[MPlayer-users] mbd, vqscale, and vpass mechanisms

[D Richard Felker III]

On Sat, Jan 03, 2004 at 03:19:37PM -0500, Elmo wrote:
> [Automatic answer: RTFM (read DOCS, FAQ), also read DOCS/bugreports.html]
> Hello,
> 
> I remember a discussion a while back about vqscale, but I don't believe 
> a conclusion was ever really reached.  Anyway, I have a few questions 
> about these mechanisms in the encoding process.
> 
> What I have typically been using as a command line to encode a dvd is:
> 
> mencoder dvd://1 -o movie.avi -oac lavc -ovc lavc -alang en -lavcopts 
> vcodec=mpeg4:acodec=ac3:vqscale=3:mbd=1 -vf crop=<foo> scale=<foo>
> 
> I figure that this will give me a constant quality of 3, which is pretty 
> much unnoticable, use the "vhq" method, which is the best quality 
> encoding, and the size will be whatever is required to get the necessary 
> quality.

vhq/mbd=1 (same thing) should not increase quality at fixed quant,
just decrease filesize. mbd=2 might improve quality and decrease
filesize.

> I don't see why people use 2-pass encoding with vqscale, basically.  If 

Since when do people do this??

> you're encoding at a constant quality, how does getting information 
> about the bitrates help the encoding quality?  Should I be doing this 
> differently to get a better overall movie, or am I misunderstanding the 
> way that vqscale works with vpass?  Also, mbd has three options, and I 
> don't really know which is the "best" in terms of quality.

mbd=2 is best. Usually encoding with vqscale isn't recommended. It
will use a lot more space and only improve quality marginally (at
best!) compared to selecting a good bitrate and using 2pass encoding.
In fact it could reduce perceived quality, since with a 2pass encode,
stills will usually get quant=2.

> Perhaps we should update the encoding-tips.txt file on the mplayer user 
> site at some point, I think it's getting outdated.

I agree.

Rich