps for big operators.
Say you have 10Gbps and 1000x100Mbps customers, you are never going to reach the full 10Gbps :) Contetion ratio of 1:10 is very mild in the consumer connection business, 1:1000 is more like it.

What actually costs A LOT is the copper, interconnectivity, in other words: Last mile. Not the bandwidth, getting the connection all the way to your home does.

Re:What we need is the death of "bandwidth caps"

[tepples]

A proper method to limit usage that will actually save the ISP money would be to figure out your 95th percentile peak hours (You know, the same way you're billed by the first tier providers you're supposed to be buying bandwidth from...) and cap usage during that time.

Good luck explaining such a plan to non-technical subscribers who value simplicity over capability.

There is no reason why Netflix couldn't precache content

Of course there is. If Netflix's agreement with its programming licensors doesn't allow precaching, Netflix would be liable for copyright infringement.

Re:What we need is the death of "bandwidth caps"

[tepples]

"Inter-web" pokes fun at non-technical users who are unaware that the Web is a session and presentation layer protocol running on the Internet.

Netflix's catalog is already shrinking

[NotDrWho]

With all their efforts concentrated on their original series, it seems like their movie and TV offerings already shrink every month already, without any compression.

Re:Netflix's catalog is already shrinking

[Anonymous Coward]

Aside from Google Play, were you have to pay to watch anything, NetFlix is the only streaming company with any sort of appreciable size of selections and options. Hulu is ok and I dropped Amazon because I got tired of only having options for weird, obscure shows from the 80s and 90s that nobody watched back then and their "new release" options lack compared to Google.

Doesn't VBR already do this?

[JoeyRox]

From the article:
"The new system will encode from the raw source material more intelligently, considering whether or not the material itself can really benefit from higher bit-rates, or whether identical quality can be maintained with less space and bandwidth."

I thought existing VBR algorithms already account for the absence of interframe changes by reducing the effective bit rate for those frames.

Re:Doesn't VBR already do this?

[omnichad]

They're probably switching to VBR. And trying to make it sound like they are doing something new so it isn't obvious how behind they are.

Re:Doesn't VBR already do this?

[kriston]

Exactly.

Netflix is trying to impress its shareholders by covering up the fact that they have blithely ignored, or have been ignorant of, established VBR technology that has been in use by the satellite TV companies for over TWENTY years.

Of course, in the satellite TV case, the window for hierarchical compression is much smaller. Netflix finally realized (duh!) that you can do this over the entire video production.

So, here is Netflix trying to cover up the fact that they were ignorant when it comes to real-world video content encoding by saying "Hey, look at this new thing we figured out!"

Re:Doesn't VBR already do this?

[omnichad]

Or Netflix doesn't save much money with this and they're only saving it to bump up share prices at the holidays so some smart execs can sell off a few shares.

Netflix = The new YIFY?

[Anonymous Coward]

So they're going to start bitstarving releases to save on bandwidth costs?

Re:Netflix = The new YIFY?

[Ravaldy]

RTFA. They claim no loss of quality so who cares about the bit rate.

Need a new TV

[Infiniti2000]

I know my current LG has already limited upgrade support for older SmartTV functions, so it's possible (likely) that my current TV will not support this upgrade. Crap.

Re:Need a new TV

[laie_techie]

I know my current LG has already limited upgrade support for older SmartTV functions, so it's possible (likely) that my current TV will not support this upgrade. Crap.

That's why I have a "dumb" TV and a smart bd player. It's cheaper to upgrade a smart bd player, then pass the decoded content to my tv.

Re:Need a new TV

[Anonymous Coward]

You think that's bad? I bought a house with a smart TV built in to it. There's no way to update the TV, so the entire house and TV are garbage now. It's a piss-off too because just last year I had to replace the house because Android no longer provided security updates for the cell-phone that was built into it.

Why 4k?

[Anonymous Coward]

Every article about video lately says everyone is going to go to 4K...who exactly are these people that will gladly sit 3 to 4 feet from their TV? I have a good size 'viewing room' with a 60" screen and there's no way I'd sit that close to the TV, it would feel uncomfortable & mess up the entire flow of the room anyway! Besides resolution is only 1 factor in regards to clarity. Generally rec rooms/brorooms are wider/longer than the 3 to 4 feet that is the maximum viewing distance for 'decent' sized TV's, anything smaller than 60" and your sitting even closer to get any benefit...just another scam like 3D if you ask me. Don't get me wrong I'm sure it would benefit a small group but certainly not the majority so it's just another way to hopefully get the masses to replace their existing good TV's...what a waste.

Re:Why 4k?

[JackieBrown]

Generally rec rooms/brorooms are wider/longer than the 3 to 4 feet that is the maximum viewing distance for 'decent' sized TV's, anything smaller than 60" and your sitting even closer to get any benefit...just another scam like 3D if you ask me. Don't get me wrong I'm sure it would benefit a small group but certainly not the majority so it's just another way to hopefully get the masses to replace their existing good TV's...what a waste.

I felt this way for the VHS to DVD conversion then even more so during the DVD to bluray conversion. I was completely wrong both times so I think I'll be shutting up during the jump to 4k and just look forward to enjoying it in 10 years when the price allows me to.

(I encode everything to 720p just so I won't start getting use to 1080p)

Re:Why 4k?

[Anonymous Coward]

4K...who exactly are these people that will gladly sit 3 to 4 feet from their TV?

WTF are you blabbering about!? I got a 49" 4K and I am sitting about 3+ meters (10 feet for muricans) away with no need to move closer.
WTF watches TV from less than 2 meters?

Re:Why 4k?

[jedidiah]

> WTF are you blabbering about!? I got a 49" 4K and I am sitting about 3+ meters (10 feet for muricans) away with no need to move closer.
> WTF watches TV from less than 2 meters?

People stuck in small apartments in Europe where there may not be 10 whole feet from one wall to the other. [snicker]

Re:Why 4k?

[avandesande]

I have a regular sized house and only have a 39"- don't want the TV to dominate the living room area.

Re:Why 4k?

[omnichad]

Because at that distance, your eyes can't resolve 4K. 350ppi is about the limit for holding a phone inches from your face. At 3+ meters, a 49" TV is going to be about the equivalent of a 3 inch phone screen.

auto-play

[Larry_Dillon]

They could save 20% of their bandwidth by having a way to disable the auto-play of the next episode.

Re:auto-play

[JackieBrown]

I love the autoplay next episode option. I keep meaning to look for a plugin that does that with KODI.

Re:auto-play

[Bengie]

I wish they allowed you to continuously auto-play. They have a limit on how many episodes right now. Many times we have Netflix on just for background noise as I read books or the wife does her thing. Instead, we need to either use Hulu or actually turn on the TV.

Re:auto-play

[CastrTroy]

There already a is a way to do this. I'm pretty sure it's been available for quite a long time. Although I don't know why autoplay is on by default. Same goes for Youtube. Their's is even worse, as I can't find a way to disable it permanently. You can disable it for a session, but if you leave and come back the next day, then the setting is back on again.

Re:auto-play

[Anonymous Coward]

I stream Netflix on my Wii, on Chrome on Linux, and on a really old version of the Android app. None of those auto-play the next episode.

Re:auto-play

[Anonymous Coward]

You are the reason why network admins cry themselves to sleep. Bandwidth hog.

Re:auto-play

[Bengie]

I have a dedicated 100/100 fiber connection for $90/m, and my ISP uses Level 3 directly to Chicago for all of their traffic. I asked my ISP about getting a Netflix CDN and they said they have plenty of bandwidth and refuse to co-lo Netflix' CDN service because it would be unfair to favor Netflix over others. 0.0001% monthly packet loss(which includes the random 2am maintenance downtimes) and 1ms stddev of jitter to 8.8.8.8 and 8.8.4.4. Even looking at my PFSense quality graph, the pings to my ISP never go above 2ms+-0.1ms. Freenet and Torrent running 24/7. About 6TiB/month. Wife lets Hulu run all day on auto-play. We fall asleep to Netflix at night.

What the hell?

[Anonymous Coward]

You shouldn't allocate the same amount of bits for My Little Pony as for The Avengers.

Who the hell is deciding which content should have shitty artifacts? I have no interest in having shitty compression shoved down my throat because somebody at Netflix deemed my show unworthy of 'good' compression., especially if it's just to use those bandwidth savings to improve shows I don't watch. Netflix is playing with fire here, but I must admit they seem to be following the standard American business model: make a good product, gain a following, then cut corners on quality and service to rake in profits on the margin until people realized you've fucked them.

Re:What the hell?

[Anonymous Coward]

Found the brony!

Re:What the hell?

[Anonymous Coward]

Who the hell is deciding which content should have shitty artifacts?

Brony here too, but that's actually not what they're saying -- they're saying that vector art can compress down a lot more easily than CGI explosions that feature a thousand pieces of shattered glass swirling around the viewer.

For a real-world example of this, look at the sizes of the scene rips of this season of South Park. Episode S19E03, "The City Part Of Town" features a few minutes of simple real-life video (stock footage of restaurant patrons.) As a result of these few minutes, the episode takes up 15-25% more diskspace than other episodes.

Middle out algorithm

[Anonymous Coward]

I bet they are using pied piper. I hear their new middle out algorithm is lossless and has the best compression ratio. Just don't give them write access to your video library....

x265 and save more than half with same+ quality

[Anonymous Coward]

use x265 and save more than half while getting the same or better quality

Re:x265 and save more than half with same+ quality

[Anonymous Coward]

There is a open source x265 decoder library out there, along with closed source binaries from different sources that works wonders even on old hardware for 1080p video.

I'm guessing this isn't what's bothering them, but royalties and patents.

Re:x265 and save more than half with same+ quality

[sanf780]

I fear they are stuck with H264 for the time being due to HW support. After all, Netflix can be installed on almost any video related appliance you can buy today, new and old! And that includes old BD players and Nintendo 3DS.

Four Years!

[ChrisMaple]

They've taken 4 years to figure out that parameters need to be optimized per film to produce optimum results for that film, and then to re-encode some films. The decisions for new parameters apparently aren't being made automatically, it's human choice. They're just beginning to consider that they may want to change parameters dynamically as a film progresses. This could and should all have been done in 2 months.

They haven't invented a new codec technology; they haven't advanced the state of the art at all.

This Breaking News Just In:

[Anonymous Coward]

Netflix admits to hating little girls and small children in general, forces them to watch shit-tier video quality for the same price

4 words to truly save bandwidth

[Anonymous Coward]

Lens flare compression algorithm ...For when Force Awakens hits Netflix

Re:Discriminative towards the little children?

[Anonymous Coward]

Stop sticking your politics into otherwise boring tech news. Gamergate punks fuck off.

Easy. Get Off Amazon.

[segedunum]

They'd be best served getting their platform off Amazon.

Obvious statement is obvious

[Anonymous Coward]

Does anyone in tech actually innovate anymore?

How's 4K streaming ever going to happen?

[L. J. Beauregard]

4K is not going to happen unless Comcast, TWC and the rest of the evil monopolies build pipes that can handle it, or they get some competition, which is not going to happen as long as crooks run the state and local governments.

Re:How's 4K streaming ever going to happen?

[Anonymous Coward]

Premium surcharges.

Re:How's 4K streaming ever going to happen?

[omnichad]

The pipes are plenty big. They just weren't paying for transit with the peers that had the content.

Re:How's 4K streaming ever going to happen?

[Anubis IV]

4K is not going to happen unless [...]

It seems a bit odd to suggest it can't happen when 4K streaming is already happening. The main selling point of Netflix's top streaming tier is that it grants you access to their "Ultra HD" (i.e. 4K) catalog. YouTube has had 4K support since late 2013 and has quietly been adding support for even higher resolutions in the last two years. You can already find content available at resolutions as high as 8K (e.g. this video). One of the production houses I follow on YouTube makes most of their animated content available at 4K and is moving more and more of their live action content over to 4K as they get more 4K cameras into the hands of their teams.

And the pipes are already good enough. Netflix's 4K content only requires a sustained 25 Mbps connection, which is orders of magnitude more common in US households than 4K TVs are.

Comcast, TWC, et al. are certainly deserving of a good raking over the coals, but when it comes to their ability to deliver 4K streaming capability...well, that bar is actually pretty low, and they've pretty much already met it. 25 Mbps plans are not particularly difficult to come by in most of the US, though I'm quite aware that some regions are woefully underserved and that prices for those plans remain unjustifiably high. For instance, the area where I live (Bryan/College Station, Texas, smack dab in the middle between Houston, Dallas, San Antonio, and Austin) is officially classified as a metropolitan area, but Internet plans here cost 34x that of peer cities not too long ago (that "34x" is sadly not a typo), simply because we don't have any adjacent urban centers.

Don't worry about 4k, HD or 100fps

[Anonymous Coward]

We won't be allowed to use that under the "Fair Use" policies the internet providers have for anything over 100 minutes a month.

Of course, it also means that the take-up of it will be abysmal, which will be blamed on "digital piracy" because P2P sharers can't afford a legal department, whilst ISPs can.

Vectors format?

[dlenmn]

How is modern animation done? If it's done on computers in some sort of vector drawing format then, unless the drawings are really complicated, it seems a little silly to convert the (lossless) vector drawings to a (lossy) video in the first place. Surely modern computers can render 2D vector drawings in real time? I assume that the vector information can be stored relatively compactly.

FWIW, it sounds like some of the animation for My little pony is done with Flash, and much as I hate flash, a flash animation would surely would take less bandwith then a normal video, right?

Re:Vectors format?

[ceoyoyo]

Sure, but it's awfully hard to protect. If they just spew the original vector drawings around the Internets, you could use them to make those ponies do whatever you want!

Re:Vectors format?

[dlenmn]

True, but it's not like regular videos are particularly protected on the internet either. I don't think that the polygons, gradients, etc. you'd be able to extract from a vector animation would be that useful for other animators; other animators can already copy the shapes and colors. What's valuable is the software and process that comes up with the shapes and their motion in the first place, and that's not part of the animation.

Re: Vectors format?

[Anonymous Coward]

Do it ms paint style. Black lines and paint
pot fill areas.

I wonder if that's plausible

Re:Vectors format?

[ceoyoyo]

3D animation uses source files that describe 3D objects and how they can move. I wouldn't be surprised if modern industrial 2D animation, such as you'd use for animated TV shows, uses a similar approach, just with 2D figures instead of 3D. The software to do the animation, at least for 3D, is pretty standard, and easy to come by. Some of the big ones are even open source. Sure you can make a from-scratch copy of Woody or Flutteryshy, but having the source files would make it a lot easier.

Re: Vectors format?

[Anonymous Coward]

I've wondered about that, if you could have a resolution-independent compression algorithm that stores color intensity maps as 2D shapes with gradient fills. Unfortunately I still have a pile of unread books on image processing, computer graphics and compression, so I have no idea if that's plausible or been tried.

It does kind of remind me of the old Apple II Graphics Magician program that stored "pictures" procedurally. Every time you loaded a file, the program recreated your drawing commands. The "shape tables" on the Apple II were really sets of drawing vectors, not bitmaps.

How about offering a 720p option?

[jetkust]

Instead of just giving the user the choice to watch at crap 480p quality or use ALL THEIR BANDWIDTH, how about a 720p option? The bandwidth jumps from .7 MB per hour with SD to 3 GB or more per hour for the next option. Why?

Re:How about offering a 720p option?

[danbob999]

Why not? 0.7 vs 3 is a small step. I'd rather have them push higher quality 1080p (as in 10GB per hour) instead.

Pinkie Pie is flatter than Emma Peel

[tepples]

The shaders in MLP:FIM are much simpler than those in The Avengers, no matter whether Emma "Black Widow" Peel is played by Scarlett Johansson or Uma Thurman. Flat-shaded animation has harder edges and less texture than photoreal-shaded animation and thus may need different compression techniques to improve the rate for a given level of distortion.

BitTorrent is the answer

[manu144x]

I don't want to sound like the devil's advocate, but why aren't they using BitTorrent technology to distribute the bandwidth problem a little. Something in the likes of Popcorn time. Each user will allocate a certain amount of hard drive space as a local cache and then use that to cache to distribute to others. Of course, it's not easy, you would need to have a very smart algorithm taking care of all this and distribute it along, but it should work much better than the standard Server => Client paradigm. Or am I missing something obvious? Piracy concerns? Copyright?

Re:BitTorrent is the answer

[omnichad]

My bandwidth is mine - don't start giving it away.

Re:BitTorrent is the answer

[godrik]

Of course, P2P technologies were built to solve exactly the bandwidth issues. My guess is that we would have NO bandwidth problems if we were distributing content with P2P technologies. Which means ISP could open the pipe to 100% for all clients all the time.

My guess is that legality is the problem. And once you tell people that the content is distributed peer to peer, they might decide that the netflix recurring payment is not that useful anymore.

What about sound compression?

[wwalker]

When are they going to fix their sound encoding?! I often have to switch to plain stereo from the default 5.1, because the higher-frequency is distorted which makes speech/dialog sound especially "tinny". Don't know if it's due to higher compression for 5.1 sound, or something else, but it annoys the hell out of me. I'm no audiophile either, so it is pretty apparent. Using PS3 hooked to a receiver, so it could be the PS3 client. Tried reporting this issue to Netflix, but there is no way to do it except by calling them.

Re:What about sound compression?

[Anonymous Coward]

I have the same issue. How do you switch to just stereo?

Re:What about sound compression?

[omnichad]

How is your PS3 hooked to the receiver? Is it playing true 5.1? Best test is to answer these two questions?

1) Is dialogue generally only coming from the center channel? It's supposed to be that way.
2) Are your rear speakers usually quiet except for special effects? It's supposed to be that way.
3) Does your receiver display a Dolby logo when audio is playing?

If you get any other result, then either your receiver's settings or your method of connection are wrong.

Poor quality, already

[DogDude]

Netflix already offer really, really poor quality video. I can't believe they're going to make it *worse*. People who do nothing but stream somehow don't notice. I actually get DVD's and rip them, and watch them that way, and the difference is really night and day. All Netflix stuff looks super grainy, and super dark. It's interesting to me that so many people just don't care. As an aside, it makes it tough for those of us who like to be able to *see* and *hear* our entertainment because it's getting so hard to get access to good quality stuff.

Shouldn't get their bandwidth for free? Hosting

[raymorris]

Every other web site in the world, besides Netflix, pays for their bandwidth, but shouldn't Netflix get theirs "free" (paid for by the rest of us) since they talk shit about the cable companies? I thought that was the general consensus among Slashdot readers, that Netflix is special and shouldn't have to pay for their own bandwidth costs like every other site in the world does.

Re:Shouldn't get their bandwidth for free? Hosting

[dave420]

It's fine you don't understand. There's nothing to be ashamed of. Just try harder in the future - you'll be better off for it, and able to participate in these discussions without choking on your feet.

My two hours will look fabulous

[paiute]

I look forward to watching my six hours of Netflix every month. Good thing that Comcast lets me watch as many of their movies as I want on Hexfinicky.

HEVC or 2 Pass ... or both.

[bobsta22]

So cant really go full HEVC, a lot of embedded devices in old TVs etc cant cope. But if it can take HEVC then do it for 60% for the same visual quality, so they need to start creating HEVC encoded versions as well as better H264. Then comes the 1 pass with 'constant quality' versus 2 pass with 'constant bitrate' (typically). If you use the 1st pass to work out where to apply savings, and encode on the 2nd pass to work at a variable bitrate that lowers space when it can, then its more than feasible to nail down 20%.

2 million hours at moderate compression

[peter303]

Im guessing. Maybe a million titles.

30 bit vs 24 bit

[Anonymous Coward]

Except 24bit color covers 16M colors, of which a normal human eye is only capable of distinguishing 10M. There's no reason to dither gradients in full 24bit color (if you're reducing the color space for compression purposes it's a different game).

Re:30 bit vs 24 bit

[BatGnat]

A black #000000 to red #FF0000 gradient, only has 256 shades, I can see the bands. Dithering or extra bits makes a big difference, depending on the circumstance.

Re:30 bit vs 24 bit

[Junta]

But if you have a pure red, green, or blue, then you have a very limited range for that color. This is why for photo realistic, it's no big deal, in nature you don't really see concentrations of just a primary color, and not usually in a gradient . It's when an artist sits down to create the content and chooses much more simplistic colors and patterns, which interestingly enough causes a unique problem for the medium when only 256 levels are available to represent any particular primary.

Now dithering up to fuzz away the bands in a gradient is pretty much imperceptible, but the compressor sees a more 'noisy' source than if the band width is reduced by a factor of 4. Of course ideally a compression algorithm simplifies the model the same way at a certain compression level, but at best it's considering the dithering to be detail that can be discarded if necessary rather than that specific pattern of detail not mattering one bit.

Netflix: DON'T COMPRESS BLACK!!!

[Thing 1]

Your dark scenes suck. Fix them with this update as well!

Re:Netflix: DON'T COMPRESS BLACK!!!

[omnichad]

Have you calibrated your TV with a test pattern and a blue filter? Don't blame Netflix.

Re:Netflix: DON'T COMPRESS BLACK!!!

[Thing 1]

No, that's not it at all. I've watched on TV, on computer monitor, and on laptop. It's not only Netflix; most existing compression decides to compress black so much that shadows end up all jagged, and the details that were present on the original source material are missing in the compressed version.

You're accurate, though, in assessing that I haven't calibrated using a test pattern and a blue filter. Any of my screens. That's not the issue though. I do blame the compression codec.

This means 1PB worth of download on every cache...

[Anonymous Coward]

I don't know if you guys are aware of how netflix works without destroying the net, but they have these cool 5U boxes full of 1-3TB HDDs which are hosted by every internet provider that can justify one to netflix (you need to have a lot of netflix traffic going through your AS). The boxes need ~4H worth of resync every night @1Gbps...

Refreshing these boxes for the re-encode is going to be quite annoying on everyone, although I assume they will spread it over the entire year...

netflix goes the DirecTV route

[roc97007]

ok, so, if I'm understanding this right, Netflix plans to keep the more profitable titles reasonably compressed, and the less profitable titles will have the living crap compressed out of them. Oh, they'll say you will hardly notice, but you know they'll do it. There's too much money at stake.

Just sayin', this is why we gave up on DirecTV. Stuff like football games (which I don't watch) were relatively uncompressed and very sharp. Stuff like the Disney channel and Nick, the reception of which was the primary use of the equipment, (having a young child and not watching much tv myself) were compressed to big splotches of color. It was so bad that even my grade school kid noticed that the video quality was crap. In the early thousands, we dumped DirecTV and switched to Netflix. Now some 12 years later, daughter still watches Netflix. Hopefully the stuff she watches won't be unpopular enough to be compressed to incomprehensibility. I guess the advantage is that Netflix doesn't do live sports, which was apparently the largest per-program usage of available bandwidth on DirecTV.

Broadcast Industry Engineer here

[Anonymous Coward]

We do this sort of mass retranscode of media especially generating proxy media in h264 for broadcast archives.
BBC, Sky... 1 petabyte is nothing when you consider that we typically deal with transcoding around at least three dozen HD streams coming in at 1 gigabyte per minute uncompressed.

So just excuse me while I yawn hard. Maybe I can slashdot my work now... suppose it's not as sexy as "omg netflix".

Proofed for Quality Control

[Anonymous Coward]

I wonder how much will be watched for quality control of the encode before released on the subscribers.

The terrorists won

[tepples]

You could move to Kansas City.

That's easier said than done now that ISIL has been successful in sowing anti-immigrant sentiment among U.S. politicians.

Exodus

[tepples]

Perhaps the only way for Seattle to get the hint is if people polish their respective resumes and then move out of Seattle in masses. Then Seattle would have to notice that, as Leia Organa might put it, the city has tightened its grip on the rights of way so much that jobs and property value have slipped through its fingers.

Scooter my daisyheads

[tepples]

If it is a capacity problem they can simply cap the speeds during commercial hours.

It appears that only satellite ISPs do this nowadays. Cellular carriers used to offer "free nights and weekends" back when voice airtime was considered expensive and Sprint was running the "Scooter My Daisyheads" commercial. But in the end, demand-based billing turned out to be too complicated for non-technical subscribers to understand, and subscribers ended up choosing plans with uniform billing because they were simpler.

Desktop in a library?

[tepples]

Go to your public library, install the "Video DownloadHelper" add-on in Firefox, and download from Youtube in glorious 1080p to your hearts content.

Which requires buying a laptop in addition to the desktop computer and mobile devices that you may already own. I don't think extensions like that are available for an iPhone or iPad. And good luck getting your library to let you bring in a desktop computer, even if you can fit it on your bike.

Comment removed

[account_deleted]

Comment removed based on user account deletion

Comment removed

[account_deleted]

Comment removed based on user account deletion

ZFS

[Anonymous Coward]

I've heard from several sources that Netflix stores everything on ZFS. Can anyone confirm this, and fill in the details? If ZFS is good enough for Netflix video catalog, it is good enough for me.

New Codec

[pebear]

They have stolen Pied Piper that's what they did. Don't let them tell you anything else!!!