Shortest path from Kubuntu 9.4 to YouTube

A while ago, I published a guide for the students on one of my courses showing how to perform certain operations in Eclipse. I tried to include a few screen captures, detailed description and try to bring them quickly to a point where true code development was easy.
The result was quite surprising. Despite the effort to make a detailed guide, the number of questions and clarifications was unusually high. It was not due to the absence of details, but, if I have to guess, I was pretending to show how a highly visual tool works, with a highly detailed and dense document. Two opposite resources. Somehow, it seems that when you are performing a visual task, you may learn faster if you have visual (rather than textual) support.
So I decided to craft some videos showing screen captures of some additional procedures (creating a debugging configuration). The premises: open-source tools, and what is more important, really quick path to publication. If I create a cool video, but it takes me two weeks, forget it. It should be under an hour even at the expense of some quality. My take on this, there is no need for George Luca's quality for a guide that you'll view once and that's it. Here is the tool combination I've used.

Screen capture

This was the trickiest part. The problem is not yet cleanly solved, but xvidcap was the solution. It put my patiente to the test, until I tweaked with some obscure option with name Manually set XDamage. Interestingly enough, it should be enable. Go figure.
I've selected MPEG as output format and avoided completely capturing at the same time video and audio. I found much more confortable to quickly draft a script of what I plan to say, capture the screen, and the add the audio on top

Audio manipulation

Tools here are up to the challenge. Audacity came to the rescue and performed perfectly. I needed, though to first install PulseAudio, the new Audio framework in Ubuntu that despite some criticism I've seen in some blogs, it works like a charm in my machine. It allows you to control all different sound "threads". For example, if you are hearing music and a video pops up in the web, you may control the volumes separatedly.
Audacity allows you also to perform fairly useful operations such as cutting silences, shifting the wave, etc. Also, you may record (as it was my case) two tracks for the same video in the same project. You simply mute one of them and export the other. Which brings me to the final touch with Audacity, exporting to MP3. It works without a glitch. An MP3 file is created and is ready to be merged.

Merging Audio/video

This is the most obscure part that I haven't figured out completely. I use mencoder which looking at the number of available options, you could think it is capable of pretty much everything. After several trials, the final command is (brace for impact):

mencoder \
  video.mpeg \
  -audiofile audio.mp3 \
  -o video.flv \
  -of lavf \
  -oac mp3lame \
  -lameopts abr:br=56 \
  -ovc lavc \
  -lavcopts vcodec=flv:vbitrate=500:mbd=2:mv0:trell:v4mv:cbp:last_pred=3 \
  -srate 22050

Does it work? Yes. Do I know what is doing? Not even the slightest idea. The audio is far from perfect, and there is some drifting in time, but it did the trick of producing the videos in less than an hour. I need to polish this last step, because the audio sounds perfect when played alone. I might post again if I improve the recipe.

As for the result...