A New H.265 Patent Pool Is Causing Concerns

[OneTimeShot]

If you're only watching 1080 you're fine with h264, but 4K has four times the pixels, and more than 4 times the data (https://support.google.com/youtube/answer/1722171?hl=en).
Also, with higher compression you can let your network sleep longer (not a huge deal for a laptop but very important for phones/tablets).
I'd imagine there're other benefits as well.

All true - although h265 largely handles 4k by knowing that the high dpi can handle more blurriness than 1080p rather than any particular compression magic - and there will always newer codecs with better algorithms... but then there is the "specification junkie" aspect of video compression. At what point is that extra 5% just not worth it for the paying punters? Other people have commented on jpeg vs jpeg2000 as an example. IMO VP8/9 is very close to "good enough" if it can develop hardware support.

[SystemCrasher]

HEVC Advance, is interested in 0.5% of gross revenue

Okay, they can count on me... I mean they can count on me doing all my future video encodings in VP9 instead. Are they covert "VP9 promotion group", btw? Because they clearly would put VP9 to advantage

[gbcox]

I'm having very similar issues. Encoding speed at good quality. It's the reason why VP9 is not an option so far fo me. Otherwise if they can speed it up, it would work fine for me. It's just unrealistically slow at this point. But I think it's a design failure. I don't think it can be sped up without fixed function hardware to do it. And maybe what they were planning for was widespread fixed function hardware, but that obviously didn't happen.

Hopefully, it wasn't a design failure, but in any event .5% is a powerful incentive to get it fixed. One major problem for VP8 and VP9 up to this point is most people could care less if they aren't paying for it - so they just go with what is perceived as more popular; but if you add cost to that calculation, that changes everything.

[Gusar]

I doubt the encoding speed of vp9 is about design. It's more about Google's priorities. They don't need to care about properly multi-threading the encoder because they're constantly encoding dozens of videos in parallel. This is actually a nice demonstration of what I mentioned in a previous post - vp9 is Google's thing, it is entirely under their control, development is prioritized according to their needs. Which might hinder its adoption by other companies.

[Nelson]

Ahahahah[/SIZE][/FONT][/COLOR]
I think that this will turn in favor to the free codec, so I'm more interested than upset for this news.

I don't know, I think it shows a remarkably shrewd and nuanced understanding of the market. The device vendors and hardware makers have a fixed cost that is very similar to h.264, AND it's a standard (which all they want in the first place...) I can't imagine devices not supporting h.265 as it's better and the terms are nearly identical to h.264 for the hardware guys. The content costs suck but they hit a different group of companies. The key would be getting wide spread and cheap VP9 or daala hardware, without that I expect the content vendors will be encouraged to negotiate individual contracts with the patent pool.

My prediction: smaller independent media providers will stick at h.264, bigger ones will have "premium" content at 4k with h.265 and all your devices will have h.265 hardware in the next 2 years.

[TheBlackCat]

No other country in the world accepts the insanity that is a software patent.

Simply not true. There are a number of other countries besides the U.S. that allow software patents.

[profoundWHALE]

VP9 is also in a really good position now, especially as there's already stuff out there with hardware vp9 decoders. But it needs a better encoder, libvpx is still too slow for home use.

To get it fast, these provide some of the best boosts:
1: Multi-threading
2: Hardware acceleration
3: Something like OpenCL/HSA magic

Personally I'm more interested in Daala. Low latency, patent free, you know, that good stuff. (I hope it gets off the ground)

[Geopirate]

I think a lot of people commenting here are a step behind. I have a 2014 model 4k television that already has a hardware H.265 decoder. I'm sure almost all of the 2015 televisions already have it and they have most likely already made the decisions for 2016 televisions as well. It's a little late for hardware vendors to switch to VP9 at this point, that's likely why this has popped up. I know people are worried about hardware decoding on mobile devices, but you're not going to see the 4k difference on a 10" or smaller screen.

[Gusar]

1: Multi-threading

I've read a comment that libvpx-git is much better in this regard. I'll have to check. The last time I tested libvpx was the 1.4.0 release which sucks speed-wise.

2: Hardware acceleration

IDK, I don't think hardware encoders can reach the quality of a good software encoder. Not flexible enough.

3: Something like OpenCL/HSA magic

OpenCL, no way. All GPU encoders suck, no exceptions. HSA could be it in theory, but there isn't anything HSA out there yet, so no way to tell whether practice will match the theory.

Personally I'm more interested in Daala. Low latency, patent free, you know, that good stuff. (I hope it gets off the ground)

Daala is still very much a research project. Too early to tell what will become of it. It's definitely exciting though, but it's meant for the more distant future.

I'm actually more interested in Thor. It has the potential to become something in a short time frame. Because it uses tech that is out there today, while Daala is about finding new stuff.

BTW, low-latency is a requirement for real-time communication, but for non-interactive use - backing up your Blu-rays, creating videos for streaming (Youtube, Netflix, ...) - latency is irrelevant. So it's not *the* feature that should define a codec, but simply be an option the encoder provides. x264 has a low-latency preset, for example.


@Geopirate: I've read quite a lot of comments on various sites that 4k TVs already have hardware VP9 decoding, so no need to switch. About phones, the only mention I've seen is the Galaxy S6, but you're right that phones are less critical - you'll be watching at most HD on them, and h264 can handle that just fine.

[speculatrix]

Some government somewhere needs to make the decision that software is inherently not patentable

Sweden may be your answer, at least partly:

https://www.prv.se/en/patents/applying-for-a-patent/before-the-application/what-cannot-be-patented/software-and-business-methods/

However I believe that the USA has tried to lean on the Swedish government to change their law to allow patents on software.