Forcing a bitrate with FFMPEG

[boyer]

Hey crew!

I'm having an issue with FFMPEG.

I'm trying to force a lowish bitrate while converting an uncompressed AVI into an FLV. Obviously I want the FLV to be as small as possible for the web.

Here is my command string:

ffmpeg -i test.avi -b 50k -qmax 5 -r 25 -s 440x360 -ab 56k -ar 22050 -acodec libmp3lame  -f flv test.flv

You'll notice I have the -b down to 50k! With this, I'm getting a reported video bitrate of 662K! WTF?

Is there any way I can FORCE the bitrate to a certain value? I'm aiming for around 200K. I'm open to multi passes.

Thanks everyone!

[txporter]

Hey crew!

I'm having an issue with FFMPEG.

I'm trying to force a lowish bitrate while converting an uncompressed AVI into an FLV. Obviously I want the FLV to be as small as possible for the web.

Here is my command string:

ffmpeg -i test.avi -b 50k -qmax 5 -r 25 -s 440x360 -ab 56k -ar 22050 -acodec libmp3lame  -f flv test.flv

You'll notice I have the -b down to 50k! With this, I'm getting a reported video bitrate of 662K! WTF?

Is there any way I can FORCE the bitrate to a certain value? I'm aiming for around 200K. I'm open to multi passes.

Thanks everyone!

You are specifying -qmax 5, which is telling ffmpeg that it cannot use a quantizer higher than 5. That is not allowing you to get to your requested bitrate. Increase that or remove it.

[boyer]

Ok, I think I get it... Can you explain how the -qmin and -qmax work in a little more detail? I figured a value of 1 would be poor, and the max value (51?) would be high... I guess I'm not fully understanding how this setting works.

[txporter]

Ok, I think I get it... Can you explain how the -qmin and -qmax work in a little more detail? I figured a value of 1 would be poor, and the max value (51?) would be high... I guess I'm not fully understanding how this setting works.

It's really the opposite. 51 would give you the lowest quality and 1 would give the highest. -qmin just tells ffmpeg to use a quantizer of at least X (whatever you specify). -qmax just says don't go above Y. I am not that intimate with details on quantization (perhaps jagabo or manono or poisondeathray will drop by ), but basically the various compression methods used by the different codec are trying to approximate the signal with some finite range of values in order to improve compression. If you are really interested, you can start with wiki's quantization page and work your way out from there.

Look here for documentation on ffmpeg options.

[boyer]

Brilliant! Thanks a lot for taking the time to help out!