x264 (MABS build) with integrated FFMS2 and LAVF

[LigH]

LoRd_MuldeR pointed me at it, and I can only confirm:

The x264 builds created by media-autobuild_suite including FFMS2 and L-SMASH Source (video codecs only = mode 6/"fullv") can't open AVI sources and eventually fall back to AviSynth, trying to use AviSource.

x264 -o *.mp4 *.avi
ffms [error]: could not create indexer
lavf [error]: could not open input file
avs [info]: trying AVISource... succeeded
...

I will try to create a full build (mode 4/"full"), wondering if possibly the limiting to video codecs only may cause issues. Could anyone else with enough insight possibly check if MABS might be misconfigured for this case? Or is it possible that there is indeed a mistake in current x264 sources?
_

In the meantime, issue #758 (https://github.com/jb-alvarado/media-autobuild_suite/issues/758) appeared, and commit c566f85 (https://github.com/jb-alvarado/media-autobuild_suite/commit/c566f85ed18a791f16c03c63da136b3f0272c731); will build again...

[LoRd_MuldeR]

Just want to add that the problem is not limited to AVI. Happens with MP4 or MKV as well.

The problem is always the same: FFMS reports "could not create indexer", and LAVF reports "could not open input file". Then x264 falls back to AVS.

Builds from here (http://msystem.waw.pl/x265/) do not have FFMS enabled (unfortunately), but LAVF input does work.

[LigH]

I just updated MABS and recompiled. x264 in 32 bit is already done, and here FFMS2 starts indexing (to RAM, I'd assume).

So I will release a new set soon.

[LoRd_MuldeR]

I just updated MABS and recompiled. x264 in 32 bit is already done, and here FFMS2 starts indexing (to RAM, I'd assume).

So I will release a new set soon.

Great. BTW, index is written to file (or reused from file) when using "--index" option.

[LigH]

Updated on MediaFire. Regards to wiiaboo for a quick response.

Fun fact: They are even smaller now.

[LoRd_MuldeR]

It is working now, thanks!

[mkver]

I didn't thought that others stumbled upon this problem, too. I actually thought that it never ever worked at all and that no one used it so no one complained. It seems that I stand corrected. Just out of curiosity: When did it stop working?
PS: That they are smaller now is probably due to the disabling of some audio codecs. (I tested an earlier version of MABS and there was code for disabling, but it didn't work.) One can reduce the size of the builds even further:
The size of my light x264 build with MABS: 17376256 B (all builds are x64).
After disabling all bitstream filters and enabling the extract_extradata-bsf (my test show this to be needed to actually get the codec parameters from transport streams; the h2645_mp4toannexb bsf are not needed for h2645 in mp4/Matroska) this is down to 17159680 B.
After also (trying to, see below) disabling subtitle decoders this is down to 17126912 B.
And looking at the list of disabled codecs I noticed that quite a few audio and subtitle decoders aren't disabled. The reason: MABS currently only disables decoders for which there is an encoder of the same name (I think I will report this). Fixing this brings the size down to 14960128 B. (Side issue: I have observed that if the input file has an mp2 audio track, lavf now emits a warning that it couldn't find the codec parameters for this stream. This might also happen for other codecs; similar things might also happen with the currently used list of codecs to deactivate. Does anyone know a way to stop lavf emitting warnings for non-video codecs?)
Finally, I also disabled some of the things that ffmpeg enables by default, but which is (to the best of my knowledge) unusable for lavf inside x264.exe: I added --disable-amf --disable-d3d11va --disable-dxva2 --disable-iconv --disable-schannel --disable-cuda --disable-cuvid. Now the size is only 13884928 B.
The relevant part of my script now looks like the following:

audio_codecs=(
$(sed -n '/audio codecs/,/external libraries/p' ../libavcodec/allcodecs.c | \
sed -n "s/^[^#]*extern.* *ff_\([^ ]*\)_decoder;/\1/p")
)
LDFLAGS+=" -L$MINGW_PREFIX/lib" \
log configure ../configure "${FFMPEG_BASE_OPTS[@]}" \
--prefix="$LOCALDESTDIR/opt/lightffmpeg" \
--disable-{programs,devices,filters,encoders,muxers,debug,sdl2,network,protocols,doc,bsfs} \
--enable-protocol=file,pipe \
--disable-decoder="$(IFS=, ; echo "${audio_codecs }")" --enable-gpl \
--enable-bsf=extract_extradata \
--disable-amf --disable-d3d11va --disable-dxva2 --disable-iconv --disable-schannel --disable-cuda --disable-cuvid
unset audio_codecs

Any suggestions/comments/improvements?

[LoRd_MuldeR]

I just updated MABS and recompiled. x264 in 32 bit is already done, and here FFMS2 starts indexing (to RAM, I'd assume).

So I will release a new set soon.

Some new commits have arrived in the official development tree. Any chance for updated x264 builds? :)
https://git.videolan.org/?p=x264.git;a=summary

(Highlights seem to include "4:0:0 (monochrome) encoding support" and "Sony XAVC" support)

[LigH]

New upload: x264 core:157 r2932 303c484

x264 0.157.2932 303c484
(libswscale 5.2.100)
(libavformat 58.17.101)
(ffmpegsource 2.30.0.0)
Win32: built on Aug 8 2018, gcc: 7.3.0
Win64: built on Aug 8 2018, gcc: 8.2.0
x264 configuration: --chroma-format=all
libx264 configuration: --chroma-format=all
x264 license: GPL version 2 or later
libswscale/libavformat/ffmpegsource license: GPL version 2 or later

[LoRd_MuldeR]

New upload: x264 core:157 r2932 303c484

x264 0.157.2932 303c484
(libswscale 5.2.100)
(libavformat 58.17.101)
(ffmpegsource 2.30.0.0)
Win32: built on Aug 8 2018, gcc: 7.3.0
Win64: built on Aug 8 2018, gcc: 8.2.0
x264 configuration: --chroma-format=all
libx264 configuration: --chroma-format=all
x264 license: GPL version 2 or later
libswscale/libavformat/ffmpegsource license: GPL version 2 or later

:thanks:

[Sparktank]

Thanks for the update!

[LigH]

New (and updated) upload: x264 core:157 r2935 545de2f (https://www.mediafire.com/file/ytyyzuy5w4czqp4/x264_c157_r2935_545de2f.7z)

x264 0.157.2935 545de2f
(libswscale 5.2.100)
(libavformat 58.19.102)
(ffmpegsource 2.30.0.0)
Win32: built on Nov 1 2018, gcc: 7.3.0
Win64: built on Nov 1 2018, gcc: 8.2.0
x264 configuration: --chroma-format=all
libx264 configuration: --chroma-format=all
x264 license: GPL version 2 or later
libswscale/libavformat/ffmpegsource license: GPL version 2 or later

[LigH]

New (and updated) upload: x264 core:157 r2935 545de2f (https://www.mediafire.com/file/6mz6lo66kkamda9/x264_c157_r2935_545de2f.7z)

x264 0.157.2935 545de2f
(libswscale 5.4.100)
(libavformat 58.24.100)
(ffmpegsource 2.30.0.0)
Win32: built on Dec 13 2018, gcc: 7.4.0
Win64: built on Dec 13 2018, gcc: 8.2.1 20181207
x264 configuration: --chroma-format=all
libx264 configuration: --chroma-format=all
x264 license: GPL version 2 or later
libswscale/libavformat/ffmpegsource license: GPL version 2 or later

[manolito]

Thanks for the new build,
for me it is the fastest of all the current r2935 builds (only tested 32-bit version).

(Probably stupid) question:
What do I do with the included ffmsindex.exe?
I use X264 under StaxRip with the latest stable ffms2 version 2.23.1. I only use the 32-bit versions.
Should I replace the 2.23.1 version of ffmsindex with the version which comes with your build? Is it compatible, is it necessary, what are the advantages?

Cheers
manolito

[LigH]

I am not sure myself how useful a separate indexer is, but MABS builds it along x264. So I assume that x264 can make use of the index files if they are provided together with a source file, and the FFMS2 demultiplexer is used in x264.

[qyot27]

ffmsindex isn't built alongside x264, it's part of the FFMS2 build. It does nothing with/for x264.

From x264's --fullhelp,
--index <string> Filename for input index file
I just tested it; that parameter makes x264 output the index from the initial pass to <output filename> rather it being used from RAM, and it also specifies a previously-generated index file to use, skipping the indexing step.

The FFMS2 version will have to match exactly for any index written by ffmsindex.exe or ffms2.dll (via AviSynth, for instance) to work with x264. And if the FFMS2 linked into x264 is static, then that only holds true for the exact binaries used there - if/when an update to ffms2.dll/ffmsindex.exe happens, FFMS2 will be newer than the one in x264 and force x264 to re-generate the index file rather than re-using it (or vice-versa; in my short test, x264 and its linked FFMS2 was newer - October 2018 - than the C-plugin build, from September).

In other words, only bother with ffmsindex if you're using FFMS2 as a plugin for AviSynth or VapourSynth, not when using the FFMS2 input on x264. If you're only using x264, stick with the indices it saves when using the --index parameter.

[LigH]

sigh

And ffms2.dll is not built by MABS.

OK, so I can omit the separate indexer safely.

[LigH]

New (and updated) upload: x264 core:157 r2935 545de2f (https://www.mediafire.com/file/6k456czv3fc4ev8/x264_c157_r2935_545de2f.7z)

x264 0.157.2935 545de2f
(libswscale 5.4.100)
(libavformat 58.26.100)
(ffmpegsource 2.30.0.0)
Win32: built on Feb 6 2019, gcc: 7.4.0
Win64: built on Feb 6 2019, gcc: 8.2.1 20181214
x264 configuration: --chroma-format=all
libx264 configuration: --chroma-format=all
x264 license: GPL version 2 or later
libswscale/libavformat/ffmpegsource license: GPL version 2 or later

[LigH]

New upload: x264 core:157 r2945 72db437 (https://www.mediafire.com/file/4btortb464dk8bc/x264_c157_r2945_72db437.7z/file)

x264 0.157.2945 72db437
(libswscale 5.4.100)
(libavformat 58.26.101)
(ffmpegsource 2.30.0.0)
Win32: built on Mar 12 2019, gcc: 7.4.0
Win64: built on Mar 12 2019, gcc: 8.3.0
x264 configuration: --chroma-format=all
libx264 configuration: --chroma-format=all
x264 license: GPL version 2 or later
libswscale/libavformat/ffmpegsource license: GPL version 2 or later
_

But wait! There's more...

New upload: x264 core:157 r2969 d4099dd (https://www.mediafire.com/file/xbapw6ra58gtxk1/x264_c157_r2969_d4099dd.7z/file)

x264 0.157.2969 d4099dd
(libswscale 5.4.100)
(libavformat 58.26.101)
(ffmpegsource 2.30.0.0)
Win32: built on Mar 13 2019, gcc: 7.4.0
Win64: built on Mar 13 2019, gcc: 8.3.0
x264 configuration: --chroma-format=all
libx264 configuration: --chroma-format=all
x264 license: GPL version 2 or later
libswscale/libavformat/ffmpegsource license: GPL version 2 or later

Interesting commitdiff remark (https://git.videolan.org/?p=x264.git;a=commitdiff;h=d4099dd4c722f52c4f3c14575d7d39eb8fadb97f):

Remove compatibility workarounds

This will break decoding with older versions of FFmpeg/Libav.

So keep your decoders up to date. I don't know how old is "too old" here...

[manolito]

So keep your decoders up to date. I don't know how old is "too old" here...

Please I need some clarifications here...
This will break decoding with older versions of FFmpeg/Libav.

Does this only apply to the X264 input, or does this mean that older player software or hardware won't be able to decode files created with this version any longer?

Out of curiosity I encoded a file with this latest r2969 32-bit version and tested if all my playing or transcoding software could handle it. No problem whatsoever, even under XP where I need to use some older decoders. And my Xtreamer box (latest firmware from 2011) also had no problems decoding the file.

Looks like a non-issue to me...


Cheers
manolito

[filler56789]

Please I need some clarifications here...

Does this only apply to the X264 input, or does this mean that older player software or hardware won't be able to decode files created with this version any longer?

Applies to the 4:4:4 [i.e., without chroma subsampling] streams generated by the newer x264 binaries:

- if( i_csp >= X264_CSP_I444 && h->param.b_cabac )
- {
- /* Disable 8x8dct during 4:4:4+CABAC encoding for compatibility with libavcodec */
- h->param.analyse.b_transform_8x8 = 0;

[manolito]

Thanks, I don't care about 4:4:4 chroma so I won't be affected...

[LigH]

I assume that libav fixed the decoding of strict video streams already a while ago, so x264 can now safely remove the compatibility changes for libav before a threshold date. But I don't know which date.

[MasterNobody]

Remove compatibility workarounds
This will break decoding with older versions of FFmpeg/Libav.
So keep your decoders up to date. I don't know how old is "too old" here...
Too old versions would mean versions before June/July of 2017. Corresponding patches where:
FFmpeg
avcodec/h264_cabac: Fix CABAC+8x8dct in 4:4:4 (http://git.videolan.org/?p=ffmpeg.git;a=commit;h=840b41b2a643fc8f0617c0370125a19c02c6b586)
avcodec/h264_mb: Fix 8x8dct in lossless for new versions of x264 (http://git.videolan.org/?p=ffmpeg.git;a=commit;h=06dda70f1e7c69a3b1684af5e6930431c62c527a)
Libav
h264_cabac: Fix CABAC+8x8dct in 4:4:4 (https://git.libav.org/?p=libav.git;a=commit;h=18d3f36d3c4d0f2c3e702f970ff8b457d7d5e39c)
h264dec: fix Lossless Decoding (Profile 244) for 8x8 Intra Prediction (https://git.libav.org/?p=libav.git;a=commit;h=79c6477c2abd8cfa41eef0c4ac39779dd8a9ec8e)

With old versions of FFmpeg/Libav you can have problems decoding lossless or 4:4:4+CABAC streams encoded with new x264.

[LigH]

Okay ...
versions before June/July of 2017
means that ffmpeg v3.4 (released shortly after) should have it fixed already, and v4.x quite certainly.

[sneaker_ger]

You can remain compatible to "old" and "new" lossless ffmpeg decoding by setting --no-8x8dct at <1% compression loss.

But: is "Remove compatibility workarounds" really in your builds? It's still only part of the "master", not the "stable" branch. Has been like that since almost 2 years now.

[nevcairiel]

But: is "Remove compatibility workarounds" really in your builds? It's still only part of the "master", not the "stable" branch. Has been like that since almost 2 years now.

Actually "Remove compatibility workarounds" was only in the experimental sandbox for 2 years, it moved to master 7 days ago, and will presumably be in the next stable push, unless they take it back out at some point, since stable is generally "just" an older master.

[LigH]

MABS update logs for x264-git report:

HEAD is now at d4099dd4 Remove compatibility workarounds

And it jumped from x264 0.157.2945 72db437 to x264 0.157.2969 d4099dd within few hours yesterday. So maybe there was a merge?

[sneaker_ger]

Actually "Remove compatibility workarounds" was only in the experimental sandbox for 2 years, it moved to master 7 days ago, and will presumably be in the next stable push, unless they take it back out at some point, since stable is generally "just" an older master.
I see.

Can confirm the changes are also already in the "official" binaries from videolan.

[LigH]

New upload: x264 core:157 r2971 98ee9d2 (https://www.mediafire.com/file/c00773j7gv5udig/x264_c157_r2971_98ee9d2.7z/file)

x264 0.157.2971 98ee9d2
(libswscale 5.4.101)
(libavformat 58.28.102)
(ffmpegsource 2.31.0.0)
Win32: built on Jul 17 2019, gcc: 9.1.0
Win64: built on Jul 17 2019, gcc: 9.1.0
x264 configuration: --chroma-format=all
libx264 configuration: --chroma-format=all
x264 license: GPL version 2 or later
libswscale/libavformat/ffmpegsource license: GPL version 2 or later

Luckily, there was quite recent activity (https://code.videolan.org/videolan/x264/commits/master).

[manolito]

Any reason why VideoLan still has r2969 as the latest stable version? Is r2971 not ready for prime time yet?

[LoRd_MuldeR]

Any reason why VideoLan still has r2969 as the latest stable version? Is r2971 not ready for prime time yet?

At this time, the x264 "master" branch is at r2984, and the "stable" branch is at r2980, both committed yesterday. VideoLAN apparently haven't updated their build since March 12th.

https://pastebin.com/HkvZdjnQ

[LigH]

Ah, I built mine the day before. Maybe I will update soon™ again.

[manolito]

Thanks for the info...

This leads to the question how well new commits are tested before they are called stable. From FFmpeg I learned it the hard way that it is a bad idea to always use the very latest "stable" build because thorough testing is not a particular strength of the FFmpeg devs.

Is this different for X264? Should I upgrade to a new stable build immediately after it is released, or is it wiser to stick with older versions until there is a real need for an upgrade (like support for new formats, important bug fixes etc) ?

[MasterNobody]

At this time, the x264 "master" branch is at r2984, and the "stable" branch is at r2980, both committed yesterday. VideoLAN apparently haven't updated their build since March 12th.

https://pastebin.com/HkvZdjnQ
If you were subscribed to the <x264-devel (at) videolan.org> mailing list, you would know (https://mailman.videolan.org/pipermail/x264-devel/2019-July/012705.html) that new binaries can be downloaded from the new URL:
https://artifacts.videolan.org/x264/

[LoRd_MuldeR]

If you were subscribed to the <x264-devel (at) videolan.org> mailing list, you would know (https://mailman.videolan.org/pipermail/x264-devel/2019-July/012705.html) that new binaries can be downloaded from the new URL:
https://artifacts.videolan.org/x264/

Thanks for the info! Maybe the link on the x264 web-site (http://www.videolan.org/developers/x264.html) should be updated to the new location then...

[LigH]

New upload: x264 core:158 r2984 3759fcb (https://www.mediafire.com/file/v73fdcspz5txe7h/x264_c158_r2984_3759fcb.7z/file)

x264 0.158.2984 3759fcb
(libswscale 5.4.101)
(libavformat 58.28.102)
(ffmpegsource 2.31.0.0)
Win32: built on Jul 19 2019, gcc: 9.1.0
Win64: built on Jul 19 2019, gcc: 9.1.0
x264 configuration: --chroma-format=all
libx264 configuration: --chroma-format=all
x264 license: GPL version 2 or later
libswscale/libavformat/ffmpegsource license: GPL version 2 or later

[filler56789]

https://git.videolan.org/ says:

MOVING TO GITLAB
We're moving projects to code.videolan.org, notably:

- x264

- VLC for WinRT

- VLC for Android

- VLMC

- NPAPI-VLC

- libdvdread

- libdvdnav

- libdvdcss

- biTStream

- DVBlast

- libdvbpsi

- libbluray

- libudfread

- libaacs

- libdplus

Regarding x264 specifically (or at least), the problem is that it's impossible to clone the x264 repository on GitLab through git :–/

=> git clone git://code.videolan.org/videolan/x264.git
Cloning into 'x264'...
warning: blah-blah-blah
fatal: unable to connect to code.videolan.org:
code.videolan.org[0: 88.191.250.9]: errno=No error

Oh, yes, anyone can download the latest tarball or zipball through a web browser, but then the compiled binary doesn't show the correct version /commit number -_-

[nevcairiel]

Did you construct that URL yourself? Because I don't see it anywhere on the GitLab page.
When you select the Clone button there, it'll give you a valid URL: https://code.videolan.org/videolan/x264.git - which works just fine with "git clone". The git protocol is being deprecated all over the web because its insecure and unscalable, in favor of HTTPS.

[filler56789]

Did you construct that URL yourself? Because I don't see it anywhere on the GitLab page.
When you select the Clone button there, it'll give you a valid URL: https://code.videolan.org/videolan/x264.git - which works just fine with "git clone". The git protocol is being deprecated all over the web because its insecure and unscalable, in favor of HTTPS.

HTTPS+git has never worked for me on Windows...
In this case specifically:

$ git clone https://code.videolan.org/videolan/x264.git
Cloning into 'x264'...
fatal: unable to access 'https://code.videolan.org/videolan/x264.git/': error setting certificate verify locations:
CAfile: R:/MAKE6T4/usr/lib/ssl/certs/ca-bundle.crt
CApath: none

OTOH, the git protocol has always worked with the old URL, git.videolan.org/x264.git

EDIT 1:

Okay, I found a workaround, and now git finally 1) "understands" HTTPS and 2) found the damn certificate because I created the directory where the stupid application expected to find it.

</CASE_CLOSED>

EDIT 2:

So far Mercurial has never shown the limitations that git shows on Windows... Until 2 days ago, I had always used git with cmd.exe and in this way it only works with the git protocol... whereas Mercurial has always worked fine on the so-called command-prompt of Windows.

[nevcairiel]

Git has always worked fine for me on Windows, but I use the official Git for Windows, which I told to use the Windows SSL library, and not whatever mingw uses with its own certificate stuff.

[filler56789]

Git has always worked fine for me on Windows, but I use the official Git for Windows, which I told to use the Windows SSL library, and not whatever mingw uses with its own certificate stuff.

Well ++ FWIW ++ just-for-the-record:

I moved from MSYS1 to MSYS2 in 2016, and since then I have used only and always a sufficient subset of the ~portable git package~ downloaded from the official Git source :)

msysgit, and that tiny subset "just worked" as well.]

Because I just want to download source-code through git, and because I still hadn't found a git server that refuses the git protocol, there was no problem for me in using just some .EXEs instead of the entire git package.

Let's be honest, at least for needs like mine, git is BLOATWARE :mad: and bloatware is pure evil :D

Not to mention that its Windows port is not as good as their creators and maintainers believe...
for example, I don't see the point
(assuming there is a (good) point)
in hardcoding exec_path = /mingw##/libexec/git-core in the Windows releases.
Why didn't they choose /usr instead? :confused:

Anyway, fortunately I discovered that placing the folders *ssl* and *git-core* in the same parent directory is what git.exe requires for dealing with HTTPS
(both under the MSYS2 environment and under cmd.exe).

[LigH]

The media-autobuild suite loads x264 nicely from the mentioned server. It runs Mercurial (hg) for x265 or git for x264, as required, in MSYS2/MinGW.

[LigH]

New upload: x264 core:159 r2991 1771b55 (https://www.mediafire.com/file/nkf884e1hse9byr/x264_c159_r2991_1771b55.7z/file)

x264 0.159.2991 1771b55
(libswscale 5.6.100)
(libavformat 58.35.101)
(ffmpegsource 2.31.0.0)
built on Jan 2 2020, gcc: 9.2.0
x264 configuration: --chroma-format=all
libx264 configuration: --chroma-format=all
x264 license: GPL version 2 or later
libswscale/libavformat/ffmpegsource license: GPL version 2 or later


Happy New Year 2020

[LoRd_MuldeR]

New upload: x264 core:159 r2991 1771b55 (https://www.mediafire.com/file/nkf884e1hse9byr/x264_c159_r2991_1771b55.7z/file)

Happy New Year 2020

:thanks:

[LigH]

New upload: x264 core:160 r3000_33f9e14 (https://www.mediafire.com/file/emu0ntukrasrn2m/x264_c160_r3000_33f9e14.7z/file)

x264 0.160.3000 33f9e14
(libswscale 5.6.101)
(libavformat 58.42.100)
(ffmpegsource 2.31.0.0)
built on Apr 16 2020, gcc: 9.3.0
x264 configuration: --chroma-format=all
libx264 configuration: --chroma-format=all
x264 license: GPL version 2 or later
libswscale/libavformat/ffmpegsource license: GPL version 2 or later

[LoRd_MuldeR]

New upload: x264 core:160 r3000_33f9e14 (https://www.mediafire.com/file/emu0ntukrasrn2m/x264_c160_r3000_33f9e14.7z/file)

x264 0.160.3000 33f9e14
(libswscale 5.6.101)
(libavformat 58.42.100)
(ffmpegsource 2.31.0.0)
built on Apr 16 2020, gcc: 9.3.0
x264 configuration: --chroma-format=all
libx264 configuration: --chroma-format=all
x264 license: GPL version 2 or later
libswscale/libavformat/ffmpegsource license: GPL version 2 or later

:thanks:

[LigH]

Updated archive: x264 core:160 r3000_33f9e14 (https://www.mediafire.com/file/105p4mb7e6qh5ek/x264_c160_r3000_33f9e14.7z/file)

x264 0.160.3000 33f9e14
(libswscale 5.6.101)
(libavformat 58.43.100)
(ffmpegsource 2.31.0.0)
built on May 18 2020, gcc: 10.1.0
x264 configuration: --chroma-format=all
libx264 configuration: --chroma-format=all
x264 license: GPL version 2 or later
libswscale/libavformat/ffmpegsource license: GPL version 2 or later

[manolito]

Thanks for the update...

The only differences I can see is a new gcc version and a newer libavformat version. X264 itself has not changed, so does the new build make a difference for real life encodes?

[LigH]

I doubt; just for the people who can't sleep without bleeding edge software.