Divx 6 bits/pixel quidance

[rcrh]

Has anyone come up with some guides for the new codec? The old accepted practice was to try to keep bits/pixel at about .22. Is this still needed with the new coded?

I've been using The Converter to encode some TV episodes (NTSC 4:3) using the HD profile. I'm ending up with bpp of around .17.

Is this reduced number because of improvements in the codec, or yet another limitation of The Converter?

Is .17 the .22?

Thanks in advance

[Sergei_Esenin]

Has anyone come up with some guides for the new codec? The old accepted practice was to try to keep bits/pixel at about .22. Is this still needed with the new coded?

Well, while bits/pixel can be a vaguely decent indicator of quality once you're familiar with a codec's properties on various material, it's never been the best way of determining what bitrate to use. You should really run a compressibility test as a better indicator.

[rcrh]

Sergei_Esenin,

Could you recommend a compressibility test?

***
BTW, I was looking for Guidance, not Quidance! DOH!

[Sagittaire]

http://forum.doom9.org/showthread.php?t=95122

[rcrh]

OUCH!

Far too much math here.

I'm going to admit that I don't understand any of this.

Thanks though.

[jggimi]

B/P*F guidance is an estimate for use prior to your first compressibility test. Only.

From that point forward, ignore B/P*F values, and instead, look at the percentage calculation returned by GKnot -- I assume you're using GK, but don't really know if that's the case.

A compressibility test produces an estimate of the maximum bitrate possible at the resolution of the test. GK will produce a B/P*F value from that, using your current resolution and frame count settings.

Obviously, if you change filters (noise reduction, etc), you should re-run the test.

Doom9 discusses compressibility tests and their results in this page of his GK/DivX 5 guide: http://www.doom9.org/gknot-divx5.htm Obviously, DivX 6 is new, and it will take time to develop a consensus among members regarding percentage guidelines for this release.

There are also two stickies about compressibility testing in this forum that I will point you to, one is active, the other is from the sticky archive:

Undersized Files. Why do I get them, and what do I do about it? (http://forum.doom9.org/showthread.php?t=24584)
Manual way to make a comp. test with DivX (can be used with other codecs too) (http://forum.doom9.org/showthread.php?s=&threadid=44414)

[rcrh]

jggimi, Thanks.

That helps.

**
At this point I'm not using GK. I have a stand alone DivX player and am therefore very concerned about ending up with an encoded file that will work. GK is a lot of work to go through when I seem to consistently end of screwing something up.

***
As you point out, it will take some time before their is consensus about the new codec.

Once again, thanks!

[Cyberia]

Why is it called B/P*F? I can't figure out where the *F comes from because it should cancel out.

(F/P)*(B/F)=(B/P)

[stephanV]

Not sure what you mean... bits per pixel = bit rate (B) / (amount of pixels (P) * frame rate (F) )

so b/(p*f) refers to the formula... bpp is usually used as unit.

[jggimi]

value = total-bits-in-video-stream / pixels-in-a-frame * frames-in-stream

[darkavatar1470]

i think pixels per frame * frames would result in pixel, so I'm going with bits per pixel :D

[Wilbert]

@all (jggimi, stephanV, etc...),

Could you please correct your formulas? It's b/(p*f) and NOT b/p*f.

b/p*f is per definition equal to (b/p)*f.

[jggimi]

I stand corrected. :stupid:

Actually, I sit corrected. I'm in a chair.