View Full Version : Xvid or Divx5
Subwars
1st November 2002, 07:03
Hi i have just been browsing the forums and been looking up about this xvid codec that people seem to be using now just wanting to know cause from what i can tell it seems to be back with divx 4 tech using the old .23 gknot package i currently encode with divx5.02 pro but is the video quality of this xvid codec better then the divx5 quality just looking for a straight forward answer is it better or not i went to look at a link that i found to goto to see comparisions at doom9.org/codecs but the page seems to be down and i couldn't find much else to really says pro's and con's of the two can i get some feedback on what really is the best codec available for quality
Thanks Subwars
:confused:
poopity poop
1st November 2002, 07:08
I would rate Xvid slightly above DivX5 because it retains more detial than DivX5, but only slightly. SBC encodings with nandub absolutly blow when compared to XVid or DivX5...too much noise.
I've done my own comparisons of DivX5 and XVid. I switched over to Xvid because it retained more of the original image. Now they are going to add b-frames which will make it absolutly an outstanding codec
Subwars
1st November 2002, 07:15
but doesn't bframes actually reduces the overall quality by taking out certain frames from the movie to make the compression better getting rid of frames to me would realy be lowering the quality???
Koepi
1st November 2002, 09:14
ifyouleaveeventhespacesawayitsmuchmorefuntoreadyoursentenceswithoutanycommaorfullstoporjustauselesssignattheendofthesentence!
Soyesifyouthinkthatbframesdegradequalityyoucouldsaysobutit'snotjustsoeasy. Butsinceyouarenewhereyoumaybedon'tknowaboutthesearchfunctionalityofthisboardwelldoesn'tmatterasIwilltellyouhowitworks:
pframesbehavedifferentthaniframesdifferentquantizerlevelshavedifferentimpactonthoseframetypesthesameisvalidforbframes.
Soifyouhavelet'ssayanIPBBPBBPBBPIsequenceandthebframequantizerratiois200%youmayalreadyseesomeveryminordegradationwhenbframequantratioissetto150itlooksalreadynottoodifferentthananIPPPPPIsequence
Butthenagainusethesearchfunctionandreadupaboutbframesinsteadblindlyasking.
Koepi
Teegedeck
1st November 2002, 11:47
Hahahahahhaha...! ROTFL
('scuse me):devil:
bb
1st November 2002, 12:31
:D
bb
Subwars
1st November 2002, 13:04
WTF excuse me umm whats your spacebar broken or something dam as if that is readable
Koepi
1st November 2002, 13:15
i dont know what happened it might be the lack of punctuation or complete sentences with some divisors within it or whatever in the initial post ah who cares i just wanted to see if i can write that way as well as you can see i succeeded!
*eg*
buba king
1st November 2002, 14:09
Your friend at the end. (http://www.penny-arcade.com/view.php3?date=2002-10-11&res=h)
btw.. IMNHO XviD is better than DivX5.
TheXung
1st November 2002, 22:17
Using a high quantizer on P or I frames has the negative effect of not only introducing artifacts/losing detail for that frame but it also degrades the quality of the subsequent frame that is predicted off of it.
Because nothing is predicted off of a Bframe, using a high quantizer on it only effects that particular frame and no other frames. Think about it this way, so if you save bits from the bframes and put it to use in the p frames, you are improving the image quality in the long run.
If you're familiar with cars, the effect is similar to a turbo on the engine. The exhaust off the car is used to power the turbo to pump more air into the engine, as the engine uses more fuel due to the increased air, more exhaust is generated to pump the turbo with more power and thus the turbo can pump more air into the engine and so on. . . If you save bits on the bframe, you can use those bits to make more accurate p frames, with more accurate p frames, not only are subsequent pframes more accurate, the bframes are as well. Of course like in a turbo, you reach a maximum gain naturally. With bframes, there is an equilibrium where taking more bits from the bframe doesn't give enough to the pframes to give better quality.
This works particularly well with low motion scenes, because it is exploiting the temporal locality of movies. If you watched a fairly still scene at 12fps instead of 24, you wouldn't notice the film being as jerky as if it were a high motion scene. If you interpolated a frame inbetween each one of the 12 fps frames, the difference would even be less noticeable. The temporal locality and redistribution of bits to the more "important" frames are the principles behind why bframes improve quality at a given bitrate.
That's one of the nice things about xvid's bframes, the number of bframes in a GOP is adaptive. Perhaps the quant ratio or final quant should be as well . . .
vBulletin® v3.8.11, Copyright ©2000-2026, vBulletin Solutions Inc.