UsefulNotes/LugTalkVideos

Capturing

1. Capture the raw DV from the camera using Kino. This uses up about 1GB for 5 minutes raw DV. You might need to load the raw1394 module and / or change permissions on /dev/raw1394.
2. Capture one talk at a time as a separate "project".
3. Insert the starting and ending caption files.
4. Generate a cross-fade effect using the FX panel.
5. Export the talk to either "Generic MPEG2" or Raw DV using the Export function. Warning, the export process will take some time.

Inserting slides

1. Take the presentation file from the speaker and export the slides as BMPs, TIFFs, JPEGs etc.
2. Start a new kdenlive project.
3. Add the Raw DV / MPEG2 file to the timeline.
4. Add the slides where necessary on the timeline.
5. Export as a high bitrate MPEG2 (8000kbps).

Audio processing

The camera I use for recording the LUG talks only records the external microphone on the left channel. This means that the audio has to be extracted from the MPEG2, converted to mono, other processing applied if necessary and then inserted back into the MPEG2 file.

1. Open the MPEG2 in avidemux and allow indexing.
2. Export the audio as a WAV or MP3 file. Audio->Save->Enter a filename.
3. Open the saved file in Audacity.
4. Click the drop-down box on the function panel at the beginning of the audio track. Choose "Split stereo tracks".
5. Click on the little cross in the corner of the function box on the track that doesn't contain useful noises.
6. Click the drop-down box on the function panel at the beginning of the remaining audio track. Choose "Mono".
7. Perform any other sound tasks, like noise reduction or filtering. A high pass filter at 443 Hz has been useful in the past.
8. Alternatively an acoustic equalisation filter can help reduce high and low level noise.
9. Save the altered audio into a new WAV file.
10. Back in avidemux, select the newly created output WAV as the source for the audio.
11. Select Audio->Main track->External PCM (WAV). Then locate the WAV file in the standard file location dialog.
12. N.B. You must do this routine after performing any cuts or inserts on the file, as the external audio track will not be edited if you edit the video.
13. From the Audio toolbar choose FFm MP2. Set the bitrate to 128kbits. (Audio should still be at 48000Hz.)
14. Set the file type to MPEG A+V PS.
15. Hit Save and choose a new filename for the MPEG2 video with new audio.

Exporting in desired formats

1. The following tools are needed:
   1. ffmpeg N.B. On Ubuntu, MP3 encoding support is disabled by default. You might need to recompile ffmpeg with MP3 encoding support enabled. This probably also applied to Fedora. See notes.
   2. ffmpeg2theora http://www.v2v.cc/~j/ffmpeg2theora/
   3. lame
2. Use ffmpeg2theora to convert the MPEG to a Theora / Vorbis Ogg file, resizing as you go:

   ffmpeg2theora -x 352 -y 264 --optimize -o output_small.ogg input.mpeg
   ffmpeg2theora -V 600 --optimize -o output_large.ogg input.mpeg

   If there is a strip of distortion at the bottom of the picture, try using -y 256 instead.
3. Use ffmpeg to convert the MPEG to a WMV / MP3 file, resizing as you go:

   ffmpeg -i input.mpeg -s 352x264 -vcodec wmv2 -b 250k -acodec mp3 output_small.wmv
   ffmpeg -i input.mpeg -vcodec wmv2 -b 600k -acodec mp3 output_large.wmv

   (Older versions of ffmpeg did not understand the .wmv extension. If that is the case then use .asf as the extension and rename the output file to an .wmv extension.
4. For widescreen use 472x264 for smaller videos and 1024x576 for larger ones.
5. Encoding to .flv for blip.tv:

   ffmpeg -i inputfile.dv -ar 22050 -b 1000k -s 1024x576 -f flv outputfile.flv

Uploading

1. Go to http://archive.org/create
2. Login when prompted. Details for hantslug account available from me or hostmaster.
3. Select the "More than 100MB files" option.
4. Follow the details given to upload the files via FTP.
5. Use the following details to complete the file's metadata.
6. Choose Creative Commons License class. Choose "No" to commercial uses of the work. Choose "No" to derivitive works. This is effectively this license: http://creativecommons.org/licenses/by-nc-nd/2.5/
7. Leave jurisdiction as Generic.
8. Give the following informatiom about the file:

   Creator's Name: Hampshire Linux User Group
   Copyright Holder: Hampshire Linux User Group
   Copyright Year: 200X
   Posted by: Your Name
   Title of Work: Title of talk
   Description: Person gives a short talk on subject. This talk does [not] require any existing knowledge of subject.
   
   See http://www.hantslug.org.uk/cgi-bin/wiki.pl?TechTalks for more details and copies of the presentation that accompanied the talk. This talk was given at the Hampshire Linux User Group (HantsLUG) meeting on 1st October 2005 at Southampton University.
   Keywords: Hampshire Linux User Group; HantsLUG; talk; LUG; some terms related to the content
   Work format: Video

9. Click Next
10. Comfirm the file type data at the end of the screen.
11. Click Next

Recompiling ffmpeg with MP3 encoding support

The ffmpeg package distributed with Ubuntu does not enabled MP3 encoding support (only decoding). You can verify this from:

$ ffmpeg -formats | grep mp3
 D  mp3             MPEG audio
 D A    mp3
 D A    mp3adu
 D A    mp3on4

The first line describes ffmpeg's ability to decode and encode the MP3 file format, the second with its ability to decode and encode data using the MP3 codec. The "D" in the text denotes the ability to decode. "E" denotes the ability to encode, and this is not present in the above output. In order to recompile ffmpeg with MP3 support on Ubuntu Dapper and Ubuntu Edgy:

• Install build-essential and quilt.
• Install the build dependencies for the ffmpeg package: apt-get build-dep ffmpeg
• Install the lame development files: apt-get install liblame-dev
• Install the ffmpeg source: apt-get source ffmpeg
• Change into the source directory.
• Edit the debian/rules file. Look for line 8 which reads:

  confflags += --enable-zlib --enable-vorbis --enable-libogg --enable-theora --enable-a52 --enable-dts --enable-dc1394 --enable-libgsm

  and add --enable-mp3lame to the end.
• Rebuild the packages (usually via fakeroot debian/rules binary).
• Install using dpkg -i.

After all this, you should now be able to encode to MP3 using ffmpeg. Verify this by checking the output of:

$ ffmpeg -formats | grep mp3
 DE mp3             MPEG audio layer 3
 DEA    mp3
 D A    mp3adu
 D A    mp3on4

The presence of E on the first two lines shows that ffmpeg can now encode to the MP3 file format and encode data using the MP3 codec.

Useful Links

• http://gentoo-wiki.com/HOWTO_Mencoder_Introduction_Guide
• http://www.v2v.cc/~j/ffmpeg2theora/
• http://ffmpeg.mplayerhq.hu/ffmpeg-doc.html
• http://ffmpeg.mplayerhq.hu/faq.html
• http://wiki.creativecommons.org/CcPublisher_2