x265 Improvements Going Forward? (Bug Fixes)

[HD MOVIE SOURCE]

Hi,
I'm new to the forums. I'm a big fan of movies and 4K Ultra HD, and I'm a big fan of high-quality video.
I have some basic questions about the codec.


1. The x265 codec, is there a list of bugs that the codec has?
2. Who chooses which bugs are worth fixing first?
3. Which current bugs affect maximum bit-rates and (to my understanding) can mux an encode because of this?

4. How long would it take to fix all the current bugs on the codec? Is it a week's worth of work or months?


Is there anyone that you could hire to fix bugs?

Thanks, I'm looking to see the codec really improve over this year.

[microchip8]

you mean x265? Because that's the most used HEVC encoder here

Multicoreware is responsible for almost everything. They choose their priorities what to implement. If there's a paying customer, they go right ahead to implement or fix

I don't know how many outstanding bugs x265 has

[benwaggoner]

Do you mean the x265 encoder for HEVC? If so, that's owned and largely coded by MultiCoreWare (MCW). MCW does custom engineering services, and most of x265's past and current development is funded by companies who pay MCW to implement features and fix bugs important to those companies. x265 is an open source (GPLv2) project, and they also accept patches from outside contributors.


On MultiCoreWare's Github: https://bitbucket.org/multicoreware/x265_git/issues?status=new&status=open.
MultiCoreWare decides what to fix when for issues outside of what's specifically funded.
The known bugs are on Github. If you've found one not listed, file it. The new 3.5 release has fixed some rate control issues, so make sure you test with the latest build before filing an issue.
It depends on how you define "bug" and "all" and "fix" ;)! Bugs that result in corrupt or spec-violating output tend to get fixed pretty rapidly. Lots of current bugs are about suboptimal performance, lack of clarity in how features work, and problems with compiling the source with various tools/platforms. Closing all 240 open issues would definitely be months. But most of them don't apply to most users and scenarios. In general, you've probably never used any software in your life that didn't have open issues.

[HD MOVIE SOURCE]

I do mean x265 yes. So, if I contacted MulticoreWare could I ask them to fix specific bugs? How would I go about contacting them?

[benwaggoner]

I do mean x265 yes. So, if I contacted MulticoreWare could I ask them to fix specific bugs? How would I go about contacting them?
If you just want them to know about the bug, use the "Issues" link above in #1 and file an issue.

If you want to pay them money to fix/improve something, I'd start with this link: https://multicorewareinc.com/contact-us/.

[HD MOVIE SOURCE]

If you just want them to know about the bug, use the "Issues" link above in #1 and file an issue.

If you want to pay them money to fix/improve something, I'd start with this link: https://multicorewareinc.com/contact-us/.

Thank you Ben.

[dahnen]

Not sure if I should've opened a new topic for this but this one seemed relevant and recent, so...

Does anyone really think x265 has a future?


I mean, of course it's open source, it's an OK encoder so it'll be continued to be used by individuals in the future but...

It has seen no real improvement in the last few years, while commercial encoders seem to be surpassing it month by month.

x265 commit history on the other hand seems to be on life support - only enough to be keep it from bitrotting.

I'm aware open source momentum seems to be moved to AV1 novadays, but I still think HEVC has its place and it's so sad that the direct descendant of the legendary x264 encoder is left out to be a mediocre tool and there's no other open source alternative to it.

My tone might read like a rant but I want to start a discussion around it thus the opening.

[benwaggoner]

3.5 was only two months ago and had some nice new stuff, but not much seems to have happened since then. But they've often gone quite for a while after a major release.

x264.readthedocs.io still hasn't been updated with the 3.5 release notes, I note. Bitrot in the documentation would be nearly as bad as in the code.

[RanmaCanada]

Not sure if I should've opened a new topic for this but this one seemed relevant and recent, so...

Does anyone really think x265 has a future?


I mean, of course it's open source, it's an OK encoder so it'll be continued to be used by individuals in the future but...

It has seen no real improvement in the last few years, while commercial encoders seem to be surpassing it month by month.

x265 commit history on the other hand seems to be on life support - only enough to be keep it from bitrotting.

I'm aware open source momentum seems to be moved to AV1 novadays, but I still think HEVC has its place and it's so sad that the direct descendant of the legendary x264 encoder is left out to be a mediocre tool and there's no other open source alternative to it.

My tone might read like a rant but I want to start a discussion around it thus the opening.

Well we are in the middle of a pandemic and many people have either lost their jobs or are still working from home. AV1 doesn't have a real future as the costs are far too high, and the compute power required is insane. Once the world gets back to normal, I am sure that things will continue, unless they decide they have done all they can and move to VCC.

[dahnen]

3.5 was only two months ago and had some nice new stuff, but not much seems to have happened since then. But they've often gone quite for a while after a major release.

x264.readthedocs.io still hasn't been updated with the 3.5 release notes, I note. Bitrot in the documentation would be nearly as bad as in the code.

Thanks, Ben.

Sorry I wasn't more clear (I was going to insert screenshot of the recent commit history here but forum messed it up slightly)

I mean, sure there was a recent 3.5 release but bulk of the commits for it made in last year (and this year only some fixes). And arguably that was a minor development considering 3.4 was almost a year before that.

To a regular joe like me it seems like Multicoreware has thrown in the towel and put x265 into maintainance mode. They don't have an ambition to improve it further to compete with commercial offerings and is thinking of keep going with their existing contracts. Maybe stop x265 development and focus on x266 instead.

I'm aware x265 is a mature software, and lack of code churn is expected, but still this puts x265 behind in competition as the time goes by.

But as I said, I'm a curious outsider and there's more knowledgeable people here and that's why I'm asking if there's more information we can learn from people in and out of the industry.