TomsMoComp Probleme with SearchEffort

[Wario]

Hi.. I love the new filter TomsMoComp, but i try to find the best and i have noticed this problem with the SearchEffort

The first picture is with TomsMoComp(1,0,1)
"0 being practically just a smarter bob" (in readme.txt)

*USE SAVE AS... FOR DOWNLOADING PICTURE*
http://www.geocities.com/marioquevillon/SearchEffort_0.jpg

The second picture is with TomsMoComp(1,30,1)

*USE SAVE AS... FOR DOWNLOADING PICTURE*
http://www.geocities.com/marioquevillon/SearchEffort_30.jpg


look the edge of object (rooftop), the edge is not clean
same result with SearchEffort 15,20,25 etc...

why i have best resultat with SearchEffort "0"

Thx for you help!

[trbarry]

I can't see the image but I'll bet I can guess. There seem to be increasing artifacts at SE levels > 5 so I recommend SE=5 for almost all purposes in Avisynth.

That is especially true I think for noisy pictures and maybe those with video text overlays.

- Tom

[Wario]

Did you try SAVE AS.. for downloading picture

and yes my source is noisy (Cable TV) and with SearchEffort 5 my probleme is not 100% resolved because I can see little artifact on edge etc.. The best result is with "0"..

But what do you mean about "smarter bob"

Thx again

[trbarry]

"smarter bob"..

Pardon my sloppy terminology there. In DScaler we've somewhat gotten into the habit of referring to any deinterlace method that relies only on a single field as a "bob" method, even if it does more than line double and doesn't even bob up and down much.

At SE=0 TomsMoComp uses an edge based line averaging technique to try to preserve diagonal lines somewhat, avoiding the stair step jaggies sometimes seen on line doublers.

At SE=1 or above it starts to also look for useful info from the prev and next video fields.

- Tom