DVDShrink quality settings review

[pieroxy]

I have put up a small review of DVDShrink with and without Deep Analysis and with the 'Quality Enhancement' set to 'Sharp' on the movie Snow White and the Seven Dwarfs.

http://nerds.palmdrive.net/video/07Feb2006/index.html

Enjoy

[jtoolman2000]

Very nice. But I am curious why you chose to back up a movie that would fit woithout anycompression. I can onlu assume you left all the extras as well as menus. Nevertheles, it is very telling. I normally always use Deep/Sharp combo on amything requiring compresssion and seem to get results that are equal or nearly as the original. I only do movie only single audio/language backups though. Have to preserve quality for viewing on superb 52 DLP.

[Methanoid]

Good review... worth also having a drag of the mouse around the "black" parts of the scenes as you see where the Deep/AEC really kicks in. Makes you wonder if there is any point to using anything but Shrink unless its a MAJOR % drop... then maybe DVD-Rb.....

[sacajaweeda]

Yeah I'd like to see some DVD-RB w/CCE reviews done up like that. That's what I use on really long movies. Other than that it's just Shrink w/DA & QE set to smooth for most everything else.

[fritzi93]

Yeah I'd like to see some DVD-RB w/CCE reviews done up like that. That's what I use on really long movies. Other than that it's just Shrink w/DA & QE set to smooth for most everything else.

Good stinkin', uh, I mean thinkin'. Sorry.

How come so many people insist on using "sharp" with DVDShrink? And what do "sharp", "smooth", etc, mean in the context of DVDShrink?

Transcoders requantize, that is, they cut down on discrete cosine transform data, there's no other way for them to achieve any "compression". Look at DCT as error correction for those frames ("P" and "B" pictures), which re-use data. From my tinkering with MPEG2Requant, in which you can vary the reduction by frame type, I wonder if the quality settings in DVDShrink refer to the same thing. What else could it be?

Normally, it's best to try to step somewhat harder on "B" frames, if there's enough DCT data to start with and there's NOT constant action/movement in the movie. More movement means more motion vector data and less DCT data available. At some point, you'll compress more than the movie will stand and get obvious artifacting. Here's a good FAQ:

http://forum.doom9.org/showthread.php?&threadid=63587

I conjecture that "smooth" refers to a somewhat less uneven reduction of DCT data by frame type. "Sharp" would necessarily mean a higher proportion of the data thrown out would be in "B" frames. If I'm correct, one should only use "sharp" at very modest percentages, and go to "smooth" when more "compression" is required.

Anyone care to comment on that?

[NiteLite]

Anyone care to comment on that?

Sure.
While your theory is sound... in real life it doesnt hold up for me.
When backing up a 60% compression of episodic series using "smooth" I notice a washed out faded look to the picture. Not a dramatic amount but it's there.
Using "sharp" appears to add back some detail separating images giving a more contrast appearance therefore a better picture closer to the original.

This is with excellent rated DVD player using calibrated analog 36" TV.
Things may be different using an HD setup.
The creator made "sharp" the default and so far I cant say he erred.
NL

[fritzi93]

I should have said max sharp, just went and had a look at settings I haven't fooled with in a while. Yes, default is sharp. If the default for Shrink is anything like the default for MPEG2Requant, it can't be anything too disproportionate in how DCT data is removed. Thanks for replying, BTW.

I recall a thread, over a year ago must have been. Yeah, I know, search for it, but how many thousands of DVDShrink threads are there? Anyway, it discussed a "stuttering" effect at very high compression. It's what I suspect you'd see when essentially all the DCT data is gone from "B" frames, yet a fair amount remains in "I" and "P" frames. I *think* the setting must have been at max sharp.

Just wondering if anyone can shed light on this, or is curious about it like me.

[fritzi93]

When backing up a 60% compression of episodic series using "smooth" I notice a washed out faded look to the picture. Not a dramatic amount but it's there.

Hmm, if I'm understanding this correctly, what you say makes sense. If, that is, more data is being removed from "I" and "P" frames than at default. Particularly from"I" pictures, as predictive and bidirectional frames depend on them. I wish now that I'd paid more attention in calculus class.

I've used "smooth" with good results, though not at so severe a compression level. But anyway, I prefer to go to DVDRB for everything that needs more compression than a few percent. JMO

[pieroxy]

But I am curious why you chose to back up a movie that would fit woithout anycompression
To test the compression settings

And what do "sharp", "smooth", etc, mean in the context of DVDShrink?
I am not really an expert when it comes to that, but my guess is that "sharp" "smooth" etc... work by adding some infos to the stream. Instead of removing 40% of data from the stream, you can remove 50% and then add 10% of recomputed data. The idea would be to analyse the 'noise' in the picture and add some data to recreate a noise that would 'look like' the noise in the original but could be compressed much further because you don't need the same noise, just a noise similarly shaped.

That said, I don't know anything about these things, so this is just a wild guess.