[MPlayer-users] black and white DVD

Corey Hickey bugfood-ml at fatooh.org
Tue Nov 4 10:17:00 CET 2003


tarass at club-internet.fr wrote:
> [Automatic answer: RTFM (read DOCS, FAQ), also read DOCS/bugreports.html]
> Le 11/04/2003 05:22 AM, D Richard Felker III a joliment écrit :
> 
>> Maybe, maybe not. There's certainly a (small) benefit in speed, since
>> libavcodec can skip looking at the (relatively small) chroma planes
>> entirely for encoding. As far as improving quality, it depends on your
>> source material. If your source is properly encoded black&white movie,
>> with absolutely nothing in the chroma planes, then it will make no
>> difference. On the other hand, if your source has noise or slight
>> coloration in the chroma planes, using the "gray" option will throw
>> all that away, saving a few bits.
> 
> 
> I tried the colour version 715305ko, the gray 715306 ko ?
> 

I don't think that necessarily means anything. Libavcodec (unless you
specify constant quantizers) tries to maintain the requested bitrate, on
average, for the whole file. There's a certain acceptable tolerance,
however, which is by default 8000kBit and changeable with the vratetol
parameter. Whenever you change an option, the ratecontrol ends up having
something entirely different and the end result may be in a different
part of the acceptable range.

Also, make sure you're using 2-pass encoding for best results.

What would be more interesting for comparison would be the PSNR, which
is an objective measurement of how close the encoded video is to the
source material (post-cropping/scaling/etc.). You'll have to encode each
test again, but this time add :psnr to your lavcopts. Mencoder will
print out a line like this at the end:
PSNR: Y:41.64, Cb:44.20, Cr:45.00, All:42.42
Tell us the PSNR lines for the second pass of each test (at least, I'm
curious). :)

-Corey



More information about the MPlayer-users mailing list