Macroblocking filter for VirtualDub or Sony Vegas or Final Cut

[choirislife923]

I purchased a DVD of my final high school choral performance from a "professional" video recording service but unfortunately, the quality was not what I expected:



Is there a good filter that will solve the macroblocking? The company squeezed 2 hours and 30 mins of video onto a single layer DVD (with an average bitrate of 3.9mbps) and, unfortunately, the result is not pleasing. I am trying to clean up the video using Sony Vegas and then reauthor it on a DVD-9, but I'm afraid with this quality (or lack of), there's nothing I can do.


Note: the faces were blanked out by myself...on the DVD, the faces are still visible, but I didn't want to post people online without their permission.

[lordsmurf]

MSU Deblock for VirtualDub

[Nelson37]

Any possibility you can get the original footage and re-encode?

At the very least make sure that whoever selected these "experts" knows that they got crap, and paid for it. I could do better than that with an almost 10-year-old, fairly poor quality video camera.

[jagabo]

MSU's Smart Deblocking filter is better than their regular deblocking filter but it's much slower.

http://www.compression.ru/video/deblocking/smartdeblocking_en.html

In the end though, you'll end up with a fuzzy picture rather than a blocky picture.

[choirislife923]

Any possibility you can get the original footage and re-encode?

At the very least make sure that whoever selected these "experts" knows that they got crap, and paid for it. I could do better than that with an almost 10-year-old, fairly poor quality video camera.

i totally agree....the audio is mono in mpeg2 (i thought mp2 wasn't an NTSC DVD standard..), all of the camera angles have different brightness and contrast levels, and on top of that, the encoding is HORRIBLE.

thanks for all the help...i seriously doubt they will let me tweak around with the source tapes, but i'll do my best with msu (smart) deblocking.

[guns1inger]

I like blockbuster for avisynth. Much faster than the MSU options, and with greater freedom and control.

[choirislife923]

just a quick update:
MSU smart deblocker, on my PC, can export-render at 2 frames per second.

MSU deblocker (regular) can render at 12.

Right now, i'm messing around with the regular deblocker to see how things go. If it removes a decent amount of macroblocking, i'll stick with it.

I don't know how to use blockbuster. Can some one give me a sample line (blockbuster.dll {source file})?

[jagabo]

Here's a simple Blockbuster script:

AviSource("blocks.avi")
ConvertToYV12()
BlockBuster(method="noise", block_size=8, detail_min=1, detail_max=10)

Blockbuster, regardless of its name, doesn't really remove macroblocks. It normally adds noise to dark areas to reduce the creepy crawly stuff you often see there. A side effect of this is to make the video less compressible.

[choirislife923]

another update:

alright, after hours of test encodes, i think the MSU deblocker is a good balance between quality and performance. the Smart Deblocker is definitely slightly better, but its not worth leaving my PC on for 3 days straight. right now, the uncompressed RGB AVI looks splended, but the mpeg2 final product looks like crap, so i'm going to have to tweak mainconcept a bit (frameserved from VirtualDub). Thanks for everyones help.

[jagabo]

i think the MSU deblocker is a good balance between quality and performance. the Smart Deblocker is definitely slightly better, but its not worth (the extra time)

That's the way I see it too.

[vhelp]

Hi choirislife923,

thanks for all the help...i seriously doubt they will let me tweak around with the source tapes, but i'll do my best with msu (smart) deblocking.

Instead, I would consider asking for a VHS tape copy. But, make sure that
he uses the "original" source tape, NOT the dvd copy.

( the reason I suggest VHS (over dvd, or even dv tape) is because its most
probable that the person would be insulted in some way, and reject your
proposal all-together. To minimize this, a comprimise might be in order..
hence, a VHS copy )

If you can get it on VHS, at least you'll have the best quality to work with,
and without any macroblock (pixelation) you have at the moment, though not
including your own after you encode it yourself

I realize that VHS is not the best source, but it really is better than what
you have right now. And, all the rest, in terms of quality can be worked
out later. Just be sure you get (or give him) a good brand quality tape,
fwiw.

-vhelp 4007

[choirislife923]

vhelp, thanks for your response. I'll call up the company to see what I can settle for...since they did screw up the right channel audio on the DVD.

but on an entirely different note: i'm done messing around with deblocking etc and i've finally settled for MSU Smart Deblock because it gave the best overall quality. regular deblock had some improvement, but it wasn't worth the wait.

i've tweaked the advance settings in Mainconcept Mpegencoder to my liking, and now I plan to encode in 1 pass VBR. What do you think? I believe since my source is so noisy and the video is of a swing choir (therefore, constant fast action dance movements for 2 hours and 30 mins), I wouldn't have much to gain with a 2 pass....which would take me 71 hours. Would you agree with me?

[guns1inger]

The problem with 1-pass VBR is you have no guarantee on output file size. 2-Pass will at least ensure you get a video that will fit. It would be a shame to put this time and effort into removing macro-blocks, just have them put back because you have to transcode the video to get it to fit.

An alternative would be to output the cleaned version to DV, then do your encoding from that. At least only have to do a single pass with the MSU plugin.

[choirislife923]

The problem with 1-pass VBR is you have no guarantee on output file size. 2-Pass will at least ensure you get a video that will fit.

are you certain? every time i do 1 pass in mainconcept with the numbers that the Videohelp Bitrate Calculator spits out (avg and max bitrates), the output file is spot-on and allows max use of the disc.

[manono]

guns1inger, I've never used MainConcept, but I don't think he's talking about a CQ encode, just an old-fashioned 1-pass VBR, which isn't much different or much better than CBR. Otherwise, I agree 100% with you.

choirislife923, you've got a difficult-to-compress 2 and a half hour video. You've devoted all this time to the project. You might as well finish it off right.

Or else, as guns1inger suggests, save it as a lossless AVI and do the MPEG-2 encoding off of that. Might save time in the long run.

Hehe, I've edited as you've been posting.

[guns1inger]

Even single pass without constant quality misses with many encoders. That the OP has hit the target to date is no guarantee his luck will hold, especially on a project like this. If mainconcept's 1 pass VBR is that accurate, I would question just how variable it actually is. I suspect the bitrate graph would be pretty flat.

[choirislife923]

alright, thanks for everyone's input.

it seems like there's more to the problem that i expected: on the original DVD they sent me, there is a part where the video and audio drops out for about 5 seconds...l'm going to call them and ask them for a new copy and then go from there...all over again.

so for right now, my restoration project is placed on hiatus.

on a lighter note, the DVD looks fine when I play it back through RF cables

[jagabo]

on a lighter note, the DVD looks fine when I play it back through RF cables

Given that you will be adding a second round of MPEG compression artifacts (even if you go with a very high bitrate) the result after deblocking probably won't look significantly better when watched on TV. For real improvement you have to go back to the original (DV?) source