How to Remove Dancing Pixels with Avisynth when Denoisers don't work?

[sanlyn]

I don't think your crop and addborders did anything screwy, but I still do that kind of thing at the end (or after your SeparateFields line, while it's still progressive). You did change the luma values because of your ColorYUV lines afterwards, so those borders are no longer a proper black.

Point taken. Sometimes I forget about that.

[sanlyn]

I'm guessing Sanlyn put the Crop at the top to lessen the encoding time slightly. Should I just leave the crop at the top and put the add borders at the end of the script?

The crop had little to do with encoding time - unless you want to count the time it takes filters and encoders to fiddle with 12 pixels of head-switching noise in each frame. So I guess you could say it helped, if only a little. Actually, it was at the top so that I could use a histogram without the borders interfering. So I just left it there. But as manono has noted, adding borders early and then making color/levels changes can change the "black" of the border. So you might want to use after any ColorYUV or Tweak work, etc. In the video I made, the borders were remasked anyway in TMPGenc.

Other than removing head switching noise, the crop centered the image. You wouldn't want to leave it cropped that way thru the script: many plugins work only with mod-16 or mod-8 frame sizes.

Hmm, I see we have questions . . .

Sanlyn I have only just now got an 8-core PC so I've been testing this script you gave me:
....
Your original script was for making a progressive video.

No, the script treats the video as progressive. "Input_Type=1" is for progressive video, not interlaced. That input mode is designed to help repair botched deinterlace jobs, which is what that clip is. The genius who performed that deinterlace went so far as to use a method that discards half the original fields and then used some kind of field-blending technique (NOT recommendeed!). There is no deinterlace statement or processor in the script.

Since Blurayndoesn't support 60 FPS for 720 x 480 I can't do that so I'll have to keep it interlaced.

It's 30 fps, not 60.

Also I highlighted the parts of your script which I can't use because it slows down the encoding time too much.

I ran the script in 2 steps. OK with me, my script was just a suggestion. There are dozens/hundreds of ways to do it. The operative word here is "script", not "scripture". ( pun intended).

I'm able to get 10-11 FPS encoding speed with your script (minus the highlighted parts). I've modified it a bit here:

Keep or modify any change at will. I got better results with NeatVideo anyway, but I usually refrain from suggesting filters that users don't have. Looked somewhat better, but it still didn't fix all the problems.

I'm using the Medium setting for McTemporalDenoise instead of Low like in your script. Is that OK?

Ok with me, it's your choice. Medium softens more.

I noticed that AddGrainC actually covers up the remaining dancing pixels in the still frames that RemoveSpots didn't remove. Is that correct? Does AddGrainC in this script just add invisible grain to cover up noise?

The fine grain added with low settings of AddGrainC is too fine to cover spots. It was added to help debanding and deblocking. RemoveSpots also helps smooth contour shimmering and to stabilize rapidly shifting chroma noise patterns. Invisible grain? This is the first I've heard of it. It doesn't really mask noise, it just gives the impression of softer blended edges. The technique is also used by DeBlock and DeBlock_QED. but in a different way.

I thought the reason you suggested TemporalSoften was to remove the flicker in the bottom right of the picture?

There is no flicker in the clip. Even if there were, TemporalSoften won't fix flicker. I think you might mean "flutter". TemporalSoften was used to smooth some of the wiggles and the shifting macroblocks and banding, which is what you refer to as flicker. TemporalSoften like most temporal smoothers does soften somewhat, which is why I used a low setting. It also helps a bit to calm the hard-edge look of exaggerrated highlights (look at the announcer).

Is TemporalSoften the latest version of TemporalSoften2?

TemporalSoften isn't a plugin. It's an Avisynth built-in.

I moved TemporalSoften above McTemporalDenoise otherwise it slows down the encoding time. That won't cause any problems will it?

TemporalSoften works only in YUY2, the others (except QTGMC) are YV12 only. I don't see that it would cause problems.

Can I assume that I don't need CPU=4 or CPU=6 because McTemporalDenoise is a better denoiser and deblocker? If so, which is better, DeBlock or Deblock QED? For whatever one is best what script should I use for it to give maximum deblocking strength?

You can turn deblock on in MCTemporalDenoise (at medium it's off by default) and set any strength you want. DeBlock is faster and stronger than QED. The "CPU" parameter you mentioned isn't that strong unless you set it that way, but every bit helps.

I assume that if there's nothing to deblock then it won't do any harm to the video and if there is something to deblock then it will remove it without harming the video?

Macroblocks and banding are big problems in this clip. Deblockers are usually specialized smoothers of one kind or other.

What does the sigma setting do?

It's explained in the documentation for FFT3D. Think of it as a strength setting. The higher, the more it repairs, and the more it blurs. Very high values can produce fog effects.

In McTemporalDenoise I noticed that "useEEDI2" is already false in Medium settings. So should I just remove that part of the code from my script then or is "true" better?

EEDI2 is partly an anti-alias filter. I think it's better with anime. Because I switch MCTD settings a lot, I leave EEDI2 = false. If it's already false in the defaults, leaving it there won't matter.

Is LSFMod(strength=50) a safe value to use for any video of this quality? Should I try a higher value?

Higher values are up to you. It sharpens everything, including grain. The default is 100.

Can you suggest any other McTemporalDenoise settings that you think will get better results?

I used settings that seemed to get some desired results. With damaged sources that are more noise and abuse than they are actual video, MCTD can be very aggressive. I tried "high" at first (results were way too soft, and most problems were still there), then "medium" (too soft), and settled on Low with some defaults turned on. The best way to tweak and learn MCTD is to just use it. You can always apply it lightly and then try to address some remaining noise with other plugins, or set it high and use other filters to recover some detail. But the biggest problem with this clip is that somebody found so many ingenious ways to mess it up.

Apart from the stuff in my script I also tried these but I can't use them because they slow down my encoding time: . . .

If speed is your top priority, go for it. Some of this video's damage can't be fixed with the tools available to us peasants. at any speed. We just do what we can with the tools we have.