comp check

[hubereevez]

"divx4 compressibility check added.
values seem to be generally 10% higher than with divx3.
so i would say lower bits/(pixel*frame) percentage values have to be choosen for divx4 also.
50-60% should be fine, of course this is not tested yet."

What are the test saying? I Rtfm and posts.....
good quality for the desire size (10Mo marge), I have with 70%....
50%, i should try?
thanks
PS : vstrip does a very good job...in GKnot....

[hubereevez]

My english is so poor that I speak like a jedi, don't I?
So I tested different compressibility %; and I find that with 60-70% (the most near 70%) with divx4, I have a good average when I encode. Can I make a "generalisation" ans say loud , ok 60-70% is good?
thanks
PS : my english is really really poor but it's too late for....arghhhhh

[manono]

Hi-

As near as I can tell, DivX4,5 and XviD come about 10% less than DivX 3.11 in the compressibility test. So I think you can actually go down to 50-55% with those codecs and still get a good movie.

So-yes, 60-70% with DivX4 will make a very good movie

[hubereevez]

Thanks manono,
Ok now I would like to know the consequences of this difference (for exemple 55% with 70%). I know that if this first pass stats (comp check) is too low, the result will not be good. And if it's above 75% i could choose an higher resolution (GKnot short guide).
But I'm not so good with the understanding of the program itself, can you explain to me (if you have the time), the difference between a movie encoded with 55% and another with 70%. The one with 55% has an higer resolution than the 70%, no? Can both of them reach the desire filesize? And which quality difference will be then?
thanks

[rui]

In fact, when using divx5 with b-frames and maybe GMC, if you try to make a manual comp check ( http://forum.doom9.org/showthread.php?s=&threadid=19927 ), your values should be much higher that the ones your are used to see with divx4, because B-frames and GMC improve the movie compressibility a lot.

[manono]

Hi rui-you found me out. I shouldn't have included DivX 5 with the others. I've had lots of experience with 3.11, enough experience with XviD, and all I know about DivX 4 is what I've read.

hubereevez-If you're getting those figures with the same movie then, yes, the 55% will be at a higher resolution. They will both reach the desired file size-resolution has nothing to do with that (unless you are above 100%). I understand that with DivX 4, you may be off by 20 MB or so either way, while using GKnot with DivX 3.11 or XviD, or XviD with it's own internal engine produce exact file sizes.

As for quality differences, that's hard to say-that's why I rely on the compressibility tests to tell me what resolution and file size to use. I prefer the objective measurements. But with the lower percentage, you might have more noise and other kinds of artifacts, or you might notice more blurring in action scenes, and it might not be quite as sharp. But sometimes it's hard for the eye to tell the differences.

[hubereevez]

Hi, manono
I made an encoding with 55% and another with 70%. It's hard to see any difference , wich objectif measuremnt can I use, the GKnot.log files, but what should I compare....
thanks to you
hubereevez

[manono]

Hi-

The 55% and 70$ are the only objective measurements I know. If it were DivX 3.11, then you could use File-File Information-DivX Quality in Nandub, but that only works with DivX 3.11