Hacker News new | ask | show | jobs
by Osiris 4889 days ago
Having done a lot of video encoding myself, I've found that H.264 offers the best quality, especially at lower bit rates.

I really do hope that VP9 can offer the quality of H.265. VP8 just wasn't good enough to compete against H.264.
2 comments
shmerl 4889 days ago
No, the point is it is good enough, even if in some cases it can't be as good because it has to work around software patents. Theora indeed can't compete in comparison, but VP8 surely can.
link
afhof 4889 days ago
Being good enough also depends on the context that the video will be used in. For example, I have an HTML5 video site where closer to 100% of the audience is either FF or Chrome. Both support VP8 out of the box, which means I only have to keep one copy of the video to serve to all users. If I had to target more browsers, such as Safari or IE, I would have to keep two copies of every video in order to work on all browsers (or worse, use flash as a work around).

I am okay with the minor loss in quality, just for the ease of managing a lot of videos and a lot of disk space.

link
shmerl 4889 days ago
Those who really lack support for WebM are mobile users who don't have dedicated hardware decoder and those who use crippled browsers like mobile Safari or mobile IE. Desktop users (at least in case of IE, not sure what's the story with Safari these days) can install needed plugins and DirectShow filters to support WebM. But the lack of dedicated hardware decoding in mobile SoCs remains a serious problem, since even when the browser supports VP8 decoding in software, it would be a degraded experience, especially power consumption wise.
link
protomyth 4889 days ago
Given how cable looks compared to over the air HDTV, the coming of 4k, and satellite channels even h.264 isn't good enough. VP8 won't be and h.265 is the industries answer. I would expect trouble with the adoption of VP9 given Google's moves with Mortorola's patents.
link
shmerl 4889 days ago
> I would expect trouble with the adoption of VP9 given Google's moves with Mortorola's patents.

How exactly? There is no indication that Google wants to make VP9 patent encumbered, since their whole idea behind VP8 was to enable high quality open video codec for the Web and beyond. If VP9 will be their natural next step, it will be open as well. If Daala comes soon enough too - it will be another option.

link
protomyth 4889 days ago
There are patents covering VP8 and I would bet VP9 also is covered.
link
shmerl 4888 days ago
That's true, and Google explicitly granted their free usage, thus the codec is open in all senses of the word. I'd expect the same story with VP9. So what's your point?

http://www.webmproject.org/license/bitstream/

http://www.webmproject.org/license/software/

http://www.webmproject.org/license/additional/

link
protomyth 4888 days ago
Google went back on what Motorola had already pledged when they bought them. What assurance does anyone have that Google won't do the same again?
link
gillianseed 4889 days ago
Google releases VP8 and VP9 with a patent and royalty free licence, so a potential patent threat will hardly come from them.
link
protomyth 4889 days ago
Motorola released their GSM required patents under FRAND licensing. Google buys Motorola and tries to use those same patents in a fight with Apple and Motorola (dragging Qualcomm in with them). Google has to agree with the FTC not to do that.

Why would Apple or Microsoft believe Google's patent and royalty free license given their behavior with Motorola's prior promises?

link
shmerl 4888 days ago
Apple and MS just don't like open codecs, period. That's the whole reason they sabotage them. It's not about belief - it's about the struggle between closed and open.
link
protomyth 4888 days ago
Apple and MS don't like legal uncertainty and additional risks in choosing technologies. Open never entered into it. People also forget some of the problem both companies had with MPEG-LA.
link
taligent 4889 days ago
Exactly the patent threat will come from the MPEG-LA or the members directly.

Because if VP9 is indeed similar to H.265 then you I would imagine a patent is being infringed somewhere. And since Google doesn't provide patent indemnification you can guarantee that some big royalties will be demanded from users.

link
shmerl 4889 days ago
> Exactly the patent threat will come from the MPEG-LA or the members directly. Because if VP9 is indeed similar to H.265

Why would they design the codec to be vulnerable to patent attacks? VP8 was designed to work around threats from H.264. They surely applied the same logic for VP9 since they intend it for practical use and not as theoretical brain exercise. MPEG-LA spits threats all the time, including against Theora and VP8, but they have no teeth to bite.

link
gillianseed 4889 days ago
I bet you said the same thing back when VP8 was released, no such patent lawsuit has occured.

Also, On2 which Google purchased holds lots of video compression patents, patents which codec's like h.264 and h.265 just as likely violates.

MPEG-LA does not offer any patent indemnification either.

link
shortsightedsid 4889 days ago
Video encoders have a number of tools. It really depends on what tools you have enabled.

The standard itself doesn't specify how an encoder behaves. That leaves a lot of room for innovation on encoders. End of the day, the qp/bitrate ratio you get is encoder implementation dependent and not sometime that's specific to the standard. This is the part where most of the patents exist.

link