Transcoding photo jpeg (.mov) using avisynth

[mofofoto]

I just got QTPro. Looks like 8 channels. Where can I send the clip if I can do it? Isn't there a way to turn off the other channels and re-save the .mov with stereo audio as well?

[poisondeathray]

There are many free hosting sites e.g. mediafire.com , megaupload.com. I don't need a huge sample, just a few seconds.

I don't think you can just disable channels. You have to downmix or it won't sound right, and I don't think QT can do that...

[mofofoto]

Hmmm, I thought I read that was a feature of QTPRO.

Can I save the sample as a reference movie so that it is smaller in size?

[poisondeathray]

Downmix audio in QT? I don't think so, let me know if you figure out how to do it. I'll fiddle with it and try to see if it works...I only use QTPro for cutting/simple editing of .mov files. I hate .mov and I hate QT.

I don't think you can reduce the size unless you re-encode it (lose quality). But aren't you doing that in vdub anyway? What is your end goal ?

[mofofoto]

I agree with you about QT and .mov.

But as far as sending you a sample under 100mb I was wondering how to save it for you hence reference movie save.

End goal is to encode this using avisynth into my encoder at multiple bit rates.

[poisondeathray]

All you do is mark in/ mark out (press I and O keys) around a tiny section, then press edit => trim to selection, file=>save as self contained movie. Only need a tiny few second clip should be ~10mb only . Make sure you be careful and don't screw up your original file! (i.e. don't save it after, or change the name)

I don't think the reference movie option contains the actual video/audio data, hence the need to save as "self contained movie"

[mofofoto]

Sorry for the delay. I had the source redelivered as a .mov 2ch stereo source. However now the audio is blank when I play back. If I drop the .mov straight into my encoder the audio is there but the video is sized incorrectly. Do I need a special script in avisynth for the .mov audio?

[poisondeathray]

QTInput("file.mov", audio=true)

Duh I feel stupid for not reading the documentation! It should work regardless of 6-channel or 2-channel

Regarding the frame size, I have no idea what your encoder does, or if the source is anamorphic or what ? Maybe just resize it 1:1 square pixels in avisynth

[mofofoto]

Still no audio

QTInput("\\blah.mov)
audio=true
converttoyuy()
autocrop(0)
lanczoresize (640,288)


Do I have the script right?

[poisondeathray]

No, audio=true has to go in the QTInput line

QTInput("blah.mov", audio=true)
converttoyuy()
Autocrop()
LanczosResize(640,288)

If you have stereo input now, and want stereo output, no need to downmix

[mofofoto]

Correct I don't need to downmix in this case but I would like to know how for other encodes.

When I do the script your way I get an error in virtualdub:

avisynth open failure
unable to open movie, err=50
line 1

[poisondeathray]

This works for me, on Apple Movie Trailers (1080p 6-ch aac), and camera videos that have m-jpg/lpcm, and audio is passed thru ok

Did you get the paths & filenames correct?

I'm using QTSource 0.0.8 , not the older version

The other option is to use ffdshow , enable m-jpg and lpcm in the configuration and use DirectShowSource(). You may need a proper mp4/mov splitter for this method, I'm using mp4splitter.ax 1.0.0.5 which seems to work on the Trailers and m-jpg/lpcm camera videos

If you can't figure it out, upload a tiny clip

[mofofoto]

I'm turning it into a vc-1, could that be an issue?
paths and file names are right.
I'm just writing qtsource as you have written it. Don't know the version.

I'll look at ffdshow now.

[mofofoto]

sending to your box

[poisondeathray]

OK , I got it

QTInput("miniclip.mov", audio=true)

works , but the audio passed through is stereo, not 8-ch. If you demux it you can probably get the full 8-ch lpcm using the script earlier - I'll check this in a second. The directshowsource() method seems to fail, but you might need a different splitter, I'll tried a few, but all seem to choke, probably because of the 8-ch lpcm stereo.

Try updating your QTSource.dll , and put the avs in the same directory as the file. It's can't be an encoding to VC-1 issue, because you haven't even got past the source decoding stage yet.

EDIT:

This works for 8-ch audio, demuxed with mp4box (yamb is the gui), but you could use ffmpeg to demux as well (but to wav). You have to change the NicLPCMSource() arguments to match your lpcm source, e.g. in this case it was 48000KHz etc....

LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\NicAudio.dll")
vid=QTInput("miniclip.mov")
aud=NicLPCMSource("miniclip_track2.lpcm", 48000, -24, 8)
AudioDub(vid,aud)

For 8-ch audio, downsampled, again audio demuxed. (Not sure about the downmix coefficents, I made them up, but it sounds OK) . You don't have to do this , because QTInput("miniclip.mov", audio=true) seems to downsample to stereo anyway and you don't need the extra demux step

LoadPlugin("C:\Program Files\AviSynth 2.5\plugins\NicAudio.dll")
vid=QTInput("miniclip.mov")
a=NicLPCMSource("miniclip_track2.lpcm", 48000, -24, 8)
fl = GetChannel(a, 1)
fr = GetChannel(a, 2)
c = GetChannel(a, 3)
lfe = GetChannel(a, 4)
bl = GetChannel(a, 5)
br = GetChannel(a, 6)
sl = GetChannel(a, 7)
sr = GetChannel(a, 8)
L1 = MixAudio(fl,sl)
L2 = MixAudio(L1,bl,0.29,0.2)
R1 = MixAudio(fr,sr)
R2 = MixAudio(R1,bl,0.29,0.2)
c_lfe = MixAudio(lfe, c, 0.2071, 0.2071)
L = MixAudio(L2, c_lfe, 1.0, 1.0)
R = MixAudio(R2, c_lfe, 1.0, 1.0)
aud=MergeChannels(L, R)
AudioDub(vid,aud)

Another option is to re-wrap it into AVI container using ffmpeg - AVI container is much easier to use, more compatible. It seems to be fine, but the clip is short to tell for certain, you would have to test on a larger sample

ffmpeg -i input.mov -vcodec copy -acodec copy output.avi

AVISouce("output.avi") will transmit everything as is, including 8-ch audio. You can even open it up directly in vdub without a script. A 2-ch downmix script would be:

vid=AVISource("output.avi",audio=false)
a=AVISource("output.avi",audio=true)

fl = GetChannel(a, 1)
fr = GetChannel(a, 2)
c = GetChannel(a, 3)
lfe = GetChannel(a, 4)
bl = GetChannel(a, 5)
br = GetChannel(a, 6)
sl = GetChannel(a, 7)
sr = GetChannel(a, 8)

L1 = MixAudio(fl,sl)
L2 = MixAudio(L1,bl,0.29,0.2)
R1 = MixAudio(fr,sr)
R2 = MixAudio(R1,bl,0.29,0.2)
c_lfe = MixAudio(lfe, c, 0.2071, 0.2071)
L = MixAudio(L2, c_lfe, 1.0, 1.0)
R = MixAudio(R2, c_lfe, 1.0, 1.0)

aud=MergeChannels(L, R)

AudioDub(vid,aud)

Cheers

[mofofoto]

ok, great. To clarify it should be 2 channel not 8ch. I wonder why my audio isn't passing through then.

I prefer not have to update dll or use ffdshow because it's a lot more work across diff. machines.

This didn't work earlier:


QTInput("blah.mov", audio=true)
converttoyuy()
Autocrop()
LanczosResize(640,288)

[poisondeathray]

Well there are 3 options that work for that test clip, at least on my system

Easiest is to drop the qtsource.dll 0.0.8 into the avisynth/plugins directory, it's autoloading even, and already 2-channel, no need to downmix. You need QT 7.0 or greater for audio, but, the others involve the extra re-wrapping step, and demuxing step (but avi is much nicer to work with IMO). You might be able do a batch process to automate it, but I'm unsure how

http://www.tateu.net/software/dl.php?f=QTSource_Alpha

[mofofoto]

I guess I don't understand why I need qtpro for audio? This would cause a problem since I don't have qtpro on the computers that I would run the encodes.

[poisondeathray]

I guess I don't understand why I need qtpro for audio? This would cause a problem since I don't have qtpro on the computers that I would run the encodes.

It says "quicktime" in the documentation for qtsource, I assumed it means pro, but it might be the free version. Did you get it working ?

Requirements:
Quicktime 6 for video only
Quicktime 7 for audio and video

[mofofoto]

no, not yet.

[poisondeathray]

Did you try the new QTSource.dll yet? or any of the other re-wrap options? I'm not sure what else to suggest, I'm on XP SP2 , and I noticed you are on Server 2003, but I doubt that would make a difference. The AVI wrapper would give you the least problems in terms of compatibility, but it takes a few extra steps

[mofofoto]

it looks like I have a QTSource.dll already from 2006.

[poisondeathray]

it looks like I have a QTSource.dll already from 2006.

The new version I linked is from 2008

Version 0.0.8 - Test - December 6, 2008

[mofofoto]

didn't work. I think I might need to update the dll across all my computers though. I only did it on one.

[mofofoto]

update. I got it to work on 5 min. tests but when i do the full feature I get an error saying:

getchannel: Attempted to request a channel that didn't exist!
Line 6

video = QTInput("\\b\ifs\data\captures_test.mov", audio=true)

flr = GetChannel(video, 1, 2)
fcc = GetChannel(video, 3, 3)
lrc = Mixaudio(flr, fcc, 0.3254, 0.2301)
bl = GetChannel(video, 5)
br = GetChannel(video, 6)
sl = Mixaudio(bl, br, 0.2818, 0.1627)
sr = Mixaudio(bl, br, -0.1627, -0.2818)
blr = Mergechannels(sl, sr)
audio = Mixaudio(lrc, blr, 1.0, 1.0)
audiodub(video, audio)

converttoyuy2()
autocrop(0)
#crop(0,140,-0,-140)
lanczosresize (640,272)


Any ideas.

[poisondeathray]

what does info() say about the channels? that script is for a 6-ch => 2ch.

You could try one of the other examples above that match (or did you change the source type?)

From above, QTInput() already transmited the 8-ch as 2-ch, so there was no need to downsample, hence the error

[mofofoto]

It sounds even though it's for 8 ch. Info confirms it's 8 ch.

Why would it work ok for 5 min. clip but not full feature?

You're saying that qt downsamples as default so there is no need for the script?

[poisondeathray]

I'm saying it did on the sample you sent a while back. NicLPCMSource() passed thru the 8-ch properly.

Not sure what the source is, but its common for some HD streams switch channels for segments like commercials e.g. 2-ch then back to 6-channel.

Are you saying you get 8-ch just from QTInput() without the downsample? If so, use the 8-ch to 2-ch script, not the 6-ch to 2-ch script. And if there is a segment where the audio switches, you have to adjust the script for that segment

If this is the case (channels changing), you could check by scrubbing foward to that point in vdub with info(). You might be able to write a conditional script which take that into account, using conditionalfilter() and getchannels(), to automatically downsample in those sections, but that is beyond my scripting knowledge. You could ask at the Doom9 avisynth subforum.

You still have the option of processing audio separately (e.g. eac3to), as mentioned back on page 1

[mofofoto]

media info says 3 channel but info in avisynth says 8ch. I'm pretty sure it's 8ch.

The script above works fine for tests even though it's written as a 6ch (just throwing away some ch.). But a full movie gives an error. this file does not go back and forth between 2 ch and 8.

I'm wondering why my script only works for small clips and not full length.

[poisondeathray]

QTInput() is still beta, and 8-ch lpcm is not that common so it's entirely feasible that 8-ch lpcm is not supported properly or not passed through properly

If you get the same error message, it definitely has to do with the audio. Line 6 refers to channel 5.

Conversely, NicLPCMSource() is designed to handle LPCM audio. You could use that option, or re-wrap with ffmpeg into avi, or use eac3to. Sample scripts are listed above

Cheers