[HEVC] x265.EXE: mingw builds

[tnti]

You can go to X265's official site to check their progress map,there are a lot left unfinish in addition their are also a lot of unknowen bugs that need to be fixed
of coures by using a patched Libav can directly code HEVC without YUV output, however it can adjust the settings of the codes

[El Heggunte]

X265 is not yet ripe
This is the reason for the standard input is not supported

I read, "This is the lame excuse for them not to add stdin support to the encoder".

P.S.: Considering they apparently DON'T WANT TO PERCEIVE the bugs and design flaws of the x265 source-code,
the less there are people willing to test their software, the better for them.

[tnti]

https://github.com/pphh77/FFmpeg
They do not want to spread the preview version

[deadrats]

You can go to X265's official site to check their progress map,there are a lot left unfinish in addition their are also a lot of unknowen bugs that need to be fixed

2 silly questions:

1) if the bugs are "unknown" then how do you know they are there in the first place?

2) if the bugs are "unknown" then how do you go about fixing them?

how about a better answer: x265 will be "ripe" enough for std::in when they get a competent business manager, someone that understands how to go about bringing a product to market that can generate a sustainable income.

one of the cardinal rules of keeping a business humming along is to identify where the demand is and fill it; the demand with x265 is in the ability to actually test the software without needing to either compile it yourself or having to rely on some third party build and in the ability to test it with a range of sources that are representative of what most consumers will use.

i don't know what role "tom" plays in this new company but he should be removed from the role of public face for the venture. thus far this "multicoreware" or whatever you call yourselves have come across as evasive, disingenuous, deceitful and purposely misleading.

thus far "Tom" has made the following claims:

1) corporate sponsor feature requests and bug reports come first, the public's will get attention when the corporate sponsors concerns have been addressed.

2) no major movie studio has requested std::in support despite the fact that most studios now shoot in compressed digital formats.

3) video encoding is inherently a linear process and gpu acceleration will only apply to certain parts of the encoding process but at the same time he claims that x265 is already "heavily multi-threaded" and completely ignores the fact that the hevc standard was designed from the ground up for easy parallelization.

4) the chosen name is atrocious and not well suited for marketing. i personally would minimize the association between older codecs and this new standard as a way of differentiation and promoting the notion that this is a superior technology. a name such as HEVC++ would have been better or even MCW-HEVC, which has the company initials and supports branding efforts. x265 is confusing as it implies an association with a technology that x265 is trying to supplant.

thus far the road the x265 team has gone down is one of a bunch of high school kids trying to start a business, where one of the friends is the front man and tries to shield the company from any criticism by offering bullshit.

there is a concept in business called called "market presence", it's the notion that sometimes it's alright to lose a few bucks, usually by putting out a loss leader or with marketing that doesn't necessarily lead to direct sales just to keep your name in the public consciousnesses; it's why companies put up billboards, advertise on the side of buses, use product placement and spend millions in various corporate sponsorships, such as getting their name on a football stadium. none of these tactics puts a dime in their pocket in the short term but it keeps the brand name in the limelight and that's what the x265 group should be doing.

add support for std::in and offer official builds.

[El Heggunte]

Still waiting for Mplayer to include the up-to-date libavcodec and libavformat

Regarding MPC-BE --- hmmm....

Originally Posted by filler56789

Standalone filter MPCVideoDec.ax under Graphstudio:

if I tick the checkbox for "HEVC" and click "Apply/OK",

the checkbox "unticks itself"

and the HEVC decoder remains disabled

Yes - we know about this bug

[tnti]

They do not want the unfinished version of the official release
If no major problems, then they should have been released
Do not forget that it is just a preview version of the RC version is yet connected to the

[Tommy Carrot]

I ran a few tests with the latest build, but honestly, an old build from august 14 was both faster and had significantly better quality for comparable bitrates. In their effort to replace the reference encoder algorythms with their own, something went awry i guess, and it's still not fixed.

[El Heggunte]

@ Tommy Carrot: many THANKS for the contribution

Regarding the buildbot x265.cc --- well, I won't deny its layout has improved a lot

[ post #1 has been updated ]

[tnti]

PotPlayer 1.5.40766
HEVC decoding support
http://cafe.daum.net/pot-tool/AZMV/10550@El Heggunte

[El Heggunte]

@ tnti: have you tested the latest PotPlayer extensively

Just for a comparison, MPC-BE is not so good as it should be:

http://forum.doom9.org/showthread.php?p=1648629#post1648629

[tnti]

PotPlayer should be no MP4 (HEVC) splitter

[El Heggunte]

Last edited by Tommy Carrot; 18th Oct 2013 at 22:47 GMT. Reason:
removed build, as x265.cc builds are now working on win XP

That's not true, sadly.

information

Currently builds x265 8bpp with MinGW (x86) and VC11 (x86 / x86-64).
Newer VC11 builds will be build static, older builds may need VC redist.

MinGW - Windows XP and later
VC11 - Windows Vista and later

And oh, BTW, http://forum.doom9.org/showthread.php?p=1648279#post1648279

@ tnti: thanks for testing PotPlayer.

[vhelp]

tnti, the need for it has been expressed many times... [rant] ..., ..., ... [/rant].

[tnti]

The latest 4066e6e I can not build out
X265 to be able to "use" stage?

X265 only be counted as Test product


I can not use PotPlayer play HEVC

[x265.cc]

The latest 4066e6e I can not build out

yep, failure at build with mingw, vc11 x86. vc11 x86-64 works fine for me.

That's not true, sadly.

VC11 builds are not yet working at WinXP, correct. I have to test out the vc11 winxp binaries myself at first.
Then i will rewrite (and enhance) the buildbot.

And oh, BTW, http://forum.doom9.org/showthread.php?p=1648279#post1648279

I'm not shure if he is correct.

this guy doesn't understand what my patch does...the switch goes to the cmake command line.

What his patch does:

-Check if CMake Version is at least "2.8.11"
-Add "-T "v110_xp"" to the "make-solutions.bat" batch file (to cmake, not cmake-gui)

Thats it. You can't use this batchfile for automated builds do to the usage of "cmake-gui"

The whole magic:

"cmake -G "Visual Studio 11 -T "v110_xp"" ..\..\source" for dynamic (win xp) builds
"cmake -G "Visual Studio 11" -DSTATIC_LINK_CRT=1 -T "v110_xp" ..\..\source" for static (win xp)builds

-> NOT TESTED

It would be much more intresting if he would post what i haven't understood.

//edit:

WinXP x32 Users possibly want to check out this (VC11 x86-32 WinXP support)

[El Heggunte]

I have just reported the most recent compiler error to the x265 mailing list;
probably Steve will give us an answer only next Monday...

[ 68%] Building CXX object common/CMakeFiles/common.dir/vec/intra-sse3.cpp.obj
F:/GITS/X265/20131019/source/common/vec/intra-sse3.cpp: In function 'void {anonymous}::predDCFiltering(pixel*, pixel*, pixel*, intptr_t, int)':
F:/GITS/X265/20131019/source/common/vec/intra-sse3.cpp:118:54: error: '_mm_cvtsi64_si128' was not declared in this scope
         pix = _mm_cvtsi64_si128(*(uint64_t*)&above[1]);
                                                      ^
F:/GITS/X265/20131019/source/common/vec/intra-sse3.cpp: In function 'void {anonymous}::intra_pred_dc(pixel*, pixel*, pixel*, intptr_t, int, int)':
F:/GITS/X265/20131019/source/common/vec/intra-sse3.cpp:177:74: error: '_mm_hadd_epi16' was not declared in this scope
         sum  = _mm_cvtsi128_si32(_mm_hadd_epi16(_mm_hadd_epi16(temp, temp), temp));
                                                                          ^
F:/GITS/X265/20131019/source/common/vec/intra-sse3.cpp:177:81: error: '_mm_hadd_epi16' was not declared in this scope
         sum  = _mm_cvtsi128_si32(_mm_hadd_epi16(_mm_hadd_epi16(temp, temp), temp));
                                                                                 ^
F:/GITS/X265/20131019/source/common/vec/intra-sse3.cpp:177:81: error: redeclaration of '<typeprefixerror>_mm_hadd_epi16'
F:/GITS/X265/20131019/source/common/vec/intra-sse3.cpp:177:74: note: previous declaration
'<type prefixerror>_mm_hadd_epi16'
         sum  = _mm_cvtsi128_si32(_mm_hadd_epi16(_mm_hadd_epi16(temp, temp), temp));
                                                                          ^
make[2]: *** [common/CMakeFiles/common.dir/vec/intra-sse3.cpp.obj] Error 1
make[1]: *** [common/CMakeFiles/common.dir/all] Error 2
make: *** [all] Error 2

[El Heggunte]

.....
//edit:

WinXP x32 Users possibly want to check out this (VC11 x86-32 WinXP support)

MWA-HA-HA-HA-HA — yes$S, it works, ^__^

[tnti]

PotPlayer need to manually add MP4 splitter
You can use MPC-BE or Strongene

[vhelp]

-addition/feature request, to report-

i have been going back in the previous x265.exe encoded videos to review them but in some of them i am not certain of the correct version used.

is there any info (identifyer) in the x265.exe hevc encoded file that i can view to determine which x265.exe build encoded the video ?

can such an indentifier be added to the encoded videos in your next release ?

or, can you advise on a method to inject an identifier (in the encoded video) code so that i can add that feature in my gui ?

[Marchand]

PotPlayer need to manually add MP4 splitter
You can use MPC-BE or Strongene

tnti

https://forum.videohelp.com/threads/359638-Which-player-would-you-recommend-for-playing...C-H265-DivX265

I'm trying to identify and rank the best or the best player (s) to which we have already anticipated in the new encoders HEVC-H265-DivX265 where I hope to be helping the community and VideoHelp users too, including myself. HEHEHE.

[Marchand]

-addition/feature request, to report-

i have been going back in the previous x265.exe encoded videos to review them but in some of them i am not certain of the correct version used.

is there any info (identifyer) in the x265.exe hevc encoded file that i can view to determine which x265.exe build encoded the video ?

can such an indentifier be added to the encoded videos in your next release ?

or, can you advise on a method to inject an identifier (in the encoded video) code so that i can add that feature in my gui ?

VHelp .... this question, I think only our partner Selur can help.

[tnti]

MPC-BE is the most convenient
PotPlayer little trouble

[vhelp]

i hex'ed around but couldn't find anything. i tried searching for:

x265, 265, h265, hm10, hevc, hvc1, ver, version, 4014edcf2157, f3523973eafb, etc., but nothing comes up.

i think i read somewhere, where you can pad certain bytes to the end of the file. i may search around for that and see if i can do that and then attempt to mux and see if that succeeds. if that does, then i will incorporate that in my gui after every encode, so that i can keep track of which version encoded the video.

[ozok]

vhelp, I'm not sure but perhaps you can try to add some identifier using the muxer you use (mp4box etc). I'm not sure even if this is possible but one solution that comes to my mind is to write tags to mp4 file. But you'll have to change this value manually everytime you update x265.exe.

[vhelp]

that i can live with. x264 does that on-the-fly for us since everything from raw h264 to a finished container is done in one step. but x265 is not at that place yet. so, for now, since only raw files are created, it is best to stamp the raw file with the id.

[El Heggunte]

Commit fabb25a is here.

[El Heggunte]

For the curious, L-Smash binaries with (experimental) HEVC support.

[vhelp]

i wouldn't mind trying l-smash. but, some of the files i test encoded contained artifacts after muxing.

does l-smash decode the raw hvc1 or the muxed ?

did you mean to post the avisynth dll file ?

the .7z file you posted does not have any readme/howto/setup/use.

[El Heggunte]

Well, the source-code I downloaded today IS "experimental". If there are differences between the muxes from MP4Box and the muxes from L-Smash, then the bugs of this latter can be detected, reported, fixed.

There exists a not-so-helpful "manual", but you can get the same info by just entering the name of the program(s) in the CLI:

Copyright (C) 2010-2012 L-SMASH project

Usage: muxer [global_options] -i input1 [-i input2 -i input3 ...] -o output
Global options:
    --help                    Display help
    --optimize-pd             Optimize for progressive download
    --interleave <integer>    Specify time interval for media interleaving in milliseconds
    --file-format <string>    Specify output file format
                              Multiple file format can be specified by comma separators
                              The first is applied as the best used one
    --isom-version <integer>  Specify maximum compatible ISO Base Media version
    --shift-timeline          Enable composition to decode timeline shift
    --chapter <string>        Set chapters from the file.
    --chpl-with-bom           Add UTF-8 BOM to the chapter strings
                              in the chapter list. (experimental)
    --chapter-track <integer> Set which track the chapter applies to.
                              This option takes effect only when reference
                              chapter is available.
                              If this option is not used, it defaults to 1.
    --copyright-notice <arg>  Specify copyright notice with or without language(latter string)
                                  <arg> is <string> or <string>/<string>
    --language <string>       Specify the default language for all the output tracks.
                              This option is overridden by the track options.
Output file formats:
    mp4, mov, 3gp, 3g2, m4a, m4v

Track options:
    fps=<arg>                 Specify video framerate
                                  <arg> is <integer> or <integer>/<integer>
    language=<string>         Specify media language
    alternate-group=<integer> Specify alternate group
    encoder-delay=<integer>   Represent audio encoder delay (priming samples) explicitly
    copyright=<arg>           Specify copyright notice with or without language(latter string)
                                  <arg> is <string> or <string>/<string>
    handler=<string>          Set media handler name
    sbr                       Enable backward-compatible SBR explicit signalingmode
How to use track options:
    -i input?[track_option1],[track_option2]...

iTunes Metadata:
    --album-name <string>     Album name
    --artist <string>         Artist
    --comment <string>        User comment
    --release-date <string>   Release date (YYYY-MM-DD)
    --encoder <string>        Person or company that encoded the recording
    --genre <string>          Genre
    --lyrics <string>         Lyrics
    --title <string>          Title or song name
    --composer <string>       Composer
    --album-artist <string>   Artist for the whole album (if different than theindividual tracks)
    --copyright <string>      Copyright
    --description <string>    Description
    --grouping <string>       Grouping
    --tempo <integer>         Beats per minute

-----

Usage: boxdumper [option] input
  options:
    --box          Dump box structure
    --chapter      Extract chapter list
    --timestamp    Dump media timestamps

-----

Usage: remuxer -i input1 [-i input2 -i input3 ...] -o output
Global options:
    --help                    Display help.
    --chapter <string>        Set chapters from the file.
    --chpl-with-bom           Add UTF-8 BOM to the chapter strings
                              in the chapter list. (experimental)
    --chapter-track <integer> Set which track the chapter applies to.
                              This option takes effect only when reference
                              chapter is available.
                              If this option is not used, it defaults to 1.
    --language <string>       Specify the default language for all the output tracks.
                              This option is overridden by the track options.
Track options:
    language=<string>         Specify media language
    alternate-group=<integer> Specify alternate group
    handler=<string>          Set media handler name
    seek=<integer>            Specify starting point in media
    safe-seek=<integer>       Same as seek except for considering random accessible point
                              Media starts from the closest random accessible point
How to use track options:
    -i input?[track_number1]:[track_option1],[track_option2]?[track_number2]:...

For example:
    remuxer -i input1 -i input2?2:alternate-group=1?3:language=jpn,alternate-group=1 -o output

-----

Usage: timelineeditor [options] input output
  options:
    --track           <integer>  Specify track number to edit [1]
    --timecode        <string>   Specify timecode file to edit timeline
    --media-timescale <integer>  Specify media timescale to convert
    --media-timebase  <integer>  Specify media timebase to convert
    --skip            <integer>  Skip start of media presentation in milliseconds
    --delay           <integer>  Insert blank clip before actual media presentation in milliseconds
    --dts-compression            Eliminate composition delay with DTS hack
                                 Multiply media timescale and timebase automatically

P.S.: L-SMASH's restrictions <= click

[vhelp]

i mentioned it because i recently learned about l-smash in this post and had trouble getting it to work--to load videos faster, that is. my last notes on this latest avisynth plugin are there.

https://forum.videohelp.com/threads/359494-LWLibavVideoSource-speeding-up-load-times?p=...=1#post2273871

see post # 8 for the actual avisynth plugin. h264 (ts) videos no longer crash when using this plugin while stepping backward/forwards/paging far. i recommend it. but be aware that re-loading the script (for editing/reviewing) are slow each time.