[MPlayer-users] TomsMoComp deinterlacer
Robert La Ferla
robertlaferla at comcast.net
Sat Feb 26 02:47:31 CET 2005
I was perusing the DScaler web site and found that there is a new
deinterlacing algorithm. I'm not sure but I think it is superior to
Greedy2Frame. Is this available for MPlayer? It sounds fantastic! I'd
love to see it in action.
Here's a link to the page and source code:
http://sourceforge.net/project/showfiles.php?group_id=7420
*TomsMoComp*
This method is the latest and greatest method from Tom Barry. This new
method improves on some of the older methods in two important ways.
1. Where there is motion, the bob pixels are generated by looking
along a series of diagonals as well as simple the ones above and
below.
2. Also when there is motion the code will try and spot slowly moving
objects and will try and maintain the detail on those objects.
The combined effect of these two searches is a big jump in picture
quality and also a big jump in CPU usage.
------------------------------------------------------------------------
* Who Should Use It? *
Everybody with a fast machine. 1 GHz is about the minimum required to
use this method.
------------------------------------------------------------------------
* The Settings *
There are only two options to set for this method:
1. Search Effort: This setting will control how much CPU is used by
the method by controlling how big an area the method will search
looking for matches. A Search effort (SE) of 0 means that all
pixels are smart bobbed and is not recommended. SE = 1 does no
motion compensation, but doesn benefit from the diagonal bob
interpolation. SE > 1 starts to do true MoComp with increasing CPU
demands. You should use the highest value that allows you to run
any other filters etc withuot dropped frames.
2. Use Strange Bob: This was some test code added by John Adcock to
try and reduce the number of errors in the diagonal searches. It
is not recommended.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mplayerhq.hu/pipermail/mplayer-users/attachments/20050225/74e21ef7/attachment.htm>
More information about the MPlayer-users
mailing list