| ▲ | H.264 Streaming Fees: What Changed, Who's Affected, and What It Means(streamingmedia.com) |
| 73 points by phantomathkg 15 hours ago | 41 comments |
| |
|
| ▲ | breve 15 hours ago | parent | next [-] |
| There's no point supporting these parasitic business models. Use royalty-free video and audio formats. AV1 for video: https://aomedia.org/specifications/av1/ And Opus for audio: https://opus-codec.org/ |
| |
| ▲ | josephg 14 hours ago | parent | next [-] | | AV1 / Opus where we can. But H264 is far better supported. For example, Safari doesn't include a software AV1 decoder. So AV1 videos only work in safari on M3 or later laptops, and iPhone 15 or later phones. H264 is the compatibility king. https://caniuse.com/av1 | | |
| ▲ | galad87 14 hours ago | parent [-] | | Safari, or better, macOS and iOS include a software AV1 decoder (libdav1d), but it's used only to decode avif, and to generate file previews in Finder. |
| |
| ▲ | WithinReason 14 hours ago | parent | prev | next [-] | | Dolby just sued Snapchat over patents for using AV1: https://www.techspot.com/news/111865-dolby-sues-snap-over-vi... | | |
| ▲ | gausswho 14 hours ago | parent [-] | | And that's when you know they're afraid. Dolby's case sounds like a desperate moonshot. |
| |
| ▲ | out_of_protocol 12 hours ago | parent | prev | next [-] | | * VP9 where AV1 is not available (default YouTube codec, almost universally hardware-supported). Also universally supported .webm is vp9+opus - which mostly used as modern .gif | |
| ▲ | ksec 11 hours ago | parent | prev | next [-] | | >And Opus for audio: https://opus-codec.org/ Just use AAC-LC, Redhat declares it patent expired and has the widest compatibility besides only mp3. At 192Kbps or above it is as good as Opus. | |
| ▲ | mdavid626 14 hours ago | parent | prev | next [-] | | AV1 lacks hw support… | | |
| ▲ | out_of_protocol 12 hours ago | parent | next [-] | | VP9 works well too and more supported (default YouTube codec) | |
| ▲ | breve 14 hours ago | parent | prev | next [-] | | My laptop has hardware AV1 encoding and decoding. My TV has AV1 decoding. My phone has AV1 decoding. None of these devices are particularly new. And don't underestimate dav1d (https://www.videolan.org/projects/dav1d.html). You can comfortably play AV1 video in software on your phone. Try it with VLC. | | |
| ▲ | petcat 14 hours ago | parent [-] | | > You can comfortably play AV1 video in software on your phone Maybe for about 15 minutes before your battery is drained to 20%. I'm not aware of any software video decoder at all that won't unacceptably heat up your phone and kill your battery. | | |
| ▲ | daneel_w 14 hours ago | parent | next [-] | | You really are underestimating just how far e.g. Apple's mobile CPUs have come in terms of raw performance and power-efficiency. | |
| ▲ | breve 14 hours ago | parent | prev [-] | | Why say maybe? Why not simply try it for yourself? | | |
| ▲ | petcat 14 hours ago | parent [-] | | I don't need to try it myself to know that software video decoding on the CPU is not a viable solution on mobile phones. | | |
| ▲ | breve 13 hours ago | parent [-] | | Of course it is. Even the iPhone 7 from 2016 can play 1080p AV1 video. Why did you spend all that money on your phone if you're not going to exercise the hardware? | | |
| ▲ | smilespray 12 hours ago | parent [-] | | Not OP, but 2hr battery time for video playback vs 20hr would be the entire point of this thread. HW decoding is an order of magnitude more efficient. | | |
| ▲ | breve 6 hours ago | parent [-] | | That isn't the benchmark set by petcat. petcat's goal post is 15 minutes to 20% battery with AV1 playback. It's baffling that petcat won't simply try it. Maybe he has awkwardly discovered he does have hardware AV1 support after all. I've still got my old iPhone 7. I'll dust it off and do the experiment. I think it'll do at least 90 minutes in VLC. | | |
| ▲ | petcat 6 hours ago | parent [-] | | Software video decoding on a CPU is woefully inefficient and will drain your battery. Full stop. > I've still got my old iPhone 7. I'll dust it off and do the experiment. I think it'll do at least 90 minutes in VLC. Your "old iPhone 7" probably wont even boot up now, let alone play 1080p AV1 video for more than 5 minutes. Go ahead, try it. | | |
| ▲ | breve 3 hours ago | parent [-] | | > Your "old iPhone 7" probably wont even boot up now, let alone play 1080p AV1 video for more than 5 minutes. None of your predictions came true. In fact, you were more than 24 times wrong about it. My 10-year-old iPhone 7 with its 10-year-old battery and a small crack in the screen, hardware which was released before AV1 was released, did boot up. I charged it to 99%. I downloaded a 4 minute music video. It's 1080p25 1609kbps AV1 video, 48khz 122kbps stereo Opus audio in an MP4 container. Using VLC 3.7.2 (which uses dav1d for decoding AV1), I played the video continuously on repeat. It took 122 minutes for the battery to go from 99% to 20%. At 20% the phone switched to low power mode and kept playing the video. I should have put it in low power mode the whole time. I'll try that next to see if it can go longer. In the meantime, what we can conclude is that the iPhone 7 is mighty. dav1d, most especially, is mighty. petcat isn't. |
|
|
|
|
|
|
|
| |
| ▲ | TheMiddleMan 14 hours ago | parent | prev [-] | | It's coming along nicely https://en.wikipedia.org/wiki/AV1#Hardware_encoding_and_deco... Also decoding on a reasonably powerful (non-accelerated) cpu is fast enough for 1080p, not ideal for battery life but still. |
| |
| ▲ | cwillu 15 hours ago | parent | prev | next [-] | | “Access Advance and Avanci have published rates for a pool asserting content royalties across AVC, HEVC, VP9, VVC, and AV1 that could push major platforms toward nine-figure annual exposure.” | | |
| ▲ | breve 15 hours ago | parent [-] | | Yes, they've made claims on AV1, claims that have never been tested in court. You need to understand that these are parasitic businesses. They didn't develop AV1. They didn't contribute to AV1. But they will make any claim they think they can get away with. Show me the court case they've won that validates their claims on AV1. | | |
| |
| ▲ | jmclnx 15 hours ago | parent | prev [-] | | I was going to suggest you missed vorbis ogg. So I went looking for a link, I found out this: > Since 2013, the Xiph.Org Foundation has stated that the use of Vorbis should be deprecated in favor of the Opus codec I never heard of Opus, so some links: https://en.wikipedia.org/wiki/Vorbis https://en.wikipedia.org/wiki/Opus_(audio_format) From what I can find, seems opus only supports audio. ogg also has a video format (ogv), odd it is suggested ogg was superseded by opus. Maybe I am missing something ? | | |
| ▲ | breve 15 hours ago | parent | next [-] | | Ogg is a container format. It contains audio and video tracks: https://www.xiph.org/ogg/ It's like Matroska: https://www.matroska.org/what_is_matroska.html Or MP4: https://en.wikipedia.org/wiki/MP4_file_format | |
| ▲ | daneel_w 14 hours ago | parent | prev [-] | | Ogg is the container, Vorbis is the audio codec, and colloquially people just called Vorbis-encoded audio "ogg" because of the ogg container. Vorbis was hit-or-miss. In some cases it did better on same or lower bitrate than MP3 encoded by LAME, in some cases worse. It also suffered an entirely new category of "chirpy/tweety" artefacts similar to what MP3 exhibits at very low bitrates, but with Vorbis they showed up even at nominal bitrates during certain complex spectral patterns. I was a vocal proponent of Vorbis back when it surfaced, but soon changed stance when realizing how unreliable it was quality-wise. | | |
| ▲ | thaumasiotes 14 hours ago | parent [-] | | > and colloquially people just called Vorbis-encoded audio "ogg" because of the ogg container. I would bet that the primary reason wasn't the container format, which nobody really cares about and most users wouldn't have been aware of, but rather the fact that the file extension was '.ogg'. | | |
|
|
|
|
| ▲ | uyzstvqs 14 hours ago | parent | prev | next [-] |
| Relevant: Have the patents for H.264 MPEG-4 AVC expired yet? https://meta.wikimedia.org/wiki/Have_the_patents_for_H.264_M... |
| |
| ▲ | ksec 11 hours ago | parent [-] | | And this is High Profile, Only a dozen left. Most of them expire this year as well. With only 5 left in 2027 to 2030, 3 of that is SVC specific and country specific. I hope Redhat do something similar work to AAC-LC and declares certain profile patent free. |
|
|
| ▲ | canpan 14 hours ago | parent | prev | next [-] |
| Patent free video is in a strange space. I recently looked into just using old formats. To be super safe. With audio we have mp3 which is good enough for a lot of cases and seems to be patent free.
But for mpeg2 (used on DVD) even though it is really old (1995?) it is still patent encumbered in some places until 2035?
https://en.wikipedia.org/wiki/MPEG-2 |
| |
| ▲ | silotis 13 hours ago | parent [-] | | MPEG plays a clever game with their standards. A standard like MPEG-2 or MPEG-4 Part 10 (aka H.264) doesn't just refer to one standard but rather a whole series of standards published over an extended period. Patents are pooled by standard with deliberate ambiguity over which parts of the standard each patent actually covers. This lets patent holders spread FUD over whether earlier parts of the standard are actually patent free even after 20 years have passed since the original publishing. In the face of patent holders threatening a costly legal battle, companies choose to continue paying licensing fees even on standards which plainly should be out of patent protection. |
|
|
| ▲ | ZeroGravitas 14 hours ago | parent | prev | next [-] |
| This author has worked on behalf of the patent parasite companies before so if you are left feeling "I need to pay these people" that may not be accurate. |
|
| ▲ | cowpig 13 hours ago | parent | prev | next [-] |
| A sure sign that this is completely, utterly, decoupled from market forces is the way that a 5M-person streaming platform (which likely has negative margins or is struggling to make profit) is expected to pay $0.45/user, while Netflix is expected to pay 1.5 cents per user. I would guess that this is because the larger the number for major players, the more incentive they would have to invest in supporting open standards (or try and get a standard of their own). This is evidence that the patent system is not doing what it's supposed to be doing imo. |
| |
| ▲ | fc417fc802 13 hours ago | parent [-] | | I sometimes wonder what the consequences would be of stipulating that patents had to be uniformly licensed to all interested parties without exception. Also that you couldn't decline to offer a license - you had to set reasonable terms and then accept all takers unconditionally. I realize that's never going to happen and would probably have lots of unintended consequences. It's just a thought experiment after all. I find it interesting to think about because R&D can still be recouped under such a model which ... is kind of the entire (supposed) point of the system. Allegedly. If nothing else we would presumably have seen mass market epaper and 3D printers much sooner. | | |
| ▲ | jburgess777 12 hours ago | parent | next [-] | | It is commonly known as ‘Fair, reasonable, and non-discriminatory (FRAND)’. Some standards organisations will only accept contributions where a patent owner agrees to license them under FRAND terms. | |
| ▲ | verall 11 hours ago | parent | prev [-] | | > I sometimes wonder what the consequences would be of stipulating that patents had to be uniformly licensed to all interested parties without exception. Isn't this the definition of FRAND which nearly the entire interview with the lawyer from the article is about? | | |
| ▲ | fc417fc802 10 hours ago | parent [-] | | Yes but I wonder what the result of applying something substantially similar to that to the entire patent system without exception might be. Also I had in mind "uniform" in addition to "reasonable", ie holders would be permitted only a single set of license terms at a single rate with zero difference regardless of size, volume, etc. Maybe being permitted to change the terms on offer at the end of every 5 year period or so. Or something vaguely like that anyway. It's really just a thought experiment about how you might kill patent trolls as well as the asymmetric advantages that large corporations enjoy while still maintaining the spirit of funding the R&D done by participants of all sizes. |
|
|
|
|
| ▲ | Hizonner 12 hours ago | parent | prev [-] |
| Aw, man, not this shit again. |
