[mythtv-users] colour "sharpness" question

[Brian Wood]

On Oct 19, 2006, at 11:42 AM, Michael T. Dean wrote:

>
> Which, I'm sure you'll agree, is an excellent resource if it's not  
> taken
> too literally.  Everything on the chart is exactly right as long as  
> the
> bitrate values at the bottom are fuzzed out.   The chart doesn't  
> factor
> in signal quality (a dirty signal will require more bitrate than a  
> clean
> signal), variable bitrates (as discussed in the text, so read the  
> text,
> too), encoder quality, and, most importantly, decoder/filter/other
> electronics' quality (that's the NTSC signal decoder used during
> capture, not an MPEG-decoder used during playback).
>
> Also, since NTSC's resolution is limited, even if you use 720x480
> capture resolution (which is a good idea for DVD archiving purposes),
> you aren't getting same same quality as you would see in a 720x480 DVD
> (even if that DVD contains interlaced video).  Now, if you can keep  
> the
> signal from ever becoming NTSC (which is impossible with a PVR- 
> x50), ...
>
>  From my experience (and from my interpretations of some good writeups
> by Cory Papenfuss), I still stand by the 4500/6000 being a good  
> bitrate
> for PVR-x50's.  However, if you're mastering DVD's from original
> sources, that chart--and the bitrate values it shows--is probably
> exactly what you'll need.  :)
>
> But, it doesn't hurt anything--except available storage--to use more
> bandwidth than required.
>

All good points and I do indeed agree.

The main points to be aware are that the function is not linear, and  
that there is a plateau beyond which all you are doing is wasting space.

Using more bandwidth than is required won't hurt anything but space  
if you are using an external hardware encoder, but if your main CPU  
is doing the encoding it will also tax that needlessly.

And of course the hardest thing to encode is noise, as you indirectly  
pointed out.