Question about Qmin and Qmax settings (VOB to Xvid)

[captvanhalen]

Hello all, sorry if this has been asked and answered before (I searched but found nothing.)

I am trying to understand how best to set the Qmin and Qmax settings. According to the docs, "setting qmin=qmax will lead to variable bitrate encoding." OK, but what is the difference between setting qmin and qmax to say 15 versus setting them to 2? And is this variable bitrate really desirable?

Also, the GUI tool tip for Qmin says "lower values mean better quality" and the one for Qmax says that "higher values mean lower quality." The latter seems counter intuitive to me -- i would think that a higher Qmax would yield higher quality. ??

So what is a good setting to use? The default of Qmin 2 and Qmax 9? (The docs recommend using no less than a 10 for Qmax -- so confusing.)

I am converting VOB files to Xvid (ffmpegx v0.0.9x) and i would like to chose the best possible settings for these to yield better quality as opposed to a smaller sized file (and longer encoding times is not a concern).

Thanks in advance

[jagabo]

In general, think of the quantizer as how much detail is thrown away. For a given frame lower quantizers retain more detail, higher quantizers will throw away more detail. That's all those descriptions are telling you.

In multipass VBR encoding the codec will use different quantizers for different frames. I don't know exactly what strategy the codec uses to determine what quantizer to for a particular frame. Presumably it's some balance between the requested average bitrate and bitrate restrictions (some profiles limit the maximium bitrate), the amount of detail within the frame, whether it's an I, B, or P frame (by default B frames get higher quantizers), how much has changed since the prvious frame, and how many motion vectors are found.

The min and max quantizer settings allow you to place limits on the range of quantizers the codec will use. There is hardly any visible difference between a frame compressed with a quantizer 1 or 2. But a quantizer of 1 will require more bits. Some playback devices may have problems with bitrate spikes with low quantizers. At the high end you are telling the codec never to allow the quality to get worse than the max quantizer will give you.