DVDshrink quality compared to CCE?

[MackemX]

these comparisions are better than stills but are these still fair comparisions?

some have different sizes and different frames and who's to say that if you sampled the next 3 seconds that the results aren't going to be the same?

I've done a few tests and I know I could take selective parts of a DVD and show mixed results. If you could slow time down and still watch the film but have time to analyse each frame (basically process information much faster than humans can now). You could then do a frame by frame comparision, marking which is the best in your eyes and going through the whole DVD. Basically you are just comparing each and every still but still viewing it in 'Bullet Time'

then you could say something like DVDShrink is closer to the original for 30% of the movie (or more pleasing on the eye), CCE 45%, and 25% looked the same within reasonable viewing distance

I did notice in one of the DVDShrink files, the nominal bitrate was something like 78000000(how many 0's is it again? ), whereas CCE is 98000000. would this not have an overall effect on quality?

I know you also get these noise to ratio (or whatever) values that people use to do tests but you don't watch those as you watch the images. It's whatever is pleasing on the viewer's eye and some like it sharp, some like it smooth blah blah. That's why TV's etc have settings such as contrast, colour, sharpness, brightness as most likes to change them. My mother for example has probably left her's the same as the day she got the TV, whereas I change mine quite often depending on what I'm viewing

if anyone is serious enough about getting the best quality then they would just do their own tests and not rely on others or others opinions . If people think something is better quality than what I think then I accept and respect that even if I don't agree as

I agree DVDShrink has improved (big pat on everyone's back involved ) but that's about as far as I would go now. I wouldn't say any particular program is better if comparing the quality of the output as long as I'm happy with the output as my perception of quality may be completely different to the next person

'To Each His Own'

[ddlooping]

MackaDude, in a nutshell:

- These comparisons are as fair as I could make them to be.
My choice of clips was based on video content that, in my opinion, caused problems to previous version of DVD Shrink.

- The DVD2DVD-R/CCE clips have different cut points, due to the fact that DVD Shrink "Start/End" works at the GOP level, and that CCE re-encoded the original stream (different GOP structure?).

- The files have different sizes due to the way each transcoder "spreads" compression over the whole title, and in the case of CCE, the way the clips were cut (see above).

- In theory, the higher the nominal bitrate, the better.
But as you said, what matters is how the backup compares to the original, and fullfil the viewer's need for quality.

These tests were mainly done to show v3.2 improvememt over previous versions, and hopefully, so others would put their prejudices aside and decide to make their own tests.

[MackemX]

Originally posted by ddlooping These tests were mainly done to show v3.2 improvememt over previous versions, and hopefully, so others would put their prejudices aside and decide to make their own tests.

I think the quality from DVDShrink is tremendous if you compare it to the rest. I also like the way the user has control over the end quality with 8 permutations

job well done mate

- The DVD2DVD-R/CCE clips have different cut points, due to the fact that DVD Shrink "Start/End" works at the GOP level, and that CCE re-encoded the original stream (different GOP structure?).

- The files have different sizes due to the way each transcoder "spreads" compression over the whole title, and in the case of CCE, the way the clips were cut (see above).

I did some quick tests with segments I cut using DVDShrink's start/end feature so I could ensure that the programs got the same video to work with. I used 5 of the settings with DVDShrink and all 4 with Recode. The original bitrate was around 4000 (trying to make them work . The results were around 2200-2400 so I set this into CCE also. CCE was better in my eyes but this is under close examination and under normal playback I doubt you would really you notice it. I was going to upload the video samples but they are too big but I had already done a bitrate graph HERE

the average bitrate is different but look at the Peak, especially Recode Basic & both DVDShrink's Smooth's. Recode is more constant whereas the smooth keeps more to the original bitrate distribution. Most of Recode's settings seem to flatten the bitrate for some reason. These are the same 3 results but shown at a different scale HERE which shows how Max Smooth keeps the original bitrate distribution but on a smaller scale

as you say, over the whole movie then the distribution may be completely different but overall DVDShrink is closer to CCE than it ever was

p.s. notice the times are 10 or 11 seconds, this is just the location of the cursor in Bitrate Viewer

[ddlooping]

Interesting test, MackemX, thanks for sharing your results.

@ ddlooping and all

comparing quality results using same sources for CCE(dvd2dvd-Roba) and DVDShrink
sources(toy story 1&2) was full ripped with dvddecrypter and was selected that parts to get a big file with dvdshrink without compression:
1- toy story1-main movie 01:20.32 3,414Mb (audio 5.1 portuguese 224Mb)
2- toy story2-main movie 01:32.10 4,887Mb (audio 5.1 portuguese 257Mb)
3- tin toy (from ts1)- 00:05.10 218Mb (audio 2 ch english 7mb)
4- outtakes(from ts2)- 00:05.29 296Mb (audio 5.1 portuguese 15Mb)
4- luxo jr.(from ts2)- 00:02.22 88MB (audio 5.1 english 6Mb)

full size ~ 8,906Mb

* the audio sizes are included in the video sizes,
they wasn't encoded with d2d/Roba or Shrink(of course)..same as source!

DVDShrink percentage using automatic compression settings:
dvd structure (all files together) 46,8%
1- 1,788Mb 48,7%
2- 2,397Mb 46%
3- 98Mb 42,9%
4- 135Mb 42,3%
5- 43Mb 45%
target: DVD-5(4.7Gb) (default)
performing deep analysis and choosing "smooth"(ddlopping recomendation)in adaptive quality error compensation

DVD2DVD/Roba target using the same source using undot and deen in the script: 4,480 cd size!
ps: each part was encoded separate... i don't remember the exact values and time needed to encode cos i did this compilation last month and author in tmpgencdvdauthor to build menus and chapters.

result of quality:
need to say?.....CCE is better for that big source (~ 8,906Mb)
as i wrote before, DVDShrink3.2 is really better but can't do a good job for that big source(more than 6Gb for my taste).

you (all) can choose what scene(frame from source)you want to compare....
---> just choose and i post (same scene from source,from DVDShrink and from CCE(D2D?Roba))!

[ddlooping]

Thanks for doing this test, jorel.

At a compression > 50%, the outcome was never in doubt though.
DVD Shrink did get better, but not that better.

[ddlooping]

MackemX, I just had a thought (yes, it happens sometimes, lol).
Would you mind including the CCE bitrate graphs in your comparisons?

[geffroman]

Viewing with a Sony XBR 40 inch in progressive 480P mode I have now compared countless movies with CCE in 2 pass mode and Shrink 3.2 with all new features on.

It is pretty consistent... I don't like ANYTHING compressed more than 40% (That would be 60% compression in the program) But if I have to go below 60%, it is a toss up until about 55% where CCE starts to be a little better (by my eye), but it has to be some really damn good original data to suffer this kind of compression and still be clean enough (by my eye).

It is to the point that it almost doesn't matter which program (using my basic settings) so the speed and ease of use and completeness of Shrink as a program means I use it most often.

I'll still split a movie to achieve best quality if I must... and sometimes CCE and DVD-RB are the perfect tool. I am happy to have all of them in my tool box.

Just one man's experience and opinion.

I mispoke a little... My personal cut off in most cases using Shrink is about 60%... If the original is really good I might go to 57% or 58%...

[ddlooping]

Very nice TV, geffroman. It must be quite revealing.

[luphy]

@geffroman,

When you say those movies that Shrink did at around 55% that still looked good....what were the average bitrates for those movies AFTER transcoding?

What have you found to be a good bitrate cutoff for Shrink before CCE starts to look better personally on your equipment?

And have you ever looked to see where along the Custom Ratio compression bar these movies fall (the ones that still look good at 55%) when using Shrink?

Thanks.

[ddlooping]

Originally posted by luphy
And have you ever looked to see where along the Custom Ratio compression bar these movies fall (the ones that still look good at 55%) when using Shrink?

Good call, luphy.

The degree to which the ratio can be lowered manually (using "Custom Ratio"), compared to the "Automatic" ratio, is usually a pretty good indicator of what the DVD Shrink output quality will be.

[ddlooping]

- "LOTR-ROTK" test: Automatic >> 60.9% - Minimum Custom Ratio >> 48.8%.
A margin of 12.1% before maximum compressability (as found by DVD Shrink "Deep Analysis") is reached.

In my opinion, the DVD Shrink backup was of very good quality.
I personally prefered it to the DVD-RB/CCE ones (quality_prec=16; quality_prec=20).

--------------------------------------------

- "Terminator 2" test: Automatic >> 59.6% - Minimum Custom Ratio >> 56.4%.
The automatic setting is only 3.2% away from maximum compressability (as found by DVD Shrink "Deep Analysis").

Even though watchable, the DVD Shrink backup exhibits a fair amount of "macroblocking".
In some scenes, even the least revealing viewing equipment would show it.
The DVD-RB/CCE backup might be slightly grainy, but it also shows less "macroblocking" in some key scenes.
In my opinion, an overal better backup.
However, both gave results that would warrant splitting the backup to two discs, or using dual-layer media, IMHO.

--------------------------------------------
This is obviously not a very extensive test, but the gap between the automatic ratio and the minimum custom ratio (maximum compressability) seems to be a good indication of what output quality to expect from DVD Shrink.

[DDogg]

ddlooping, I think most would agree that the maximum reduction possible is dependent on the source complexity. In one of my more manic phases I once calculated the Average BR of the original source and then did a 1% sample via CCE OPV at Q28. Depending on the source complexity, the derived ABR of the sample might show a reduction from 15% (think SPR) to 60% (think Matrix 1). When these reduction rates were then used in DVDShrink as a maximum they seemed to provide somewhat of a parallel result. That would seem to correlate with your findings, or at least I think it does. Also, just wanted to mention not everybody agrees with my thinking and methods, but there may be some merit in it.

I did not spend much more time on it as the process was way too cumbersome and required advanced use of CCE which seemed to be contrary to what the average DVDShrink user would want, but I did wonder at the time if the author might be able to do something similar to predict the maximum level of compression a source might allow as part of the pre analysis step. I'm curious how the program arrives at the maximum compression value that you mentioned?

[ddlooping]

Thanks for sharing your tests results, DDogg.

I am affraid I have absolutely no idea how DVD Shrink "decides" what the maximum compressability of a title is.
Here is a post by dvdshrink, related to v3.0, that might help:
http://forum.digital-digest.com/show...540#post118540

[geffroman]

Honestly I rarely look at bitrates... Long before I understood bitrates I struggled to discover how I could tell in advance what kind of compression I could get away with... For me I would view the original DVD and look specifically for dark scenes with shadowed facial tones and anywhere there was a dark RED... Then I would compress several times to see the result... I found original discs were crap in many cases and any compression just made them worse... A very clean original compresses to 60% pretty easily with Shrink or CCE in 2 pass RB mode and looks great... A crappy disc looks like hell at 70%... I could make a chart and review bitrate numbers to memorize a numeric value with each viewing so that I had a way to gauge these things without the endless reviewing of finished data... but at this point I just know what to expect by skipping through the original disc and give it a once over...

[colemar]

An objective benchmark:

http://forum.doom9.org/showthread.ph...352#post623352

[geffroman]

After reading that thread I feel even better about my post here!