Hi!
> From: [EMAIL PROTECTED]
> I decided to compare using the --reduce-hf switch of mpeg2enc to other
> methods for reducing noise and the bitrate, after having ignored it
> previously. I was pleasantly surprised. To take an example, I
> converted noisy DV footage to DVD format with -q 5 and -b 8500:
Out of curiosity what threshold (-t) and depth (-d) values did you
use with yuvdenoise? Or did you go with the defaults (5 and 3
respectively)?
> no denoising : 8300 kb/s (mostly hitting the upper bound)
> yuvdenoise : 7700 kb/s
> --reduce-hf : 7400 kb/s
> yuvdenoise + yuvmedianfilter: 6000 kb/s
> yuvdenoise + --reduce-hf : 4900 kb/s
> all of the above : 3600 kb/s
>
> Note that while --reduce-hf or yuvdenoise alone is only a modest
> improvement, together they reduce the bitrate substantially. But
Can I say "I told you so?" now? <grin>
I have found the combination of light to moderate (-l 1 or -l 2 with
yuvdenoise) and -N to mpeg2enc to produce excellent results without
noticeable softening of the image and the bitrate's perfect for
DVDs in the ~80-90 minute range. To get a 2hr6m movie to fit
I did fall back to '-q 6' the other day.
> here's the thing - I can't see the difference between yuvdenoise alone
> and yuvdenoise with --reduce-hf. There's no free lunch, so can
--reduce-hf|-N
Reduce high frequency resolution - useful as a mild noise reduction
It does say "mild" but, as you mention, it is extremely effective
and without apparent side effects.
> somebody please tell me what sorts of artifacts I need to look for? I
I thought the effect was supposed to be a very slight reduction in
sharpness.
> filtering) look like, but unless my eyesight is worse than I think,
> --reduce-hf is a lot more subtle. I was assuming that fine detail
> would suffer, but I don't really see that. But then again my camera
> doesn't really capture too much fine detail.
I think all that is telling us is that there is information present
which is not visible to the human eye (at least via a TV or monitor)
but which of course the encoder "sees" and has to deal with.
Thanks for the chart - it's quite similar to the one I'd done for
myself a while back when I was experimenting with the options.
Steven Schultz
-------------------------------------------------------
This SF.net email is sponsored by:Crypto Challenge is now open!
Get cracking and register here for some mind boggling fun and
the chance of winning an Apple iPod:
http://ads.sourceforge.net/cgi-bin/redirect.pl?thaw0031en
_______________________________________________
Mjpeg-users mailing list
[EMAIL PROTECTED]
https://lists.sourceforge.net/lists/listinfo/mjpeg-users
