Como si fuera grabado en el mármol

Enjoying streamer and ffmpeg, I wanted to try this little transformation of jpeg pictures. My example is short (3 seconds), but sufficient to be afraid of the result. Again, the origin of my tour in the field of video understanding is our friend's presentation wisdoming about streets of Douala on YouTube.com. He made me curious about how to do a better use of my webcam. Not sure yet if I got the true result, hum !

Here a script of that production :

1. Fist of all, I produced 48 pictures with a simple webcam ; I grabbed them using streamer :
   streamer -q -c /dev/video0 -b 24 -f jpeg -s 320x240 -r 12 -t 00:00:04 -o CC00.jpeg
   
   Notice I did 12 frames per second (argument -r 12), with 4 seconds, you have indeed 48 pictures ; the result goes in jpeg format to files named CC...jpeg (the files are named automagically CC00.jpeg, CC01.jpeg, and so on).
   
   
2. With The Gimp, I produce an embossed effect on each jpeg file.
   
   
3. Then I use jpeg2yuv :
   ls *.jpeg|jpeg2yuv -f 12 -I p >resultat.yuv which gives a very very big file, because of my poor understanding of this command or, less probably, of a bug ; so I had to delete the process. No matters, I received more than my four seconds of raw video.
   
   
4. I reformat this yuv file in a mov file : ffmpeg -t 4 -s qvga -i resultat.yuv result.mov.
   
   
5. I register my voice with 3 messages and do a speed distorsion, obtaining a 3 to 4 seconds wav file ; for that purpose, I have this excellent tool named Audacity.
   
   
6. I do the mix of both files :
   ffmpeg -i result.mov -i CC.wav CC.mov
   
   
7. And now, I can listen this video with voice included, with mplayer for instance.
   

Finally, yes, it could have been better, but it is a first trial of my own ; I wanted to share with you these huge possibilities of free software included in Linux ; you should try them before you decide leaving the zindow$ world. Next time when I do it on YouTube, I'll tell you.