Dealing with grainy "sky" via seesaw

[Moti172]

Hi
I'm trying to encode an old movie. Some parts of it, especially the skies are very grainy.
What is the preferred filter to use with seesaw in order to remove the grainy parts and sharpen the movie?

I have already tried:
b=mcnr_simple2(frames=3,thY=24,thC=20,thY2=-1,thC2=-1,removdirt=true,LPrad=2.0,LPlosens=1,LPhisens=5)
but I think it too strong.
and
b=a.degrainmedian(mode=0).VagueDenoiser(threshold=2, method=1, nsteps=6, chromaT=0.8)

Thanks

[Mug Funky]

you might want to try applying a heavier noise reduction to the brighter areas - depending on how the film was transferred, it's likely there'll be more noise in the bright parts.

so maybe use your existing denoiser plus:

mt_merge(denoised, clean,denoised.levels(128,.5,255,128,255,coring=false).fity2uv(),u=3,v=3)

that's just a guess... it'll perform half the denoise strength on most of the picture, but apply full strength to the highlights.

[Pookie]

Also, test out Mvdegrain2 from MVtools . Example usage in MVtools documentation. Slow, but pretty amazing.

[Jeremy Duncan]

First use a Temporal denoiser like FluxsmoothT(8). Then apply a spatial denoiser after this if the temporal denoiser isn't enough..

[Terranigma]

First use a Temporal denoiser like FluxsmoothT(8). Then apply a spatial denoiser after this if the temporal denoiser isn't enough..

Fluxsmooth, like dedot, can sometimes cause unexpected results such as creating halos (From personal experiences). He/she'd be better of using TTempsmooth.

[Moti172]

Thanks. I'll give it a try.

[Pookie]

http://img129.imageshack.us/img129/3783/103tm2.th.png (http://img129.imageshack.us/my.php?image=103tm2.png)
http://img50.imageshack.us/img50/4718/103xszm5.th.png (http://img50.imageshack.us/my.php?image=103xszm5.png)

[Moti172]

I guess that this is the correct code:

backward_vec2 = last.MVAnalyse( isb = true, delta = 2, pel = 2, overlap=4, sharp=1, idx = 1 )
backward_vec1 = last.MVAnalyse( isb = true, delta = 1, pel = 2, overlap=4, sharp=1, idx = 1 )
forward_vec1 = last.MVAnalyse( isb = false, delta = 1, pel = 2, overlap=4, sharp=1, idx = 1 )
forward_vec2 = last.MVAnalyse( isb = false, delta = 2, pel = 2, overlap=4, sharp=1, idx = 1 )
a = last.MVDegrain2( backward_vec1, forward_vec1, backward_vec2, forward_vec2, thSAD=400, idx=1 )

b = a.DeGrainMedian( mode=1 )
SeeSaw( a, b, NRlimit=4, NRlimit2=5, Sstr=1.5, Slimit=9, Spower=9, Szp=16, sootheT=80, bias=40 )

What's the differences between MVDegrain2 and mcnr_simple2 ?

[foxyshadis]

Two very different ways of doing essentially the same sort of thing. mcnr has the advantage that it's a lot more flexible, scaling better between really terrible to light touches, but mvdegrain is faster in my experience.

[Terranigma]

foxyshadis, could you perhaps recommend some initial values you'd use with mcnr_simple2 for very grainy sources, that's about even to mvdegrain2 with a blksize of 4 and thsad value of 800?

[foxyshadis]

No, because I use mvdegrain for everything, ever since it came out. Only used mcnr a few times. =p

[Terranigma]

No, because I use mvdegrain for everything, ever since it came out. Only used mcnr a few times. =p

hahaa, ok :p