Did DVD Shrink v3.2 achieve it's main goal? I don't think so... like, friendly, achieve, because

andy

I caught this statement from ddlooping in the quot;Instant Copy looks better than CCEquot; thread, here.

I hadn't seen this remark before, and found it interesting.  I thought it could bear further discussion on its own.  Now, I like DVD Shrink a lot, and I know ddlooping's done a lot to help further it's development, so let's keep it friendly...

Here goes:

Originally posted by ddlooping
Surprising because I didn't think any transcoder would get good results at 50%.

v3.2 main goal was to equal or better Instant Copy.
We managed to achieve this during testing of the last few beta versions.
None of the beta-testers tried a title at 50% compression though.
At this point, all of us would split to two discs, use CCE, or wait for dual-layer. [/B]

Now, this statement about DVD Shrink v3.2's main goal really threw me.

Has this been achieved?  Unless this can be backed up with something I haven't seen yet, I would say that it definitely has not.

Many of you have seen the Matrix Reloaded comparison I did, here:

Shrink 3.2 vs. InstantCopy 8 vs. CCE. Guess who won? It wasn't Shrink...

InstantCopy prevailed here.  I think DVD Shrink is getting close, though.  The one thing I like about Shrink when comparing various frames, is that, in the ones where I think it betters InstantCopy, it preserves more detail and doesn't wash it out like InstantCopy sometimes does (see Snapshot 17, Tom Cruise asleep, below).

But, then you get snapshots where Shrink botches portions of a scene, particularly elements of the scene that are in motion (see Snapshots 1, 4, and 8, below).

I see this all too often when comparing transcode snapshots.  Boy, if Shrink could correct this, then he'd reign supreme!

Onto the comparison...I did some testing with Vanilla Sky recently; this was about a 78% transcode for this comparison.  See my Matrix Reloaded thread above for more information on methodology and tools used.  ACDSee Classic is a good viewer for flipping between images.

Oh, and I had to take snapshots at 2 frames past chapter stops; when I did them at chapter stops, 1/2 of the Shrink snapshots were binary equivalents of the snapshots from the original DVD.  What is with that??!

These snapshots are InstantCopy 8 versus Original versus DVD Shrink v3.2 Maximum smoothness.

Snapshot 1: Car in motion. Shrink mushes elements (brick wall down to top of car) of right-hand side of scene.  Winner: InstantCopy 8.

Snapshot 4: Tom Cruise in motion. Shrink mushes right-hand side of Tom Cruise (foream, painting at forearm, book in right hand, painting at feet).  Winner: InstantCopy 8.

Snapshot 8: Tom Cruise in motion. Shrink botches Tom Cruise face, mushes underside of bridge in background. Actually, it looks like the Shrink rendering of underside of bridge is more complex than the original frame!  Winner: InstantCopy 8.

Snapshot 17: Tom Cruise not in motion (asleep). Shrink retains more detail and essence of scene; InstantCopy washes it out somewhat.  Winner: DVD Shrink v3.2.

Here are the snapshots (self-extracting archive)...

Update: fixed - now includes all snapshots.
snapsVSky.exe

Note: only download below if you didn't get the #17 snapshots earlier.
snapsVSky17.exe

Thanks for reading!

p.s. - Oops, I excluded Snapshot 17 from the package above.  It's been fixed.  Also, a separate link for downloading just Snapshot 17 if you had already downloaded the first package with it missing.

If you are writing a transcoder, and you want to have good quality, wouldn't you try to preserve the i-frames, and especially the chapter stops as much as possible?Will d/l your comparisons, thanks for doing them

Originally posted by b0b0b0b
If you are writing a transcoder, and you want to have good quality, wouldn't you try to preserve the i-frames, and especially the chapter stops as much as possible?Why? I think the goal should be to have best possible quality throughout the complete movie. Having some frames identical means that you have to save some bits more on some other frames. Thus you get a bigger (probably visible) variation in quality.

JFerguson,
thanks...i can't find snapshot 17 in the file and i have some doubts,please answer me:
in the link of the Matrix Reloaded comparison, you are using acdsee to compare the pictures but you get the snapshots from powerdvd and this player can quot;dopequot; the pictures right? was used powerdvd to get the quot;snapsVSkyquot; snapshots too? ACDSee Classic is a good viewer but IMHO i don't trust in snapshots from powerdvd
and
take a look in this pictures from dvdshrink with 68,1% compression and the original from the source(snapshots from vdubmpeg2),see the differences(i can't see):
my.php?loc=ing42...=maxsmooth.jpg

my.php?loc=ing42amp;image=smooth.jpg

my.php?loc=ing42amp;image=sharp1.jpg

my.php?loc=ing45amp;image=maxsharp.jpg

my.php?loc=ing42...=dvdsource.jpg

i'm not trying to show quot;what is bestquot; but how was did and how was get the pictures.

i want to test but still can't use IC8, is crashing in my system.

Hi all.
Originally posted by JFerguson
Oh, and I had to take snapshots at 2 frames past chapter stops; when I did them at chapter stops, 1/2 of the Shrink snapshots were binary equivalents of the snapshots from the original DVD.  What is with that??!

JFerguson, DVD Shrink spreads compression across frames in a different manner than Instant Copy.
The frames that were identical to the original were probably P-frames.
It'd be interesting to see what the corresponding IC frames looked like.

However, would you mind uploading short clips (3~5 sec) containing the frames you captured (similar to these)?
I can provide webspace if required.
For us to really be able to judge the quality of a backup, we would have to see it quot;in actionquot;.
This is what ultimately matters after all.

P.S: I know this is all very subjective, but my preference usually goes to quot;betterquot; slow scenes, as opposed to fast ones.
When looking at a close-up, for example, I find it much more distracting to have it all smoothed-out/washed-out or with quot;crawling noisequot;.
Artefacts in fast scenes very often don't register when viewed at normal speed.

Personally I agree with ddlooping in that screenshots are not a good way to compare quality, not at all.  To a point they show the quality of a transcoder but not very accurately.  I mean how it looks in motion is what really matters i would say.  That is not to totally dismiss jferguson's findings, although at 78% compression there is almost no visible difference in quality that I can see in any of those snapshots.  I dont think that either instantcopy or shrink is better than the other, they both have strengths and weaknesses.  They both produce very good quality backups in my opinion.  IC seems to handle seamless branching titles much better whereas shrink has the still image option and reauthor mode(not to mention being a bit faster).  I personally use shrink on most dvd's that dont require much compression at all because of its quality relative to its speed.  For seamless branching i use IC and for stuff that requires heavy compression i use DVDRebuilder.  They all have their uses, and the difference in quality between ic8 and shrink is so little that you could starve on the difference.  So lets just agree to disagree and say that each tool has its purpose.

hmm well ddlooping himself showed with terminator that ic8 can give better results depending on the movie. also the fact there is 4 settings for AEC in dvdshrink means you may have to do a movie 4 tames to get what you consider the best out of it doesnt help.  

since there are thousands of movies and people will never agree 100% on which looks better to them on any given 1 why worry ?

the goal of beating ic for that reason is a bit silly. the goal should have been to increase quality so that it rivalled ic most of the time but being a lot faster and more accurate final size.  

To add to the confusion I just did a transcode of quot;The Butterfly Effectquot; full DVD with both IC8.3 and Shrink 3.2(MaxSharp).

My results of full movie preview plus extra features on a 55quot; HDTV with A/B comparison from 2 Progressive Scan DVD players. Made one preview, then swap discs between players for second preview to make sure that differences between players wouldn't affect the tests:

- IC8.3, oversized to 4570 with PowerToys hack, selected High Quality and 1 audio track(AC 5.1) for main movie, the rest was left untouched. Overall video compression 61.91%.
Impressions: The quot;Mosquito Effectquot; killed quot;The Butterfly Effectquot;, the colors were OK but than random pixel or shifting effect really made the picture looked like a capture from DSS rather than a DVD, macroblocks made the extra features un-watchable, overall not as good as I expected with this new version.

- DVDShrink 3.2, AEC MaxSharpness, main audio track only and 68% compression on main movie, all audio tracks and 51.9% on extras/menu.
Impressions: Looks as good as original, nice colors and even sharpness, extras had a few macroblocks but not as bad as IC8.3.

Final Conclusion: New version of Shrink surpasses most transcoders or encoders out there. This is not the only transcode in which Shrink is a winner, I also tried a few weeks ago when it first came out with John Wayne's quot;The Alamoquot;, this is a really sharp/noise and lengthly movie and once again Shrink was the winner, even against DVD-RB with CCE.

Originally posted by jorel
JFerguson,
thanks...i can't find snapshot 17 in the file and i have some doubts,please answer me:
in the link of the Matrix Reloaded comparison, you are using acdsee to compare the pictures but you get the snapshots from powerdvd and this player can quot;dopequot; the pictures right? was used powerdvd to get the quot;snapsVSkyquot; snapshots too? ACDSee Classic is a good viewer but IMHO i don't trust in snapshots from powerdvd
and
take a look in this pictures from dvdshrink with 68,1% compression and the original from the source(snapshots from vdubmpeg2),see the differences(i can't see):
my.php?loc=ing42...=maxsmooth.jpg

my.php?loc=ing42amp;image=smooth.jpg

my.php?loc=ing42amp;image=sharp1.jpg

my.php?loc=ing45amp;image=maxsharp.jpg

my.php?loc=ing42...=dvdsource.jpg

i'm not trying to show quot;what is bestquot; but how was did and how was get the pictures.

i want to test but still can't use IC8, is crashing in my system.

Ok, I fixed snapshot #17.  It's in there now.

Yes, using ACDSee to compare.  I don't know if PowerDVD can affect the quality of a frame when it gets captured to a bitmap, but if it does, I would assume it does so uniformly.  And yes, I did use PowerDVD v4 for capturing these Vanilla Sky snapshots.

I looked at your snapshots at 1x zoom in ACDSee.  They look pretty good (especially Maximum sharpness).  I would order them best to worst as follows:
1. Maximum sharpness (this looks near original!)
2. Sharp
3. Smooth
4. Maximum smoothness

Honestly, maximum smoothness looks awful (lots of pixel bouncing) compared to the Sharp ones, and even to some degree against the default (Smooth).  I don't know what's with that, but I don't use Maximum smoothness anymore.  I generally use Maximum sharpness.

Hope that helps.  Were these from quot;A Bug's Lifequot;?

Originally posted by dragongodz
the goal should have been to increase quality so that it rivalled ic most of the time but being a lot faster and more accurate final size.  

I believe we scored, lol.

Originally posted by ddlooping
Hi all.

JFerguson, DVD Shrink spreads compression across frames in a different manner than Instant Copy.
The frames that were identical to the original were probably P-frames.
It'd be interesting to see what the corresponding IC frames looked like.

I don't know much about the various types of frames.  Unless it's more conservative to leave one alone then change it, then like  
infoscapeone alluded to above, I would think that you would distribute the the transcode across all frames for better distribution.

InstantCopy touched all frames, so in this case the DVD Shrink ones looked better.

Actually, I didn't think about it at first, but when I started seeing all these frames that looked better than InstantCopy...it was like, quot;wait a minute...quot;, then I went back and did the binary compare.However, would you mind uploading short clips (3~5 sec) containing the frames you captured (similar to these)?
I can provide webspace if required.
For us to really be able to judge the quality of a backup, we would have to see it quot;in actionquot;.
This is what ultimately matters after all. Yeah, I can probably do that.  It might take me a few days though...P.S: I know this is all very subjective, but my preference usually goes to quot;betterquot; slow scenes, as opposed to fast ones.
When looking at a close-up, for example, I find it much more distracting to have it all smoothed-out/washed-out or with quot;crawling noisequot;.
Artefacts in fast scenes very often don't register when viewed at normal speed.

Yes, it can be difficult to see these deviations when things are played at normal speed.  I just don't understand why they get introduced to the degree that they do with Shrink.  I really do not see this with InstantCopy or CCE encodes.  I don't know if it's by design, or just an area for improvement.  But boy, if it could be improved...it wouldn't leave much room for arguing quality.

There's always room for improvement, JFerguson.

However, I don't know how much is left and long it'd take to quot;fillquot; it.
dvdshrink is a very busy guy.

I'm just grateful Shrink came out with an update.  I'm not an expert but why would anyone compaint about Shrink. It's loaded with features and produces excellant copies.  I haven't tried many of the other's but I can't believe I'd get a much better product by buying any of the other products.

I'd gladly donate to Shrink.

ps:  Thanks also to the producers of Decrypter.