For only 5 or 6% reduction, should I use CCE/HC or Shrink?

[stanley_a]

I'm trying to backup a movie (black and white, with some color).
Everything has been removed. This is just the movie, the audio file and a subtitle. I can't remove anything else.
DVDShrink says 96,6%.
DVDRebuilder+CCE says 93,4%.

My question is. Having to make such a small reduction, which program gives me the best result VISUALLY?
Even at this level, do the encoders still produce a better image than a transcoder?

And here's the log of the Rebuilder Preparation Phase:

[12:17:02] Phase I, PREPARATION started.
- DVD-RB v1.28.2
- AVISYNTH 2.5.7.0
- CCE encoder selected.
- Source: MOVIE
- VTS_01: 2.346.135 sectors.
-- Scanning and writing .D2V & .AVS files
-- Processed 178.448 frames.
-- Building .AVS and .ECL files
- Reduction Level for DVD-5: 93,4%
- Overall Bitrate : 5.442/4.353Kbs
- Space for Video : 3.955.206KB
- HIGH/LOW/TYPICAL Bitrates: 5.633/1.571/4.353 Kbs

[minidv2dvd]

i think you already know the answer, but don't want to admit it. if the choice is cce or shrink for anything regarding quality, the only answer is cce. will you see a difference without enlarging an image? maybe not, but that wasn't the question.

[fritzi93]

Either way entails *some* quality loss, as you probably know.

1) DVDShrink transcodes, i.e. removes pixel residual data (DCT coefficients). The purpose is error correction, since there is not an exact match from frame-to-frame using ONLY previous picture + motion vectors in a GOP. That's all transcoding does, remove data to reach a specified file size, although DVDShrink is quite good at allocating the reductions. At low percentages, the result may be indistinguishable from the original. An action movie typically has more motion vector data than, say, a sedate drama, and less residual data, therefore it likely will not transcode as well.

2) DVDRebuilder re-encodes, i.e. re-compresses all data. (Actually, the encoder does that, whether CCE, HC or other.)

Which would be better in your case? Well, the size reduction required is not great, although the bitrate is not very ample. Try both methods and see if you can tell the difference.

[EDIT] I seem to recall a thread on Doom9 quite some time ago in which one fellow maintained that at very low percentages, DVDShrink could theoretically produce a better result, for instance in the region of ~1% for a "talking heads" movie, adequate bitrate, lots of excess B-frame residual data. This assumes the compression required doesn't touch any I or P-frames. Since DVDShrink is biased towards B-frames, in the nature of the case, any artifacts introduced cannot accumulate from B-frame errors. No other frames reference B-frames, only I and P pictures are referenced. Blah blah. :P

[filmboss80]

With such a small amount of reduction, DVDShrink should suffice, and I think it would be faster. I doubt your eyeballs could detect any noticeable quality loss.

[Tedness]

Yeah DVDShrink does a excellent job at quality reduction, but if you're down to less than 70% then thats where I've noticed a big difference on my HD tv.