Ok your CQ values explain what I've felt looking at your samples: new moon is ok, kpax should be better (I did not look Matrix).
Actually, I did not use Tmpgenc since years, but for what I remember, I always targeted a CQ of 70 (or upper). Between 65-70 its good, under this I find the result too bad for my eyes.
This is the template of the script I used when I did KVCD:
Code:
DGDecode_Mpeg2Source("PATH\NAME.d2v",cpu=4,idct=7)
ColorMatrix(d2v="PATH\NAME.d2v")
#TDeint(tryweave=true)
#FieldDeinterlace(blend=false)
#LeakKernelDeint(order=1)
LanczosResize(...)
LimitedSharpenFaster()
LRemoveDust_YV12(17,1)
#MergeLuma(Blur(0.2))
#MergeChroma(Blur(1))
ScriptClip(" nf = YDifferenceToNext()" +chr(13)+ "unfilter( -(fmin(round(nf)*2, 100)), -(fmin(round(nf)*2, 100)) ).TemporalSoften( fmin( round(2/nf), 6), round(1/nf) , round(3/nf) , 1, 1) ")
Addborders(...)
The 3rd commented lines are if the source is interlaced (not your case here)
The 2 "Merge..." lines are only if the CQ is still too low withtout them. I uncomment first the Chroma , then the Luma if it is not enought.
You can find all my filters here:
http://www.kvcd.net/downloads/Avisynth_All_Plugins.rar
Note: why are you adding uneven borders ? 128/144 ?