digitalFAQ.com Forum

digitalFAQ.com Forum (https://www.digitalfaq.com/forum/)
-   Capture, Record, Transfer (https://www.digitalfaq.com/forum/video-capture/)
-   -   Diagonal line noise on VHS video? Luma noise? (https://www.digitalfaq.com/forum/video-capture/6760-diagonal-line-noise.html)

bilditup1 10-15-2015 02:14 PM

Very cool stuff. It's been educational, as always :)
What's your opinion of santiag as a general AAer for progressive video (not deinterlaced - 'natively' progressive)? Or as an AAer/deinterlacer for interlaced animation?

sanlyn 10-15-2015 04:38 PM

Tough question to answer. It depends on the particular video, how badly it's damaged, how its structured, etc. Two of the most difficult glitches to clean are aliasing and dot crawl. Filters for both use similar methods, and all of them can kill detail. Some are designed for interlace/telecine, others for progressive only. QTGMC uses some tricky pattern matching, motion interpolation for smoothing, resizing, masking, etc. QTGMC is not primarily an anti-alias filter. Helping to smooth aliasing is a side effect. MCTemporalDenoise uses SangNom by default if you activate edge cleaning. There are medium-strength aa's (maa, naa, santiag), stronger ones (SangNom, which has various settings), and very strong (supaa). And a lot in between. Usually you experiment with the filter and its settings to see how much you can smooth without destroying everything.

If you look at the typical UTube video where interlaced originals have been deinterlaced by discarding fields, you can usually forget about cleaning up aliasing.

lordsmurf 10-15-2015 05:03 PM

1 Attachment(s)
Both aliasing and dot crawl are solvable. You can't 100% correct it, but at least 80-90% of it can be cleared up. The side effects are minimal.

From my MultiScript:
Code:

# santiag(strh=2,strv=2) # antialiasing
# santiag(strh=4,strv=4) # antialiasing alternate for toons

... and you can stack them.

I've played with DeCrawl (Avisynth) before.

And I've used DotCrawl in VirtualDub (attached).

sanlyn 10-16-2015 02:38 AM

There are many ways to clean up or minimize aliasing and dot crawl. Unavoidably there are other losses involved, so it takes experimentation and compromise to get suitable results. There s no one-filter-fits-all solution.

bilditup1 10-16-2015 04:59 PM

Thanks for chiming in guys. I guess you'd characterize the effects of santiag to be similar to maa? And this is mostly used either early in the filter chain to fix aliasing in the source or later in the chain to fix aliasing caused by (over)sharpening? That's what I've been doing, but don't know if one is better than the other. Looks like yet another thing to test.

lordsmurf 10-16-2015 06:27 PM

Quote:

Originally Posted by bilditup1 (Post 40386)
And this is mostly used either early in the filter chain to fix aliasing in the source or later in the chain to fix aliasing caused by (over)sharpening?

1. Don't oversharpen.
2. It's used where needed. I generally have it later in a script. But placement is determined by not only the initial error, but any subsequent errors reintroduced from filtering. So it depends. Video is not one-size-fits-all, and this is no exception.

bilditup1 10-17-2015 01:03 AM

Quote:

Originally Posted by lordsmurf (Post 40387)
1. Don't oversharpen.
2. It's used where needed. I generally have it later in a script. But placement is determined by not only the initial error, but any subsequent errors reintroduced from filtering. So it depends. Video is not one-size-fits-all, and this is no exception.

LOL, well nobody ever means to oversharpen, but it can happen, and it's not always a matter of turning it down a notch or using Repair() or something. And to clarify - I have used AAers in both cases, but was wondering what the use cases were for these specific filters; iow, whether one is used more heavily to fix errors pre- or post-filtering.


All times are GMT -5. The time now is 02:05 PM

Site design, images and content © 2002-2024 The Digital FAQ, www.digitalFAQ.com
Forum Software by vBulletin · Copyright © 2024 Jelsoft Enterprises Ltd.