[mythtv-users] Pixelation Issues

Rod Smith mythtv at rodsbooks.com
Tue Mar 6 00:53:10 UTC 2007


On Monday 05 March 2007 19:21, Brad DerManouelian wrote:
> On Mar 5, 2007, at 4:18 PM, D wrote:
> > What are the ideal settings for recording SDtv? I know that it will
> > differ depending on HW setups, but I feel like I must have something
> > set up wrong. I've got a pvr-150 encoding to MPEG-PS @ 720x480 at
> > 2200-2400 kbps and I'm getting pixelation on movement. Should I up the
> > encoding rate? Should I try a different encoding option such as DVD-2?
> >  What settings do you guys run? What settings would provide the best
> > possible image?

My Hauppauge PVR-USB2 is stuck 6000kbps. This seems to be a driver/MythTV 
disconnect concerning the API used to change the encoding rate. At that rate, 
I don't see a lot of pixellation at that rate.

Many moons ago I had a standalone TiVo, which supported encoding rates that 
ranged from (IIRC) ~1GB/hour to ~4GB/hour. Since 2200kbps is ~1GB/hour, that 
means your setting is equivalent to the standalone TiVo's low-quality 
setting, which may be a useful reference point. Of course, this could be 
comparing apples and oranges, since the encoding hardware may be different. 
It's certainly possible to get much higher quality video at a ~2200kbps rate; 
that's in the neighborhood of what DirecTV and Dish Network use for their 
broadcasts, for instance. These companies, though, use much more expensive 
encoders than a TiVo or a typical MythTV box uses. It's also possible to 
improve the quality by doing multi-pass encoding, but you certainly won't get 
that with on-the-fly encoding from a consumer-grade video encoding card. You 
CAN do multi-pass encoding when transcoding after the fact, though.

> I run at 640x480 at around 6400 kbps and then transcode down from
> there to about 4400 kbps if it's something I don't care that much
> about. I thought 2200 looked pretty awful when I was recording at
> that bitrate.

One extra point: When you transcode using MythTV's built-in options, you'll be 
doing one of three things:

1) Cutting out sections of the recording without changing the encoding
   method used. This will reduce the file size but shouldn't affect the
   quality of video, except for brief periods around your cut points.

2) Converting from whatever encoding method you used to MPEG-4. MPEG-4
   has a reputation for producing higher quality at any given bitrate
   than MPEG-2, so in theory you can reduce the size of an MPEG-2 video
   while retaining most of the quality. (A conversion from one lossy
   format to another WILL degrade the quality slightly, though, even
   if you INCREASE the file size in the process. Hence my use of the
   phrase "most of the quality.")

3) Converting from whatever encoding method you use to RTJpeg. This
   encoding method produces larger files than MPEG-4, if you try to
   match the quality. I'm not sure how it compares to MPEG-2. RTJpeg
   consumes less CPU time, though, so it's sometimes used on systems
   with weak CPUs and frame grabber cards. My own experiences with it
   have been pretty negative.

After encoding with an MPEG-2 card, chances are you'd do #1 or #2, and what 
you (Brad) describe is definitely #2. You might even be able to squeeze a 
6400 kbps MPEG-2 recording down to less than 4400 kbps without causing too 
much degradation in quality, although of course this is a subjective matter.

Overall, for long-term storage with an MPEG-2 card, it's probably best to 
encode with a high bitrate (say, 6000-10000 kbps) and then transcode to 
MPEG-4 with a significantly lower bitrate (perhaps 2000-4500 kbps). If you 
expect to watch the show soon, I'd recommend just encoding with a moderate 
bitrate (4000-6000) and then watching it without transcoding it. Of course, 
different people have different ideas of quality, so you should experiment 
yourself. Disk space and CPU speed (for transcoding to a different format) 
can also be important considerations.

-- 
Rod Smith
http://www.rodsbooks.com


More information about the mythtv-users mailing list