Screen capture for a tutorial

[rich]

Staun asked for a few notes on making a screen capture video tutorial, so in typical long-winded fashion this is the procedure I use.

First thing is, I am using linux. In the main, it's PCLOS 2011, but a couple of recent efforts have been using Bodhi linux, a lightweight Ubuntu derivative. The procedure I use is the same for either.

Since the video will end up on YouTube, their advice on frame size is either 1280x720 for HD or 640x480 for regular. I try and keep everything to a minimum and record an area 1.5 x 640x480 ie. 960x720 then as a final procedure, shrink the video. Also to keep size down, I record at 15 frames per second (fps), adequate for a screen capture.

The tools/applications I use.

ffmpeg as a command line (cl) application. There are front ends for it, such as recordmydesktop_gui, but I find cl more flexible. This only works for linux, Windows users could try Camstudio which is free & opensource or maybe AutoScreenRecorder-Free (but beware of the ASK toolbar).

Avidemux for the video editing. About as basic as you can get but does the job. It also uses ffmpeg. There is a Windows version

Audacity for sound recording. There is a Windows version.

Now the technical bits. I find it very difficult to make a sensible running commentary during recording so I leave the audio 'till last. The video recording is a monster command. It would be longer if I included sound. Obviously, I do not type it out each time, I have it set up in a menu "shortcut".

ffmpeg -f x11grab -xerror -r 15 -s 960x720 -i :0.0+80,60 -vcodec mpeg4 -b 1000 -qscale 5 -bt 256k -y `date +%s`.avi

the parameters are

-f forces the format, x11grab captures the screen, -xerror ignore errors, -r 15 is the frame rate

-s 960x720 is the capture area and -i :0.0+80,60 defines the top left corner.

-vcodec mpeg4 -b 1000 is the codec and bitrate, -qscale 5 -bt 256k are quality settings

-y will ask if there is an output file with the same name but should not be required since `date +%s`.avi should give a unique 'number'.avi

The thing about this command is it is a reasonable trade-off between file size and accuracy. The avi is compressed but has keyframes every 12 so editing in Avidemux is reasonably accurate.

Typically I end up with 4 or 5 avi files and using Avidemux, I chop out any "ragged" lead-in and "non-relevent" middle bits. Avidemux will then "append" all the separate files and you can play the whole thing through before adding sound.

I record audio separately because I find it easier to redo just the commentary rather than re-record whole chunks of video+audio and it is the bit I find most difficult.

I record the audio using Audacity while playing the video in Avidemux. Using the cinema clapper-board method, I tap the microphone when I start the video. Then crop the recorded audio at that mark and crop again the end at the video to the length given by Avidemux. Finally save the audio in 2 channel mp3 format, 56 kbps - good enough for YouTube.

Avidemux will mux this audio with the video and for the final result I select divx avi as the output format. Then go into the filters and resize to 640x480, also apply a little sharpen and save to a "filename.avi". This gives a more compressed file than the source, typically 10 minutes of video comes out at 16 to 20 MB.

This might not be the best method but it works for me. Any questions, let me know.