Paradox? 76% Comp. Test vs. Ugly Result :)

[CdinT]

Movie = Panic Room Zone 2 PAL
The result for comp. test were simply amazing.
I know that this was a very dark movie with clean image but this is just unbeliebable :)

With Divx 5.02, bidirectional frames, 640x272, lanczosresize and 5.1 AC3 (344MB) I've got 74.9% compressibility test if the target size was 701 MB! This means that the actually movie file has 350MB and the bitrate is 456kbs. To be sure I repeated the comp. test using 15% and it gave 76.6%!
Finally I encoded the movie and as I expected it looked like crap, lots of moskito noise, problems on scene changes and uniform colored areas. I mean 456kbs...

So my question is how was this possible? Extremely low bitrate but hight comp. test.

PS: In the end because I didn't want to throw away the ac3 sound I did the movie at 720x304 with 1 Pass Quality Based quant. 2 and it came out at a whooping 813MB (for the video file) :)

[Evil Andy]

Sounds like a case of bad luck to me. With the standard 5% comp test gknot 1st pass encodes 5% of the movie to try and get a feel of how compressible it is. In this case it would appear that ALL of that 5% was taken from extremely compressible parts of the film AND so was your 15% test. Statistically improbable but not impossible.

[jggimi]

Well, consider this a teaching moment. You now know, without a doubt, that "Compressibility is not Equal to Quality."

If you think about it, you were using 456kbps for video, and 448kbps for audio.

[jonny]

My little experience with 2.35 aspect ratio movies...

I never like resizing those movies to 7xx x 304

Resizing the y to 304/288 is a lot of details gone.

My target with this movies is comp.test near 75%-80% and, for resolution something like 7xx x 432.

I correct the aspect ratio with the player and i play the movie with no post processing (only film effect), IMO, in this way, you get really good results with 2.35 movies.

(note also that if you play it with 1024x768 screen resolution, all the 432 lines are visible and there is... in approximation... only an x resizing)

[Ruppl]

So my question is how was this possible? Extremely low bitrate but hight comp. test.

@ CdinT
The Problem you have is your source! I´ve encoded "Panic Room" too, the movie is very dark most of the time. DivX5.02 is very bad for dark sources, cause the codec is very fast feeded...(feeded-> more quality isn´t possible). :scared: (<-hope I´m not wrong with this...)

I would solve this Problem over the Avisynth Plugins "Tweak" and "Levels", for a better gamma.

[CdinT]

jonny: I don't like to change the aspect ratio because I share this with my friends and not everybody knows how to modify AR in player.

Ruppl: the final encode 1 pass qmode looks very good so I guess there's no need for tweaking. I'm only sorry that I can't fill the cd's with directors commentary or something else because ussualy I like to use every available space to my encodes (701MB). The only extras is a 2 min. teaser again very compressable (I hardly managed to make a 9mb avi from it :) )

jggimi: "Well, consider this a teaching moment. You now know, without a doubt, that "Compressibility is not Equal to Quality."
You are so right. I did movies at 55% whith that looked amazingly sharp and with vivid colors and no artefacts. So this must be it.

Evil Andy: I don't think if I do a 100% comp. test I'll get different results. I think jggimi is correct.

Thanks.

[Ruppl]

the final encode 1 pass qmode looks very good so I guess there's no need for tweaking.

@ CdinT
But with little more gamma your Video will be more compressable. But do what you want! When i encoded "Panic Room" the final movie was about 583MB with two vbr-mp3´s....