Is Divx 5.05 any better than Divx5.02??

[Snilly]

Howdy all,

I've yet to upgrade to a newer version of Divx5.02 since the 5.03 version seemed to be unstable and get a lot of bad press. From what I've noticed a lot of you did the same thing.

Divx 5.04 was replaced with Divx5.05 so quickly it's a non-factor in my opinion.

Soooo my question is, since most of the traffic I've seen on Divx5.05 has been discussing the merits of multipass vs two pass encodes...and the consensus seems to be 2 pass is fine and dandy....Is there any reason to move to 5.05?

Outside of spending a few hours tweaking EKG for each encode (if you are into that) are there any benefits to picture quality etc...?

Thanks!

Snilly

:sly:

[Babeinternational]

Multi-passing is definitely something to do. Your quantizers are fine-tuned, so is your filesize, so you gain quality.
As for EKG, it is a very interesting tool that allows you to adjust your movie to your subjective eye.
Of course you have to play your file, and take decisions.
But you can suppress artifacts when you see them, and just that is a major improvement.
And if you're lazy and impatient (no EKG and just 2 passes), there are still many bugs in 5.0.2 that were corrected in 5.0.5.
So no real reason not to upgrade...

[makikes]

But if you're still insterested in making 2-pass copies you can always use the original 2-pass mode instead of the nth-pass mode. So, going for 5.0.5 is a win-win solution.

[bond]

divx505 is much better than divx502
try and you will see!

[frauz]

Originally posted by bond
divx505 is much better than divx502
try and you will see!

... but I'm still not quite sure about improvements in latest version. On the other hand, 5.02 seemed workin' better (however not worse), especially in dark scenes where these famous forum-around-speaking macroblocks never appeared.

I suggest to open a thread on comparison between these 2 codec versions (oops there I am :p) where anyone is STRONGLY invited to post comments on personal experience.

ps: I'm expecially talking about low bitrates for 1 CD movie (<2h30m), GordianKnot settings like Doom9's guide (GMC and B-frames enabled and bla bla bla :cool: ).

ps: I LOVE THIS SITE, THANK YOU ALL FOR EXISTING!!!

[makikes]

doom9 already did a remarkable job in his codec comparison (http://www.doom9.org/codecs-103-1.htm). Further discussion would have to be based on new features, or else it would be a repetition and a waste of time.

[DaveQB]

any more comments ??

i echo 'Snilly' with what he said.

i am on 5.0.2 now and tempted to update to 5.0.5 and try out these multi-pass options.

any bugs found so far ?? n freezing of encoding apps etc? :D

thanx

[leadman584]

Divx 5.05 is by far the best release to date from the folks at DXN. It is possible to configure Xvid to give slightly improved results, but I'm too lazy to spend hours configuring a codec. Especially after spending hours setting up my .avs file to feed optimum signal to the codec. Divx 5.02 was inferior to even the default set-up of the December stable release of Xvid.

[Snilly]

Are you suggesting that running Divx5.05 with default options and using 2 pass produce these results?

I was under the impression there has been a fair amount of debate to use n pass and/or EKG to get the real benefits of Divx 5.05.

I agree with your comments on XVID. I'm impressed with the quality of the videos I've seen but do not plan to spend SBC like time configuring my rip. Thats why I started this thread as I believe spending a lot of time with SBC, XVID, or the EKG tool in 5.05 will yield slightly better results with a significant investment in time.

What I would have liked to see with 5.05 is a codec that produces noticeably superior results to 5.02 without the need to do more than was done to set up a 5.02 rip.

Snilly :sly:

[DaveQB]

hmmm ok doesnt seems to be any negative feed back on 5.0.5 atleast so i'll give it a go

(after imaging my boot partition of course ;) )

*mental note: must spend more time at looking at Xvid, its the best for encoding on the fly*

[EagleDM]

in MY experience (notice that i said mine)

Divx 5.02 is better than divx 5.05 when using multipass.

But, Divx 5.05 decoder is WAY better than the divx 5.02 decoder, my old movies (made with divx 5.02) looks better now.


As long as you use Divx 5.05 with the old 2-pass method, it should be equal or better than 5.02.

[frauz]

Originally posted by EagleDM
in MY experience (notice that i said mine)

Divx 5.02 is better than divx 5.05 when using multipass.

...

As long as you use Divx 5.05 with the old 2-pass method, it should be equal or better than 5.02.

It seems to me too that new DivX 5.05 multipass fails in quality when done in automatic (not with EKG) and more than 2 times. I agree with you when you say that the old 2-pass method (I'm still talking about 5.05) looks better

[VideoJedi]

I've always used Nandub (SBC divx3.11a).. trying divx5.05 2-pass now.. wow.. I am seriously going to give up Nandub. 5.05 seems sharper. I didn't really get into 5.02 and 5.03. I think I have found a suitable replacement for 3.11a.

[DaveQB]

no one has mentioned the Quality Pass method here
supposedly better, just harder to pick the size.

i found that 93% (an optimal setting) in 5.0.2 resulted in a 4.0 setting
and now 93% results to a 3.0 setting and acts like it too,
the same movie

in 5.0.2 QB @ 93% ~ 550mb
in 5.0.5 QB @ 93% ~ 1gig

[MyC sPiTTa]

I've just put 5.02 back on after trying 5.05. I did a few movies with 5.05pro just using the default 2-pass mode, and I found the quality to be very poor. I thought I was just imagining it (you know the feeling) until I put 5.02 back on and the quality is MUCH better. I dunno, maybe I had 5.05 set up wrong but I'd followed the guide to the letter. The main problem I found with 5.05 was massive amounts of "bubbling" effect - you know, motionless backgrounds (usually of uniform colour) would move and flicker. There were also a lot of jpeg-like artifacts around the edges of objects. And the shrpness was generally lower.
I'd like to use 5.05 but I bet the (supposed) quality improvements are not worth the effort of me trying to resolve these problems. 5.02pro always semmed to to a good job for me.

[DevilsChild]

Never used 5.0.2, so I can't compare the two. 5.0.5 works great for me, especially the latest one from DivX Labs (which is an improvement over the regular 5.0.5).

I always use the 6-pass mode, with 128 kbps MP3 audio, for a full length movie on 1 CD.

[bond]

Originally posted by MyC sPiTTa
The main problem I found with 5.05 was massive amounts of "bubbling" effect - you know, motionless backgrounds (usually of uniform colour) would move and flicker. There were also a lot of jpeg-like artifacts around the edges of objects. And the shrpness was generally lower.this could have been my describtion for divx5.0.2 :D

in my opinion divx5.0.5 is far superior than 5.0.2 (bringing a much sharper and more "quiet", nicer picture)...

but i suggest everybody to use the original 2-pass as it produces a much more sharper picture in low motion scenes OR move the bitrate modulation slider to low motion (~0.2) to get similar results

and it was widely agreed on this forum that more than 3 passes are worthless (btw a good codec should be able to do the same work in 2 passes imo :rolleyes: )

and i wont use the divx labs codecs as they are for testing only and i am sure that there is the great possibility that you cant playback these encodings with future official codec releases

[SeeMoreDigital]

Very interesting Mr Bond....

and i wont use the divx labs codecs as they are for testing only and i am sure that there is the great possibility that you cant playback these encodings with future official codec releases

I've done quite a few tests now with DivX 'Kauehi' and the images produced are better than DivX 5.0.5.

Just to put your mind at rest. I've found that the Sigma Xcard plays back the DivX 'Kauehi' encodes more smoothly than the DivX 5.0.5 encodes. Very strange!

[r0cket]

you forgetting about more flexible rate controle in 5.0.2. also you can set minimum quantizer there.
btw 5.0.2 has no problem with Motion Vectors reuse unlike 5.0.5.

[Selur]

about the bluring, did you check that you disabled any smoothing&co settings in the decoder?

Cu Selur

[Soulhunter]

@DaveQB
_________________________________

in 5.0.2 QB @ 93% ~ 550mb
in 5.0.5 QB @ 93% ~ 1gig
_________________________________

> in 5.02 QB100% = Quant2 <
> in 5.05 QB100% = Quant1 <


@All

I've made some tests with MultiPass, and I have to say Im not happy with the results...
So I stay by the old 2Pass Mode.
But 5.05 is still better than 5.02 because 5.02 causes "some" problems in "some" situations (flickering blocks)! With 5.05 this problems are gone....

Between, DivX 'Kauehi' works well... But is it a Pre-Build of DivX5.06 or DivX6 ???

Bye

[MyC sPiTTa]

Ok, I've reinstalled 5.05pro on a fresh windows install and encoded Final Fantasy. It looks pretty good so maybe I had it set up wrong before. Then again FF is probably the sort of movie which encodes pretty well anyway.

[iparout]

I had 5.0.5 Pro installed but I re-installed 5.0.2 because 5.0.5 Pro 2-pass mode never gave undersized files, even at 160% compressibility, which makes me think that it uses Q1 frames at that compressibility. And since you cannot set the lowest quantizer for a two-pass encode, I went back to 5.0.2 Pro, cause I prefer to have better mp3 audio than Q1 frames in my movie. The difference between Q2 and Q1 are NOT noticable IMHO.

And what is also very important, I noticed that the macroblocks in the dark areas of a movie are less noticable in a 5.0.2 Pro encode (that was based on tests I did with two movies).

[gpower2]

I agree with iparout, I also want to control the quantizers, because it has great difference at the compression. Why did they disable manual quantizer control? Will it come back at future versions? I believe that is what they said when they disabled it at 5.0.3 which by the way sucked big time!

I now use XviD because I have much better control of the quantizers and the compression level, but I would return to DivX if it had quantizers...
CU

[Soulhunter]

@gpower2

I thought the old commandline function is still possible...

Commandline:
-dr 12,2,2000,10,20

But it seems not work anymore... :(

Bye

[gpower2]

@Soulhunter

That's exactly my point...

CU