SLUG_Wiki: AviToVob

Notes useful for those wishing to convert arbitrarily formatted .avi files to .VOB files and then using these to create a DVD playable on any commercial DVD player.

Toolchain

Transcode: http://www.transcoding.org

ffmpeg: http://ffmpeg.sourceforge.net/ (when building, configure to use and make shared libs!!)

Mplex: mplex is a part of mjpegtoolshttp://inferno.slug.org/cgi-bin/wiki?action=edit&id=AviToVob http://mjpeg.sourceforge.net

Dvdauthor: http://dvdauthor.sourceforge.net

cdrecord/cdrtools: ftp://ftp.berlios.de/pub/cdrecord/alpha/

growisofs/DVD+RW-tools: http://fy.chalmers.se/~appro/linux/DVD+RW/tools/?M=D

Mencoder: mencoder is part of mplayer http://www.mplayerhq.hu

Sox: http://sox.sourceforge.net

MKVToolnix: http://www.bunkus.org/videotools/mkvtoolnix/

OGMTools: http://www.bunkus.org/videotools/ogmtools/

The Script

 #!/bin/sh

 # ***************************************************************
 # This is a batch processing script for normalizing and converting
 # a mixed collection of .avi files into .mpg files that can be fed
 # to dvdauthor to create dvd's that will play perfectly on nearly
 # all NTSC dvd players and analog/digital televisions.
 #
 # A special feature of this script is the overscan compensation
 # based on laborious trial and error. Because I went to this
 # trouble your subtitles and/or supertitles will be visible
 # on even the most badly overcompensated television screen, but
 # you will not see deformed edges on a television that has 'normal'
 # overscan.
 #
 #
 # NOTE: This script takes it's input filenames from the
 #       command line. Globbing is permitted, i.e.:
 #
 #         ./avitovob *.avi
 #       or
 #         ./avitovob file1.avi file2.avi ...
 #
 # This script requires transcode, mplayer, sox, and toolame.
 #
 # Performance on my 2.8 GHz system is 30-40 fps conversion.
 #
 # copyright 2004 Phil Ehrens <phil@slug.org>
 #
 # This script is licensed for public use as described here:
 # http://www.gnu.org/licenses/gpl.txt
 #
 # Valuable contributions by Adam Di Carlo <adam@onshored.com>
 # The current version of this script can always be found at:
 # http://inferno.slug.org/cgi-bin/wiki?AviToVob
 # ***************************************************************

 # this block writes out the ffmpeg.cfg file with some
 # possibly useful values.
 #
 # The trell option slows encoding down significantly,
 # but is a big quality boost. You can also add:
 # cmp = 3
 # subcmp = 3
 # And quality will increase further, but the speed of
 # encoding will be VERY slow.
 #
 # Note that the lines in this
 # block must begin in text column zero or the script
 # will exit at this point!

 
cat > ffmpeg.cfg <<_EOF
[mpeg2video]
mbd = 2
trell = 1
vqcomp = 0.7
vqblur = 0.3
_EOF

 # We now support ntsc and pal

 DVD_TYPE=ntsc

 # Common export aspect ration for all files. 4:3=2 16:9=3

 EXPORT_ASR=2

 # the -j option here is intended to account for a phenomenon
 # of the NTSC standard and analog TV sets called 'overscan'.
 # the black borders created by this option will generally NOT
 # be visible when viewing on a TV, since they are outside of
 # the effective picture area.
 #
 # There is an additional interesting side effect of -j that can
 # be exploited. Using values that are *not* mod(8) seems to
 # *improve* the output quality quite a bit.
 # Try using -j -18,-34,-22,-34.
 # For letterboxed source with ASR 4:3, use -j -8,-34,-8,-34.
 # Using non mod(8) values will slow down transcoding by about
 # 20%.
 # -----------------------------------------------------------
 # If you see strange colors or noise at the top and bottom
 # of the source video, consider adding --pre_clip 4,0,4,0.
 # This is particularly useful for rescuing bad rips from VHS.
 #
 # Note that the -j top and bottom values should be made
 # larger (i.e., from -18 and -22 to -22 and -26) when using
 # --pre_clip to maintain aspect ratio.
 # -----------------------------------------------------------

 OVERSCAN_COMPENSATION="-j -16,-36,-16,-36"

 # Which audio stream to use from mkv and ogm file

 AUDIO_INDEX=1

 # Which subtitle stream to use from mkv and ogm files

 SUBTITLE_INDEX=1

 # Script will loop over input file(s) on the command line.
 # Input files are never deleted.

 [ ! -z "$1" ] && files="$@";

 for arg in $files ;

do

 # strip the .avi, .mkv, .mov, mp4, or .ogm file extension.

 file=`echo $arg | sed -e 's/\.[amo][vkgop][ivm4]$//'`
 ext=`echo $arg | sed -e 's/^.*\.//'`

 # test for file existence

 if [ ! -f "$file.$ext" ];

 then

 echo "file '$file.$ext' doesn't exist" >&2
 exit 1

fi

 ## ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 ##
 ## New input handling code!
 ##
 probedata=`mplayer -vo null -ao null -frames 0 -identify "$arg" 2>/dev/null \
            |grep ID_VIDEO`

 eval "$probedata";

 geometry=${ID_VIDEO_WIDTH}x$ID_VIDEO_HEIGHT;
 asr=0;
 framerate=$ID_VIDEO_FPS;
 index=0;
 [ $framerate = 23.976 ] && index=1;
 [ $framerate = 25     ] && index=3;
 [ $framerate = 29.970 ] && index=4;

asr=`bc -l << _EOF
define asr(w,h) {
if (w/h >= 2.0) return (4);
if (w/h >= 1.6) return (3);
if (w/h >= 0.0) return (2);
}
asr($ID_VIDEO_WIDTH,$ID_VIDEO_HEIGHT)
_EOF
`

 probedata="-g $geometry --import_asr $asr -f $framerate,$index"

 avi () {
   MPLAYER_OPTS=""
   probedata=""
 }

 mp4 () {
    MPLAYER_OPTS=""
 }

 mov () {
    MPLAYER_OPTS=""

}

 # we use the same naming convention as ogm so we can clean up
 # Use mkvmerge -i $arg and mkvinfo -v $arg to choose the audio
 # and/or subtitle stream.
 mkv () {
    mkvextract tracks $arg 3:$file.mkv-t1.ass
    MPLAYER_OPTS="=\"-sub $file.mkv-t${SUBTITLE_INDEX}.ass\""
    # If the container has vobsub titles you need to do this
    #MPLAYER_OPTS="=\"-vobsub $file -vobsubid 0\""
 }

 # use ogminfo to get audio and subtitle stream info
 ogm () {
    ogmdemux $arg
    MPLAYER_OPTS="=\"-sub $file.ogm-t${SUBTITLE_INDEX}.srt\""
 }

 # execute the type spec
 ${ext};
 ## ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

 # --------------------------------------------------------------
 # At this point you can extract and make use of an existing
 # 5.1 channel ac3 audio track by doing this and skipping all
 # of the audio processing steps between here and the transcode
 # invocation:
 #
 #        tcextract -i $arg -x ac3 > $file.ac3
 #
 # If you are doing this, remove the -p, -b, and -m options from
 # the transcode invocation, and replace the word "raw" in the -x
 # option with the word "null".
 # --------------------------------------------------------------
 #
 # dump the audio to a .wav file using mplayer. You may need to use
 # the -aid option if your source has multiple audio tracks.

 echo ""
 echo "**********************************************************"
 echo "If mplayer crashes now it's because you have a very recent"
 echo "version that no longer supports the '-vc dummy' option."
 echo "please change the line to use '-vc null' if it crashes!"
 echo ""
 echo "     - thanks to Curt Howland for the heads-up on this!"
 echo "**********************************************************"
 echo ""
 echo ""

 mplayer -ao pcm \
         -vo null \
         -vc dummy \
         $arg > /dev/null 2>&1

 # if the sound turns out to be 8 bit, then sox needs
 # extra options to handle it correctly.
 # thanks to Kenneth Stailey for this patch!

 file audiodump.wav | grep -qs 'PCM, 8 bit'
 if [ $? = 0 ]; then
    B=-b
    W=-w
 else
    B=
    W=
 fi

 # if the incoming sound is not sampled at 48 KHz, we
 # upsample the sound to 48000.
 # We handle the fact that sox will abort if the input
 # frequency is 48000. Note that if it aborts, there will
 # be a stub 44 byte long output.wav file to clean up.

 if sox $B audiodump.wav -r 48000 $W output.wav resample ; then
    mv -f output.wav audiodump.wav 
 else

 # otherwise sound was already 48 KHz

    rm -f output.wav
 fi

 # if converting from 30 fps to 25 fps, you may need to use '-I 3'
 [ $DVD_TYPE = "pal" ] && DVD_OPTS="--export_fps 25,3"
 [ $DVD_TYPE = "ntsc" ] && DVD_OPTS="--export_fps 29.970,4 \
                                    -Z 720x480,fast \
                                    $OVERSCAN_COMPENSATION"

 # first, make sure no zombie named pipe is hanging around

  rm -f stream.yuv

 # -----------------------------------------------------------
 # When encoding ANIMATED material, adding the temporal
 # denoiser using the option '-J hqdn3d' will produce
 # significant improvements in image quality and an
 # impressive decrease in file size. In some cases it means
 # the difference between getting 6 episodes on a dvd versus
 # 10! The quality will actually be BETTER!
 #
 # Note the gamma boost option '-G 0.9'. This is used to
 # restore the depth and contrast to material that is
 # washed out or has otherwise lost it's "punch". If things
 # still look washed out, try using 0.8. If things come out
 # too dark using the default 0.9, remove the option.
 # -----------------------------------------------------------
 # Example of how to hard code vobsubs with .sub and .idx
 # files:
 #
 # -x mplayer="-vobsub $file -vobsubid 0",raw \
 #
 # Similarly, the -sid option can be used to choose a soft
 # subtitle stream from a .ogm or .mkv sourcefile.
 # -----------------------------------------------------------
 # When ALL the files are 16:9 aspect ratio, you can set the
 # --export_asr to '3', and use -j -16,-36,-16,-36. This will
 # result in a slight quality improvement.
 # -----------------------------------------------------------
 # To create a 2-pass invocation (for huge quality improvement)
 # simply make two identical calls to transcode, but with the options
 # '-R 1,2pass.log' and '-R 2,2pass.log' in the respective
 # invocations.

 transcode --nice 20 \
           --print_status 500 \
           -G 0.9 \
           -J modfps=clonetype=3 \
           -x mplayer$MPLAYER_OPTS,raw \
           $probedata \
           -y ffmpeg \
           -p audiodump.wav \
           --import_asr $EXPORT_ASR \
           --export_asr $EXPORT_ASR \
           --export_prof dvd-$DVD_TYPE \
           $DVD_OPTS \
           -o $file \
           -m $file.ac3 \
           -i $arg

 rm -f stream.yuv *.wav *.ogm-* *.mkv-*

 # mplex supports constant sync offset correction.
 # '-O -300ms' would, for example, start audio 300 ms
 # earlier than otherwise.

 mplex -f 8 -o "$file.mpg" "$file.m2v" "$file.ac3"

 rm -f *.m2v *.ac3 ;

 # now you have .mpg files, all ready for dvdauthor.
 # like so:
 #
 #        dvdauthor -t -o mydvd \
 #                  -c 0,11:30 file_01.mpg \
 #                  -c 0,11:30 file_02.mpg \
 #                  -c 0,11:30 file_03.mpg
 #
 #        (and possibly -v ntsc+4:3+720xfull if you get errors
 #        and want to be certain that nothing funny happens.)
 #
 #        dvdauthor -T -o mydvd
 #
 #        mkisofs -dvd-video -o mydvd.dvd.iso mydvd
 #        growisofs -dvd-compat -Z /dev/dvd=mydvd.dvd.iso

 done

 # end of script

Notes

ogg and aac audio to wav

 # operates on $file.ogm-a2.ogg forexample
 #oggdec $file.ogg -o audiodump.wav <-- ogg to wav
 # I have some .mkv containers with aac and h264
 #faad -o audiodump.wav $file.aac <-- AAC to wav

gamma oddity

: I just discovered something interesting about Transcode... if you hardsub directly in Transcode and use trellis and heavy denoising, the gamma option is disabled!! After much testing I found that using '-x mplayer="-vf eq2=0.9"' causes the gamma adjustment to be applied as a preprocessor, and so works properly.

AviToPsp

 mencoder -oac lavc -ovc lavc -of mpeg -mpegopts format=dvd -vf scale=720:480,expand=-64:-32:32:16\
  harddup -srate 48000 -af lavcresample=48000 -lavcopts vcodec=mpeg2video:\
  vrc_buf_size=1835:vrc_maxrate=9800:vbitrate=5000:keyint=18:acodec=ac3:\
  abitrate=192:aspect=16/9 -ofps 30000/1001 \
  -o movie.mpg movie.avi

 lmin=1:mbqmin=1:vqmin=1:dia=-4:predia=-4:preme=2:vmax_b_frames=0:dc=8:sc_threshold=-1500
 :vrc_buf_size=1835:vrc_maxrate=9000:vrc_minrate=600:cmp=256:precmp=256:subcmp=256:mbcmp=256:
 vqcomp=0.75:turbo:aspect=16/9 -sws 2 -oac toolame -toolameopts br=128 -af volnorm

http://www.mplayerhq.hu/DOCS/HTML/en/menc-feat-vcd-dvd.html