Some tests / info related to "making" MP4

[MemTronic]

the following may be of some help for someone...
(I had to figure this out by myself... and it took a while...)

Pure xvid.avi : 573 MB (2 B-Frames / VHQ1 / Chroma M+O / no QPel / no GMC)
Pure audio.aac (MP2) : 121 MB (generated via faacGUI v1.16 all settings: default)
Pure audio.aac (MP4) : 121 MB
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

mp4creator60 v0.9.8.4 EP -c"xvid.avi" -rate=25 -H "video.mp4"
==============================================================
=> MP4ERROR: MP4WriteCountedString: Length is 352 = too large
(seems to be ISMA-related! but this is only a guess...)


mp4UIv0.94 (no ISMA! / no hinted Tracks / optimize on / inerleave on / MTU:-1)
=================================================================
(If ISMA-compliance is enabled it results in a corrupted .mp4 file!!!)

muxed.mp4 (MP4 audio) : 644 MB (should be around 694 MB!)
muxed.mp4 (MP2 audio) : 690 MB (impressed!)

mpegable DS version 1.06 plays both muxed.mp4 flawless...

but shows for MP2 / MP4 audio-properties
----------------------------------------
Channels: Stereo
S. Rate : 44100 kHz
Bits : 16
Bitrate : - kbit/s (yes... empty!)

...MPEG-2 AAC-LC sounds very good!
...but with MP4-Audio I can't hear anything!

===============================================================
muxed.mp4 (xvid.video + MP2 AAC-LC) = 690 MB
muxed.avi (xvid.video + MP3 CBR 128KBit) = 694 MB
-------------------------------------------------
System : AthlonXP 2.1 GHz / Epox 8RDA+ / 2x256MB 333-DDR (twin-bank) / WinXP-SP1
Player : BSPlayer v0.86 Build 493

muxed.mp4 (mpegable DS v1.06) = max cpu-usage : 27% (average: 22%)
muxed.avi (ffdshow 24-04-2003) = max cpu-usage : 24% (average: 19%)

[hans-jürgen]

Originally posted by MemTronic
the following may be of some help for someone...
(I had to figure this out by myself... and it took a while...) Cough... ;)

Pure audio.aac (MP2) : 121 MB (generated via faacGUI v1.16 all settings: default) OK, but please don't call this file MP2, because that would be a totally different thing (MPEG-1/2 Layer II). FAAC produces AAC files with either the MPEG-2 or the MPEG-4 flag in its header simply by choosing the appropriate switch, either in the GUI or on the commmand line.

mp4creator60 v0.9.8.4 EP -c"xvid.avi" -rate=25 -H "video.mp4"
==============================================================
=> MP4ERROR: MP4WriteCountedString: Length is 352 = too large
(seems to be ISMA-related! but this is only a guess...) Hmm, never heard of a length problem for mp4creator... Maybe you should search the MPEG4IP forum for that and ask if you don't find anything about this problem.

mp4UIv0.94 (no ISMA! / no hinted Tracks / optimize on / inerleave on / MTU:-1)
=================================================================
(If ISMA-compliance is enabled it results in a corrupted .mp4 file!!!) As far as I know, this file muxer has not been updated to the latest ADTS header revision of the MPEG, so it will produce garbage with AAC files that signal MPEG-4 in the header.

mpegable DS version 1.06 plays both muxed.mp4 flawless... Wow, more good news every day... ;)

Bitrate : - kbit/s (yes... empty!) I don't know which decoder they use, but the file length/bitrate for the AAC audio part normally is mentioned in the header of a MP4 file. What does MP4UI say or MP4Info (part of the MPEG4IP package)?

...MPEG-2 AAC-LC sounds very good! Thank you... :) Maybe you should lower the quality setting then to -q 90 or even 80, so the audio part will get smaller for the whole MP4 file. But keep in mind that there hasn't been any tuning for lower bitrates than ~128 kbps/stereo (i.e. -q 100) during the development of FAAC yet and that it lacks of important encoding tools like joint stereo or resampling until now.

...but with MP4-Audio I can't hear anything! OK, this might be caused by the outdated MP4UI version. Maybe you should also consider two other authoring tools that are available for free now: the IBM Toolkit for MPEG-4 based on an installed Java Virtual Machine (best results with v1.4 or newer) and thus being independent of your OS and mp4tools from the french ENST. Both apps can even create MPEG-4 Systems content which means that you can use the MPEG-4 features that go beyond simple encoding of elementary video and audio streams (e.g. subtitles). They also can choose from several audio and/or video tracks, thus enabling different language versions within a MP4 file. Both tools work together with their own players (M4Play by IBM, Osmo4 by ENST) and most of the MPEG-4 Systems content created with their authoring tools should also be playable by the EnvivioTV plugin for Windows Media Player, RealPlayer and QuickTime. IBM as well as ENST have also created web forums for user support lately.

http://www.alphaworks.ibm.com/tech/tk4mpeg4/

http://www.comelec.enst.fr/~dufourd/mpeg-4/index.html

[MemTronic]

"but please don't call this file MP2, because that would be a totally different thing"
=> just wanted to be short in this line of description (I name the tools used!)
__ ...and later I write it in full length = MPEG-2 AAC-LC... SORRY for confusion!

"...so it will produce garbage with AAC files with MPEG-4 in the header"
=> Kind of this is mentioned some lines below...
__ (see the filesize and that I can't hear anything)

"Wow, more good news every day..."
=> Hope so... I don't saw any mpegable (playback) tests...

"What does MP4UI say or MP4Info (part of the MPEG4IP package)?"
=> Sorry... not tried out (got the tools of course)

"Maybe you should lower the quality setting then to -q 90 or even 80..."
=> Just wanted to see WHAT comes out with DEFAULTS! ...and was impressed! ...really!
__...and it is really FAST!!! (132 min of audio encoded in 10 min 30 sec) ...hehe

"it lacks of important encoding tools like joint stereo..."
=> I hate joint stereo!!! ...my speakers are wide spread in my room...
__ (can really hear the difference with MP3)


THANKS alot for pointing me to some very interesting places!!!
...especially the IBM section (Java)


EDIT:
I forgot something very important! ... for the xvid.video MPEG-Custom was used (hvs-better.matrix)



so long...

MemTronic

[tiki4]

Just to mention that:

I downloaded the patched version of mp4creator60.exe from RareWares yesterday as well. I had the same strange problems with it. Didn't test it fully, but maybe you have to set the -mpeg-version= switch now in all cases. Did the patch break something in the program?

tiki4

[MemTronic]

"but maybe you have to set the -mpeg-version= switch now in all cases"
=> Sorry but AFAIK = the mpeg-switch is for AUDIO only!
__ I get this ERROR = when trying to convert the VIDEO!



so long...

MemTronic

[MemTronic]

This time with all examples from : http://www.comelec.enst.fr/~dufourd/mpeg-4/index.html
...which really is a very interesting project! (thanks again hans-jürgen)

NONE of them work with mpegable v1.06...

so I only can come to the conclusion: mpegable currently is in a very basic state...


so long...

MemTronic

[symonjfox]

Yes, Patched version of MP4creator60 is buggy.

I turn back to 0.7.8.3 until new relases will be ready

PS: for playback I'd advise you to use 3ivX codec ... yes, sometimes makes some artifacts, but it plays nearly everything and well.

[hans-jürgen]

Originally posted by MemTronic
"...so it will produce garbage with AAC files with MPEG-4 in the header"
=> kind of this is mentioned some lines below...
__ (see the filesize and that I can't hear anything) Kind of, yes, but you don't mention the reason (ADTS header revision for MPEG-4 AAC LC files).

"Wow, more good news every day..."
=> Hope so... I don't saw any mpegable (playback) tests... I once tried the mpegable player (also freeware) and found out that it couldn't play back audio-only MP4 files. The Dicas support confirmed that they don't support this neither in their player nor in the DS filter, so maybe there is a general problem hidden somewhere. But it's good to know that the DS filter can play MP4 files with MPEG-2 AAC audio combined with video content.

"What does MP4UI say or MP4Info (part of the MPEG4IP package)?"
=> Sorry... not tried out (got the tools of course) I forgot to mention that you didn't use the -optimize or -interleave switch with mp4creator which might be the reason for your problems with the file length (just guessing...). MP4UI is more or less just a GUI for mp4creator, and it seems to use these switches on default when I look at your settings.

"Maybe you should lower the quality setting then to -q 90 or even 80..."
=> Just wanted to see WHAT comes out with DEFAULTS! ...and was impressed! ...really!
__...and it is really FAST!!! (132 min of audio encoded in 10 min 30 sec) ...hehe Yes, one of the main advances in the recent FAAC development was a significant speedup (more than 10x faster than v1.10 and still 50% faster than PsyTEL FastEnc, i.e. about on par with Nero AAC with fast settings).

"it lacks of important encoding tools like joint stereo..."
=> I hate joint stereo!!! ...my speakers are wide spread in my room...
__ (can really hear the difference with MP3) It is an important coding method to further reduce the bitrate without loss of quality especially with multichannel files, so it's one of the next things that will be implemented in FAAC. :) It will also help to reduce the artifact that FAAC still shows on some samples, because it's related to stereo image and/or time resolution of the filterbank.

THANKS alot for pointing me to some very interesting places!!!
...especially the IBM section (Java) Did you already test their toolkit? They also have a lot of test samples on their site.

NONE of them work with mpegable v1.06... so I only can come to the conclusion: mpegable currently is in a very basic state... Maybe you should also try the 3ivx DS filter or EnvivioTV which is probably the most advanced plugin for WMP etc. concerning MPEG-4 Systems content.

[hans-jürgen]

Originally posted by tiki4
I downloaded the patched version of mp4creator60.exe from RareWares yesterday as well. I had the same strange problems with it. Didn't test it fully, but maybe you have to set the -mpeg-version= switch now in all cases. Did the patch break something in the program? Which problem, which AAC codec, which settings for which app?

[MemTronic]

"Kind of, yes, but you don't mention the reason (ADTS header revision for MPEG-4 AAC LC files)."
=> You are right!

"you didn't use the -optimize or -interleave switch with mp4creator"
=> I simply forgot to mention the following in the test
__ (was a little tired and must have deleted that lines accidently!)
__ so the parameter settings are the LAST I tried! ...sorry really!
__ and I think that I now understand your "Cough..." or don't I?

TEST-ADD:
1. tried with and w/o -H
2. tried with and w/o -optimize
3. tried with and w/o -interleave (but as -I)
4. tried with and w/o ALL of them...

"Did you already test their toolkit? They also have a lot of test samples on their site."
=> Just finished... hehe

ALL "their" examples work pretty well in IBM Player but read the following...

Converting with IBM-Toolkit-For-Mpeg4-2003-04-25 with ALL SET DEFAULT
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
the already converted .mp4 (with mp4UI no ISMA)
-----------------------------------------------
Adding E:\video.mp4#video
MPEG-4 Video -- 7904.5s, 640 x 272, ISMA profile 1
The output is ISMA profile 1 compliant
Finished

mpegable v1.06 = plays well
Osmo4 = choppy
IBM player = Exception quant must be 0


xvid.video
----------
Adding E:\video.avi#video
MPEG-4 Video -- 75886.8s, 640 x 272, ISMA profile 1
The output is ISMA profile 1 compliant
Finished

choppy playback over mpegable v1.06 and Osmo4 = unwatchable!
IBM player = Exception quant must be 0


Now the AUDIO-Part:
~~~~~~~~~~~~~~~~
Creating D:\audio_MPEG2.mp4
Adding E:\audio_MPEG2.aac
MPEG-4 Audio -- 7904.6s, 44100 Hz, 2 channel(s), ISMA profile 0
The output is ISMA profile 0 compliant
Finished ...and plays well in Winamp ...hehehe
----------------------------------------------
Creating D:\audio_MPEG4.mp4
Adding E:\audio_MPEG4.aac
MPEG-4 Audio -- 0ms, 44100 Hz, 2 channel(s), ISMA profile 0
The output is ISMA profile 0 compliant
Finished ...and 1.924 Bytes in size!!! ...ho ho ho



so long... and at first glance = it still seems to be much room for MP4 development

MemTronic

[Animaniac]

Originally posted by symonjfox
PS: for playback I'd advise you to use 3ivX codec ... yes, sometimes makes some artifacts, but it plays nearly everything and well.

You can set it up to use alternate video decoder filters instead of the 3ivx one. I have it set up to use ffdshow, and everything plays nicely.

[MemTronic]

but this thread concerns to things NOT tested so far...

1. Latest mp4creator / IBM-MP4-Toolkit / Osmo4
2. Latest FAAC / FAAD2
3. mpegable DirectShow filters

...I never wanted to repeat already known things here...
...and to stay with the truth = I "just" wanted to test mpegable...

BUT = "just" creating a valid .mp4 file out of a xvid.video ended in this little test of MP4 creation tools...

AND for me it was really surprising that a creation-tool which is called "outdated" (mp4UI094)
=> was/is the only one producing a "valid"??? .mp4 !!!
(valid = a file which can be played back via actual decoders)

so long...

MemTronic

[hans-jürgen]

Originally posted by MemTronic
"Kind of, yes, but you don't mention the reason (ADTS header revision for MPEG-4 AAC LC files)."
=> You are right! Please, try to remember this part: mp4UI in its current version SUX, IST DEATH!! :devil: You might mail the developer that he has to adopt his GUI to this last MPEG revision, i.e. using mp4creator v0.9.8.3 or newer source code, because I don't have the time today.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
the already converted .mp4 (with mp4UI no ISMA)
----------------------------------------------- See above...

Now the AUDIO-Part:
~~~~~~~~~~~~~~~~
[...]
Creating D:\audio_MPEG4.mp4
Adding E:\audio_MPEG4.aac
MPEG-4 Audio -- 0ms, 44100 Hz, 2 channel(s), ISMA profile 0
The output is ISMA profile 0 compliant
Finished ...and 1.924 Bytes in size!!! ...ho ho ho See above...

so long... and at first glance = it still seems to be much room for MP4 development See above... ;)

[MemTronic]

ahem... this time I seem to NOT get the point...

@Hans-Jürgen...

"mp4UI in its current version SUX, IST DEATH!!"
=> Everybody tells that... I believed that... and nearly went crazy trying out every parameter-combination with mp4creator... hehe

BUT... the REST of your post = I am too dumb / tired (14 hours awake and reading about "Xmto" since 3 hours) to understand...

WHAT has the video.avi conversion via IBM-MP4-Toolkit to do with mp4ui ???

WHAT has the Audio-conversion via IBM-MP4-Toolkit to do with mp4ui ???


seems I am too tired...


so long...

MemTronic

[hans-jürgen]

Originally posted by MemTronic
"mp4UI in its current version SUX, IST DEATH!!"
=> Everybody tells that... I believed that... and nearly went crazy trying out every parameter-combination with mp4creator... hehe OK, today I had the time to reread this thread and found that there might be a switch you've overlooked in mp4creator:
-use64bits Use for large files

Does your long file work with that switch perhaps? You have to keep in mind that MPEG4IP is mainly aiming at streaming MPEG-4 content over the internet (since Cisco is the company behind this project ;) ). So maybe they do not care too much what happens with their tools while ripping a movie to MP4. Another switch that belongs to this purpose of streaming is -hint which you will definitely not need, because it prepares the MP4 file for streaming with RTSP instead of HTTP.

BUT... the REST of your post = I am too dumb / tired (14 hours awake and reading about "Xmto" since 3 hours) to understand... Maybe you would like to have a look at the Wiki of http://www.audiocoding.com/ when you're "on board" again, e.g. on the pages for "AAC", "MP4" or "mp4creator", also "PsyTEL". There I tried to summarize the problems with different MPEG-x headers that have come up during the last months.

WHAT has the video.avi conversion via IBM-MP4-Toolkit to do with mp4ui ???
WHAT has the Audio-conversion via IBM-MP4-Toolkit to do with mp4ui ??? The video conversion is OK, as far as I know (I don't know much about the video part, as I'm all interested in AAC/MP4 audio for now). But you wrote in your last post that you took the already damaged MPEG-4 audio file muxed with mp4UI again and tried to use it with the IBM toolkit which makes no sense, because the IBM muxer can't repair anything that mpUI has broken. Remember the small file size with MPEG-4 AAC? That was caused by the outdated mp4UI (or better: the outdated mp4creator source code inside of mp4UI) dropping many bytes while muxing, because it couldn't read the new MPEG-4 ADTS headers and so threw away too much stuff. That's also the reason why mpegable can't decode this damaged audio part and so you can't hear anything.

seems I am too tired... I was too tired and/or busy yesterday to inform the mp4UI developer about these issues, but I've sent a mail to him today, so maybe there will be a new version soon.

[tiki4]

Sorry hans-jürgen,

I was away for a while. I didn't find time yet to do some further tests with the patched version of mp4creator60. I just did a quick test with a transcoded music video from a DVD (3 min or so). Movie was XviD with B-frames but no GMC or QPel. Audio was decoded with BeSweet 1.5b18 to stereo WAV. I encoded the audio with NERO 5.5.10.28 AAC Plugin to MP4 (MPEG4) and AAC (MPEG2) and extracted the audio from the MP4 with mp4creator60 0.9.8 (from RareWares). Then I tried to mux both again to an MP4 container including the video. Somewere there I got this strange problem. I will test that again this evening and try to further define where the problem lies. I'll post what I found tomorrow.

tiki4

[MemTronic]

OK, today I had the time to reread this thread and found that there might be a switch you've overlooked in mp4creator:
I have tested this switch yesterday at night... and the SAME ERROR!

The video conversion is OK, as far as I know (I don't know much about the video part, as I'm all interested in AAC/MP4 audio for now).
YES... looks like to me...
I tried to rell you about VIDEO-conversion ...hehe
I tried to tell you = that (at least to me) mp4Ui is/was the ONLY app producing a PLAYABLE .mp4 VIDEO!!!

With every other app (mp4creator60 EP / IBM-Toolkit) = I was NOT able to produce a PLAYABLE video.mp4 file!

mp4creator = ERROR (see above)
IBM-Toolkit = NO error but choppy like hell PLAYBACK (see above)

so I made a AUDIO-only conversion ...and IBM-Toolkit FAILS like mp4UI


that was it I tried to provide as INFO... but it seems (to me) YOU were also a little tired when reading this thread... :D


...and really thanks a lot for the 2 links...
I am very interested in this "advanced" features of MP4 and OSMO4 looks promising!



so long...

Memtronic

[MemTronic]

after testing with overwrite -mpeg4-video-profile=3 also and ending up wit hthe same ERROR...

I tried the "logical" combination of 2 tools...

FINAL RESULTS to produce a valid .mp4 with actual tools:

Mp4creator60 v0.9.8.4 EP used to generate valid .mp4 containing MPEG4-Audio
=> Mp4creator60.exe -c"audio_MP4.aac" -rate=25 -I -optimize "d:\video_MPEG4.mp4"

(think the rate param is obsolete but was a leftover and it works)


mp4UI v0.94 used to ADD xvid.video.avi to the MPEG4-Audio.mp4
(settings see first post)

the FINAL.mp4 (containing MPEG4-Audio + xvid.video) = 690 MB

mpegable = plays flawless WITH sound (nice it works!)
IBM-Player = plays sound well but I can't see video
OSMO4 = does not hang but I can't see video nor hear any audio

EDIT:
3ivx D4 v4.04 = plays FINAL.mp4 flawless with sound
(as Animaniac pointed out (thanks) = if you set the 3ivx media splitter to "allow unsuported decoders" ffdshow can be used for video playback
and 3ivx's AAC-Decoder is used for the MPEG4-Audio...

...hope this brings up some clearance related to LATEST efforts in using MPEG4-Audio in .mp4


so long...

MemTronic

[tiki4]

Well, in order to complete my reports,

I get the above described error with the patched version of mp4creator when I try to add video to an MP4 file. I can do everything with audio, but adding video blows the app. Video was always XviD with b-frames. I don't know where the problem is, but I don't have any problems with the RareWares' mp4creator 0.9.8. As I said, it happens if you try to add video to a new file or to a file already containing audio (produced by latest Nero). My command lines were something like:

mp4creator60 -create=out.avi -rate=25 file.mp4

or (if I add to an existing file with audio):

mp4creator60 -create=out.avi -rate=25 -interleave file.mp4

After that the file is destroyed. I can't open it in any player and 'mp4creator60 -list file.mp4' fails with 'Can't open file.mp4'.

tiki4

[hans-jürgen]

Originally posted by MemTronic
Pure xvid.avi : 573 MB (2 B-Frames / VHQ1 / Chroma M+O / no QPel / no GMC)
~~~~~~~~~~~~~~~~~

mp4creator60 v0.9.8.4 EP -c"xvid.avi" -rate=25 -H "video.mp4"
==============================================
=> MP4ERROR: MP4WriteCountedString: Length is 352 = too large
(seems to be ISMA-related! but this is only a guess...)

mp4UIv0.94 (no ISMA! / no hinted Tracks / optimize on / inerleave on / MTU:-1)
==============================================
(If ISMA-compliance is enabled it results in a corrupted .mp4 file!!!)
OK, to come back to the original problem: Did you also try to mux your xvid.avi to MP4 with an older unpatched version of mp4creator like tiki4 did? And if yes, did it work or not?

Your assumption that the ISMA option (only available in mp4UI) might have something to do with it could be true, because I found a related thread in the MPEG4IP forum when searching for "MP4WriteCountedString" there:

http://sourceforge.net/forum/message.php?msg_id=1848860

It seems that there are two possible reasons for this error message, one being that the ISMA compatibility implemented in the mp4creator source code can't handle long strings, and so the advice is to disable this part in the mp4creator code for now, until a fix is available for this problem. Maybe the ISMA option in mp4UI does exactly this, I don't know. And maybe older mp4creator versions also had this part disabled while the new patched version has not, I don't know either. But Enrico Palmeri will surely answer this, as soon as his (ridiculous) 5-days waiting period for Doom9.org access is over... ;)

The other reason why .xvid and/or .divx files won't work with mp4creator is that they can contain other video streams than plain MPEG-4 (which was the first version of DivX, as far as I know). So the simple solution for this is to rename a .divx file to .avi, and mp4creator should be able to mux it to MP4 without further problems then. But in your case this doesn't work of course, because you already have an .avi here. I only mention it for other readers that might have stumbled over this issue in the past.

Another remark to the other options on your quoted mp4creator command line: -rate=25 is only correct if your video has a frame rate of 25 of course. You can get this information with avidump, another tool from the MPEG4IP package. This switch is one of the three essential parameter for mp4creator when muxing a video file (besides input and output file names), so you should find out what video frame rate your file has before muxing the MP4 container.

And to hint a track (video or audio) with the -h switch only makes sense if you plan to stream the MP4 file either with RTSP or HTTP, because this adds time stamps to each track and so increases the file size which is unnecessary for plain movie ripping. Furthermore you should also add -interleave for the audio track if you decide to hint it, because this adds a payload in order to synchronize the audio with the video stream.

The last step you should not forget is to optimize the complete MP4 file, because this works like a "virtual defragmentation" for a large movie file that puts the important header informations at the beginning of the MP4 container, so reading this file from a HDD and/or streaming it will be easier and with less or no choppy playback.

Many of these informations can be found in one of the available docu files and/or articles of the MPEG4IP project, but I agree that they are hard to find sometimes, especially the ones hidden in the CVS directories. ;) I will probably add a list of the docus/guides/articles known to me in the Sticky thread about "important links" soon.

[MemTronic]

but this lukilly is NOT true! ...hehe

WHY?

THIS is what the author of mpUI has answered to my mail:
"thx. hope that i have this week the time to update mp4UI. let's see...

markus

p.s. really don't know why u get an error with mpeg4ip but not with mp4UI. i didn't modify the source. probably an error in the current release."
...and SO I was not searching any further like you Hans Jürgen...
(If someone who has deeper knowledge of the source-code doesn't know...)


Did you also try to mux your xvid.avi to MP4 with an older unpatched version of mp4creator like tiki4 did? And if yes, did it work or not?
=> YES... but didn't post because for a longer time there only was silence here...

Ok... tried MPEG4IP_0.9.7.7 with the same ERROR...


...and so I only can say:
THANKS a lot Hans-Jürgen for your efforts...
...and I will post again some "test" results AFTER mp4UI IS UPDATED!



so long...

MemTronic

[symonjfox]

Originally posted by hans-jürgen
Furthermore you should also add -interleave for the audio track if you decide to hint it, because this adds a payload in order to synchronize the audio with the video stream.

The last step you should not forget is to optimize the complete MP4 file ...
Hi, I've a small question about it:

Should I use -interleave and after -optimize or just -optimize?
I tell you this because I never used interleave, and AFAIK it should work just with AAC audio. Am I right?

[hans-jürgen]

Originally posted by symonjfox
Should I use -interleave and after -optimize or just -optimize?
I tell you this because I never used interleave, and AFAIK it should work just with AAC audio. Am I right? Yes, -i or -interleave only makes sense for a hinted AAC audio track, but probably doesn't hurt when used together with the video track (it will possibly be ignored then). But it's important to use -optimize as the last step for the complete MP4 file, not at the beginning of the muxing process.

[hans-jürgen]

Originally posted by MemTronic
"p.s. really don't know why u get an error with mpeg4ip but not with mp4UI. i didn't modify the source. probably an error in the current release."
[...]
Ok... tried MPEG4IP_0.9.7.7 with the same ERROR... So this contradicts the theory that the patching of Enrico Palmeri would have broken something in the source code of mp4creator. As far as I know, he did not change anything in the video part and only added the -mpeg-version switch for the audio part. But of course there have been several updates from the MPEG4IP developer himself that might be the cause for your problem. Another possibility could be compiler issues again: Roberto uses VC6 for all MPEG4IP tools, as far as I remember, while Markus uses Visual Studio.NET for mp4UI (which is the reason I can't use his app on Win95).

...and I will post again some "test" results AFTER mp4UI IS UPDATED! Yes, and maybe Enrico can provide some additional results soon, so we can figure out what is going on.

[hans-jürgen]

Originally posted by MemTronic
OSMO4 = does not hang but I can't see video nor hear any audio I forgot to mention that this is probably caused by the lacking AAC plugin for Osmo4 that is based on FAAD2, too. You can download it from RareWares now, look into the AAC section.

EDIT:
3ivx D4 v4.04 = plays FINAL.mp4 flawless with sound (as Animaniac pointed out (thanks) = if you set the 3ivx media splitter to "allow unsuported decoders" ffdshow can be used for video playback and 3ivx's AAC-Decoder is used for the MPEG4-Audio... Another possibility would be the new CoreAAC DS filter, also available at RareWares now. Its advantage is working multichannel support (with FAAC multichannel encodings), because it's based on FAAD2 and not on FAAD like the 3ivx audio decoder.

By the way, yesterday I've edited my posting about important MPEG-4 links in the Sticky thread, so everyone should be able to find some documentation files and places for creating MPEG-4 content now.

[MemTronic]

(worked on the latest HTML-DB for eMM to add opera-support [opera IMHO still sucks a lot])

caused by the lacking AAC plugin for Osmo4
=>Yes... I have figured it out (by reading the readmy.txt after a while ...shame)

You can download it from RareWares now
Sure will have a look (right after this post)

CoreAAC DS filter, also available at RareWares now
THIS one I will (of course) try out also!
...but I really believe = waiting until mp4UI update brings up more room to the CoreAAC developers to improve that nifty thing...



so long...

MemTronic

[ChristianHJW]

Originally posted by MemTronic THIS one I will (of course) try out also! ...but I really believe = waiting until mp4UI update brings up more room to the CoreAAC developers to improve that nifty thing...

We currently dont see any need to update the filter, as we hardly get any sensible user response if it performs well for them or not. One guy was reporting that 5.1 AAC didnt work from MP4 container correctly, there were problems if he was trying to seek in the file. Normal playback and muxing worked fine for him together with the 3ivX MP4 parser/splitter filter, and seeking in normal Stereo files was fine also. We dont think the filter can be to blame for the seeking problems, as its working fine for 5.1 AAC in MKV files.

I personally expect that the long awaited Nero Digital (http://www.nerodigital.com/) release, including an AAC encoder plugin for Nero with enhanced 5.1 support ( menno and Ivan said 160 - 180 kbps is achievable for a 5.1 movie soundtrack, and shall be undistinguishable from the source AC3 ) will boost its usage by a high percentage ( hopefully ).

Christian

[bond]

Originally posted by ChristianHJW
We currently dont see any need to update the filter, as we hardly get any sensible user response if it performs well for them or not. Nero Digital will boost its usage by a high percentage ( hopefully )i am sure that coreaac will become very important in the future as an opensource aac decoder when people will start using aac as their preferred audio format.

i think until now there are 3 major steps (and a long way) still to go:
1) implementation of aac+ (for example in nero) to allow 64kbs movie-soundtracks ;)
2) fine tuning of aac (comparable to lame)
3) a 100% stable container format which can handle aac probably. as there arent any news from the .ogm side and .mp4 isnt really supported in most of the apps, matroska will be the way to go ;)


so plz dont forget coreaac and dont let the project die just after its birth...

[tiki4]

Well,

so far, CoreAAC performs really well. There seem to be some problems with the 3ivx splitter, but I think they'll get eventually solved. Main problem at the moment seems to be that the encoders aren't up to 5.1 AAC. I tried FAAC binaries from RareWares (1.16) and wasn't too happy. Nero seems to have matrix coupling bug (I think hans-jürgen said s.th. like that) and the output doesn't sound too nice. I wasn't able to compile faac cvs version in Windows so far as some of the source files in the common subdirectory are missing (same for faad2). It compiled in Linux, but I didn't test it yet and also faad2 again didn't compile. No need to go into details with that right now. I patiently wait for the next version of Nero with a hopefully fixed WAV input plugin and AAC encoder plugin (there must be a reason I paid money for the plugin, though).

Take care,

tiki4

[hans-jürgen]

Originally posted by MemTronic
...but I really believe = waiting until mp4UI update brings up more room to the CoreAAC developers to improve that nifty thing... I just read that mp4UI will probably be updated on this weekend to the latest ADTS header revision and the added -mpeg-version switch in the RareWares version of mp4creator, so keep an eye on its homepage: http://www.mediacruiser.de/mp4UI/

[hans-jürgen]

Originally posted by tiki4
Main problem at the moment seems to be that the encoders aren't up to 5.1 AAC. I tried FAAC binaries from RareWares (1.16) and wasn't too happy. Did you update the Nero WAV input plugin in your version? Otherwise it won't recognize multichannel WAV files at all. What kind of problems did you have/hear with FAAC v1.16, which settings did you use?

for Nero seems to have matrix coupling bug (I think hans-jürgen said s.th. like that) and the output doesn't sound too nice. In fact Ivan said this on Audiocoding.com and also confirmed that PsyTEL has the same bug, I only repeated this.

[tiki4]

Well,

I referred to the same thing in both sentences. I replaced the WAV input plugin of NERO. There is unfortunately a problem of reading WAV files that are larger than 2GB. So you have to split a full movie into more than one part to get it into NERO. Then I tried something like 256 KBit CBR and 'streaming' VBR. The result didn't sound too nice. Something like dumb and hollow (sorry, I don't know better English words).

FAAC on the other hand doesn't make it easier to control the bitrate. I didn't get the full sense of that quantizer quality switch which seems to range from 0 to 100? I was up on compiling a CVS version of FAAC but it didn't compile as is in Windows. It compiled in Linux but I just was too lazy to compare for now. I still hope for the fixed NERO plugin from Ivan.

Cheers,

tiki4

[hans-jürgen]

Originally posted by tiki4
I referred to the same thing in both sentences. I replaced the WAV input plugin of NERO. There is unfortunately a problem of reading WAV files that are larger than 2GB. So you have to split a full movie into more than one part to get it into NERO. Oops, I hope Ivan is aware of that one, too...

Then I tried something like 256 KBit CBR and 'streaming' VBR. The result didn't sound too nice. Something like dumb and hollow (sorry, I don't know better English words). OK, so you weren't happy with the sound of the Nero AAC plugin, I see... Did you use a multichannel WAV for this test? Because the older Nero AAC codecs are all buggy and mix up the channels. Only the newest Nero versioin from Saturday (available on their FTP server) has the fixed plugin.

FAAC on the other hand doesn't make it easier to control the bitrate. I didn't get the full sense of that quantizer quality switch which seems to range from 0 to 100? No, you can also go higher than -q 100 if you want, e.g. -q 125 should result in ~190 kbps with a normal sounding stereo input file with 16bit and 44.1 kHz sample rate. I created a new Wiki page for FAAC yesterday that should explain the different switches a little bit better than the help screen:

http://www.audiocoding.com/wiki/index.php?page=FAAC

I was up on compiling a CVS version of FAAC but it didn't compile as is in Windows. It compiled in Linux but I just was too lazy to compare for now. I still hope for the fixed NERO plugin from Ivan. The compiling errors are a main theme at the Audiocoding.com forum, that's why I provided a shortcut on the FAAC page to a forum search. And I also put an excerpt from an older Readme.txt there concerning CR/LF issues, but this will probably not solve your compiling problem. Thanks for reporting this anyway.

[Enrico Palmeri]

Could you send me a small sample file which causes the reported bug with my special version of mp4creator?
plese include the command-line which causes the bug to occur.
Regards
Enrico Palmeri

[tiki4]

@hans-jürgen:

Downloaded new Nero today, will test again...

tiki4

@Enrico:

By the way, the bug is also there in the DeXt compiled version of mp4creator60.exe 0.9.8.4.

[hans-jürgen]

Originally posted by tiki4
By the way, the bug is also there in the DeXt compiled version of mp4creator60.exe 0.9.8.4. And what source code did he use for his compile, the one from the MPEG4IP CVS (without added -mpeg-version switch) or the one with Enrico's patch?

[MemTronic]

have send a mail to Enrico yesterday evening...

NO answer until yet...

(so that no one starts to think I am no longer interested in this problem to be fixed)


so long...

MemTronic

[Enrico Palmeri]

I have never received any question?

[tiki4]

@hans-jürgen:

I didn't compile, DeXt compiled it. It's here (http://peque.metropoliglobal.com/dext/). I didn't test this compile very long because it gave the same error when adding video to a new mp4-file like:

mp4creator60 -c=video.avi -rate=25 temp.mp4

By the way: New NERO gives quite good audio quality when transcoding from a 6 channel WAV input. But why didn't they change the WAV plugin? One still has to backup the old and install the one from rarewares.

Cheers,

tiki4

P.S: One can playback multichannel AAC with the newest CoreAAC.ax compiled by John33 and using the 3ivx splitter for MP4. However seeking is still broken, I hope they can fix this. Matroska makes still problems on my machine. The DS filter produces funny output with XviD video and the latest mkvmerge seems not to work at all on my XviD AVI files. Hm. Let's try Gabest's new splitter filter...

[hans-jürgen]

Originally posted by MemTronic
have send a mail to Enrico yesterday evening...

NO answer until yet... Was this a very long mail with a file attachment of your xvid.avi perhaps? Then maybe his mail account hasn't accepted it...

(so that no one starts to think I am no longer interested in this problem to be fixed) By the way, yesterday Markus Brenner released a new version of mp4UI, so please test your problem files (both video and audio) with that one, too.

The new v0.95 can now use Enrico's added -mpeg-version switch as well as the new -old-aac-file-format switch which enables to mux non-standard compliant AAC files produced with PsyTEL AACEnc_MPEG4 (58 bits ADTS header instead of the new 56 bits) into a working MP4 container file. And the bug with extracted AAC files from a MP4 container should also be fixed, the old version produced *.bin files instead of MPEG-2 AAC LC files when using the batch mode for this extraction.

[MemTronic]

I have never received any question
=> I used the one from your profile of this forum!
__ (so I can't see the address and the forum acted the way it would work)

just have send you a simple text-mail to which you can respond to...
...and attaching files to this mails is impossible?


yesterday Markus Brenner released a new version of mp4UI, so please test your problem files
=> SURE I will do that! ...thanks hans-jürgen...


EDIT:

First impressions (tests?)...

Test.video (for settings see earlier posts here) = 605,0 MB
Test.aac (made with faac v1.16 as MPEG-4) = 89,2 MB


mpUI v0.95
==========
ISMA : OFF
Optimization: ON
AAC Mode : Original
Use Interl. : Checked
MTU : -1


RESULTS:
~~~~~~
with add HINT tracks:
FINAL.mp4 = 716 MB and NONE-watchable [no picture / sound only] (using last setup as mentioned above)

NO HINT tracks:
FINAL.mp4 = 691 MB and PLAYS FLAWLESS and SEEKING is NO PROBLEM (same like with .avi) !!!


TIP:
HINT tracks are required ONLY for STREAMING purposes (so leaving them off saves a lot MB!)



so long...

MemTronic

[hans-jürgen]

Originally posted by tiki4
I didn't compile, DeXt compiled it. I know, but you can tell me if the help screen shows the "-mpeg-version" option or not. ;)

I didn't test this compile very long because it gave the same error when adding video to a new mp4-file like:

mp4creator60 -c=video.avi -rate=25 temp.mp4[/B] I get a little bit irritated when you use the word "like" while describing command lines: is this the command line you used and which failed or not? Because I already mentioned that the -h switch should not be used for movie ripping, so if you overlooked it it may be one reason for your problems (see also MemTronic's last test with the new mp4UI). The same is true for the video frame rate: only the one used in the source AVI file should be set with -rate, if you don't know it, then don't set this switch. And sorry for being such a nitpicker, but it's almost impossible to rule out errors in the apps vs. errors made by users if they don't tell exactly what they did.

By the way: New NERO gives quite good audio quality when transcoding from a 6 channel WAV input. But why didn't they change the WAV plugin? One still has to backup the old and install the one from rarewares. I wish I knew that, too...

P.S: One can playback multichannel AAC with the newest CoreAAC.ax compiled by John33 and using the 3ivx splitter for MP4. However seeking is still broken, I hope they can fix this. I don't know why that DS filter even has to use an external splitter, because the 3ivx solution is based on OpenQuickTime, as far as I know, so it should be possible to integrate it without the need for just another DS filter. And like I already mentioned, MPEG4IP (mp4creator, mp4player) and Osmo4 (player) / mp4tool (authoring tool) from the ENST are all available as open source, so there are other solutions for splitting the bitstreams of a MP4 container, too, that probably stick to the MPEG-4 standard at least as well as 3ivx.

[hans-jürgen]

Originally posted by MemTronic
Test.video (for settings see earlier posts here) = 605,0 MB
Test.aac (made with faac v1.16 as MPEG-4) = 89,2 MB By the way, the latest v1.17 with an added ABR option is available at RareWares. It also sounds better, not sensational, but a little bit.

mpUI v0.95 And Markus has updated this version to 0.95a, it seems. The readme only mentions "small bugfixes", and the MPEG4IP source code he used is v0.9.8.5 from the CVS.

ISMA : OFF I think here's probably the reason why your problem xvid.avi is accepted by mp4UI, because mp4creator doesn't have such an option. Maybe I should mail Markus again...

Use Interl. : Checked Should be unnecessary, only for hinted audio tracks

with add HINT tracks:
FINAL.mp4 = 716 MB and NONE-watchable [no picture / sound only] (using last setup as mentioned above) And does the mpegable player/DS filter still show this error, or is it working now, too?

NO HINT tracks:
FINAL.mp4 = 691 MB and PLAYS FLAWLESS and SEEKING is NO PROBLEM (same like with .avi) !!! Fine... and thanks for testing! :)

[tiki4]

@hans-jürgen:

Sorry, but I have to comment that.

When I write 'like' I just meant that I have used exactly that command line, except for the name of the AVI file (I can't remember that and it doesn't matter). I live in the PAL world, so I always have 25 fps source and set -rate=25. I never used hint and I saw that it isn't necessary from your posts. I also gave up '-interleave' because of your comments but it never made a difference. I will have a look again, if that '-mpeg-version' switch is there, but I didn't check because problems came up when adding video in the first place, so I kicked it off my system pretty fast.

CU,

tiki4

[shitowax]

Originally posted by hans-jürgen
I don't know why that DS filter even has to use an external splitter, because the 3ivx solution is based on OpenQuickTime, as far as I know, so it should be possible to integrate it without the need for just another DS filter. And like I already mentioned, MPEG4IP (mp4creator, mp4player) and Osmo4 (player) / mp4tool (authoring tool) from the ENST are all available as open source, so there are other solutions for splitting the bitstreams of a MP4 container, too, that probably stick to the MPEG-4 standard at least as well as 3ivx.

The 3ivx splitter isn't a simple MP4 splitter as it can also split all derivative of Quicktime files (.mp4, .mov, .m4a, .3gp, ...)
Now, you're right, nothing prevent anybody to recode from scratch a splitter using mpeg4ip, OQT or whatever ... You could also code a "all in one" filter like Envivio ... I don't recomand it tho, as it is not in the coding "spirit" of DirectShow ;)

[MemTronic]

latest v1.17 with an added ABR option is available at RareWares
=> and again something for testing... ho ho

And Markus has updated this version to 0.95a
=> and again something to test... mhhh...

I think here's probably the reason why your problem xvid.avi is accepted by mp4UI, because mp4creator doesn't have such an option
=> YES! ...I come to the same conclusion ...hehe

Should be unnecessary, only for hinted audio tracks
=> It sure does NOT hurt... and as long as I don't know for sure...

And does the mpegable player/DS filter still show this error, or is it working now, too?
=> SORRY! ...deinstalled it and not tested again until yet!

and thanks for testing!
=> I am as much interested in those things working as you! ...hehe
__ SO = thanks go back to you!

...and I really don't know when I have enough time again for further testing...
(currently working on a user friendly HTML-Cruncher-Obfuscator-Compressor ...hehehe)

so long...

MemTronic

[hans-jürgen]

Originally posted by MemTronic
"latest v1.17 with an added ABR option is available at RareWares"
=> and again something for testing... ho ho Well, if you are interested in the ABR option, I have to tell you that it doesn't work yet. ;) I just found it out today doing a comparison between FAAC, PsyTEL, Lame, Musepack and Vorbis at ~128 kbps with one sample. But like I mentioned before, FAAC v1.17 sounds a little bit better than v1.16 (not much) which is already good enough to run circles around MP3 and OGG at this bitrate and with the sample I used (complex symphonic music with a lot of stereo ambience). I even had to test it at -q80 (because I didn't believe it), and that still sounds quite good (~100 kbps at 16 kHz). So for multichannel AAC sound this is really a good (and free) alternative, because the one artifact I can hear with my other standard test sample will probably not show up in multichannel mode, since there is a separate centre channel that will take over all mono-like signals from the stereo image (which are causing the problem in normal stereo mode until now).

[MemTronic]

Well, if you are interested in the ABR option, I have to tell you that it doesn't work yet.
=> 1. nice that I don't have to figure this out myself! ...hehe
__ 2. well... good development sure needs it time! ...so for the moment I think people can live without ABR (not that it is a "bad" !!!)

I just found it out today doing a comparison between FAAC, PsyTEL, Lame, Musepack and Vorbis at ~128 kbps
=> For music (mostly ambient / synthetic sounds) = I used Musepack at about 192KBit/s until today... (too lazy to compare Vorbis with it!)

So for multichannel AAC sound this is really a good (and free) alternative
=> THIS "sounds" really very interesting!!! ...think NOT only for me! ...hehe

...after I have finished the most important parts of my current project = I'm interested in investigating that further!

...but sure will follow this thread in the meantime...


so long... and keep up busy

MemTronic

[hans-jürgen]

Originally posted by MemTronic
2. well... good development sure needs it time! ...so for the moment I think people can live without ABR (not that it is a "bad" !!!) Knik only implemented it because of my nagging for some kind of CBR/ABR setting that seems to be needed by the MPEG4IP people for live streaming purposes. I'm still not sure why they think they need this, because the MP4 container is capable of holding and transporting VBR frames (and AAC always was a VBR format, too). There must be some kind of buffering in the transport layer, i.e. RTSP probably, and in the player app to compensate for any huge jumps in the bitrate, but AAC also uses modern coding tools to prevent such jumps and doesn't produce unexpected results even with VBR, so this danger is not as big as e.g. with MP3 or MP2/MPC.

By the way, using the ABR switch in FAAC 1.17 doesn't hurt at the moment, because it simply will be ignored (although signaled in the DOS box) and FAAC encodes with its default VBR setting at -q 100 -c 16000 with exactly the same result as not using the ABR on the command line (FAACGUI doesn't show this new switch anyhow). With my Mahler test sample it resulted in 122 kbps/stereo, while Lame 3.92 with --alt-preset 128 thought 121 kbps would be enough and Musepack at --radio needed 144 kbps, PsyTEL with -streaming wanted 138 kbps and Vorbis with -q 3.99 used 127 kbps.

=> For music (mostly ambient / synthetic sounds) = I used Musepack at about 192KBit/s until today... (too lazy to compare Vorbis with it!) Well, for my taste you can spare the time anyway, because it shows its typical hard and noisy sound at any bitrate while narrowing the stereo image even at 128 kbps and not only at lower bitrates (maybe -q 6 will prevent this, because they switch to L/R stereo then, as far as I know). Musepack on the other hand did sound very good with the --radio preset (I'm only talking about one sample here), so if you can live with its unexpected high bitrates for mostly tonal content and with the non-existing support by the industry (not part of any standard or hardware yet), you can also use it for bitrates at ~128 kbps. I don't know how the situation is for multichannel support, but I think this might happen in that legendary Stream 8 version while I'm not sure that it can keep up with AAC at lower bitrates than 128 kbps/stereo which are in need for several channels of course.