Unbreakable (R1, DISC 1). Big undersize -> 3.99GB vs. 4.38GB. What happened?

[JFerguson]

Undersized by 400MB. Here are the various project files:

http://www.chicagophoto.net/JeffStuff/Unbreakable/UNBREAKABLE.RBD
http://www.chicagophoto.net/JeffStuff/Unbreakable/REBUILDER.INF
http://www.chicagophoto.net/JeffStuff/Unbreakable/REBUILDER.ECL

I loaded REBUILDER.INF into RB-Opt and it gave TargetSectors=2267954, very close to my project file's 2267400. DVD-RB put the bitrate at 3977, so this isn't like the only other BIG undersize I had (here (http://forum.doom9.org/showthread.php?t=106699)), which had high bitrate at 5200. I loaded the output into DVD-RB just to make sure I wasn't seeing things. Here's a sample snapshot of target sizing, followed by a sample snapshot of actual sizing:

http://www.chicagophoto.net/JeffStuff/Unbreakable/Unbreakable.Target.jpg

http://www.chicagophoto.net/JeffStuff/Unbreakable/Unbreakable.Actual.jpg

Anyone have any ideas on what may have happened with this one?

Thanks...

[wmansir]

Unbreakable is one of the most easily compressed movies I've ever encountered, probably due to the abundant low light, low motion scenes. The undersize could be due to the encoder just failing to reach the AVG bitrate assigned for some segments because it has reached the MAX bitrate it can use to encode video of that complexity.

It is also possible that by stripping the audio/extras the main movie fit without needing to be recompressed. Recent versions of DVD-RB will retain the original video if the reencoded segment would be larger than the original. Force Reencoding, under MODE menu will overide that, but is only needed if you are altering the video via filters or resize.

PS. Your links are broken, I tried entering the URL manually but got 404 on the last 2.

[JFerguson]

Ok, I just noticed that this title is multiangle, with some extraneous video for Spanish, French playback.

I'm going to preprocess this and delete the angles and see what happens. I don't think that multiangle should affect sizing, but it's a waste of bitrate, so I'll try it anyway...

[JFerguson]

@wmansir:

The movie definitely needs compressing. Shrink has me at 5365MB (5171MB main movie) with stripped audio and just 40MB of extras.

Fixed the links - thanks. I'm kicking off this retry in just a few minutes...

[jdobbs]

You may want to run Bitrate Viewer against it and see what the average Quant is... if it is very low it means the encoder was probably saturated and it couldn't throw more bitrate at it.

[JFerguson]

Bitrate Viewer, here?

http://www.tecoltd.com/bitratev.htm

Do I load an individual VOB file, or is there a better way to get a read?

[jdobbs]

Yeah, but I don't think I realized it was "shareware"... I thought it was freeware.

You can either load the VOB or the M2Vs in the D2VAVS directory.

[JFerguson]

Ok, here are the average Q values for the 7 VOBs in the main titleset:

5.77
5.80
5.82
5.75
5.47
5.85
7.63

Are those low?

Kind of moot at this point. After I ripped angles 2 and 3, and saved just the DD 5.1 audio, the movie fits at 100% with some minimal compression of the menus (DVD Shrink).

If there's anything yet you want me to try with this title, though, let me know...

[Boulder]

They are certainly not low..if you had something close to 2, it could be saturation. What encoder did you use? If you used CCE, did you by any chance enable the adaptive quant matrices?

[jdobbs]

They're good values in terms of quality -- but definitely not what I would expect as a result of saturation.

[Rippraff]

What encoder did you use? If you used CCE, did you by any chance enable the adaptive quant matrices?
It's all in the rbd file he posted above. ;)
CCE Trial is used and AQM is off.
But I'm wondering where [CCEOptions] are:confused:

@jdobbs
Or is there only an entry when you change the default values?

Cu Rippraff

[JFerguson]

Yes, using CCE Trial v2.70.02.01. I pretty much leave everything at default, so there shouldn't be anything in my configuration that's outside the norm.

Again, I can experiment with this a little bit if interested...

[Boulder]

Try encoding with a high-bitrate matrix and see if there's any change - alternatively you could try with CCEAQM=1 and 3-4 passes.

[JFerguson]

Just got another one, with Hoodwinked, R1. Do these Q-values look more in line with saturation?

2.70
2.92
2.44
1.83

[Boulder]

Yep, you should definitely use a matrix that will compress less, or enable adaptive quants and do 3 or 4 passes.

[JFerguson]

I'll probably just leave it. The movie was only 80 minutes long, so the average bitrate was pretty high to begin with.