Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

H.266/Versatile Video Coding (VVC)

newsletter.fraunhofer.de

Post reply on HN
Print view View original

361–370 of 435 posts

Re: H.266/Versatile Video Coding (VVC)

#361

Post by IfOnlyYouKnew »

pedrovhb wrote: post #126
sp332 wrote: post #26

Earlier quoted context omitted.

I don't know much about H.266, but some of the advances in H.265 depended on players having enough RAM to hold a bunch of previous decoded frames, so they could be referred to by later compressed data. Newer codecs tend to have a lot more options for the encoder to tune, so they need a combination of faster CPUs and smarter heuristics to explore the space of possible encodings quickly.

I wonder if instead of heuristics, machine learning could be used to figure out the best parameters.

In the limit case, compression and AI are identical.

Once you get to an AI that has full comprehension of what humans perceive to be reality, you can just give them a rough outline of a story, add some information on casting, writers, and Spielberg's mood during production, and they'll fill in the (rather large) blanks.

That's a bit exaggerated, but I remember reading about one such algorithm a few days ago (by Netflix, maybe?). It was image compression that had internal representations such as "there is an oak tree on the left".

It would then run the "decompression", find the differences to the original, and add further hints where neccessary.

Re: H.266/Versatile Video Coding (VVC)

#362

Post by marta_morena_25 »

Unklejoe wrote: post #14

It's interesting that they are able to continue improving video compression. You'd think that it would have all been figured out by now. Is this continued improvement related to the improvement of technology? Or just coincidental? Like, why couldn't have H.266 been invented 30 years ago? Is it because the computers back in the day wouldn't have been fast enough to realistically use it? Do we have algorithms today tha…

"You'd think that it would have all been figured out by now."

Would you? Video compression is one of the few things that we will work on for the next 1000 years and still be nowhere near finished. The best video compression would be to know the state of the universe at the big bang, have a timestamp of the beginning and end of your clip and spatial coordinates defining your viewport. Then some futuristic quantum computer would just simulate the content of your clip...

So yeah, sure we are done with video compression :). This is of course an extreme example of constant time compression that may or may not be ever feasible (if we live in a computer simulation of an alien race, then it is already happening).

But the gist is the same. Video compression is mostly about inferring the world and computing movement not by storing the content of the image.

For instance by taking a snapshot of the world, decomposing it into geometric shapes (pretty much the opposite of 3D rendering) and then computing the next frames by morphing these shapes + some diff data that snaps these approximations back in line with the actual data.

We are all but in the very infancy of video compression. What should surprise you is why it takes us so long to get anywhere.

Re: H.266/Versatile Video Coding (VVC)

#363

Post by ClumsyPilot »

zanny wrote: post #282
donatj wrote: post #18

Earlier quoted context omitted.

Honestly, the first step in this is getting the ffmpeg av1 library to a good usable place. It's currently so slow as to be near unviable. I'd happily switch when it becomes a usable option.

Is there any indication that rav1e is so inefficient as to have substantive double digit percentage speedup left to be realized? Because even if you cut encode times in half, they already take hours per minute of video on a quad core. AV1 is unlikely to ever be practical for "muggle" encode use, at least in this decade. It will only be worth committing that much compute workload to making a smaller file if the recipi…

HEVC encoder on 20 series nvidia cards is actually really good

Re: H.266/Versatile Video Coding (VVC)

#364

Post by giantrobot »

fwip wrote: post #321
pedrovhb wrote: post #126

Earlier quoted context omitted.

I wonder if instead of heuristics, machine learning could be used to figure out the best parameters.

Sure. Machine Learning is just "heuristics we don't understand."

Or more typically Monte Carlo heuristics.

Re: H.266/Versatile Video Coding (VVC)

#365

Post by naikrovek »

beervirus wrote: post #343
pcl wrote: post #316

Earlier quoted context omitted.

Yes — many codecs can be optimized for decoding at the expense of encoding. This is appropriate for any sort of broadcast (YouTube, television, etc). Also, in many applications, it’s suitable to exchange time for memory / compute. You can spend an hour of compute time optimally encoding a 20-minute YouTube video, with no real downside. Neither of these approaches are suitable for things like video conferencing, where…

Surely videoconferencing doesn’t actually use 60 FPS...

Why not? It's not full-motion video, it's literally talking heads. Talking heads are easy to push to 60fps on a relatively low bitrate.

Re: H.266/Versatile Video Coding (VVC)

#366

Post by rasz »

nickik wrote: post #193
otterley wrote: post #12

Earlier quoted context omitted.

No, because the market is more than happy to pay a few cents or dollars per device to get better compression and lower transmission bandwidth. This observation has held true consistently in the 3 decades since compressed digital media was invented.

Now audio uses open codecs for the part and new video will to.

Nah, Dolby excels at injecting itself where not needed. Blurays support 8-channel LPCM (uncompressed) audio, yet Dolby managed to push its proprietary junk codecs in there.

Re: H.266/Versatile Video Coding (VVC)

#367

Post by ww520 »

proverbialbunny wrote: post #283
sparkling wrote: post #80

Earlier quoted context omitted.

So compressing a 1080p video with H266 will not result in similar file size/quality improvements as a 4k video? How much are we looking at for 1080p, 10%?

>How much are we looking at for 1080p, 10%? We don't know yet. There are no public technical details (that I know of) for H266 yet, but if I recall H265 gave the same 50% reduction in bandwidth claims, and for years people stuck with H264, because it was higher quality due to dropping off less subtle parts of the video you really want to see. Only in the last couple of years has H265 really started to become embraced…

H.265 is patent laden. H.264 is much better in avoiding getting sued.

Re: H.266/Versatile Video Coding (VVC)

#368

Post by crazygringo »

superkuh wrote: post #312

I'd rather have slightly larger files that don't take hardware acceleration only available on modern CPU to decode without dying (ie, h264). Streaming is creating incentives for bad video codecs that only do one thing well: stream. Other aspects are neglected. And it's not like any actual 4K content (besides porn, real, nature, or otherwise) actually exists. Broadcast and movie media is done in 2K then extrapolated a…

Huh? TV and movies are widely shot with 4K cameras these days.

What is 2K? I've never even heard of a "2K" camera. Where did you get the idea things are being filmed in "2K" and being scaled to 4K?

Genuinely curious where you're getting this information from. Or are you confused because 1080p refers to the vertical resolution while 4K refers to the horizontal resolution?

Re: H.266/Versatile Video Coding (VVC)

#369

Post by realusername »

pault wrote: post #125
realusername wrote: post #44

Earlier quoted context omitted.

Software patents are not valid everywhere, x264 is brought up by VideoLAN in France, where software patents don't apply like in the rest of the EU.

So if a company in france writes open source software that infringes on a US software patent, and a company in the US bundles that source code in a product, is the US company liable for damages?

Yes, that's exactly how it works, the US company is responsible for following the US laws.

Re: H.266/Versatile Video Coding (VVC)

#370

Post by elithrar »

awill wrote: post #324
MikusR wrote: post #49

Earlier quoted context omitted.

The constantly growing list of H265 patent pool organizations with various licensing plans made even Apple to join AV1 bandwagon.

that's not why Apply joined AV1. They joined AV1 because they were forced to. Netflix, Google, Amazon are all on the AV1 bandwagon. If Youtube, Netflix and Prime Video all use AV1 for future higher quality streams, Apple cannot avoid supporting it.

It works both ways: if popular devices don’t support it in decode, encoding your library in a format that is slow(er) to decode and/or more taxing on battery life isn’t a clear win.
Post reply on HN