Difference in Features "Xvid Vs DivX"

[saad]

well...i want to know the main feature differences in these codecs ...can someone help me out !

If someone knew a good link about "Difference in Features, Xvid Vs DivX" ..please put here

thnx in advance

[saad]

wht i have found here lately in the site search is

-http://www.doom9.org/index.html?/codecs-103-1.htm
-XviD is - in contrast to DivX - capable of using different quantizers for each macroblock
- B-frames cannot be played with the current build of either XviD or DivX

but the main features are not highlighted or i might not have thoroughly read it.....
:>

[sysKin]

The above points are ancient and both untrue.

[DeathTheSheep]

DivX
------
- No "Cartoon Mode" in DivX as far as I know.
- No Chroma/Edge Optimization there either (or very different than XviD's)...?
- No support for Zones
- No N-VOP tweaking
- Max B-frames: 3 consec.
- No Trellis quantization

XviD
------
- No awesome psychovisual enhancements in different levels/applications like DivX
- No colorful/super user-friendly GUI
- No "commercial" availability (uhhh...OS, remember? ;))
- Has VHQ (for B-frames)

OK, sorry. Quick list. G2G

[communist]

- No "commercial" availability (uhhh...OS, remember? ;))
If you only want to encode content to MPEG4 and sell / rent whatever you can use XviD just fine as long as you pay the required license fees.

DivX:
- no support for custom quant. matrices

Also the GUI is subject to personal taste, I for one cant stand the bloated DivX gui that still fails to offer the amount of configurability as seen in XviD.

[Sagittaire]

DivX
------
- No "Cartoon Mode" in DivX as far as I know.
- No Chroma/Edge Optimization there either (or very different than XviD's)...?
- No support for Zones
- No N-VOP tweaking
- Max B-frames: 3 consec.
- No Trellis quantization
G2G

You know the divx code ... ???

IMO H263 optimized is trellis but it's perhabs false
Psy for DivX and lumismask for XviD are HVS adaptative quantisation
I think that chroma optimisation and cartoon mode are only pre-process filter
More than 3 bframe is totally useless for MPEG4 ASP

In fact the principals differences between XviD and DivX are:
- Custom Matrix support for XviD
- More complex GMC for XviD (1WP for DivX and 3WP for XviD)

[DeathTheSheep]

Ah yes, of course!

[Mr_Schizo]

You know the divx code ... ???
Psy for DivX and lumismask for XviD are HVS adaptative quantisation
I think that chroma optimisation and cartoon mode are only pre-process filter

DXN's Psy mode doesn't use adaptive quantisation.AFAIK it uses some kind of DCT modifikations (lambda values) like it's planned for XviD in the future.
You are right about the chroma optimizer but cartoon mode isn't a pre-process filter. It just modifies the skipping values of the codec and allows to skip more blocks.

[Manao]

If by features, you're talking about mpeg4 features only, then :

XviD has GMC 3 points where DivX has only 1 point : making it almost useless, some would say, I'd rather say it *might* sometimes help, but in any case less than 3 points
XviD supports custom matrices while DivX doesn't : that's a shame, because it's easy to do, but I guess they don't want to trouble their users with it.
XviD supports any number of bframes while DivX supports only 3, which is enough, XviD will hardly use more than 3 in a raw anyway.
DivX supports all interlaced stuff, while XviD supports only interlaced DCT : that's a big difference if you're encoding interlaced stuff. Else, it doesn't matter.
XviD has adaptive quantization, DivX doesn't. But there, results achieved by adaptive quantization can be done otherwise ( adaptive RD lambdas for example, as it is planned to do in XviD ). Once again, DivX is closed source, so we can only presume what they are doing.


Now, talking of the implementation itself :

XviD and Divx both have RD decision : XviD calls that VHQ, DivX has quality levels that might ( assumption there, since closed source ) use RD too.
XviD has trellis while DivX has H263 optimized quantization : here again, it's the same thing branded differently
XviD has chroma motion. DivX doesn't say anything on that, but it can be assumed they have it too, enabled by higher quality modes ( once again, assumption only, but should be a correct one too )
XviD has cartoon mode. DivX can't know if its input is a cartoon. But there, does cartoon mode helps that much ?
XviD has chroma optimization, which is only a preprocessing filter. DivX has denoising, which is too.
DivX has a true CBR mode ( with VBV constraint ). XviD doesn't export the VBV settings but allows the use of DivX's profiles. I never checked if the profiles were completely respected though. So in the overall, I trust DivX STB compatibility more than XviD's ( but XviD in the hand of the knwoledgeable guy can be configured to be compatible enough )
DivX handles several processor while XviD doesn't.
Their GUIs both please and displease people. Personnally, I don't like GUIs :D

In the end, it can be summed up as : XviD has custom matrices and is open source, DivX handles interlaced content and several CPUs. All the rest is imho not important.

[Wilbert]

It would be nice if the mods can add this info to the faq :)

[DeathTheSheep]

Yeah....this is good stuff. Very interesting and thorough. Perhaps these should be verified first, though...? But otherwise, this is definately a frequently asked question.

[Kopernikus]

...
XviD has trellis while DivX has H264 optimized quantization : here again, it's the same thing branded differently
...


Souldn't it be H.263?

[Manao]

Souldn't it be H.263?Yep, it's corrected.Perhaps these should be verified first, though...?That would be great if DigitAl56k would comment our 'assumptions'.

[stephanV]

I think there's also a difference in multipassing implementation wise.

- XviD uses a constant quant for the first pass while DivX uses an ABR method.
- XviD by default uses a fast first pass, while with DivX the encoding speed is determined by the selected mode. (of course you could use 'balanced' or 'better' for a "fast" first pass.)
- XviD supports only 2-pass and DivX supports multiple passes.


/me waits for the spray of comments that XviD only needs 2 passes to get it right ;)

[bond]

i didnt follow it exactly, but it seems that with the xvid cli you can do multipasses too, its just not there in the vfw

and yes, why use more than 2 passes, if you only need 2 :p

[Sagittaire]

i didnt follow it exactly, but it seems that with the xvid cli you can do multipasses too, its just not there in the vfw

and yes, why use more than 2 passes, if you only need 2 :p

If quant difference between first and second pass is very high then 2 pass mode for XviD isn't very good (because quant predictibility is more difficult).

but it's possible with XviD vfw to make first pass with higher quant. In fact if you choose for first pass quant the same average quant than last pass then Multipass will be useless.

For example:
encoding between 75% and 100% for compressibility (q2 encoding for reference) it's better to choose q2 for first pass
encoding between 50% and 75% for compressibility (q2 encoding for reference) it's better to choose q3 for first pass
encoding between 25% and 50% for compressibility (q2 encoding for reference) it's better to choose q4 for first pass

[V-tec]

encoding between 75% and 100% for compressibility (q2 encoding for reference) it's better to choose q2 for first pass
encoding between 50% and 75% for compressibility (q2 encoding for reference) it's better to choose q3 for first pass
encoding between 25% and 50% for compressibility (q2 encoding for reference) it's better to choose q4 for first pass

I' dont have understood in what manner yo can decide quantizer in the 1st pass, perhaps with "zone" options?

regarding XviD *VS* DivX features, for bframes I have noticed that DivX seems to be smarter on B or P decision.
If in XviD I set 1 max consecutive B-frame, generally the output video is composed by some I frames and the rest are 50% P and 50% B.
This is not dependent to bitrate, i other word generally XviD create IPBPBPBP...PBPB streams

In DivX the number of B-Frames decrease if the bitrate increase, in extreme mode and insane mode over a certain bitrate de codec stop using B-Frames.

[Teegedeck]

You can set a lower b-frame sensitivity to reach a similar behaviour to that kind of 'extreme' mode in XviD.

Also, please search for information on XviD's mode-decision and VHQ and you will learn that I/B/P decision in XviD is anything but dumb. Mode decision should in no way correlate to bitrate; a voluntary decision to use fewer b-frames in order to sacrifice efficiency for quality (as in DivX's 'extreme' mode or through XviD's b-frame sensitivity) is a different thing.

[Caroliano]

Multithreaded XviD! (http://forum.doom9.org/showthread.php?t=104257)
It is already there (in 1.2).

And about B-frames: you have to do it by hand in Xvid, while Divx do it automaticaly. And Xvid always calculate where an B-frame is more efficient than an p-frame or the mode-decision thing is something diferent?

And also remember the problem of Xvid with N-vops (was it already solved for 1.1 or 1.2?).

[communist]

And also remember the problem of Xvid with N-vops (was it already solved for 1.1 or 1.2?).
What problem?

[Caroliano]

What problem?this (http://forum.doom9.org/showthread.php?p=745570#post745570) and this (http://forum.doom9.org/showthread.php?t=100346&highlight=xvid+n-vop).

I would also like to have an clip with B-frames and real N-vops. In the 2nd link also has the B-frame placement problem.

[HarryM]

I want a simple resizer/denoiser in XviD, like DivX - very usefull for direct capturing from TV card.

I use ffdshow-VFW with "XviD" profile, but this isn't the ideal solution for me.