MeGUI: General Questions and Troubleshooting Thread

[tebasuna51]

Can work also.

But the problem with ffaudiosource (also with ffmpeg) is the output decoder precission: 16 bits int

---[NoImage] input format: Microsoft WAVE Signed 16-bit little-endian 48000 Hz 5.1-channel

BassAudio and NicAudio decoders output 32 bit float samples, and like this is transferred to encoder.

---[NoImage] input format: Microsoft WAVE Floating-point 32-bit little-endian 48000 Hz 5.1-channel

If there are other functions, like downmix or samplerate change, is better also work with 32 bit float.

[Sharktooth]

32bit float has more or less the same "quality" as 24 bit int. if 32bit INT is going to be implemented in ffmpeg/libavcodec i'd prefer 32bit INT over 32bit float.

[tebasuna51]

Originally Posted by Sharktooth

... i'd prefer 32bit INT over 32bit float.

Of course.

But the 32 bit float is the internal format of the decoders and you can't improve the quality with a conversion to 32 bit int.

Also many AviSynth functions can only operate with 16 bit int or 32 bit float but not with 32 bit int.

I think the best option is preserve the same sample format througth the whole process.

BTW, you can implement ffaudiosource in MeGUI for AAC (and other formats not covered by NicAudio), is always better than DirectShowSource and you don't need more AviSynth plugins (BassAudio.dll, Bass.dll, Bass_aac.dll).
If a user want use BassAudio always can use the .avs method.

[LigH]

One of the main advantages of float as internal sample format would also be that you don't need to be afraid of overflows until you finally render the sample down to integer. Having a 24 bit mantissa, it won't lose much precision even for quiet scenes or calculation results (the mantissa is independent of the exponent). And some audio functions are easier to implement if they are based on normalized float values (-1.0 .. +1.0), applying them on integers might require forth-and-back conversion to float.

I speak from my experience of e.g. the "Booster" dynamic compressor in BeSweet, implemented as simple functions on normalized float samples.

[Sharktooth]

thanks for the explanation. i think we will go for the ffaudiosource way for coherency since we already have ffms and mencoder/ffmpeg for other tasks.
let's hope developers will increase the resolution in future updates.

[LigH]

Hooray -- x264 is now also available with integrated audio encoders!

Sh** -- MeGUI doesn't know that, and encodes my manually created custom AviSynth scripts with audio output without adding "--acodec none", so the "qtaac encoder" gets active, but is unstable, and my AVC encode crashes. And furthermore, I would have preferred NeroAacEnc anyway.

[nurbs]

Originally Posted by LigH

Hooray -- x264 is now also available with integrated audio encoders!

When did that happen? I can't see anything on the mailing list or in the git changelog.

[LigH]

MeGUI 0.3.5.24 -- x264 1745-1 (Jeeb)

x264.exe --fullhelp

x264 core:107 r1745+311 2317302
Syntax: x264 [options] -o outfile infile

Infile can be raw (in which case resolution is required),
  or YUV4MPEG (*.y4m),
  or Avisynth if compiled with support (yes).
  or libav* formats if compiled with lavf support (yes) or ffms support (yes).
Outfile type is selected by filename:
 .264 -> Raw bytestream
 .mkv -> Matroska
 .flv -> Flash Video
 .mp4 -> MP4
 .3gp -> MP4 (branded '3gp6')
 .3g2 -> MP4 (branded '3gp6' and '3g2a')

...

Audio:

      Audio options may be used if audio support is compiled in.
      Audio is automatically opened from the input file if supported by the demuxer.

      --audiofile <filename>  Uses audio from the specified file
      --ademuxer <string>     Demux audio by the specified demuxer [auto]
                              Supported and compiled in demuxers:
                                  - auto, lavf, avs, lsmash
      --atrack <integer>      Audio track number [auto]
      --acodec <string>       Audio codec [auto]
                              Available settings:
                                  - auto (select muxer default codec and its default encoder)
                                  - copy (copy source audio without transcoding)
                                  - none (disable audio)
                                Set audio format only and automatically choose encoder
                                  - raw, mp3, aac, ac3, alac, mp2, vorbis, amrnb
                                Force to use specified audio encoder
                                'ff' prefix indicate they are supported via libavcodec
                                  - raw, lame, qtaac, (ff)aac, (ff)ac3, (ff)alac, (ff)mp2, (ff)vorbis,
                                    (ff)libopencore_amrnb
      --abitrate <float>      Enables bitrate mode and set bitrate (kbits/s)
      --aquality <float>      Quality-based VBR [codec-dependent default]
      --asamplerate <integer> Audio samplerate (Hz) [keep source samplerate]
      --acodec-quality <float> Codec's internal compression quality [codec specific]
      --aextraopt <string>    Pass extra option to codec [codec specific]
                              Should be comma separated "name=value" style

      Encoder specific helps:
      * raw encoder help
        Directly pass the decoded PCM samples (in native endian) to muxer.
        All audio options except for --acodec and --audiofile are ignored.

      * lame encoder help
        --aquality        VBR quality [6]
                             9 (lowest) to 0 (highest)
        --abitrate        Enables CBR mode. Bitrate should be one of the values below
                           - for 32000Hz or 44100Hz or 48000Hz
                             32, 40, 48, 56, 64, 80, 96, 112, 128, 160, 192, 224, 256, 320
                           - for 16000Hz or 22050Hz or 24000Hz
                             8, 16, 24, 32, 40, 48, 56, 64, 80, 96, 112, 128, 144, 160
                           - for 8000Hz or 11025Hz or 12000Hz
                             8, 16, 24, 32, 40, 48, 56, 64
        --asamplerate     Output samplerate. Should be one of the values below
                             8000, 11025, 12000, 16000, 22050, 24000, 32000, 44100, 48000
        --acodec-quality  Internal algorithmic complexity [0]
                             9 (poor quality) to 0 (best quality)

      * qtaac encoder help
        --aquality        VBR quality [63]
                          Cannot be used for HE-AAC and possible values are:
                             0, 9 ,18 ,27 ,36 ,45 ,54 ,63 ,73, 82, 91, 100, 109, 118, 127
                          0 is lowest and 127 is highest.
        --abitrate        Enables bitrate mode
                          Bitrate should be one of the discrete preset values depending on
                          profile, channels count, and samplerate.
                          Examples for typical configurations
                           - for 44100Hz or 48000Hz with 1ch
                             LC: 32, 40, 48, 56, 64, 72, 80, 96, 112, 128, 144, 160, 192, 224, 256
                             HE: 16, 24, 32, 40
                           - for 44100Hz or 48000Hz with 2ch
                             LC: 64, 72, 80, 96, 112, 128, 144, 160, 192, 224, 256, 288, 320
                             HE: 32, 40, 48, 56, 64, 80
                           - for 44100Hz or 48000Hz with 5.1ch
                             LC: 160, 192, 224, 256, 288, 320, 384, 448, 512, 576, 640, 768
                             HE: 80, 96, 112, 128, 160, 192
                          The lower samplerate, the lower min/max values are applied
        --asamplerate     Output samplerate
                             LC: 8000, 11025, 12000, 16000, 22050, 24000, 32000, 44100, 48000
                             HE: 32000, 44100, 48000
                          Samplerate greater than input is not supported
        --acodec-quality  Encoder's internal complexity [0]
                             0 (medium), 1 (high), 2 (highest)
        --aextraopt       Profile and bitrate mode
                             sbr  : enable HE-AAC encoding [0]
                             mode : bitrate control mode [abr]
                                    "abr", "cbr", "cvbr"

        --aquality/--abitrate setting may be changed inside codec due to its
        limitations and extreme resampling settings (e.g. 48000->8000) may not work.
        If something goes wrong, it will result in a failure of codec initialization.

      * (ff)libopencore_amrnb encoder help
        Accepts only mono (1ch), 8000Hz audio and not capable of quality based VBR
        --abitrate        Only one of the values below can be acceptable
                             4.75, 5.15, 5.9, 6.7, 7.4, 7.95, 10.2, 12.2

...

[nurbs]

So it's a patch, I see. I didn't check if there was a difference between 1745 and 1745-1 and I never noticed it since my scripts don't have audio. Thanks.

[Sharktooth]

add --acodec none in the custom commandline options of the x264 preset config.

[LigH]

... until it gets officially supported by (or "banned" from) MeGUI.
__

P.S.: Added a tracker task.

[ZZZERO]

It's been a month or so since I encoded something. I used to get around 33fps encoding with x264 to mp4 on both film and interlaced sources. All of a sudden I'm getting less than 10fps with film sources, but my interlaced stuff still encodes at 33fps. Has something changed recently in the way megui handles film sources? Here's one of my scripts that encodes really slowly

LoadPlugin("C:\Program Files\megui\tools\dgindex\DGDecode.dll")
DGDecode_mpeg2source("E:\800FL\800FL.d2v", info=3)
LoadPlugin("C:\Program Files\megui\tools\avisynth_plugin\ColorMatrix.dll")
ColorMatrix(hints=true, interlaced=true, threads=0)
fft3dfilter(interlaced=true)
LoadPlugin("C:\Program Files\megui\tools\avisynth_plugin\TIVTC.dll")
tfm(order=1).tdecimate()
crop( 4, 6, -8, -2)

LanczosResize(576,432) # Lanczos (Sharp)
#denoise


Could it be a cpu utilization problem? I have an i5 quad core and I let x264 automatically set the number of threads. It looks like film encodes are not using a very high percentage of cpu power when compared to an interlaced encode.


Here's a typical script for one of my interlaced encodes.

LoadPlugin("C:\Program Files\megui\tools\dgindex\DGDecode.dll")
DGDecode_mpeg2source("E:\PR23\PR23.d2v", info=3)
LoadPlugin("C:\Program Files\megui\tools\avisynth_plugin\ColorMatrix.dll")
ColorMatrix(hints=true, interlaced=true, threads=0)
fft3dfilter(interlaced=true)
Load_Stdcall_Plugin("C:\Program Files\megui\tools\yadif\yadif.dll")
Yadif(order=1)
crop( 12, 6, -12, -6)

LanczosResize(576,432) # Lanczos (Sharp)
#denoise

[luke83]

Hi to everyone and sorry if my english won't be well but I'm italian

Is possible to add cleanup job manually? I meant without one-click feature.
That's my procedure, I:
1) Open File indexer to make the d2v file
2) Open it with AVS creator and save the AVS script
3) Select the profile in the video encoder settings
4) Set the right bitrate with bitrate calculator and insert it in the profile
5) Click Auto-encode to generate all the jobs automatically.
6) At the end of the job I have to clean up by myself.

These passes works good but I'd prefer cleaning up to start through jobs

Thanks in advance

[Sharktooth]

@luke83: 4 is useless. Auto-Encode window will ask you for a target bitrate OR filesize too.
also, it's not clear to me what you mean with "i have to clean up". clean up what?

@ZZZERO: ensure you're using the same encoding preset as before.

[luke83]

Originally Posted by Sharktooth

@luke83: 4 is useless. Auto-Encode window will ask you for a target bitrate OR filesize too.
also, it's not clear to me what you mean with "i have to clean up". clean up what?

Clean the working folder where the files are (.d2v, .avs .ac3 ecc...)

When I use the "One-Click" encoding, one more job appears at the end named:"Clean up". This job deletes all the file in the working directory.
When I don't use "One-click", Megui doesn't add that final job. In this case the last job is "Muxing".

I want to know the way to add "clean job" manually. So I haven't to open working folders and delete all the files by myself

[ZZZERO]

I'm using the same preset as before. In fact, I'm using the same preset for film encoding and for interlaced material. The only difference between the encodes are these script lines:

Film script:
LoadPlugin("C:\Program Files\megui\tools\avisynth_plugin\TIVTC.dll")
tfm(order=1).tdecimate()

Interlaced script:
Load_Stdcall_Plugin("C:\Program Files\megui\tools\yadif\yadif.dll")
Yadif(order=1)

Both are generated from the MEGUI analyze function.

The film script results in 10fps, the interlaced script gives me 33fps. I don't have an old film script so I don't know if MEGUI has changed some parameter in the scripts it creates, but I do know that I used to get at least 33fps when encoding film sources in the past. Now I'm getting a quarter of that speed and I don't know why.

[Sharktooth]

@ZZZERO: inverse telecine is a more complex process than just deinterlacing. it's pretty normal.

@luke83: the clean up job cant be added outside one-click encoder.

[Carpo]

When looking at a CSI disc in DGIndex and DGIndexNV both say that the disc is interlaced, yet when i go to make the avs file in MeGui it says that the stream is progressive, so which is right? Which is it? should i trust MeGui and let it make a .avs for progressive?

[nurbs]

From the DGIndex manual:

Frame Type - Displays Interlaced or Progressive depending on the value of the MPEG2 progressive_frame flag. Please be aware that this merely describes how the frame was encoded; it says nothing about whether the content of the frame is progressive or interlaced. It is common for progressive video to be encoded as interlaced, and vice versa.

[Carpo]

Thanks Nurbs, helps me to understand a bit more. Yes RTFM is a good thing, maybe i should do that after coffee :-)