Question on compressibility test

[augi]

I have read numerous threads in many forums and all say what % you should use after a compress. test and what quality you'll get and mpla mpla mpla...I can't understand some things:
1)Jonny's compressibility tool, doesn't use a bitrate as input.So the compress. test doesn't depend on the bitrate you use?
2)It uses of course the resolution (taken from the .avs file) so it depends on the resolution...?
3)When I set the compressibility % (in avisynth) to be 5% , it takes randomly frames from the movie or it analyzes the movie from it's start to 5% of it?
4)Afer the test is done and I load the log file to Gknot, it gives me a bits/pix*frame value taken from the log file.What does that mean?Is it the max value that the codec used during the compressibility test's pass?And why MY b/p*f value ,in the given resolution,should be between 60-80% of this calculated value?Is this accurate?
If anyone knows what is actually happening, plz answer me...

[jonny]

1) - 2)

First note that CompTest and DivX5Enc are totally different programs:

I've written CompTest because GKnot refuse reading the log files generated with DivX 5. What CompTest do is a 1° pass with mix-max quant=2 and some adjustments to the final log file. Next, all the calculation are done by GKnot and GKnot take care of the bitrate/resolution used.

I've integrated a new compressibility test in DivX5Enc because the results are not correct if you enable bframes. This compressibility test works in the way i describe in my site.



3)

Both methods use x frames every y frames (example 13 frames every 260 frames if you use CompTest and .avs saved at 5%)



4)

Don't get stressed for all the parameters you see. Basically every compressibility test compare the target size with the size you'll obtain if you encode at 100% quality.
Good comptest values are 60%-70% for me... but this is not objective, you must do some experiments and use what is the best for you.


I don't respond to your e-mail because the questions are basically the same :)

Cheers
jonny