What does DVD Rebuilder do DVD Shrink HQ doesn't?

[FredThompson]

Currently, I use DVDStripper followed by DVDShrink 2-pass HQ mode to inhibit blockiness. "Regular" transcoding such as that with older DVDShrink builds would choke on scenes like the gauzy shades at the beginning of "Matchstick Men." The 2-pass HQ mode does a pretty good job of preventing those blockies.

I see in the various stickies about DVD Rebuilder that it can provide quite a bit of manual control over bitrate allocation. OK, understood.

Assuming manual tweaking isn't going to be used (except menus and still images will probably be left uncompressed for "simple" titles) and our goal is to create a DVDR-sized backup of a >4.3G source image, what are the benefits of using DVD Rebuilder instead of DVDSHrink 2-pass HQ mode?

[Boulder]

The biggest difference is probably the fact that Shrink is a transcoder whereas DVD-RB uses mostly encoders. ReJig can be used too so it would be the option to do Shrink-like DVDs.

Shrink uses stuff that is already there like motion vectors and quantizers. When using DVD-RB with encoders, everything is created from scratch and everything's based on the video itself.

When the need for compression is very low, I'd say something like max 10%, transcoding might produce similar or higher quality than re-encoding.

Then again, you can use all the AVS stuff with DVD-RB which is a huge plus:)

[FredThompson]

That's why I'm asking. I typically see comments about 10% being the threshold where people recommend encoding again. When I've run the common 2-hour dual-layer disc through DVD Shrink HQ, I can't really see a quality difference. Granted, 4G/hour is an incredibly high bitrate, at 4X teh rate of MPEG2 satellite in North America.

I can certainly see where scripts could come in real handy with poor transfers or format conversions, though.

[Boulder]

Dealing with truly interlaced DVDs produces much better quality with DVD-RB than any transcoder, forgot to mention that.

The video pulsing can sometimes be very annoying in transcoded movies, especially if the video is noisy.

[FredThompson]

I've never seen that. Can you suggest some commonly-available titles where this happens?

Then again, does DVD Rebuilder support the RoBa method (a la DVD2SVCD) for CCE? Maybe the time to complete would be comparable to DVDShrink HQ.

[Rockas]

Then again, does DVD Rebuilder support the RoBa method (a la DVD2SVCD) for CCE? Maybe the time to complete would be comparable to DVDShrink HQ.

No Rebuilder doesn't support Roba but is has its own Q calculator called "One pass VBR (w/analysis)" under CCE Options (Menu Options)

keep it UP

[FredThompson]

Can you give an approximate comparison of time for this one-pass VBR method vs. DVDSHrink HQ?

Is it roughly the same, 50% longer, 300% longer? I'm just wondering about a rough comparison.

[Boulder]

Originally posted by FredThompson
I've never seen that. Can you suggest some commonly-available titles where this happens?

Can't tell you any specific titles to check, but any dirty source should show the pulsing quite easily, provided that the compression is high enough, of course.

Then again, does DVD Rebuilder support the RoBa method (a la DVD2SVCD) for CCE? Maybe the time to complete would be comparable to DVDShrink HQ.
As Rockas mentioned, there's the internal OPV method. I'd still recommend using RB-Opt for OPV as it also allows tweaking parameters for CCE, for example going to 15-frame GOPs which gives a nice compressibility boost while keeping within DVD standards. I don't know which one predicts the Q value more accurately, but I've had good results with RB-Opt.

[Boulder]

Originally posted by FredThompson
Can you give an approximate comparison of time for this one-pass VBR method vs. DVDSHrink HQ?

Is it roughly the same, 50% longer, 300% longer? I'm just wondering about a rough comparison.

I can't say anything specific, but without any heavy AVS scripting, the encoding should easily go faster than realtime with any modern CPU. So the time difference should be rather small. Of course, predicting takes a certain time.

[FredThompson]

I'm using 2.4 GHz or faster here. Shrinking runs me around 1.5 hours or so. I guess it's time for me to do some testing. As long as the nightmare days of 16-hour encodes are in the past...

[jptheripper]

mine encodes are 2 hours with shrink, 28 hours with rebuilder on my slow p3, and the quality difference is so great with cce that it is worth the wait.

[TheSeeker]

With a 2.4 ghz proc. you can probably expect a 1+2 pass cce multipass encode take about 2.5-3.5 hours approximately. Maybe as much as 4 if its a longer movie. But this time increases quite a bit if you use any sort of avs filters. As far as RB's advantage over DVD Shrink. Mainly its quality and its options. The ability to use the numerous avs filters and custom matrices is essential sometimes to getting a nice quality output. That is if we are talking about low bitrate situations. That is another thing. Try and make your decision for transcoder vs. encoder more on the avg. bitrate that will be attained as opposed strictly to the percentage of reduction. For example I recently backed up the first season of Friends. man those dvds are horrible!! at only like 15 percent reduction the avg. bitrate was a lousy 2400 or so. It was absolutely crap and if I had run in through shrink it would have turned out horrible. So first figure out some approximate avg. bitrates and then you can make an informed decision on either shrink or dvd rb. I highly recommend rebuilder and its now almost the only thing that I use. Oh one other thing. If you ever think that these movies you are backing up might be watched on a larger hdtv or something then i would go with rebuilder for most encode because you WILL see many artifacts that you cant see with a smaller regular tv. Believe me. I was using shrink for most of my encodes and I just recently perchused an hdtv and now alot of my movies have alot of blockies. So pretty much use shrink for movies that are high bitrate and low compression, and use dvd rb for everything else. Including any interlaced material. You could even use procoder with dvd rb instead of CCE because its said to handle interlaced and low bitrate sources much nicer than cce does.

[TheSeeker]

Originally posted by jptheripper
mine encodes are 2 hours with shrink, 28 hours with rebuilder on my slow p3, and the quality difference is so great with cce that it is worth the wait.

Thats using alot of filtering yes? Or your doing like 6 passes.

[Boulder]

Originally posted by TheSeeker
For example I recently backed up the first season of Friends. man those dvds are horrible!! at only like 15 percent reduction the avg. bitrate was a lousy 2400 or so. It was absolutely crap and if I had run in through shrink it would have turned out horrible.

That's definitely one DVD release which needs some serious AVS processing..a real piece of crap IMO.

You could even use procoder with dvd rb instead of CCE because its said to handle interlaced and low bitrate sources much nicer than cce does.
ProCoder's biggest problem is that you can't use custom quant matrices which in turn doesn't do good on low bitrate encodes as the standard matrix is not suitable there. On high bitrate interlaced encodes, ProCoder is the best you can use.

[FredThompson]

There's an encoder comparison over at VideoHelp with screen grabs. I realized a single frame is a horrible way to compare encoders and also the caliber and depth of the people who post over there but: http://www.videohelp.com/forum/viewtopic.php?t=235665 (Kindly ignore the troll trip I stumbled over...)

The source in this comparison is DV but which type? Who knows? COuld be 4:2:2, could be 4:1:1.

I've bounced between ProCoder 1.5 and CCE 2.66 a number of times. With 2 passes, ProCoder sure seemed to give far better results than CCE. Of course, it also took quite a bit longer. Maybe ProCoder's 2 pass mode does a far beter job of analysis.

Currently, when I use DVD2SVCD to process DV I've shot (either PAL or "fixed" NTSC.) I use the RoBa plugin because the quality is quite high and the speed is nice.

All of that was background my next question:

When using DVD Rebuilder and ProCoder 1.5 for interlaced material, how does the result compare with CCE 2.66 and the one pass VBR with analysis mode, both using dynamic cell bitrate setting?

Yes, I know, I'm already running my own tests. Am hoping to tap the experience of folks here as well. If the general recomendation of experienced users is heavily in one direction, it's probably pretty good advice.

(Nice GUI for DVD Rebuilder, too.)

[rte5]

Originally posted by FredThompson
I've never seen that. Can you suggest some commonly-available titles where this happens?

I think I've seen this blinking on some of the "Are You Being Served" disks with the most episodes. It's really annoying. The background is blinking. Rebuilder did a great job on them, but no subtitles here and rebuilder doesn't save closed captions.

[jdobbs]

I run a system with a Athlon XP 3200+ processor, 512MB 3200 memory, and a fast 250GB hard drive. It normally takes about 90 minutes to do a typical movie using OPV. A two pass encode takes about 2 hours and a three pass takes 3. I consider any more passes a waste. I'm dumbfounded when I hear of 16 hour encodes. I'm not sure what kind of processor could take that long. A 28 hour encode has to be running on a steam powered processor (sorry jp ;) )

[Rockas]

The computer I use for my backups it's only an Athlon xp+ 2000 with 1GB RAM and two hdd of 80GB each (I use both on rebuilder like: Source on E: - Work on D: - Output E:) I know it's a bad computer in our days but it never take more than 3 hours (in really big movies) to have my work done.... by the way I use 2 passes on rebuilder (1+1 to CCE).

JP... you really must change your machine :)


keep it UP

[FredThompson]

My 2200 with a shared drive and Shrink HQ takes about 2 1/2 hours for a full job. Speed seems about the same as DVD ReBuilder with CCE. Is there a way to run ProCoder with a predicitve single pass?

[robot1]

ProCoder2 has an OPV mode (CQ mode), but it has some problems:
the scale goes from 0 to 1 in steps of .01, but the really usable values are from 0.90 to 1 (1 is the best).
For every .01 change in this parameter, there is a great change in final size, so you can't predict well the final size.
I use it for my DV conversions, when size isn't an issue, setting it to 1, but never used for DVD backups.

Anyway, if you want to test OPV and Procoder, there's a way in DVD-RB: select CCE mode, and in the CCE Options sselect CCE 2.50 and One Pass VBR mode. In setup you have to use ECLPro in place of ECLCCE for the CCE 2.50 path. I don't think the final size will be good, but if you test, please report results.