compressibility test: are these normal results?

[cereal killa]

I keep getting 0.1 - 0.15 on my comp-test (enemy at the gates, 2010 year we make contact), where i expect values around 0.3

This normal?

When u use a bitrate 100% of comp-test, this should mean u get the best possible result for that resolution?

By default it tests 5% of movie, which 5%? the first?

Which rule should you respect generally the most: the one about bitrate=60%<>80% of comp test or bitrate=0.17<>0.25?

I use divx5 pro with bi-directional encoding & gmc set on as default in G-knot.

[cult]

5% of the movie,it gets the 5% of the whole movie,check the avisynth script.Same happens with 1%.

[cult]

Also the rule is bitrate=60%<>80% of comp test
The other one was introduced at the first versions of gknot before comp test to be introduced,and is left now more as a memory to these days :)

[jggimi]

You may also be confused between bits/pixel*frame value and the load percentage produced by a compressibility test.

Use b/p*f as the resolution estimation tool prior to your first comp test.

Once you have a comp test load % value, ignore b/p*f.

It is recommended to re-test once you make additional changes in filters or resolution.

[cereal killa]

So I should first select a good resolution (based on the old 0.25 b/p*f rule, for 2 cds?), then do the test, see that i fall in a good percentage, if not change the resolution and test again. Correct?

I did not know it made a difference what resolution you had selected to do the test.

tnx

[cereal killa]

I did some more reading on this subject, and found out that whenever you go above 80% of comp test you have a chance of ending up with a too small file size. I don't understand, why not 100%?

So thinking a bit further on it, it's quite possible that a movie like 2010 The year we make contact (1h51) is not compressable on 2 full cds, because the codec is maxed out? I always took for granted that you could chooce whatever filesize you wanted, the more the better, no limit.

Plz correct me if I'm wrong somewhere

[cult]

yes cereal killa,you will end with a movi with smaller size.Thats true.Believe me.I am using gknot since it came out.Try resizing a movie to 3xx,xxx and choose 2-3 cds.Make a comp check and you ll get a result more than 100%,Then do a normal 2pass encode with these settings.Check the size of the movie you ll get.It will be alot smaller than you wanted to.
After 600 encoded movies (like me),you can predict the number of cds ,the filters that may be needed,etc.Of course you can always be wrong,2001 is an example.

[jggimi]

So I should first select a good resolution (based on the old 0.25 b/p*f rule, for 2 cds?), then do the test, see that i fall in a good percentage, if not change the resolution and test again. Correct? yesI did not know it made a difference what resolution you had selected to do the test. I didn't either, until recently, and having tested with it, it's true. The difference is slight, but it does appear.So thinking a bit further on it, it's quite possible that a movie...[won't fit on].. 2 full cds, because the codec is maxed out? True -- for a particular resolution. There's a sticky thread in the DivX 5 Encoding forum on this very subject. If the test shows 80% load or above at your chosen resolution, you may increase the resolution, or choose a smaller file size.

Note: It is recommended that you do not increase resolution larger than the values you get after cropping the original 720x480 or 720x576 video stream.

For short content where size is not an issue, like trailers or featurettes, I sometimes eliminate cropping and resizing in the .avs script and do a single pass with Quant 2 (100% quality). I end up with a 720x480 .avi, which I then play back in a player that can adjust aspect ratio to 16:9 or 4:3, such as BSPlayer.

[TheWEF]

Originally posted by cereal killa
I did some more reading on this subject, and found out that whenever you go above 80% of comp test you have a chance of ending up with a too small file size. I don't understand, why not 100%?

because divx5's filesize prediction is not very exact when encoding at the limit. 80% is on the save side. with divx3 you can go up to 100%.

but don't forget that 2-pass encoding is completely sensless at 100%. one pass at quality 100% will do the same thing in halve the time... you get the best possible quality and maximum filesize.

wef.

[cereal killa]

tnx alot these have been great answers

cereal killa

[aris]

quote:
"----------------------------------------------------------------------So I should first select a good resolution (based on the old 0.25 b/p*f rule, for 2 cds?), then do the ..."

Can someone please enlighten me about this "Old 0.25 bpf" rule? :-)
Is it some sort of "minimal-quality-required-for-a-decent-movie-is 0.25bpf?"
Thanks!

[manono]

Hi-

Look in the bottom left-center of GKnot. Do you see "Bits/(Pixel*Frame")? It's a guide line (and a pretty useless one if you ask me) for pointing you towards a proper resolution for your compression test, the results of which should give you a better idea of the correct resolution for the audio and file size.

[aris]

*That* I know.
In fact, it's rather useless for me, since I have an inherent distrust towards a TV-card's real-time resizing capabilities (I've simply noticed it hurts quality in both my TV-OUT cards) so I encode all my movies, even the 4:3, at 640xsomething (depending on AR) resolution. That way no resizing takes place during playback and the avi file is displayed as is, the only resizing having taken place is that done by GordianKnot.

We had a nice thread about it lately.

I still do the compression test, for the sole reason of checking if I didn't under- or over-invested a CD or two. Always 640 by something.

And I still want to know what the "25" in the "25bpf rule" refers to.

[manono]

Hi-

OK, so you wanted to know only about the number and not about bpf. Here's a quote from Doom9's DivX3-GKnot Guide (http://www.doom9.org/gknot-basics-divx3.htm).

It's a bit of guesswork as you cannot predict compressibility of your movie but here's some hints:
* It's not a good idea to go below 0.20 bits/pixel*frame
* If you go above ~0.35 you might not be able to reach your desired filesize
* If you go for 1 CD stay close to 0.20
* If you go for 2 CDs stay around 0.27
* If TV bits/pixel*frame is below ~0.13 you should raise your bitrate (and go for more CDs).
And some preferred 0.25 as a guideline for 2 CD rips. So yes, it's some sort of "minimal-quality-required-for-a-decent-movie-is 0.25bpf?"