Divx 5.2 vs Divx 5.1.1

[nicco]

We started to talk about this in another topic, so I started a new dedicated one.
Comparsion between divx5.2 and xvid or other codecs are no very good for divx (see other topics), furthermore I can't see any improvements in term of quality from 5.1.1 to 5.2, but at this first approach dvix5.2 may looks even worse, and I red that many people have this opinion.
What do yuou think?

[Enigmax]

I hold an opinion like you. DivX does not get better in last versions.

For my, the best compressions obtained with DivX: the version 5.0.5, in the Original Pass mode.

I wait improve in the next version.

But... I Love XviD :)

Greetings

[nicco]

5.0.5 maybe is a bit old, I refer to 5.1.1;)

[Enigmax]

Look:

DivX 5.11 and 5.2 whit de Home Theater Profile, no psy, no MV, B-frames.
DivX 5.05 in Original 2 Pass Mode, no psy, no MV, B-Frames.

Bitrate 700 kbps:

DivX 5.2:
http://enigmax.en.eresmas.com/DivX52.jpg

DivX 5.11:
http://enigmax.en.eresmas.com/DivX511.jpg

DivX 5.05:
http://enigmax.en.eresmas.com/DivX505.jpg

Captured whit BS Player and FFDSHOW. No postprocessing.

Greetings

[Enigmax]

DivX 5.2:

http://enigmax.en.eresmas.com/52.jpg

DivX 5.11:

http://enigmax.en.eresmas.com/511.jpg

DivX 5.05:

http://enigmax.en.eresmas.com/505.jpg

DivX 5.2:

http://enigmax.en.eresmas.com/_52.jpg

DivX 5.11:

http://enigmax.en.eresmas.com/_511.jpg

DivX 5.05:

http://enigmax.en.eresmas.com/_505.jpg

2 minutes and 16 seconds
700 kbps

5.2 = 12.066.816 bytes
5.11 = 12.052.480 bytes
5.05 = 12.060.672 bytes

Results: DivX 5.2 is the worse.

EDIT: From the film "Tomorrow Never Dies", chapter 2.

Greetings

[nicco]

DivX 5.11 and 5.2 whit de Home Theater Profile, no psy, no MV, B-frames.

How many passes? Stndard or Slow? Qpel on? What kind of b-frames in 5.2? Maybe you should post the dvx-command line for more accuracy
;)

Divx 5.05 frames have all a lower quantize (so better quality, of course)

But these good tests shows that from 5.1.1 to 5.2 there is almost no differences....

P.S.How can you overprint the frame's caracteristics on it? It would be very useful for me too, tnx:)

[Enigmax]

More pictures:

DivX 5.2:

http://enigmax.en.eresmas.com/52000.jpg

DivX 5.11:

http://enigmax.en.eresmas.com/511000.jpg

DivX 5.05:

http://enigmax.en.eresmas.com/505000.jpg B-Frame.

Whit the Home Theater Profile are activated, no QPEL. Standard Mode.

All the tests whit the same configuration, except 5.05 (only change the Original Pass Mode, Slowest of course).

Sorry, I did not intend to copy the CLI.

Zoom the pictures, the differences of images are more evidents.

Greetings

[Enigmax]

How can you overprint the frame's caracteristics on it? It would be very useful for me too
Use FFDSHOW Decoder, and enable OSD.

For me 5.11 looks better than 5.2.

Greetings

[nicco]

Talking about divx5.2, can you help me on this movie?
It's a DV frameserved from premiere, then avisynth

AVISource("D:\video.avi",audio=false,pixel_type="yuy2")
TomsMoComp(0,15,1)
Convolution3D(Preset="movieLQ")
SpatialSoften(2,2,4)
ConvertToYV12()

Divx encoding
900Kbps
3-pass @ stan-stan-slow
home theatre profile
No psv,qpe,gmc

***

It looks not very good to me, I tryed to encode it with 5.1.1 (same settings) and xvid (similar settings,900Kbps), but with almost the same result
Can you give me any suggestion?

[Enigmax]

Excessive resolution. You force the codec to use bigger quantizers.

Greetings

[nicco]

This is a 720x576 (PAL)
Should I resize it for example to 640 to 480?

On FFDSHOW, how can I enable it to be the decoder of my divx files while haveing divx installed?

[jggimi]

Start...(All) Programs .... ffdshow ... configuration

Select Codecs, set DivX 5 to "libvacodec" to turn on ffdshow.

[Enigmax]

Apply the test of compressibility to search the optimal resolution (or rise the bitrate).

Greetings

[Sharktooth]

... or use VP6 :)

[nicco]

Problem fixed; just needed to update FFDSHOW to 20040718 version.

Apparently DX 5.2 has very high 'Merit" and takes over, so FFDSHOW updated to fix that.
;)

@enigmax
What do you mean for

Apply the test of compressibility to search the optimal resolution

Thanks

[piscator]

Originally posted by Enigmax
Results: DivX 5.2 is the worse.


I admit I didn't try DivX 5.2 yet. But coming to this conclusion on a few snapshots with lower quants for 5.0.5 is imo premature. I'm sure there are other frames in which DivX 5.1.1 and DivX 5.2 have lower quants. The same goes for looking for differences per frame (just 1/25th of a second for PAL).

For me DivX 5.1.1 yields better results than 5.0.5. Especially in high-motion scenes or dark scenes. I'm usually targetting at 80% compressability and using 4-pass@stan/stan/stan/slowest. Other settings are: use MV, PsyVis Fast, GMC, bidirectional encoding.

For one interested in filters and CLI settings, I usually use:

Applied Filters: Crop(x,x,x,x)
LanczosResize(x,x)
UnDot()

Video Encoder: DivX 5.1.1 Pro, 4-pass
+ comptest 1-pass quality-based Q2, standard, PsyVis Off, GMC, bidirectional encoding
-b1q 2.0 -psy 0 -key 300 -b -g -sc 50 -pq 5 -profile 0 -nf
+ credits: 1-pass quality-based Q20, slowest, PsyVis Off, GMC, bidirectional encoding
-b1q 20.0 -psy 0 -key 300 -b -g -sc 50 -pq 192 -profile 0 -nf
+ first pass: multipass 1st-pass, standard, write log, use MV, PsyVis Fast, GMC, bidirectional encoding
-bvn1 XXX -psy 2 -key 300 -log "XXX" -mv "XXX" -b -g -sc 50 -pq 5 -profile 0 -nf
+ Nth pass: multipass nth-pass, standard, write log, use MV, PsyVis Fast, GMC, bidirectional encoding
-bvnn XXX -psy 2 -key 300 -log "XXX" -w -mv "XXX" -b -g -pq 5 -vbv 100000000,3145728,2359296 -profile 0 -nf
+ Last pass: multipass nth-pass, slowest, write log, PsyVis Fast, GMC, bidirectional encoding
-bvnn XXX -psy 2 -key 300 -log "XXX" -w -b -g -pq 192 -vbv 100000000,3145728,2359296 -profile 0 -nf

greetz,
Piscator

[Enigmax]

DivX 5.11 yields better results from 5.05 in high motion scenes, but not in low motion.
The difference quantization is for a different compression applied by the frame.
The Frame quant sequence in 5.05 are 4(P) , 8(B) ,4(P) , 8(B), 4(P), 8(B).

But in the case of 5.11 or 5.2 are most equally.

I have the opinion, DivX 5.11 works far better that 5.2.

DivX 5.2 it has more sound bad in the image.

Greetings

[Sagittaire]

DivX 5.1.1 (Standard, slow & slowest) is very better than DivX 5.0.5 for PSNR ... and visual quality is very better for slow or high motion with less artefact (blocking and ringing) for 5.1.1 ...

http://jfl1974.free.fr/Video/DivX5.JPG

Source: DivX Network


The question is: "DivX 5.2 is very better than DivX 5.1.1?"
For me it's "NO" ...

[SeeMoreDigital]

I have to admit I'm not very happy with DivX5.2.0

Although the explanation (in the new DivX guide) of how 'adaptive 1B-VOP and mB-VOP should help with the encoding process, I really don't see any visual improvement at all :confused:

When you weigh up encoding speed together with the finished encode I'm with some of you guys when you say DivX5.0.5 is a better product - and maybe even DivX5.0.2!

Personally speaking, I'm sad to say I think there's very little point conducting further tests using DivX5.2.0...

In fact I've removed it from my PC altogether as I can't see myself ever using it :(


Sorry DivX

[nicco]

I really don't see any visual improvement at all

I fully agree with you, it has no improvements from 5.1.1 at all and other codecs like VP6 and especially RV10 are a step above. (This is also the overall opinion in the posts I red)

I'm very disappointed :(

I don't think Divx would like to become "obsolete", so I hope divx will give us an answer on a few days, and I'm waiting for a real new amazing version as soon as possible

:rolleyes:

[SeeMoreDigital]

The thing that really narks me is why DivX did not offer any beta versions of the 5.2 codec for testing!

There are plenty of people on this forum who could have helped DivX with their product.... and who can keep a secret....


Cheers

[Sharktooth]

Yeah... i'm still asking myself if they have a QA (quality assurance) team and if they have it who are the moles in QA...

[nicco]

The thing that really narks me is why DivX did not offer any beta versions of the 5.2 codec for testing!


I don't understand too:scared:

5.1.1 had two betas which were tested by hundreds of peoples!

I hope someone of divx team will reply in this topic...

[nicco]

We compared single frames with the same quantizer, but becouse our final movie is not a frame, I encoded the same movie in Divx and xvid with both 2-pass (stan-slow for dvix, GMC, 900Kbps, mB-frames)
and I noticed that divx tends to give an higher quantizer to frames (both P and B) this leads to an overall worser quality of the same video, and the final size is almost the same.
We can give the same quantizer to frames for compare them, but becouse when we encode a full movie with 2 passes we have to let the codec decide, I think it's another bug in Divx5.2...

[SeeMoreDigital]

I'd be interested to see what the XviD guys could come up with if they created their own version of 1B-VOP and mB-VOP 'adaptive' seeking and encoding!


Cheers

[piscator]

Originally posted by Enigmax
DivX 5.11 yields better results from 5.05 in high motion scenes, but not in low motion.

The difference quantization is for a different compression applied by the frame.
The Frame quant sequence in 5.05 are 4(P) , 8(B) ,4(P) , 8(B), 4(P), 8(B).


Also low-motion in 5.11 is better for me than 5.05. Most probably because of slowest encode mode; I found that improved quality quite a bit, though of course not encoding time ;) But probably too because I usually employ a quite generous bitrate (800-1200 kbps depending on source), so the difference between the various codecs is harder to spot.

At the time I made my own test on several sources, it also gave better results than XviD. But that was more than a half year ago and before XviD 1.00. At that time I found XviD could sometimes give better results tweaking with matrices and some encoding settings. I didn't like that because I like a 'single setting gives reasonably quality for all' policy, because of limited time :D

But with all the nagging om 5.2, I guess it's more worthwhile for me to try the newest XviD.



Originally posted by SeeMoreDigital
The thing that really narks me is why DivX did not offer any beta versions of the 5.2 codec for testing!


As you can no doubt guess --at least I do :D-- this has to do with the new license/sale policy of DivX 5.2. No unlimited use of the codec anymore at the cost of gain spyware, but only 'limited use' or 'buy now' policy. And of course this message is sold under the benefit 'DivX 5.2: 100% adware free' :devil: == marketing... :rolleyes: Though half a year free use is quite generous, I wonder if that will also hold for newer versions or if the trial will be extended for newer versions. I know, I'm quite suspicious on these things and usually for a very good reason :D

greetz,
Piscator

[denissio]

Hi!
I have to admit that I agree with those of you who prefer 5.05 to higher versions, and "for a very good reason". I've been encoding movies since the release of 5.0 version and I was quite happy with DivX encodings (v. 5.02 to 5.05) until 5.1 was released. This version was a kinda unpleasant surprise to me, mostly due to a "new high-quality psychovisual mode". It seems to me that they sacrificed the details to a less amount of blockiness in high-motion scenes making the entire movie look very blurry.
But the 5.05 is ok to me though for low bitrates (< 600kbps) I'll probably stick with RV10 and for high bitrates to XviD or VP6.
Maybe you'll be interested to read my post in DivX forum here http://forums.divx.com/viewtopic.php?forum=5&topic=59303

[DigitAl56K]

Originally posted by SeeMoreDigital
The thing that really narks me is why DivX did not offer any beta versions of the 5.2 codec for testing!



Hey guys - we're already talking about beta releases for future versions. It's definately something we should have done for 5.2, but we had so many projects running concurrently we didn't have any time in the schedule to support a beta program.

I'll keep you all posted on beta information as we move forward - thanks for the feedback :)

[SeeMoreDigital]

Originally posted by DigitAl56K
Hey guys - we're already talking about beta releases for future versions. It's definately something we should have done for 5.2, but we had so many projects running concurrently we didn't have any time in the schedule to support a beta program.

I'll keep you all posted on beta information as we move forward - thanks for the feedback :) Well I know I have a tendency to come on a bit strong... but, should you decide to create private builds, I would certainly like to be considered as a beta tester!


Cheers

[Soulhunter]

Originally posted by SeeMoreDigital
...but, should you decide to create private builds, I would certainly like to be considered as a beta tester!Me too, me too... :D


Bye

[Fr4nz]

Originally posted by DigitAl56K
Hey guys - we're already talking about beta releases for future versions. It's definately something we should have done for 5.2, but we had so many projects running concurrently we didn't have any time in the schedule to support a beta program.

I'll keep you all posted on beta information as we move forward - thanks for the feedback :)

Any news about this?? :D

[AsTimeGoesBy]

Hello! - Initally then getting the 180-day-trial i have appreciated that the slow-mode is much faster than at previos versions.
But now in meantime after encoding some DVDs I also have my doubts about versio 5.21.

At v5.05 and v5.11 mostly i have choosen a bitrate around 2Mbit (+/-300) that nether has caused real grave quality problems.

Now at v5.21 i have _sometimes_ really strong quality problems, when/at:
- 2pass
- high definition preset
- slowest mode
- psy-effects: slow
- light pre-processing
- key frame within 750 frames
- 1750kbps at 720x320 with 25fps

Parts of my newly encoded 'Pirates of the Cariibean' looks really ugly. Please see the examples (http://homepage.hispeed.ch/audiold/Cariibean.html) if you like.
Moreover i have noticed, the kind how DivX v5.21 sets the keyframes is quite strange.
Please see/check this dialoque of two guards (divx-avi, 1.4mb) (<= link removed afterwards, sorry).
The 1st camera shot absorbs 35 keyframes within 2 seconds for a speaking but standing guard. But the 2nd guard (also speaking) gets no keyframe.


I have already posted a similar post at the DivX forum - but no reply.
So perhaps the user of this forum may help me to get my DivX files good again.

______
Edit:
I must re-tink my position... - The quality of some of my new 5.21 encoding is so bad that i guess there was another negative influence too. - Nethertheless some doubts still remain.