The Sopranos, prob. been done to death but have to ask

[mrwhitethc]

The Sopranos Season 1, I'm trying to get the best quality without splitting each disc to 2 discs. So far this is what I've gotten to be close to the original.

DVD Rebuilder 0.72
CCE 2.5
VBR Bias 25
Quality Prec 32
2 Pass
UnDot().Deen()
DGMpgdec120b6.dll

Everything seems to come out well except in the psych office with the wood grain walls. Even when the camera is still you can see the Mpeg blocks moving on them. Anyone else see this? If not what settings should I use? So far I can't seem to see them anywhere else. I know this series has noise just like sex in the city so hopefully if this helps here it'll help there as well.

[TheSeeker]

I would try a lower Qual prec. Your giving more importance to flat color areas by making it higher than the default. I would try 16 and see what happens. Maybe go up to 20. But not too much more than that for regular movies.

EDIT: Maybe try it without the deen() as well.

[mrwhitethc]

Thanks for the fast response, I'll try that tonight and post back with the results.

[borgraf]

also, try tweaking the bitrate of that segment with RB-Opt (http://forum.doom9.org/showthread.php?s=&threadid=75202)

[mrwhitethc]

Has anyone else tried tackling these series with stellar results or did everyone just leave DVD-RB to it's defaults and just take that as it was as good as it gets? All the help so far very much appreciated, i'm going to try TheSeeker's suggestion first and try a lower Qual Prec of 20 without the use of Deen(). And do all HBO series has that faint noise running through them? I noticed Sex and the City had it as well. Maybe once we figure out what works this needs to be a sticky or something.

[TheSeeker]

Oh you might want to try Procoder 2 as well.. I have heard it is better with interlaced and low bitrate video. I know it is better at encoding my Dragonball GT dvd's that Im doing right now.

[sweetness]

Hi mrwhitethc
Ya that dvd is aroung 52 to 55% reduction, that's tough on any encoder. what i did was i used 3 passes with RB-Opt to added KVCD Notch matrix. looks ok and anyone else who watched it couldn't really see anything.
And yes the sopranos is a little noise even the second season.
If you're still not happy after trying different settings you just might have to split the thing.(even good brand media are cheap now.)

[mrwhitethc]

the good with the bad, very little moving on the wood in the psych office but now around hard edges, jpg artifacts, is the only way I can describte it. Before edges seemed clean most noticable around the opening text but also around characters. Gonna give procoder a shot and see how that turns out.

[mrwhitethc]

sweetness did you leave all the other DVD-RB settings default like the qual prec, vbr bias and just change the matrix?

[mpennel]

This was one of the first DVDs I ever tried to copy a couple of years ago, and I've pulled my hair out. I finally decided that good enough was good enough, and left it alone. I thought about breaking it up into 2 discs instead of one, but never bothered...Just my 2 cents worth.

[sweetness]

my settings are qual prec 16(20 is good too) and vbr bias 30 and 3 passes.

instead of waiting for the whole movie to encoded try pausing after a few segments are encoded and then view them with media player. this way if you don't like it you can abort faster.

mpennel by be right, you will just end up pulling your hair out over this.

[mrwhitethc]

I tried ProCoder 2 left it to Auto with EclPro and Mastering Quality and so far that is the worst one of the bunch. Everything in the foreground looked great but the backgrounds were horrible, mpeg blocks moving everywhere. So scratch that one off the list. Is there any merit to lowering the res 704x480, right now DVD-RB is telling me I'm getting a bitrate of 2,800 high and 2,100 low which is hovering around SVCD bitrates. Maybe bring another filter into the mix to smooth out some of the blocking?

[TheSeeker]

Yea if you havent tried it yet I would try and add deen() to the mix. THat should smooth out the blockies. Strange that you had such an experience with Procoder 2. I have always liked it...

[Trahald]

HBO stuff has so much grain in it it is a bear to reencode. you can try higher settings on undot.deen or a stronger filter but comes with the penalty of smoothing the detail also. usually when i process that stuff i deal w/what i get. i use 16 for qual prec w/ undot().deen()

[mrwhitethc]

Actually I did have undot().deen() when I used procoder I never took it out from last last time I tried VBR 10, Qual Prec 4. Trahald so would you say, like sweetness mentioned, that qual prec 16 and vbr 30 would make the most sense? Is there any sense in making the res 704x480? Or maybe using something like unfilter(-5,-5) or adding something else to the AVS script? It seems that everything in the foreground for the most part is about as sharp and clear as it should be it's really the backgrounds that are blocky and you can really tell it's been MPEG'd sometimes. I've tried to use RB-Opt but I'm running CCE 2.5 so I can't change the Matrix settings unfortunately.

[rte5]

I settled on DVDshrink with Deep Analysis and high quality, since it looks about the same as rebuilder and you can keep the closed captions.

[jdobbs]

Originally posted by rte5
I settled on DVDshrink with Deep Analysis and high quality, since it looks about the same as rebuilder and you can keep the closed captions. I have doubts about that... but to each his own. :)

[mrwhitethc]

very true indeed, with 50% compression there has to be a lot of quality loss with the trancoding method. Though I do have a question I remember DVD2One had 2 modes one was variable and one which dropped the bitrate down as a whole. This was suppose to allow for better quality on longer movies. Is there any truth to this and is this available in DVD-RB?

[jdobbs]

That would be an apples to oranges comparison because DVD2One works within the compressed domain. It's "variable" (as I recall) relates to how aggressive it is for different parts of the movie (as opposed to a uniform reduction).

In DVD-RB/CCE variable bitrate (VBR) is used for all DVD-RB encodes. So bits are allocated based upon an analysis of the demand. RB also uses dynamic bitrate allocation based upon the originally encoded stream.

Unquestionably a full scan, and resulting reencoding with VBR is the best way to get the highest available quality. Only the uninformed would argue to the contrary.

[mrwhitethc]

I wasn't trying to compare DVD2One with DVD-RB, I just remember DVD2One had 2 modes one for everyday movies and one for very long movies. I was wondering if the Dynamically Assign Cell bitrates was one such setting, would it be better to not do this for longer movies?

[jdobbs]

That's what I'm trying to tell you. It doesn't apply.

[mrwhitethc]

man remind me next time to drink more coffee :) I re-read your post and saw exactly what you were saying now after a cup and a half of the good stuff. I did find something hopefully that will help, I'm using CCE 2.5 and in the Doom9 guides it says that qual prec for 2.5 is X*64/100 so setting it to 16 or 20 was actually setting it to 10.24 and 12.8 respectively. I'm going to try it at 32 which actually means 20.48 with undot().Deen() in the mix. Hopefully with this it might start to be looking like what everyone else has. Is there any merit to deinterlacing this? If my thinking is correct you would loose some sharpness and detail but encoding just the frames and loosing the interlaced fields should help compression or have I know drank too much coffee? Plus going to back to the original I'm seeing just how much compression there is already on these around edges and such looking at them on my monitor.

[jdobbs]

I think you may have that backwards. v2.50 uses the 1-100 scale. So 16 would translate to 24.

[mrwhitethc]

Wouldn't it just be easier to say in the guide for CCE 2.5 X*1.5 = Qual Perc Value? :D Now that that is cleared up, any thoughts on deinterlacing?

[jdobbs]

I can't comment on the guide(s) -- as I didn't write them. As for deinterlacing --- I recommend against it.

[mrwhitethc]

It's really looking like UnDot().Deen() and vbr 30 qual prec 20 is the best approach. I'm just going to have to accept the output with such a low bitrate, I'll probably up it to 4 passes just for my peace of mind. Can anyone comment on why another post said that vbr 10 and qual prec 4 was the best for majority of movies? I know to each his own but are those low settings suppose to capture the most detail when high bitrates are available like above 3500? Anything really damaging going with those low settings? Thanks for everyones input on this series. Too bad all these HBO discs have so much noise, I just did a test of Star Trek Voyager, 4 eps at about 8gb per disc each and with the default settings in DVD-RB and 3 passes it came out very nice. Source material was very clean and the backups had only a very slight hint of blockiness every so often not bad for 50% reduction.

EDIT: Or in my case with CCE 2.5 qual prec 12, thanks jdobbs :D

[Boulder]

If it's interlaced material, you can't just use UnDot().Deen() . You must first bob, apply all filters and then re-interlace the video. Otherwise you'll ruin it.

For a particularly tricky material, I recommend resizing to 704x480/576. If you have a display device that has an overscan area, you could use

AssumeTFF() # assuming TFF video
Bob(0,0.6) # or better yet, a smart bob
Crop(0,4,720,568)
UnDot()
Deen()
LanczosResize(672,544)
AddBorders(16,16,16,16)
SeparateFields()
SelectEvery(4,0,3)
Weave()

for PAL or

AssumeTFF() # assuming TFF video
Bob(0,0.6) # or better yet, a smart bob
Crop(0,4,720,472)
UnDot()
Deen()
LanczosResize(672,448)
AddBorders(16,16,16,16)
SeparateFields()
SelectEvery(4,0,3)
Weave()

for NTSC.

[mrwhitethc]

Thanks everyone for all the help, I tried Boulder's method and did it at 4pass, CCE set to natural and the KVCD Notch Matrix and the blocking was slightly decreased, I think that is how I'm going to leave it. It took a whole day to complete but I think it is not going to get better than this at 50% reduction. Thanks again everyone.

[Boulder]

On second thought, The Sopranos is very likely progressive. Maybe you have the same situation as someone had with Six Feet Under, see that thread.

Also it's been verified by many that you don't need to do more than 2 passes so that'll save you a lot of time. I use OPV myself. You might also want to try RemoveGrain().RemoveDirt() instead of UnDot().Deen().

[mrwhitethc]

Ok I'll give the Remove Grain & Drit a try, and here I thought I was done! :D Also I happened to be reading about someone compressing a movie off laserdisc, long story short they indicated a very low Q value because of the softness of the picture. Now since this is very low bitrate 2.1-2.5 would there be any benefit to soften the picture slighty, and if so what filter would accomplish this. Also I'm now using CCE 2.70, with what seems like an insane amount of options, I'm assuming that I should do 3 pass since when I set it too 2 pass, in DVD-RB it does a VAF sweep, then does a final pass and moves on or am I missing something.

[Boulder]

A slight UnFilter might help, try something from (-7,-7) to (-3,-3).

DVD-RB probably calculates the vaf pass as one so three passes would mean vaf + 2 passes, which has been found as the maximum you should ever need to do.