Compressability check with xvid

[sherpya]

I've tried gknot compressabilty check with xivd stats, but It gave me
strane values (not at all but it tells me that I should encode an anime movie at too low resolutions).
Maybe gknot doesn't support correctly xivd? (Help says it supports divx5, anyway I've tried with xvid stat to see what happens)

[glenn]

No, GKnot doesn't support xvid. GKnot only supports DivX 3.11 (sbc) and DivX 5. Read the manual before posting questions.

[manono]

Hi-

When you load the XviD Compress Test .stats file into GKnot, it gives you lower values than those done with DivX 3.11-Nandub. You might try running the Compress Test with Nandub if you understand those values better. It could be that your anime movie is hard to compress and you'll have to use some light filtering or go to 2 CDs to make it look good at the resolution you want. As near as I can tell, 50% in XviD=60-65% in DivX 3.11

[sherpya]

I was able to done a great vhs rip, I've used two filters for quality... then I guessed width/height (512 width and corresponding height - was full PAL interlaced 720x576). Xvid and DivX comes from the same source I've supposed a similar stat file... In effect opening xvid stats as nandub stat give me bad values (illegal) but opening as divx5 stat gives me only non acceptable values (not bad at all).
So I suppose support for xvid will not be a big deal for the author.

@glenn
As I've said in the previous post, gknot's help doesn't says
to support xvid, so your reply is not really so usefull.
I've only tried xivd stats on gknot to see what happens.

[Teegedeck]

You can do compressibility-tests for XviD in GKnot AFAIK. Only once, when using an experimental DLL with B-frame-support, I got some strange values.

[TelemachusMH]

How do you find out the compressability with xvid encodes then? What program does everyone use?

TelemachusMH

[stax76]

How do you find out the compressability with xvid encodes then?

you can use DVX

Regards,
Dolemite

screenshots
http://www.planetdvb.net/dvx/screenshots.htm

homepage
http://www.planetdvb.net/dvx/

discussion of the current release candidate version
http://www.planetdvb.net/board/index.php?act=ST&f=22&t=31

discussion in this forum
http://forum.doom9.org/showthread.php?s=&threadid=31368&highlight=dvx

[manono]

Hi-

Since I wrote that a long time ago, I started getting XviD compress test results using GKnot. It works fine. I have found that after the second pass, when using DeBugView and Moonwalker's XviD Analyzer utility, that the final results have actually been a little bit better than the results of the compress test. I don't know why, but I'm not complaining. But I'm not using B-Frames, and don't know how that effects things. Or you may be able to adapt jonny's method (http://forum.doom9.org/showthread.php?s=&threadid=44414) of getting compress test results with DivX 5.03 to XviD.

[jonny]

@manono:

The problem with b-frames is that, sometimes, there is an oversized b-frame to discard (to have good results).
So with b-frames you must actually discard 2 frames (artificial keyframe+oversized b-frame) in each snip.
This happens when the codec doesn't detect a scene change between 2 snips, the result will be that the last frame of the first snip will by oversized.

The manual method i've proposed is really accurate without b-frames (discarding all the keyframes or discarding only 1 artificial keyframe every snip doesn't make a big difference)

Problems came with b-frames enabled because there is no way (manually) to discard bad b-frames... so the predicted size will be oversized and the resulting comp.test value will be lower than the real value (i think something like 10%-15%).

Edit: i've not made experiments with xvid in this field (but in the future i'm planning to add xvid support in a new project)