Posted by Jukka Aho on 06/15/07 17:55

Den Murray wrote:

> I'm in the US and I have a Cyberhome DVD player - it's region-free,
> but that is incidental to my question.
>
> 1) When I play a PAL DVD from the UK or Australia my DVD player reads
> the PAL encoded DVD but outputs an NTSC signal - to allow my TV to
> correctly display the picture. All is good except that the picture
> can be a little jerky if there is fast movement - I assume that this
> is part of the frame dropping that occurs during the conversion.
>
> 2) I also have an external video standards converter - a TenLab TR21
> - that will convert a PAL signal to NTSC. If I play a PAL DVD and
> output PAL from the DVD and run it through the video standards
> converter and output an NTSC signal to the TV the picture is also
> fine - perhaps of marginally poorer quality than in (1) above, but
> not jerky.
> Can someone explain to me what is different between the two ways of
> getting an NTSC signal to the TV from the DVD and why the difference
> in jerkiness and clarity?

The DVD player probably resorts to frame-doubling (not dropping!) to
keep things in sync. Every 6th frame (or so) will be doubled. Hence the
jerkiness.

The external video standards converter might use frame-blending,
instead - i.e., it it captures, stores and mixes several adjacent frames
together, in a moving window, with varying degrees of opacity. (Think of
the "onion skinning" techniques that cartoon animators use:
<http://en.wikipedia.org/wiki/Onion_skinning>.) This will give a
smoother impression of motion as nothing is ever duplicated straight
away, but at the same time, everything that moves will get faint shadow
images around it from the previous (and following?) frames.

You could try capturing the output of you standards converter to a
computer and analyzing the individual frames there (and perhaps making a
direct comparison to the original MPEG-2 frames on the DVD.) Of course,
a PAL test disc with some suitable test material would come handy here.
(A ball rolling across the view, or a timecode display burned into the
picture, or something similar.)

Professional standards converters may also apply sophisticated motion
estimation techniques. You might want to take a look at these documents:

"Motion Science - A New Option for Standards Conversion"
<http://studio-systems.com/broadfeatures/JulyAug2001/Motion/32.htm>

"The Engineer's Guide to Standards Conversion"
<http://www.snellwilcox.com/community/knowledge_center/engin
eering_guides/estandard.pdf>

"The Engineer's Guide to Motion Compensation"
<http://www.snellwilcox.com/community/knowledge_center/engin
eering_guides/emotion.pdf>

--
znark

[Back to original message]