On Thu, 18 Dec 2003 [EMAIL PROTECTED] wrote:
> In trying to make a VBR MPEG1, there is a tradeoff between higher
> filesize/bitrate vs. quality.
That's always true - it's not VCD specific (and it sounds like you
are creating a XVCD).
> To get the same visual quality (i.e. no artifacts, no halo around text, etc),
> is it preferable to increase the bitrate or reduce the -q parameter to
> mpeg2enc (i.e. increase quality factor)?
Slight misunderstanding/confusion detected ;)
You can set the bitrate to 9000 kbits/sec but with a high '-q' the
encoder will only use a fraction of the maximum bitrate. Lower -q
and watch the encoder come closer to the maximum rate.
You set the maximum bitrate and then adjust -q so that the Average
is ~10% lower than the Peak rate.
> I just used a VBR bitrate of 1150, "-K tmpgenc -E -10 -R 0 -q 8" to make a
> 130 minute DVD source fit a single CD-R. The quality is not exactly great,
I can believe that :)
> but it is not bad either. There are halos around letters (subtitles),
> the picture is not very sharp/clear.
After discarding 80 to 90% of the data what did you expect?
Sharpness/detail cost bits and with the length of video you're
attempting to put on a single CD you're almost lucky to get any
picture at all.
> Comments?
Let's do some arithmetic ;)
Assume 800 megabytes for a CD-R (that's a little low but there is
overhead of multiplexing audio/video and the VCD data structures to
take into account).
That's 800000000*8 or 6400000000 bits.
Divide that by 130 minutes 6400000000/(130*60) and get 820512.
Total of 820512 bits/sec for both the audio and video. Not sure
what audio rate you're using but for purposes of discussion I'll use
128000 bits/sec for MP2 audio. That leaves a total of 692512 bits/sec
for video. That is so low that there's not a lot of bits for
sharpness/detail.
Subtitles are trouble because they present sharp edges or contrast
against the background and that uses lots of bits.
The other problem comes from downscaling - during the scaling from
DVD (720x480) to VCD (352x240) you're discarding 50% of the vertical
resolution (from 480 lines to 240) and over 50% of the horizontal.
That is going to lose detail on the text of the subtitles.
You can minimize that loss (at the expense of using more bits) by
doing a high quality scaling.
What are you using to do the downscaling? 'y4mscaler' has a wide
variety of choices to choose from - some are quick but lose more
detail, others are slower but produce higher quality output.
Something like
y4mscaler -O preset=VCD -S option=sinc:8
(and you can change the 8 to something a little higher such as 10 or 12,
but it will use more cpu time, if desired) will do very high quality
scaling.
Given a video bit budget of only 692512 bits/sec there's not a lot
of room for sharpness of subtitles.
When you set the bitrate to 1150 what did mplex report as the
Average and Peak bitrates? If the average was less than ~700 then
you can decrease -q by 1, say to 7 and see if the video will still
fit.
Now if you could go to a 2CD set (pick a nice convenient fade to black
scene to switch discs) that would make for a much better looking VCD
(at that point a CVD would be a good choice of format).
At the moment I'm in the process of creating a 90min CVD - almost have
the video down to the right size. Quality's fairly good too
considering the source material wasn't the highest quality (it's from
a DVD but the original tape/film is noisy).
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
