Log in

View Full Version : TMC causes aliasing?

dbzgundam
7th June 2004, 00:56
I'm sure people know about this already, but unless I use the -1 SE value, some edges of my footage become aliased, and distorted. Although, if I don't use 0-30 values, not everything is deinterlaced. I'm dealing with Hybrid content, and will just leaving it at 29.97fps.

MPEG2Source("H:\IRC\BakaVille\d2v.d2v")
Telecide(order=1,post=4)
TomsMoComp(1,-1,1) <------ With positive SE values, aliasing occurs.
LanczosResize(640,480)

Positive SE values:
http://www.esenet.com/baka/aliased.png

-1 SE Value:
http://www.esenet.com/baka/lackinterlace.png

And here is an example of why I can't use the -1 value. And it ONLY seems to effect the fake film scratches that were used in the video:
http://www.esenet.com/baka/falsefilm.png

Any solutions to both problems? It can't be IVTC'ed so that's out of the question.
Guest
7th June 2004, 12:49
Why do you say it can't be IVTC'ed?

Any interpolation is going to mess up diagonals.
AS
7th June 2004, 14:10
Well, I happened to have dealt with Tenbatsu! Angel Rabbie OVA so...



Telecide(order=1,post=4)
TomsMoComp(1,-1,1)
LanczosResize(640,480)



All you're doing here is disguising the artifacts from postprocessing using a supersampling (mostly) in vertical direction.

The best solution is to do correct field matching for all the frames, giving you the best quality and resolution. It only took an hr or two for me, with the help of yatta.
dbzgundam
7th June 2004, 20:36
Originally posted by neuron2
Why do you say it can't be IVTC'ed?

Any interpolation is going to mess up diagonals.

The OVA contains various amounts of hybrid pans and special effects.

The best solution is to do correct field matching for all the frames, giving you the best quality and resolution. It only took an hr or two for me, with the help of yatta.

I've never been able to get YATTA to work.
scharfis_brain
7th June 2004, 23:53
please read the tmc-documentation. It explains the -1 SE.
-1 SE does a height doubling using edge directed interpolation.
-1 SE is not deinterlacing.

btw. applying a deinterlacer after telecide isn't good...
dbzgundam
8th June 2004, 02:39
Originally posted by scharfis_brain
please read the tmc-documentation. It explains the -1 SE.
-1 SE does a height doubling using edge directed interpolation.
-1 SE is not deinterlacing.

btw. applying a deinterlacer after telecide isn't good...

I have terrible skills with Deinting, I'm using a script someone else gave me.