Doom9 has released a brand new codec comparison featuring: 3ivX, DivX3, DivX5, ffvfw, NeroDigital, RV9, VP6 and XviD. Check it out here (http://www.doom9.org/codecs-203-1.htm).
So where do I get the latest/stable release of XivD ?
Is it Beta 3 ?
Koepi's XviD binaries (http://koepi.roeder.goe.net/xvid.shtml)
Doom9 has released a brand new codec comparison featuring: 3ivX, DivX3, DivX5, ffvfw, NeroDigital, RV9, VP6 and XviD. Check it out here (http://www.doom9.org/codecs-203-1.htm).
Link is currently dead. Bet it's been slashdotted
doom9 wasn't fair towards the ffvfw.....check out my comments here
http://virtualdub.everwicked.com/index.php?act=ST&f=3&t=5515 (http://virtualdub.everwicked.com/index.php?act=ST&f=3&t=5515)
either he didn't know what the settings do,or he made it look bad intentionally..(to make xvid a winner).....[edit;or he used settings that work well on linux... ]
xvid is not a bad codec,but how many of you tried ffvfw?
was it slower than xvid...was it worse in image quality....?
use ffvfw on defaults(unlike doom9),change only bitrate/quant.matrix/quantizers....(in a same fashion u used it on xvid)
so...?
doom9 wasn't fair towards the ffvfw.....check out my comments here
http://virtualdub.everwicked.com/index.php?act=ST&f=3&t=5515 (http://virtualdub.everwicked.com/index.php?act=ST&f=3&t=5515)
either he didn't know what the settings do,or he made it look bad intentionally..(to make xvid a winner).....
xvid is not a bad codec,but how many of you tried ffvfw?
was it slower than xvid...was it worse in image quality....?
use ffvfw on defaults(unlike doom9),change only bitrate/quant.matrix/quantizers....(in a same fashion u used it on xvid)
so...?
doom9 used the settings recommended by the authors of the codec.
Menno
He provided very interesting link for ffvfw; there are some newer versions of ffdshow than at SourceForge.
Also, he says:
But be aware that you are not only entitled to see things differently, you might actually see things differently and can in all honesty disagree with my findings(...)
To me, ffvfw feels like an early XviD version. It certainly had potential but it's not quite there yet. As the underlying libavcodec is more than just an MPEG-4 encoder it remains to be seen which part will be optimized in the future.
I don't really know, but after your post ffvfw sounds interesting to try.
doom9 used the settings recommended by the authors of the codec.
yep, doom9 was in contact with all developers of the codecs he used in his comparison and he used the builds and settings he was told by the codec developers themselves...
yes,he was.......
my bad......
i have corrected the vdub forum posting and i offer my sincere apologies...
but even so,everyone should try ffvfw if xvid fails for them in some areas...(for example i dislike xvid's ME techniques,although i like MPEG matrix)
and enthusiasts should try many codecs regardless to what tests say....for me,ffvfw performance was stellar....
and i really dunno why so "big" difference between the milan's ffvfw defaults and ffmpeg dev. recommendations....i can only say that milan's defaults work fine....i have tried many different settings and didn't find any settings which offered significant improvement over deaults (i did that some time ago)
everyone who tried ffvfw raise your hands....what do YOU say?
did it worked for you?
i wish i saw 2 or 3 hands and not xvid fans only....
I tested a build from Nov 13 2003, I presume it's slower as well?
"The athos's ffvfw 09-27-2003 build is also 25-50% faster thanks to gcc/nasm assembly optimizations." ...my tests prove that this is true....09-27 is MUCH faster than new unoptimised builds.....
I only made a very short test and the quality seemed to be comparable to Xvid, encoding speed was slightly slower though. And to tell the truth, I'm a little lost in the motion estimation setup dialog. Is there a guide (with more extensive explanations on speed and quality of different settings) available somewhere?
Also there appears to be no average quantizer option and I keep forgetting the quantizer-quality setting relation.
(Besides, would it be too much to ask for proper punctuation? )
Wow thats interesting.
I can't wait to hear everyone who has their feelings hurt start complaining though. Looks like i4004 is off to a head start . . .
Well, I think it's fair that someone who has an opinion about the test can say it, so others can discuss it. I see no problem with that.
So, if he thinks that ffvfw is better than the test showed, he is welcome if he can prove it right.
yes,RIV@NVX that is my aim....i'll do my best....(i'm planning an QPEL test now..QPEL test means i must post short .avi clips...otherwise you won't see MotionEstimation performance...)
continuum;that's correct....Nov 13 is also slower,as it's not optimized likewise..(Nov13 outputs same quality,but ffvfw-20030927.exe is faster...MUCH faster...ie. the only difference is speed...)
Is there a guide (with more extensive explanations on speed and quality of different settings) available somewhere?
with some builds you'll find short help files by milan with explanations...(in the codec folder)...
you should look for ffmpeg's web to find out more on this...
quantizer?
2-2 looks excellent and eats bitrate...31-31 looks like crap.....
i usually go for 2-6 to 2-11 (obviously 2-11 for higher resolutions...i just did Thin Red Line with nandub;2-9 (162min on 2 cdr's) looks pretty fine...now i'll use some portions for QPEL test...i'll dload last xvid now....) i usually use "MPEG" quantizer matrix as it's sharper than h.263
I only made a very short test and the quality seemed to be comparable to Xvid,
offcourse it is!now get that speedy version and see how it flies....(and now i'll make QPEL test to show few points (i hope))
it seems like ffvfw is a tad slower than nandub,but faster than xvid...
i did the test,but i'm now looking for some web-space to host 7.55MB of clips and one html page....(i must find some folks that have accounts at croatian ISP as they have 5MB of web space each...)
if someone here is willing to host,let me know....
i did divx4,nandub,ffvfw and xvid...interesting results i got....
one forum member from here provided the space,i'm uploading the stuff now...
as it's not clear if this will stay online (that member said "The account may be closed in future because I registered it using fake info") i will try my best to make a mirror on a croatian web (as i hinted above..i need 2 people's free-web space..i believe i can do it..) if needed (ie. if this site fails..)
so you'll use this web i'm uploading to now...new post on this forum is due, to post that link....
thanks for listening...now you'll be able to do some viewing too...(take both mine and doom9's tests with grain of salt...
"thanks" flies to Latexxx.......
I am actually somewhat suprised at how ffvfw did in the doom9 comparison. To be honest I have used it quite a bit and it has performed rather well often at times outperforming XviD. It can be blazing fast too. If it were used under linux you would see what I mean, the vfw version does seem to be slower. Under linux if you used compariable options to ND, it would be able to easily beat it in speed. Of course as soon as you enable things like SATD it will be much slower of course. Anyhow, in the end you have to make up your mind what you like best.
ok,so 2 surprised ffvfw users is not a bad score! (hehe)
some topics from this thread are continued here:
http://www.hydrogenaudio.org/show.php/showtopic/17017 (http://www.hydrogenaudio.org/show.php/showtopic/17017)
i did the test,but i'm now looking for some web-space to host 7.55MB of clips and one html page....(i must find some folks that have accounts at croatian ISP as they have 5MB of web space each...)
if someone here is willing to host,let me know....
i did divx4,nandub,ffvfw and xvid...interesting results i got....
i think netfirms.com gives you like 20mb or something ....