what kind of noise is this or just too poorly transferred DVD

[Valky]

Well, since I have now tried almost every de-noise filter there is and many many XviD settings I just cant get these Robin of Sherwood episodes to work.

I have never seen so much noise in DVD-transfer even if know this is old 80's tv-serie. Still I am wondering if the picture is just too faded or burned and maybe I just need to do some sharpening also. It's just that there isn't much to sharpen except that noise..
I have been satisfied with some out-scenes, but since these episode includes these dark forest and castle-scenes a lot, the overall quality is really poor.

Here is one example if you want to download and look of these horrible scenes. I started it at the mainscreen so that you can easily notice how much noise DVD contains on that blue night screen:

http://www.sunpoint.net/~valky/robin.rar

It contains 15mb vob-file. This scene has been particulary ugly when watching using TV-out. No matter what codec I have tried.

Even if I have been managed to get decent quality with DustV5 GoldDust-filter, it still takes too much time for every 50min episode (at least 12 episodes) and after that the picture is veeeery blurry cause it is that already. Also lowering resolution to 512X in these 4: episodes blurs the picture even more, but just been trying to get decent bitrate/pixels when aiming for 350mb/45min.

I guess nobody other hasn't ordered these DVD's from England and had some luck with them?

[sh0dan]

Since you already experimented with Dust I will not go further into that.

The clip you supply is fairly dark, and _very_ noisy.

For AviSynth 2.5 I'd suggest this filter.

#ASYNTHER MPEG2Source (mpeg2dec.dll)
MPEG2Source("D:\dev\avisynth2\avisynth_2_1_test files\robin\robin.d2v")
convolution3d(preset="movieLQ")
temporalsoften(3,8,8,mode=2,scenechange=10)
coloryuv(gain_y=64,off_y=-16,opt="coring")
#histogram()


I don't know if the coloryuv settings are any good - it depends on the rest of your source - it might overbright some scenes. Doing color/brightness corrections on a source a debatable - at least I think you should lower the black areas to be purer black. (gain_y=8,off_Y=-8, opt="coring")

But if you do a bicubic resize down to 512x384 after this, you might get good enough results (you should also remember to crop before conv3d of course).

[Valky]

Hmm..I am not sure what I did wrong here, but with your suggestion the picture goes totally messed with Divx 5.02 and seems to be little bit drunk with 3ivx codec too :)

My Scipt:

#
LoadPlugin("C:\avs25\MPEG2Dec3.dll")
LoadPlugin("C:\avs25\Convolution3DYV12.dll")
LoadPlugin("C:\avs25\BicublinResize.dll")
#
MPEG2Source("E:\dvd_robin\koepala.d2v")
#
crop(10,2,-8,-2)
convolution3d(preset="movieLQ")
temporalsoften(3,8,8,mode=2,scenechange=10)
coloryuv(gain_y=64,off_y=-16,opt="coring")
#
FastBicubicResize(512,384)
#


And I guess I did everything like you suggested..?

[sh0dan]

What's this: "FastBicubicResize(512,384)" ?

What happends if you use the "ordinary" BicubicResize?

[sillKotscha]

Originally posted by sh0dan
What's this: "FastBicubicResize(512,384)" ?

//build into 'Bicublinresize' :)

FastBicubicResize is slightly faster than SimpleResize but results are IMO equal...

regards Sill

edit: d'oh now I know what you mean... his picture is messed up 'cause he simply missed the bicubic coefficients

[kxy]

Originally posted by sh0dan
The clip you supply is fairly dark, and _very_ noisy.

For AviSynth 2.5 I'd suggest this filter.

#ASYNTHER MPEG2Source (mpeg2dec.dll)
MPEG2Source("D:\dev\avisynth2\avisynth_2_1_test files\robin\robin.d2v")
convolution3d(preset="movieLQ")
temporalsoften(3,8,8,mode=2,scenechange=10)
coloryuv(gain_y=64,off_y=-16,opt="coring")
#histogram()


I don't know if the coloryuv settings are any good - it depends on the rest of your source - it might overbright some scenes. Doing color/brightness corrections on a source a debatable - at least I think you should lower the black areas to be purer black. (gain_y=8,off_Y=-8, opt="coring")



Couple newbie questions, if this has already being answered please be nice enough to point me to the right thread so I can read up on it.

1. Why did you use convolution3d(preset="movieLQ") instead of convolution3d(preset="movieHQ"), is it because it is a noisy source?

2. In regards to temporalsoften(3,8,8,mode=2,scenechange=10), can you explain why the particular numbers are used?
And how is temporalsoften different compare to Undot()?

The only thing that I picked is from the avisynth manual.

Starting from v2.5 two options are added to TemporalSoften:

An optional mode=2 parameter: It has a new and better way of blending frame and provides better quality. It is also much faster. mode=1 is default operation, and works as always.
Added scenechange=n parameter: Using this parameter will avoid blending across scene changes. 'n' defines the maximum average pixel change between frames. Good values for 'n' are between 5 and 30.


What does opt=coring do in coloryuv? I didn't find that in the manual. http://www.avisynth.org/index.php?page=ColorYUV
What about Autogain, you think it will be a good idea to add right after gain_y=64,off_y=-16?