bkev said:
Was the script custom made for this film...
Yes, it doesn't work on Blues Brothers. ;) Joking aside, it should work good on other sources with similar chroma noise, tested it on my JSC ESB capture and it worked very good, as with all derainbowing and dotcrawl filtering it isn't bulletproof, there's even minor artifacting from it in the sample I just posted. g-force script is made to be applied after IVTC, most derainbowing/dotcrawl filtering are actually most effective or even needed before IVTC.
source = last
sigma = 40
filt = source.FFT3dFilter(sigma=sigma,plane=4,bt=3)
idx1 = filt.MSuper(pel=2,sharp=1)
bw_vec1 = MAnalyse(idx1,isb=true, delta=1,overlap=4,plevel=0,dct=1,chroma=false)
fw_vec1 = MAnalyse(idx1,isb=false,delta=1,overlap=4,plevel=0,dct=1,chroma=false)
bw_vec2 = MAnalyse(idx1,isb=true, delta=2,overlap=4,plevel=0,dct=1,chroma=false)
fw_vec2 = MAnalyse(idx1,isb=false,delta=2,overlap=4,plevel=0,dct=1,chroma=false)
idx2 = source.MSuper(pel=2,sharp=1,levels=1)
pcorr = source.MDegrain2(idx2,bw_vec2,bw_vec1,fw_vec1,fw_vec2,thSAD=800,thSCD1=150,thSCD2=75,plane=4)
corr1 = Median1(source,pcorr,filt,chroma="process")
idx3 = corr1.MSuper(pel=2,sharp=1,levels=1)
corr1.MDegrain2(idx2,bw_vec2,bw_vec1,fw_vec1,fw_vec2,thSAD=400,thSCD1=150,thSCD2=75,plane=4)
Median1(corr1,last,filt,chroma="process")
MergeLuma(last,source)
#######################################################################################
Function Median1(clip input_1, clip input_2, clip input_3, string "chroma")
{# median of 3 clips from Helpers.avs by G-force
chroma = Default(chroma,"process") #default is "process". Alternates: "copy first" or "copy second"
Interleave(input_1,input_2,input_3)
chroma == "process" ? Clense(reduceflicker=false) : Clense(reduceflicker=false,grey=true)
SelectEvery(3,1)
chroma == "copy first" ? last.MergeChroma(input_1) : chroma == "copy second" ? last.MergeChroma(input_2) : last
Return(last)
}
Dunedain said:
Very nice, msycamore. :) Getting the faces right when color correcting the Star Wars trilogy is always the hardest thing (dark_jedi and You_Too know all about that :) ). It's so hard to restore that accurate natural fair look with the sources available. Leia isn't too far off, but still red, Daine Jir, more reddish yet, but perhaps that can be adjusted a bit. :)
Thanks! Haven't seen their accomplishments but yes the skintones are of course the most crucial part when color correcting, there are several reasons why my first upload ended up looking like it did, uncalibrated monitor, lack of care and in a way of getting it over with so that I could continue with THX 1138 which I still after all this time haven't finished. :( But the positive response I got with this one even though the DVD looked like shit (IMO) makes me want to do a much improved one, especially when I know how much better it can look.
Didn't know who the hell Daine Jir was but now I know after some google search, you learn something new everyday :) I guess every stormtrooper in the film have a name these days as well... I agree that the colors in the sample are far from perfect but I don't go scene-by-scene, instead just reel-by-reel to find a middle ground. The reasons for the screencaps are only to show the excellent derainbowing by g-force, not the colors as they varies quite much, had I taken a screenshot from a second later or earlier it would perhaps look too green or to yellow, that's why screenshots doesn't work so good with this source. I'm glad you like the improvements though. :)
Just an example of the variations in levels I mentioned earlier... the same settings that I used on the previous video sample:
^^ brightness and contrast looks quite ok, we cut to Han and back to Leia and boom the levels are changed:
^^ the colors are actually much better here, but the brightness and gamma is way off. This happens quite often in this telecine, sometimes even in the middle of a scene. This makes setting a correct white and black point quite difficult sometimes.
Darth Mallwalker, when you have time for it, could you please check this scene on your copy:
Will try to find the numbers for you, curious if it is a mastering defect present in all copies of this LD. Thanks.