[MPlayer-users] PSNR Questions

[D Richard Felker III]

On Sat, Feb 15, 2003 at 06:17:02PM +0000, Ian wrote:
> [Automatic answer: RTFM (read DOCS, FAQ), also read DOCS/bugreports.html]
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
> 
> D Richard Felker III wrote:
> 
> >How are you measuring the PSNR of the original?? Like I said, you have
> >to use the same reference to compare PSNR.
> 
> I've been experimenting with a 3 or 4 pass TV recording method whereby I 
> capture at maximum bps for a high-quality "original" mpeg4 or mjpeg, 
> then do a 2- or 3-stage reduction to 1Mb/s mpeg4.  I left the PSNR 
> option on for both stages and just observed that the figures were higher 
> for the low-bw output than the max-bw "original".

Of course; this is a nonsense comparison!

The PSNR values when you capture measure how well the encode
approximates the original signal from the TV card.

The PSNR values when you reencode to 1mbit mpeg4 represent how well
this new encode approximates the first encode (when you captured).

There's a very good chance that this second PSNR value will be better,
since you're encoding data that's already quantized, and thus the
second encode's quantization will be able to approximate the data
better.

On the other hand, this is a sign that your first 'high quality'
encode during capture is probably not high enough quality. Try using
vqscale=1 on the command line.

Oh, and if you really want to compare the PSNR values, do this:
Capture first in a lossless format (like huffyuv). Then encode from
the lossless capture to the same parameters as you would use for
capture (the "high quality/original" mpeg4 or mjpeg). Then, encode
from this mpeg4/mjpeg file to a new 1mbit mpeg4 file. Finally, compute
PSNR values for both encodes compared to the original lossless
capture. You'll see that PSNR for the first one will be much higher
than for the second.

> Incidentally, the quality of this procedure was not significantly 
> (visibly) better than an original grab at the low-bw rate, hence my 
> interest in an objective measurement.

PSNR is a good objective measurement if you know how to use it. As
I've been saying all along, you're drawing conclusion from it that
make no sense.

Rich