I need to be able to programmatically transcode mpeg-2 files to .mp4, .mp3, .wmv, .rm (optional), and .flv (optional), and hopefully generate a thumbnail as well. I found the Java Media Framework, but it frankly looks pretty crappy. This will be running a Linux server, so I could shell out to ffmpeg using Commons Exec - does ffmpeg do everything I need to do? FFmpeg seems pretty daunting, which is why I'm having trouble finding this information, but it definitely seems to be a jack-of-all-trades. Any suggestions?
A:
The popular transcoding applications for Linux are ffmpeg, transcode and mencoder. Both transcode and mencoder use ffmpeg and all three can handle the tasks that you require, including FLV transcoding and video thumbnailing. ffmpeg is probably the most popular of the three, so you might find better online support. Also worth mentioning is that ffmpeg supports multithreaded transcoding.
I would recommend using ffmpeg.
codelogic
2009-01-01 00:35:35
ffmpeg definitely seems like the way to go, I'll be looking into it more for sure.
dancavallaro
2009-01-01 01:06:55
+1
A:
An interesting web service from encoding.com will transcode files for you.
Brandon
2009-01-01 00:45:32
That looks like a good service, but we really don't have that many videos to transcode, it's really in batches of 10 or so, and I'm looking to automate a workflow, so it would be easier to keep everything in-house.
dancavallaro
2009-01-01 01:04:47
+3
A:
Ffmpeg is the best and easiest. To output/convert video:
ffmpeg -i {input}.ext -r {target_frame_rate} -ar {target_audio_rate} -b {target_bitrate} -s {width}x{height} {target}.ext
And your screenshot:
ffmpeg -i {input}.ext -r 1 -ss 00:00:04:005 -t 00:00:01 -an -s {width}x{height} {target_name}%d.jpg
15 fps is standard for flv and audio sample rate should be 44100 for flv. Options to use: -r specifies a frame rate of 1 fps (one frame used as the screenshot), -ss seeks to the position you want hh:mm:ss:fff, -t is the duration (one second to match your one fps), -an tells ffmpeg to ignore audio, and -s is the size of the screenshot. The %d is necessary as it will be the digit incremented on how many screenshots you use. With the above, %d will always be the number 1 (one frame, one screenshot). Good luck.
Thanks very much, that was a helpful taste of what ffmpeg can do. I'll definitely be looking at ffmpeg more closely..
dancavallaro
2009-01-01 01:07:34
Out of interest, is using -an when screen-capping of any use? It makes sense that you'd ignore the audio but I'm wondering if there's a performance gain of some description?
Adrian Lynch
2009-07-24 13:01:14