On 15 Aug 2003, Florin Andrei wrote:
> I'm trying to generate DVD-compatible MPEG2. I used these commands to
> create the video and audio streams:
>
> mpeg2enc -v 0 -f 8 -b 9800 -F 4 -n n -o "test".m2v -q 7 -a 2 -I 1 -4 2
> -2 1 -P
That's probably a little too high on the bitrate. Yes, the
DVD specs say 10.8Mb/s is the max - but don't forget to include
the overheads.
>
> mp2enc -v 0 -r 48000 -b 224 -s -o "test".mpa
9.8 + .224 = 10.024 BUT then add in ~2.5% overhead for the extra
information that's in the stream and you get 10.274 and that's
probably a little too close to the limit to account for any
peak values.
Try -b 9000 and see what happens or see below for some more
ideas.
>
> The generated files seem to be ok. In fact, the quality of the video
> stream is pretty good.
> Then i tried to mux them together, using this command:
>
> mplex -f 8 -o test.mpeg2 test.mpa test.m2v
> (i tried to add -V, but no change)
Try adding "-r 12000" and seeing if at least the mplex run
completes with out errors.
>
> What could be the cause of it? What needs to be changed to avoid the
> error?
Bit rate is too high. It's also probably slightly over the DVD
maximum if the multiplexing overhead is taken into account.
> The source is an amateur video, made with a DV camcorder, NTSC 29.9 fps
> interlaced, 720x480. It has lots of motion (camera sweeping left and
> right a lot), but doesn't have noise.
>
> INFO: [mplex] mplex version 2.2.1 ($Date: 2002/02/04 19:06:14 $)
Hmmm, current version is 2.2.2 I believe - perhaps a cvs update
and recompile might help but I suspect the problem's really
that mplex needs to be told to use a higher maximum bitrate.
Try higher values of '-r' with mplex. A test run using a very
high value should give Average and Peak values - if these are too
high then the video will have to be encoded again using a less
aggressive bitrate.
The other thing you can try, if you haven't already, is use the
alternate quantization matrices that have been added to the
encoder. The savings can be substantial while maintaining
quality. "-K tmpgenc" uses the TMPGEnc tables which "-K kvcd"
uses the KVCD.NET tables.
Did you run the data thru 'yuvdenoise -l 1' first? That might
help reduce the peak bitrate needed but it is a mild setting
that shouldn't have an effect on the quality,
> ++ WARN: [mplex] Audio c0: buf= 2587 frame=000631 sector=00000210
> ++ WARN: [mplex] Stream e0: data will arrive too late sent(SCR)=1427456
> required(DTS)=1427432
That sure looks like mplex wasn't given a high enough bitrate
to work with.
Try a larger value of -r to mplex and see what the average and
peak values are, I suspect they'll be slightly over the limit
Cheers,
Steven Schultz
-------------------------------------------------------
This SF.Net email sponsored by: Free pre-built ASP.NET sites including
Data Reports, E-commerce, Portals, and Forums are available now.
Download today and enter to win an XBOX or Visual Studio .NET.
http://aspnet.click-url.com/go/psa00100003ave/direct;at.aspnet_072303_01/01
_______________________________________________
Mjpeg-users mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/mjpeg-users
