x265 HEVC Encoder

[asarian]

Originally Posted by sneaker_ger

Oh, I'm sure the difference will be noticable to consumers. Simply because authors will make it so, not because it's superior. Just like in the past authors used different mixes for AC3 and DTS tracks on the same DVD. It's all about marketing.

Having an HD Amp myself, trust me, you can hear de difference! Try listening to Blade Runner, the Final Cut, with full HD audio, or the same movie, just with DD5.1, and I guarantee you that everything below HD audio will sound crap from there on in.

[jd17]

Originally Posted by asarian

Having an HD Amp myself, trust me, you can hear de difference! Try listening to Blade Runner, the Final Cut, with full HD audio, or the same movie, just with DD5.1, and I guarantee you that everything below HD audio will sound crap from there on in.

That is not a very good argument against what sneaker said.

How do you know both mixes are based on the same mastering?
How do you know the DD-mix is not willingly different/worse than the HD mix?
Are you comparing identical speaker layouts (both just 5.1)?
Are you sure there is no dynamic compression in the DD-mix?
Are the volume levels identical?
Is the DD-mix limited in bitrate (384/448) or does it use the full 640kbit/s potential?

If you want a comparison of codec audio quality only, try encoding your own DD-mix from the HD-audio stream, considering all the factors above.
Even then, you cannot be sure in case of AC-3, because people claim that the original, commercial encoder is superior to those available freely...

[asarian]

Originally Posted by jd17

That is not a very good argument against what sneaker said.

How do you know both mixes are based on the same mastering?
How do you know the DD-mix is not willingly different/worse than the HD mix?
Are you comparing identical speaker layouts (both just 5.1)?
Are you sure there is no dynamic compression in the DD-mix?
Are the volume levels identical?
Is the DD-mix limited in bitrate (384/448) or does it use the full 640kbit/s potential?

If you want a comparison of codec audio quality only, try encoding your own DD-mix from the HD-audio stream, considering all the factors above.
Even then, you cannot be sure in case of AC-3, because people claim that the original, commercial encoder is superior to those available freely...

Blade Runner is just one example. Let's not derail this thread too much, but I can assure you, that once you go HD audio, you'll never want to go back, ever. And why should that surprise anyone?! A typical DTS-MA/TrueHD track is often larger in size than an entire DVD! With bitrates around 10x higher as regular DTS, the superiority of HD audio is a no-brainer, far as I'm concerned.

[excellentswordfight]

Originally Posted by asarian

And why should that surprise anyone?!

Cause 48Khz 16bit audio is very much sufficient for human hearing, and if you have enough experience of audio compression you know that there is several codecs that can achive very close to transperent audio with say 64kbps per channel (and most formats are above that!). I've encoded several surround tracks form lossless to multiple formats, you be surprised how little it affects hearable fidelity.

With that said, I'm sure that the HD-track sounds better for you, but most of that doesnt come from the higher specs (sample rate, bit depth and losslessness)

[jd17]

Originally Posted by asarian

With bitrates around 10x higher as regular DTS, the superiority of HD audio is a no-brainer, far as I'm concerned.

Nobody questions that lossless - in theory, or better measurably - is always better.
However, this is a forum that pretty much evolves around compression efficiency, i.e. saving bits where we don't see or hear it.

This is why I (and plenty others here too) take these lossless audio streams and compress them to 200-600kbit/s AAC or opus.

And while these codecs are obviously far superior to the ancient AC-3, you'd be surprised how good 640kbit/s Dolby Digital really is.
Accordingly, if the same master (and volume) is used, you would most likely fail in telling TrueHD apart from Dolby Digital, if you don't know which is playing.

[asarian]

Originally Posted by jd17

Nobody questions that lossless - in theory, or better measurably - is always better.
However, this is a forum that pretty much evolves around compression efficiency, i.e. saving bits where we don't see or hear it.

This is why I (and plenty others here too) take these lossless audio streams and compress them to 200-600kbit/s AAC or opus.

And while these codecs are obviously far superior to the ancient AC-3, you'd be surprised how good 640kbit/s Dolby Digital really is.
Accordingly, if the same master (and volume) is used, you would most likely fail in telling TrueHD apart from Dolby Digital, if you don't know which is playing.

'CD Quality' is 44.1 KHz/16 bit audio, generally considered 'very good' for music. Movies are a different thing, though. AC3 (aka DD 5.1) yields reasonable results at (almost always) 640kbps. But the difference between DTS-MA/TrueHD vs. AC3 is, well, unimaginably high. Believe me, I was in the same camp as you, one day, thinking you wouldn't be able to tell the difference... until I got an actual HD Amp. The fullness of the track when you switch to HD audio is staggering -- to the point where I can barely bear to listen to AC3 any more. In fact, for every Blu-ray I order (unless it's some sort of vintage deal), I first check to see whether it comes with HD Audio. If not, I simply don't buy it.

I'm not familiar with compressing audio 'to 200-600kbit/s AAC or opus' myself, so I'll take your word on those codecs.

[nevcairiel]

Originally Posted by asarian

But the difference between DTS-MA/TrueHD vs. AC3 is, well, unimaginably high.

Except on a technical level it really is not.

At least DTS has a "core" you can use for a direct comparison, since it has to be from the same master. If you were to decode both the core and the full HD stream to PCM and send that PCM to your "HD Amp", I'm positive that in most cases you wouldn't even be able to tell the difference. What these devices do is cheat you by playing with volume and EQ settings that play towards how people perceive audio. Just a slight bit more volume for HD, and most people already perceive it as "better", and there is more such tricks.

Or take a HD track and re-encode it as AC3, just to ensure its the same master, and then decode both back to PCM and send that to the Amp so the Amp does not know what the original format was.

There is a lot of trickery to try to sell you on "HD" stuff, because they had to sell you something, and in fact the audio quality difference have been minimal for years. If one really goes deep into it on a technical level, you'll eventually find that out. If you just blindly trust the Amp, then sure, HD probably sounds better to you, but not because its HD audio, but because the Amp cheats you.

[Atak_Snajpera]

Originally Posted by nevcairiel

Except on a technical level it really is not.

At least DTS has a "core" you can use for a direct comparison, since it has to be from the same master. If you were to decode both the core and the full HD stream to PCM and send that PCM to your "HD Amp", I'm positive that in most cases you wouldn't even be able to tell the difference. What these devices do is cheat you by playing with volume and EQ settings that play towards how people perceive audio. Just a slight bit more volume for HD, and most people already perceive it as "better", and there is more such tricks.

Or take a HD track and re-encode it as AC3, just to ensure its the same master, and then decode both back to PCM and send that to the Amp so the Amp does not know what the original format was.

There is a lot of trickery to try to sell you on "HD" stuff, because they had to sell you something, and in fact the audio quality difference have been minimal for years. If one really goes deep into it on a technical level, you'll eventually find that out. If you just blindly trust the Amp, then sure, HD probably sounds better to you, but not because its HD audio, but because the Amp cheats you.

That whole TrueHD/Lossless audio topic reminds of 192KHz 24bit bullshit promoted by some well known companies. (Super Audio-CD , DVD-Audio and others). Facts are simple. 48Khz 16bit is more than enough for humans. Adults have hearing range up to ~18KHz ,so it still below 24Khz. The same story with bit depth.
TrueHD can sounds better than AC3 640kbps only if was encoded from better source or your amp is doing some tricks boosting artificially volume for some frequencies. (bass/trebles). If 128kbps OPUS/AAC is transparent for 2.0 then I see no reason why 320kbps wouldn't be enough for 5.1.

[jd17]

Originally Posted by asarian

Believe me, I was in the same camp as you, one day, thinking you wouldn't be able to tell the difference... until I got an actual HD Amp. The fullness of the track when you switch to HD audio is staggering -- to the point where I can barely bear to listen to AC3 any more.

I get the feeling that you do not really understand where we are coming from...
We don't doubt you hear a difference!
But that difference is not based on codec superiority, but mastering, volume, dynamic compression and so forth.

However, we are turning in circles here, it's all been said already. You either do some legwork yourself and compare apples with apples, or you keep believing.

BTW, lossless-capable AVRs have been around for at least 10 years... Do you really think you are the only person here who owns one?


Sorry for the off-topic loop, let's get back to x265.

[benwaggoner]

Originally Posted by LigH

No. I believe because GPU features would not speed x265 up. At least not without risking a loss of quality or losing the independence from hardware and software platforms (portability).

GPU features are not magical general speed-ups for every case of use, sometimes they just don't match the requirements.

I could see a 20-25% speedup for high quality encoding using GPU and fixed-function encoder features with the new, deeper Intel encoder/decoder APIs for preanalysis and such. Not because it would natively produce high quality output, but because it could help determine some optimal encoding parameters.

[benwaggoner]

Originally Posted by jd17

Nobody questions that lossless - in theory, or better measurably - is always better.

I'll exactly argue against that. Information is a difference that makes a difference. A difference people can't discriminate in a proper double-blind test isn't a difference that matters.

Lossless or near lossless for sources makes sense, to the degree the extra information can eventually result in a detectable difference in derived content. Lossless, if used at all, is only used for archiving deep in studios.

[jd17]

I agree with you. That's why I inserted "in theory".

[Selur]

Question regarding:

--qp-adaptation-range

is this a suboption of '-aq-mode 1+', '--hevc-aq', both, or is it independent of both?

Cu Selur

[benwaggoner]

Originally Posted by Selur

Question regarding:

--qp-adaptation-range

is this a suboption of '-aq-mode 1+', '--hevc-aq', both, or is it independent of both?

I believe it only applies to --hevc-aq.

[kosta1000]

hi guys. Where can I find changelog for x265 3.0 ???

on official site, changelog exists only for version 2.9

another question : When Ryzen 2 get out around this June, would this processor have much better performance in x265 encoding, while AMD put 256-bit AVX2 now, along with many other notable improvments.

[Selur]

hi guys. Where can I find changelog for x265 3.0 ???

Probably hasn't been written yet,... iirc there hasn't been a 3.0 final/stable flag last tag was '3.0_RC'.

[benwaggoner]

Originally Posted by Selur

Probably hasn't been written yet,... iirc there hasn't been a 3.0 final/stable flag last tag was '3.0_RC'.

The best bet for the moment is to read the documentation that got added to —fullhelp in the checking.

https://bitbucket.org/multicoreware/.../branch/stable

[agressiv]

Looks like 3.0 is finally released! Congrats to the team! Here are the release notes:

New features
-------------
1. option:: '--dolby-vision-profile <integer|float>' generates bitstreams confirming to the specified Dolby Vision profile. Currently profile 5, profile 8.1 and profile 8.2 enabled, Default 0 (disabled)

2. option:: '--dolby-vision-rpu' File containing Dolby Vision RPU metadata. If given, x265's Dolby Vision metadata parser will fill the RPU field of input pictures with the metadata
read from the file. The library will interleave access units with RPUs in the bitstream. Default NULL (disabled).

3. option:: '--zonefile <filename>' specifies a text file which contains the boundaries of the zones where each of zones are configurable.

4. option:: '--qp-adaptation-range' Delta-QP range by QP adaptation based on a psycho-visual model. Default 1.0.

5. option:: '--refine-ctu-distortion <0/1>' store/normalize ctu distortion in analysis-save/load. Default 0.

6. Experimental feature option:: '--hevc-aq' enables adaptive quantization
It scales the quantization step size according to the spatial activity of one coding unit relative to frame average spatial activity. This AQ method utilizes
the minimum variance of sub-unit in each coding unit to represent the coding unit’s spatial complexity.

Encoder enhancements
--------------------
1. Preset: change param defaults for veryslow and slower preset. Replace slower preset with defaults used in veryslow preset and change param defaults in veryslow preset as per experimental results.
2. AQ: change default AQ mode to auto-variance
3. Cutree offset reuse: restricted to analysis reuse-level 10 for analysis-save -> analysis-load
4. Tune: introduce --tune animation option which improves encode quality for animated content
5. Reuse CU depth for B frame and allow I, P frame to follow x265 depth decision

Bug fixes
---------
1. RC: fix rowStat computation in const-vbv
2. Dynamic-refine: fix memory reset size.
3. Fix Issue #442: linking issue on non x86 platform
4. Encoder: Do not include CLL SEI message if empty
5. Fix issue #441 build error in VMAF lib

[Selur]

4. option:: '--qp-adaptation-range' Delta-QP range by QP adaptation based on a psycho-visual model. Default 1.0.
6. Experimental feature option:: '--hevc-aq' enables adaptive quantization

Sadly still unkown how they are related to the normal aq-modes,..

[benwaggoner]

Originally Posted by Selur

Sadly still unkown how they are related to the normal aq-modes,..

--hevc-aq overrides whatever --aq-mode is set to. I suspect that --aq-strength may get overridden itself, with --aq-adaption-range being the equivalent. Or both parameters could be used together, ala the interaction of CRF with maxrate/bufsize.