On Tue, 20 Jul 2004, Marc Gregoire wrote:

> > > F:\dvdauthor/jpeg2yuv.exe -n 1 -f 25 -I p -j flower.jpg > 
> > flower.yuv 
> > > F:\dvdauthor/mpeg2enc.exe -q 4 -n p -a 2 -f 8 -I 0 -4 1 -2 
> > 1 -D 10 -h 
> > > -F 3 -o flower.m2v < flower.yuv

        I didn't comment on it earlier but it would appear that you're using
        a really old windows version of mpeg2enc.    Over time there have been
        quite a few changes that stood a good chance of breaking the Win32
        build but no one's mentioned anything - that either means mjpegtools
        still builds ok on win32 (and no one's said anything) _or_ it doesn't
        build but no one has noticed...

> > > the resulting m2v file generated by mpeg2enc is 57 KB for the 
> > > ... while the m2v file generated by the other encoder is 173 KB.

> >     other encoder is using about 3x the bitrate.  Naturally with that
> >     kind of disparity one would expect to visual differences ;)

> >     Try using -q 3 or perhaps -q 2 with mpeg2enc.   I'm not 

> Yes, I know that the other program is generating 3x as big files and I know 
> that's why the image quality is much better, but I'm unable to let mpeg2enc 
> generate still mpegs of that size and quality. 

> Btw: is there any way I can query the bitrate of an .m2v file,
> so as that I can check the bitrate of the other encoder.

        Other than a bitstream analyzer ($$$ unless someone knows of a less
        expensive or even free one) the only thing I know of is to mplex the
        .m2v file with NO audio and look at the bitrate summary.

                mplex -f 8 -o /dev/null flower.m2v

        Towards the end of the report you'll see lines like:

   INFO: [mplex] No. I Frames    :      833 avg. size 43149 bytes
   INFO: [mplex] No. P Frames    :     3280 avg. size 34467 bytes
   INFO: [mplex] No. B Frames    :     8169 avg. size 26840 bytes
   INFO: [mplex] Average bit-rate :  7188400 bits/sec
   INFO: [mplex] Peak bit-rate    :  7210400  bits/sec

        If you see something about "D frames" then that is a _really_ ancient
        version of mplex ;)

> I have tried like you suggested to use -q 3, and I also tried -q 2 and -q 1. 
> With -q 1 the size of the file becomes 64 KB (far from the 173 KB) and the 
> artifacts are still there. :(

        That doesn't seem right.  I'd have expected a big jump in filesize
        going from -q 4 to -q 2.  The only thing I can think of is
        that the default bitrate of mpeg2enc is restraining the size.  The
        default is 7500Kb/s.   You might try raising that with "-b 9800" and
        see if an extra 25% or so makes a difference.

        I did take a look at the two images you mentioned (PNG would be lossless
        so you wouldn't have to worry about introducing artifacts during the
        re-creation of the .jpg files).    Yes, I see a bit of fringing around
        the edges of the flowers.   I can't see the difference in crispness 
        that was mentioned though.

        Could you make available the original flower.jpg file?  It's probably
        too big to post to the list - so a mail item to the list with a link
        would work well.  I'm curious what recent versions of mpeg2enc will do
        with the same

        Cheers,
        Steven Schultz



-------------------------------------------------------
This SF.Net email is sponsored by BEA Weblogic Workshop
FREE Java Enterprise J2EE developer tools!
Get your free copy of BEA WebLogic Workshop 8.1 today.
http://ads.osdn.com/?ad_id=4721&alloc_id=10040&op=click
_______________________________________________
Mjpeg-users mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/mjpeg-users

Reply via email to