I have been working with ffmpeg and libav for several years and also have found no decent recent API-level tutorials. Sometimes I just have to dive into the source to figure out what is going on and how to use it. Also, reading the source for the CLI programs (which use the underlying libraries) can also be informative.
In fact, ffmpeg recommends that you just read source code, either for real OSS projects or their examples. Libav provides Doxygen documentation, but no step-by-step tutorials. IMO this is typical of open source projects: the source may be good, but documentation is often lacking.
Taking a step back, you might consider OpenCV, which is well documented (books exist!) and has an easier API. Lastly, ppm is such a simple format you could probably write a 50 line function to save the pixels yourself, and I found a motion extraction program here.
