Now I know about ffmpeg with x11grab patch, xvidcap, and ScreenKast. From all these three alternatives, ScreenKast is the best one in feature comparison. By the way, it is totally based on KDE while my desktop is GNOME. I don't want to run KDE application in GNOME environment because I have to waste lots of main memory to hold 2 low-level desktop interface.

To solve this problem, I find better solutions similar to ScreenKast but rely on X11 or GNOME. I found BVidCap which is designed for Ubuntu 6.06. It is just a shell script wrapped ffmpeg+x11grab to capture such a frame and/or encode the video to theora format automatically. Well, it is still not what I want.

The closest one is vnc2swf in C and Python or pyvnc2swf. Well, I like to have my screencasts in video format instead of SWF. However, vnc2swf led me to vncrec which is a small tool for recording VNC session just like ScreenKast do. The recorded session then can be converted to other video format using transcode as simple as below.

transcode -x vnc -y xvid -i file.vnc -o file.avi -z -k

Some might say the output avi produced by transcode is not as good as they get from mencoder. The result getting from above command is worse. Better result can be obtained by exporting still image from the recorded session and then assembly them together into a video file using mencoder as a series of image. To automate all of these steps, I found record-visual which also patched vncrec to record audio concurrently.

As of now, I didn't try record-visual because I have to build it on Ubuntu manually and I don't want to install them manually. So I'm trying to patch them especially for Ubuntu.

I will let you know the result after testing in a few days.

Tags: screencast, ffmpeg, x11grab, xvidcap, screenkast, bvidcap, vnc2swf, pyvnc2swf, vncrec, vncrec-audio, record-visual