[HN Gopher] Decoding Netflix's AV1 Streams
___________________________________________________________________
Decoding Netflix's AV1 Streams
Author : singhkays
Score : 67 points
Date : 2025-10-01 18:36 UTC (4 hours ago)
(HTM) web link (singhkays.com)
(TXT) w3m dump (singhkays.com)
| CharlesW wrote:
| Fascinating, thank you for this analysis! Currently pining for
| release of an updated Apple TV, which will have an SoC capable of
| hardware AV1 decode.
|
| It'd be great to hear from someone at Netflix about the
| unexpected Bojack Horseman results. I'd bet that Netflix just
| isn't yet taking advantage of AV1 features designed especially
| for this kind of animation and synthetic content.
| alfalfasprout wrote:
| What benefit would a new appleTV have other than reduced
| bandwidth usage?
| adzm wrote:
| The Bojack example shows that it's not just reduced
| bandwidth; in some cases you can get higher quality with
| their AV1 encodes. Additionally if you are thinking of
| average bitrate, that's ignoring variable bitrates (and the
| extension, the per-shot encoding params that Netflix
| utilizes.) That is to say, high complexity scenes get more
| bitrate than low complexity, so very noisy scenes can look
| way better using AV1 even while still using lower bitrates,
| but notably some peaks have similar max bitrates.
| toast0 wrote:
| The promise of new codecs is reduced bandwidth _and_ higher
| quality. Probably a new device also has a faster processor
| and /or more ram in general, which helps with incidental
| jank. Apple TVs are well regarded and maybe have less jank
| than other products (I don't have personal experience, I'm
| interested, but my Apple computer is a IIe so I expect
| account management issues), but it's nice to get a new Roku
| every once in a while as bloat/jank seems to creep up on
| them.
| CharlesW wrote:
| The credible rumors beyond AV1 decode include: Wi-Fi 7 via
| Apple's home-grown N1 chip, a CPU fast enough to support the
| next-gen Siri release on-device, a RAM bump, improved pass-
| through for high-end audio formats, potentially a camera (the
| new square selfie sensor would be perfect for this) for
| easier group/family FaceTiming, and more aggressive pricing.
| isatty wrote:
| The appletv is such a good device that I'm paying it
| whatever the price may be.
|
| TVs have horrible UIs and are generally ad ridden garbage.
| Not using anything android based because of the same reason
| and slow.
| zamadatix wrote:
| I wouldn't mind if they enabled the 120 Hz support with the
| new chipset. I like my TV but the framerate matching feature
| causes a few seconds of black screen each time I switch
| between and also drops the UI to 24 FPS. Would be nice if
| everything just ran at 120 FPS all the time, and since
| 24/30/60 are perfect frame multiples that's nice to. NTSTC
| content at 23.976 I'd hope the player would just speed up at
| that point, but even if not... judder at 120 Hz is better
| than at 60 Hz.
|
| Also 6 GHz Wi-Fi would be nice. I had to run a cable to my 2
| because the 5 GHz airspace where I am is too crowded to
| stream high quality movies via Infuse without occasional
| hitching. Same with the seek speed. Meanwhile my iPhone gets
| 2.9 Gbps of goodput at solid jitter on 6 GHz Wi-Fi.
|
| There's probably some updates to the HDR standards. For me at
| least though the current one already supports what my TV
| does.
|
| Also apps seem to assume "because hardware decode isn't
| available don't serve AV1" sometimes. As silly as that is
| with the CPU power in the AppleTV, at least that problem
| would go away with hardware support and they'd stop trying to
| serve a "compatible" SDR h.264 stream. Despite internet
| pessimism, sometimes the quality is also raised with more
| efficient codecs rather than just "the same quality at less
| bandwidth".
| CharlesW wrote:
| > _Also apps seem to assume "because hardware decode isn't
| available don't serve AV1" sometimes._
|
| This isn't a completely unreasonable decision, since the
| current 2022 model's software AV1 decode apparently can
| only sustain 4K AV1 decode (although it handled 1080p
| content fine in my test) for as little as 45 minutes before
| thermal throttling kicks in.
| pkroll wrote:
| "NTSTC content at 23.976 I'd hope the player would just
| speed up at that point, but even if not... judder at 120 Hz
| is better than at 60 Hz."
|
| I'd bet money when TVs are advertised at 120 FPS, they're
| really 119.88 FPS, so no judder showing 23.976 FPS and the
| other NTSC-off display rates.
| zamadatix wrote:
| Some content is truly exactly 24 FPS or 30 FPS though, so
| whichever path the TV goes (i.e. NTSC rate or integer
| rate) the same problem will exist. I suppose some TVs
| might have extremely fancy film mode detection which
| catches the occasional frame difference, but I doubt mine
| does :D
| mdasen wrote:
| As someone who tried Roku, Android TV, and Fire TV before
| switching, better hardware offers a vastly better experience.
| The Apple TV's hardware is fast. The UI doesn't lag. Things
| feel smooth as butter. Yes, maybe the Apple TV doesn't need
| more, but more can be helpful.
|
| In terms of AV1 support, YouTube often only does 4K with AV1
| so that's an issue for people.
|
| Personally, I'd love to see an Apple TV that was great for
| gaming. New Apple processors have hardware ray tracing and
| decent gaming performance.
|
| I think it's also likely that Apple will try and make an
| Apple TV that will support next-gen Siri and on-device AI
| stuff. Yes, you can complain about Apple's AI delays, but
| Apple's probably looking toward an Apple TV that can support
| their AI models.
|
| In some ways, "what benefit would a new <insert-thing> have?"
| Sometimes we don't know until we have it and people start
| using it.
| pnw wrote:
| Apple just hasn't been able to get traction with gaming on
| Apple TV. Gaming on Apple TV is so small I couldn't even
| find an analyst report breaking down the market size.
|
| I don't think new graphics hardware solves the problem.
| Beyond the friction of the unit not shipping with a
| controller, tvOS lacks good discovery for games and there
| is no ad infrastructure comparable to mobile. Most game
| developers aren't looking to invest in small, closed
| platforms with bad discovery. It's hard enough to make
| money on Apple's mobile platforms.
| adzm wrote:
| > the unexpected Bojack Horseman results. I'd bet that Netflix
| just isn't yet taking advantage of AV1 features designed
| especially for this kind of animation and synthetic content.
|
| While the percentages look scary, it's only a slight difference
| (60kbps!) and still around 1mbps average, but with a
| significant quality boost (very crisp lines and near perfect
| quality). I bet Netflix could encode at nearly half that
| bitrate and stay similar to HEVC in quality, but I'm pleased
| they seem to have made a good tradeoff here.
|
| It's actually quite amazing the quality that AV1 delivers at
| such low bitrates across the board. I've said it before, but
| AV1 is almost magical. Which I think is behind the lack of
| enthusiasm for VVC/h266; is anyone even using that? I've yet to
| actually see it in the wild.
| CharlesW wrote:
| Ditto. Between AV1's momentum and the imminent release of
| AV2, it sure feels like VVC will find limited adoption
| outside of broadcast applications.
|
| https://aomedia.org/press%20releases/AOMedia-Announces-
| Year-...
| bcraven wrote:
| Some of the torrent releases out of MVGroup are done as h266.
|
| I think as technical tests as much as anything, but they're
| interesting to see.
| jpalawaga wrote:
| I feel like this was copy edited by ChatGPT and it really grates
| me. I couldn't help but lose focus after I started seeing
| telltale signs of AI.
|
| While the topic matter is interesting, I feel like obviously
| synthetic content falls into the "that which was not worth
| writing, is not worth reading either" trap.
|
| If the authors tone is extremely ChatGPT-esque, I apologize in
| advance.
| janice1999 wrote:
| It's the emoji bullet points and headers that makes me
| instantly close a page.
| hnuser123456 wrote:
| "It's not just X, it's Y!"
| __float wrote:
| Heh, didn't have to go far in this article to find this
| exact construction:
|
| > The data shows it's not just an incremental improvement;
| it's a demolition.
|
| Complete with extra bold to emphasize the second half,
| sigh.
| bashtoni wrote:
| Came to say pretty much the same thing. This slop is unreadable
| for me at this point.
|
| I keep getting a paragraph or two into something, read one of
| the terrible "It's not just word - it's massive hyperbole!"
| sentences, see that there are several more in subsequent
| paragraphs and can't continue.
|
| However bad the author's original writing that generate this
| output was, it can't be as awful as this.
| pnw wrote:
| The "it's not this - it's that!" phrasing is everywhere now
| and is driving me insane.
| binaryturtle wrote:
| I agree that this text in its current style is very hard to
| read. Feels like the text was ballooned up to 3 or 4 times its
| original length with pointless "side content"? Lots of
| distracting noise basically. AI or not AI, this is not very
| good.
|
| ... and so I'll continue to stick with AVC, thanks! :-)
| galaxy_gas wrote:
| I would 100000% rather read the author's own writing even if
| English is their 10th language
|
| Rather than this inflated slop that look like I am trying to
| reach word count in a paper and one sentence becomes 15 useless
| ones
|
| Edit: This is not so much commentary on AI than it is the core
| of your post is a few tables. Just post the tables and one or
| two sentence of conclusion and that is all ! It is so tedious
| to read through dozens of paragraph of autogenerated
| _unnecessary_ nonsense -- that contribute nothing of value to
| the data
| re wrote:
| It's unfortunate because there's plenty of good, technical,
| perfectly readable content in the author's blog archive from
| 2021 and earlier without the overwrought purple prose.
|
| https://singhkays.com/archives/
| galaxy_gas wrote:
| I wonder why there is 4 year gap there. Their English is
| not even bad ! Its better than mine and definitely does not
| need any kind of GPT copyediting
| jajuuka wrote:
| I definitely appreciate the concise nature of the article.
| Doesn't waste my time at all. Whether that's due to the
| authors writing ability or some summery tool, I could care
| less.
| z0r wrote:
| This didn't trip my AI detector; I instinctively skimmed to
| look at the numbers and conclusions. Your comment made me go
| back up to the top and read the opening paragraphs and I see
| what you are saying. It is always painful to realize you are
| reading AI product. I think it is less of a problem with this
| blog post because it is just presenting a handful of tables of
| numbers and a few graphs, but it seems I am already
| unconsciously training myself to ignore florid AI writing.
| alwa wrote:
| Intolerably ChatGPT-esque. Which is a shame, it seems like a
| nifty little DIY experiment.
|
| I think what stands out to me is this cartoonishly punchy,
| faux-dramatic framing.
|
| That, and specialist terms that seem to be thrown in there in
| an empty way, just to signal subject-matter expertise that's
| not even expected of a DIYer's experiment report:
|
| > _It's a multi-decade, billion-dollar street fight over bytes
| and pixels, waged in the esoteric battlegrounds of DCT blocks
| and entropy coding_
| delta_p_delta_x wrote:
| LLMs tend to produce ridiculous similes, idioms, euphemisms,
| and analogues that fall apart upon the slightest scrutiny, as
| though they are genuine English constructions. I despise it.
| No one in reality writes like that. Who the hell thinks
| 'bytes and pixels' are a battleground? if you want to talk
| about licensing problems, _just say they are licensing
| problems_ , don't accept shitty similes from LLMs that
| humanise decidedly non-human constructs. More so if it's a
| technical discussion: just get to the bloody point.
|
| I like using real idioms that have percolated through culture
| ('birds of a feather', 'white elephant', 'nip in the bud',
| etc), not stupid contrivations.
|
| As someone who sweated through hours and hours of English
| essay-writing in school, LLM output that is misrepresented as
| genuine human writing is annoying and highly disrespectful of
| the reader's time and effort. The moment I saw the stupid,
| contrived headers and dozens of emojis, I closed the tab.
|
| I refuse to waste my time reading the output of a matrix
| multiplication done in some server farm when I could do the
| latter myself.
| vanviegen wrote:
| Well, journalists often write like that. I presume that's
| where the LLMs learned. I always find that style annoying,
| but especially so in a technical post like this.
| wonnage wrote:
| Probably the only overlap between AI slop and actual
| journalistic writing is the obsession with em-dashes
| porridgeraisin wrote:
| And hallucinations.
| Wowfunhappy wrote:
| I don't know which publications you're reading, but the
| ones I read do _not_ write like this!
| landl0rd wrote:
| > article about compression
|
| > uses slopbot 9000 to explode his point into ten times the
| "prose"
|
| > mfw
| delta_p_delta_x wrote:
| I laughed. Now only if this was a real greentext.
| adadtttt wrote:
| The adverts on this website are very annoying
| Spare_account wrote:
| Why don't you use an adblocker?
| keane wrote:
| Was once out in a remote area on an 800 kbps DSL connection.
| YouTube couldn't stream, Prime Video couldn't stream. Netflix
| worked fine. Years later, I remain impressed at their uniqueness.
| newman314 wrote:
| Does anyone know what was used to produce the graphs?
| nirewen wrote:
| Inspecting the page, I can see some classes "dw-chart" so I
| looked it up and got to this:
| https://www.datawrapper.de/charts. Looks a bit different on the
| page, but I think that's it.
| input_sh wrote:
| Do you mean charts? If so, it's Datawrapper:
| https://www.datawrapper.de/charts
|
| One of the quite expensive paid plans, as the free one has to
| have "Created with Datawrapper" attribution at the bottom. I
| would guess they've vibe-coded their way to a premium version
| without paying, as the alternative is definitely outside
| individual people's budgets (>$500/month).
| encom wrote:
| Okay, so AV1 has lower bitrate. I can encode any video format at
| arbitrary bitrates, but that metric is not useful on its own. An
| article about _how_ AV1 requires less bits for the same or
| improved perceptual quality would have been far more interesting.
| mijkal wrote:
| This article is a good primer on how AV1 works (and compares
| with VP8 and VP9): https://www.red5.net/blog/av1-vs-vp9-vs-
| vp8-comparison-for-l...
| shmerl wrote:
| _> Device Support: Hardware decoding for AV1 isn't on every
| device yet._
|
| By now - it should be in most devices that's aren't outdated by
| even average standards. And it's worth mentioning that for
| devices that don't have hardware decoding, dav1d does an
| excellent job of decoding it on the CPU.
|
| The problem is more with hardware encoding. That's indeed only
| present in only recent generations (or a couple) of hardware and
| even with that, AMD for example have an aspect ratio limitation
| bug in their AV1 hardware encoder (which requires adding black
| bands to work around) that's only fixed in RDNA 4 which is not
| available in their APUs, so it won't be fixed in APUs until their
| UDNA is used for them (they didn't fix it in RDNA 3.5 chips).
| kllrnohj wrote:
| Interesting charts, but this is all completely meaningless
| without image quality comparisons. I can easily use 50% less
| bandwidth than Netflix's H264 streams as well, with H264 even, by
| just cranking up the compression & dropping the bitrate.
|
| Presumably nothing jumped out at the author as being worse, but
| come on how can you have a whole section on why AV1's regression
| on Bojack is actually a good thing because the quality is way
| higher, and then _not show any quality comparisons_?
| galaxy_gas wrote:
| Taking screenshot of netflix on my device results in black
| square, I dont know if this applies on lower levels of Widevine
| but if it doesn't then the quality will be much lower as
| Netflix do not serve 720+ video unless there is protected DRM
| path
| kllrnohj wrote:
| It'd have to be the good ol' "take a picture of the screen"
| route without getting into the more gray areas.
| mattkrick wrote:
| The ads made this unreadable. Is this a thing? Get to hackernews
| front page & then inject your post with as many ads as possible
| to "cash in"?
| Spare_account wrote:
| Why aren't you using an adblocker?
| ElijahLynn wrote:
| Very good read, love some of the humor in this article. It helped
| me get through to the end!!
|
| Also, If anyone was wondering where AV1 stands in comparison to
| VP8 and VP9... I just looked it up after a few years of not
| paying attention and I guess Google donated VP8 and VP9 to the
| alliance for open media foundation (AOMedia) in 2015 and they
| created AV1 and released it in 2018.
| MrRadar wrote:
| Yeah, AV1 is primarily based on what Google was working on for
| their own successor to VP9, what would have been VP10, with
| technology contributions from Mozilla/Xiph's Daala and Cisco's
| Thor codecs.
| dilyevsky wrote:
| Looks like latest chatgpt model is not aware of massive battery
| draining on mobile
| 1oooqooq wrote:
| the worst screen tearing in old analog movies with camera
| panning.
___________________________________________________________________
(page generated 2025-10-01 23:02 UTC)