Compressibility Check extremes

[Brian_92gsr]

I'm encoding two shows, season 2 of Deep Space Nine, and The Shield. The Compressibility test for The Shield resulted in 0.680. None of the shows I've done before have been anywhere near this high. With two 62MB AC3 tracks encoded to 700MB, the Bit/Pixel is only 35% of the compressibility test result. Is there anything that could have thrown this off, or what reason could there be for this to be so high? On the other end of the extreme, DS9 compressibility is around 0.15, which seems quite low.

Should I always shoot for a Bit/Pixel of 50-60% of the compressibility test even at these extremes, or should I go higher for DS9 and lower for The Shield?

Also, I know DS9 is hybrid film/video, and I'm using the following script:

Telecide(guide=1,gthresh=30)
Decimate(mode=3,threshold=1.0)

The film percent in dvd2avi was around 92-94% for The Shield. Should I use forcefilm or the same script above?

[hakko504]

Always aim for 50-60% of compressibility test, even if that means going to 2CD. BPP values up to 1 exists depending on your dsource. With 92-94% I'd use the regular IVTC mode of decimate instead of the hybrid.

[manono]

Hi hakko504-

BPP values up to 1 exists

I've seen a bunch over 1 (and at 512x384 resolution!). What does that mean anyway? That the first pass size will be larger than the DVD source vob files? If so, that's pretty strange.

[hakko504]

I just meant that I've seen values as high, or as you say, even larger than 1.

A DVD can have BPP values of up to 1.18 (FILM 9.8Mbit/s) but usually it is about 0.32(PAL/NTSC 3.5Mbit/s) ~ 0.60 (FILM 5Mbit/s)

[manono]

OK, it correlates with the DVD's bitrate. I never knew that. Thanks.

[hakko504]

I think you misunderstand me: bpp is defined as bitrate/(pixels*number of frames per second) Or, for a
NTSC DVD: 3.5e6/(720x480x29.97)=0.33
FILM DVD: 5e6/(720x480x23.976)=0.60
PAL DVD: 3.5e6/(720x576x25)=0.32

And for a 512x384 DivX of 1Mbit/s:
DivX: 1e6/(512x384x29.97)=0.17

[manono]

You're right. I didn't understand (big time). :) I still have other questions about it, but this isn't the place. I guess one thing we can conclude about it is that Brian_92gsr's figure for The Shield of 0.680 isn't out of the ordinary.