How to record screencasts
From Inforail
Contents
Generic requirements
- Everybody loves high-res videos, please take that into account
- The ones I make are usually 960x720
- Try to keep the audio noise down, make a few test recordings and play with the sensitivity of the microphone before you begin
Hints
Making screencasts is very difficult, and it gets frustrating too. If you make a mistake, you will be tempted to stop the recording and start from scratch. After several iterations your morale is down. There are some ways to deal with that.
- Milestones - divide the scenario into little bits (under 7 minutes, maybe even shorter), as soon as you're done with one bit - save it. Continue recording the remaining parts into other files. If something goes wrong, you will have to redo just the last part, not the whole thing.
- Repeat - if you didn't say something correctly or something didn't go as expected, do not stop the recording - just repeat your last actions. Once you're done, remove the faulty part in post-processing.
- Relax - you are not on national television, slang is acceptable.
- Silence instead of "aaaahhmmm" or "uuuuuh" works very well.
- Mistakes are a part of the learning process. Do not remove the parts in which the code does not compile or does not work correctly - leave that in the video, such that others can see how you've debugged the issue. It is very helpful, because they will be making similar mistakes.
- Re-record audio only - sometimes the picture and the actions are right, but the audio is bad because you either forgot a word, or you pronounced it incorrectly. If this happens, you can replace a subset of the video's audio track, by re-recording it.
- Extract the soundtrack into a separate file
- Open it with an editor like Audacity
- Select the problematic chunk and replace it with silence (via Generate in the main menu)
- Record a new chunk in a separate track
- Render a new WAV file, merging the existing tracks
Windows
- CamStudio - the recording tool, open-source, has all the features you could ask for.
- Enable the 'follow mouse' feature
- To zoom the screen into a specific area, rely on Windows' built-in screen zooming tool (Win + NUM_PLUS, Win + NUM_MINUS)
- VirtualDub - cut videos into chunks, remove unnecessary parts of the video
- Make sure you enable the Direct stream copy feature, otherwise things will be much slower
Linux
- To zoom in and out, rely on Compiz - it does it very well
- recordMyDesktop (with a GUI front-end, ex: GTK-recordMyDesktop) - this tool works out of the box on my distro, it picks up the microphone correctly, there is no wizardry with Pulse-audio or ALSA. It just works.
- Disable Encode on the fly to increase performance
- Enable Follow mouse
- To specify a custom video size, enter this in the Extra options - --width 1280 --height 720 (adjust the numbers to your needs)
- Avidemux (with GTK front-end) - an equivalent of VirtualDub
Note that by default (at least on my platform), recordMyDesktop encodes the video in a way that Youtube cannot handle. To deal with that, I transcode it with ffmpeg, ex: (replace the size and the paths)
/usr/bin/ffmpeg -i "/home/alex/advice-01-input.ogv" -f avi -r 29.97 -vcodec libxvid -vtag XVID -s 920x720 -maxrate 1800k -b 1500k -qmin 3 -qmax 5 -bufsize 4096 -mbd 2 -bf 2 -flags +4mv -trellis -aic -cmp 2 -subcmp 2 -g 300 -acodec libmp3lame -ar 48000 -ab 128k -ac 2 "/home/alex/advice-01-output.avi"
YMMV
