x264 rev2409 tMod
Sources on GitHub: https://github.com/astrataro/x264_tMod,
and on BitBucket: https://bitbucket.org/astrataro/x264_tmod.
Ad: new binary host is http://tmod.nmm-hd.org.
Download: x264_rev2409_tMod.7z
Updates:
Only regular updates as always. Find more details on official x264-devel mail-list.
Nothing changed in tMod, except for some minor patch updates for x264 plain branch.
Edited on UTC 11:30 17 Mar 2014:
fixed a bug (patch/audio-support) that Matroska video length was not written after encoding, causing total length not available in MediaInfo or other players and editors. Thanks to reekilynn for reporting.
Edited on UTC 06:00 18 Mar 2014:
Automatically invoke AutoLoadPlugins() if available when using avs demuxer to open non-avs scripts. For AviSynth+. Patch from BugMaster.
天山牛棚 
Thanks as usual. I always waiting for your build 🙂
Hi there,
I am trying to make tMod on Ubuntu 12.04, and ffms and lavf refuse to show up.
I googled, and it said that I need to compile ffmpeg to fix this problem.
According to Ubuntu compile guide of ffmpeg, a “lib264” is in the guide.
So…I am a little confused how I can combine this ffmpeg and tMod together, so we can use ffms and lavf in tMod… Any hint of how to do that?
Thanks in advance.
Beining
ACICFG
Building ffmpeg for x264’s lavf/ffms support does not require libx264, as x264 use lavf/ffms for demuxing and decoding, not for encoding.
If you’d like to build *full* version of both ffmpeg and x264, try to:
1. build and install ffmpeg without further encoding libraries — you don’t need libx264 so don’t specify
--enable-libx2642. build and install ffms ( if you need it in your x264 )
3. build and install x264 with libavcodec/libavformat/libswscale/libffms2/… — by default x264 will be built with them as long as you don’t use
--disable-lavf --disable-ffms, if those libraries were not found, try to usePKG_CONFIG_PATH=xxxxxxto indicate pkg-config path. On Ubuntu it’s normally not needed.4. build ffmpeg again, with
--enable-libx264this time.P.S. if you build ffmpeg with libx264-tMod, you need to make some modification on ffmpeg. I wrote this patch quite a while ago, but it may not suit for current ffmpeg any more as I didn’t build ffmpeg with x264 for quite a long time:
diff --git a/libavcodec/libx264.c b/libavcodec/libx264.c index 35c0030..d5f89da 100644 --- a/libavcodec/libx264.c +++ b/libavcodec/libx264.c @@ -41,6 +41,7 @@ typedef struct X264Context { char *preset; char *tune; char *profile; + char *device; char *level; int fastfirstpass; char *wpredp; @@ -365,8 +366,8 @@ static av_cold int X264_init(AVCodecContext *avctx) if (x4->nal_hrd >= 0) x4->params.i_nal_hrd = x4->nal_hrd; - if (x4->profile) - if (x264_param_apply_profile(&x4->params, x4->profile) < 0) { + if (x4->profile || x4->device) + if (x264_param_apply_profile(&x4->params, x4->profile, x4->device) < 0) { int i; av_log(avctx, AV_LOG_ERROR, "Error setting profile %s.\n", x4->profile); av_log(avctx, AV_LOG_INFO, "Possible profiles:"); @@ -475,9 +476,9 @@ static const AVOption options[] = { { "crf_max", "In CRF mode, prevents VBV from lowering quality beyond this point.",OFFSET(crf_max), AV_OPT_TYPE_FLOAT, {.dbl = -1 }, -1, FLT_MAX, VE }, { "qp", "Constant quantization parameter rate control method",OFFSET(cqp), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, VE }, { "aq-mode", "AQ method", OFFSET(aq_mode), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, INT_MAX, VE, "aq_mode"}, - { "none", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_AQ_NONE}, INT_MIN, INT_MAX, VE, "aq_mode" }, - { "variance", "Variance AQ (complexity mask)", 0, AV_OPT_TYPE_CONST, {.i64 = X264_AQ_VARIANCE}, INT_MIN, INT_MAX, VE, "aq_mode" }, - { "autovariance", "Auto-variance AQ (experimental)", 0, AV_OPT_TYPE_CONST, {.i64 = X264_AQ_AUTOVARIANCE}, INT_MIN, INT_MAX, VE, "aq_mode" }, + { "none", NULL, 0, AV_OPT_TYPE_CONST, {.i64 = X264_AQ_NONE}, INT_MIN, INT_MAX, VE, "aq_mode" }, + { "variance", "Variance AQ (complexity mask)", 0, AV_OPT_TYPE_CONST, {.i64 = X264_AQ_MIX}, INT_MIN, INT_MAX, VE, "aq_mode" }, + { "autovariance", "Auto-variance AQ (experimental)", 0, AV_OPT_TYPE_CONST, {.i64 = X264_AQ_AUTOMIX}, INT_MIN, INT_MAX, VE, "aq_mode" }, { "aq-strength", "AQ strength. Reduces blocking and blurring in flat and textured areas.", OFFSET(aq_strength), AV_OPT_TYPE_FLOAT, {.dbl = -1}, -1, FLT_MAX, VE}, { "psy", "Use psychovisual optimizations.", OFFSET(psy), AV_OPT_TYPE_INT, { .i64 = -1 }, -1, 1, VE }, { "psy-rd", "Strength of psychovisual optimization, in <psy-rd>:<psy-trellis> format.", OFFSET(psy_rd), AV_OPT_TYPE_STRING, {0 }, 0, 0, VE}, --can’t download, only page of jibberish appears.
DL well here though
Got a question: since your avs4x264mod can open mov/mp4 without indexing them first via LSMASHVideoSource(), why don’t you include this awesome feature in your x264 builds?
Currently the only way is somewhat messy: a temporary avs script with LSMASHVideoSource needs to be created and then piped with avs4x264mod. The other option *would be* to use x264_64_tMod –demuxer lavf but unfortunately it doesn’t work with 100-200GB lossless mp4/mov files produced via x264_64_tMod –fragments –no-remux, it only sees ~100frames even if –frames #### option is specified.
tMod invokes LSMASHVideoSource for mov/mp4 if you use
--demuxer avs:Actually lite version does this by default as it doesn’t contain lavf/ffms.
thanks, but this isn’t direct, it will use avisynth which autoloads lots of plugins and stuff, thus wasting OS memory and time. I mean why isn’t it possible to statically build with lsmashworks included?
trying to compile for OSX with numerous errors. Do you have a compiled OSX version?
somewhile later….
Hope this would help you in a way…
https://github.com/superwbd/ffmpeg-x264-tMod-OSX
Yes, binaries of them. Hope they won’t blow up your Macbook…
Thank you for the Binary download links. The problem I seem to have with those binaries is that they are linked against a home-brew version of FFMPEG (and an older one at that). Do you know of a build that is not?
Well…..I will probably redo that later today and see whether it works or not.
Well….I think I am the second person in the history who decided to build x264 binary and decided to make them public…. And both of these 2 persons are in the same comment area…
Whatever, check the Github for updates. This time I unlinked FFmpeg from Homebrew, and used a standalone binary of FFmpeg. Hope this would make a difference…
I really appreciate the time and effort you guys put into this. Thanks again!
Hello. I’ve just discovered this specific build, and i’m very interested by it. I’ve until now make my own build including 2 patchs, but i’ve seen they’re allready in your release. And, as i’m encoding animes, i’m of course very interested by all the new patches. But, for my personnal use, i’m also modifying one very little thing (allowing weightp=2 with blu-ray). So, i’m using turtoise git. Once i’ve clone your github repository, created my own personnal custom build branch, how can i compile your 2 differents versions (little and ffmpeg) ? I’m currently using to compile the x264 official realease the method described on doom10 (msys + mingw versions provided by komisar).
Hum… Not even able to compile, error on lsmash_importer.h… Will try disabling lsmash.
lite version is pure x264 with l-smash, and no ffmpeg/libav/ffms linked in; while full version is compiled with those packages. there’s no difference in source code. you can build full version with doom10’s guide, or lite version by simply not compile ffmpeg/libav.
btw, the only hacky thing for building tmod is compiling l-smash library. Though it’s not difficult by patching l-smash with this patch.
Thank for your answer. I’ve been able to compile disabling l-smash and all ffmpeg/audio/etc… As i’m doing only either mkv or raw h264 and without audio, allways from an avisynth script, what i’ve got is fine for me.
I’m only interested in the new aq method patches. Do you have by any chance any feedback on them, or any “user manual” ? I’ve no idea how they should be used. Is it either aq2 or aq3 ? Can they be used together ? What parameters are advice for each kind of footage ? If used, is it intended to be with also aq-mode ? If yes 1, 2 or even 3 ? Are they more efficient on 8bit and finaly not very interesting on 10bit ? I’ve a lot of questions i’m unable to find answers. Apparently these patches were made by japanese people to improve anime quality encode. Good, it’s what i’m looking for. As people took and spend time to make them, i intend to believe they are interesting, but, it’s so much easy to screw things when you don’t realy know how to use theim, that you may finaly wrongly discard theim thinking they don’t improve quality, or even they give worse result.
Hi, it seems to have issues on i586 machine
[ 30s] filters/video/x86/yadif_sse_template.c: In function ‘filter_line_sse2’:
[ 30s] filters/video/x86/yadif_sse_template.c:234:9: error: can’t find a register in class ‘GENERAL_REGS’ while reloading ‘asm’
[ 30s] filters/video/x86/yadif_sse_template.c:242:9: error: can’t find a register in class ‘GENERAL_REGS’ while reloading ‘asm’
[ 30s] filters/video/x86/yadif_sse_template.c:234:9: error: ‘asm’ operand has impossible constraints
[ 30s] filters/video/x86/yadif_sse_template.c:234:9: error: ‘asm’ operand has impossible constraints
[ 30s] filters/video/x86/yadif_sse_template.c:242:9: error: ‘asm’ operand has impossible constraints
[ 30s] filters/video/x86/yadif_sse_template.c:242:9: error: ‘asm’ operand has impossible constraints
I can’t find anywhere to report this to you….so…
Hello,
I’m trying to patch libx264.c for building ffmpeg with tMod, I added the modifications from your patch but I’ve the following error:
libavcodec/libx264.c: In function ‘X264_init’:
libavcodec/libx264.c:420:41: error: incompatible types when assigning to type ‘int[3]’ from type ‘int’
libavcodec/libx264.c:422:41: error: incompatible types when assigning to type ‘int[3]’ from type ‘int’
CC libavcodec/loco.o
CC libavcodec/lossless_videodsp.o
make: *** [libavcodec/libx264.o] Erreur 1
Any help would be much appreciated. 🙂 Cheers and thank you.
Thx for removing my previous reply!
Thx for your help, very sweet.
Sorry, I wasn’t able to view my first reply. Dunno why.
You can ignore the 2nd reply.
ps: I’m still waiting for help, of course. 😀
on Ubuntu 14 (64 bit) I’ve compiled l-smash and on make I get an AVI output error. So I disabled avi-output and now i get the following:
mp4_lsmash.c:(.text+0x139f): undefined reference to `mp4sys_amr_create_damr’
collect2: error: ld returned 1 exit status
make: *** [x264] Error 1
any suggestions on compiling for Ubuntu 14/64bit?
I haven’t updated codes for the new l-smash API, so please use the versions of libs no later than the ones in the documents of provided windows binaries. They are proved to work. I need to spare some time to do the API updates.
Thank you for this great build of x264. I was wondering if you can add a patch to prevent the computer from going to sleep while in the middle of encoding. More information can be found here https://www.doom9.org/showthread.php?t=169713
Thank you!