On Mon, 8 Dec 2003 [EMAIL PROTECTED] wrote:

> I am trying to encode one of my commercial DVDs onto a XVCD using dvd::rip.
> 
> The DVD title is 132 minutes. Using the following settings for mpeg2enc

        Whew - that's a lot to try and get any quality from and still
        fit on a CD-R.

        Are you using the standard VCD frame size (352x240 or 352x288)?

> -K kvcd -q 8 -N 1.0 -b 2000

        Leave out the -N if  you're using -K kvcd.  The kvcd tables already
        have the high frequencies "rolled off" (not the right term actually -
        sampled to less detail might be better) and -N amplifies that effect.

> Most of it is good, but some sequences are blocky and there are many 
> artifacts i.e. halos around people, letters (subtitles rendered with 
> transcode's extsub), etc.

        How did you scale the DVD data to the VCD size?  With y4mscaler or
        some other method?   Depending how the scaling is done can cause
        increased halo/ghosting.   If you're using y4mscaler 0.5.0 then
        "-S option=sinc8lan" will probably give the highest quality, if you're
        using y4mscaler 0.6.1 (which was released this weekend) then that
        becomes "-S option=sinc:8" (unless I forgot the new syntax already ;)).

        -q 8 will produce more blockiness than -q 6.

> If I bump -q down to -q 6, I get mplex buffer underruns. If I take out 
> the -K kvcd, the filesize shoots up to twice.

        Then specify the correct bitrate to mplex.  When you're creating 
        nonstandard (XVCD/XSVCD/genericMPEG) formats you have to give mplex
        a better idea of the maximum bitrate.   Either give an arbitrarily
        large value (knowing it will never be approached) or add the video
        rate and audio rate and multiply by 1.03.   "-r 3000" should work
        fine.  Or take the 2000 you specified to mpeg2enc, add in 224 for
        the audio and then a couple percent for overhead - 2224*1.03 = ~2300

        Oh, you'll probably also need '-V' to mplex to tell it that the
        bitrate is variable rather than constant (be nice if mplex could
        figure that out for itself ;)).

        Depending how big the image ends up being you might try '-K tmpgenc'
        to use the TMPGEnc tables - the bitrate won't be quite as low as
        with kvcd but the quality will be better (hopefully).

> Any magic incantations to mpeg2enc to get the optimal quality at lowest filesize? Or 
> is it pretty much trial and error?

        Yes. 

        ;)

        If you're using the cvs version of mpeg2enc you can also try adding
        "-E -10" to the options - for clean source (such as from a DVD)
        that option can be quite valuable.  Might save enough bits that you
        could use a lower -q (and odd values of -q are allowed - you could
        try -q 7 if desired).

        Good Luck.

        Steven Schultz



-------------------------------------------------------
This SF.net email is sponsored by: IBM Linux Tutorials.
Become an expert in LINUX or just sharpen your skills.  Sign up for IBM's
Free Linux Tutorials.  Learn everything from the bash shell to sys admin.
Click now! http://ads.osdn.com/?ad_id=1278&alloc_id=3371&op=click
_______________________________________________
Mjpeg-users mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/mjpeg-users

Reply via email to