DivX v5.04 is now out

[SiXXGuNNZ]

http://www.divx.com/

Release Notes (http://www.divx.com/divx/divx_win_versions.php)

[Snollygoster]

Can someone explain what this EKG thing is and how do we use it using our current tools?

[Dark-Cracker]

hi,

hum now the 1pass 100% quality use the quantiser 1.

++

[JohnMK]

Where does one download an update? I have a 5.04 beta, upgraded from a 5.02 Pro version.

EDIT:
http://www.divx.com/divx/pro/upgrade.php

[zyrill]

to clear out this issue once and for all: the so called "upgrade" version is the full pro version without the gain spyware - so if you have a key (if you don't - buy one!) download that version and just fill out the registration. one info though: since it is NOT ad-supported this version will not encode unless you registered it.

[Bez]

Originally posted by Snollygoster
Can someone explain what this EKG thing is and how do we use it using our current tools?

The EKG is explained in the official DivX guide on DivX.com.

http://www.divx.com/support/divx/guide.php. (Scroll down about 2/3 of the way in.)

:bez

[draggoon01]

Originally posted by Dark-Cracker
hi,

hum now the 1pass 100% quality use the quantiser 1.

++

now, did they actually do something different, or just re-number their previous settings.

[Ewi]

Originally posted by Dark-Cracker
hi,

hum now the 1pass 100% quality use the quantiser 1.

++

Are you sure? If you have tested it, OK... (I did not). But the only thing I can find about this is this:

During 1 pass quality-based, Quantizer can now be set to 1 with CLI

IIRC in the betas before the final the slider at 100% was quant 2 and you can set the quantizer manually to 1 (via CLI). But this can be wrong...

[ang3l]

With the new Codec 5.04 final Version

In "Multipass, 1st pass" mode... if I choose " write Log file" and "write MV file" ... it is not created the file MV to make to go swifter the second pass, but alone the file LOG.


It happens also you?

[draggoon01]

so anyone have any thoughts on quality of this codec? i haven't messed with ekg, but with nth pass i get marginally better results than 502. i've never used xvid so can't compare. for quality based pass though, the files are 10% smaller, but also seem a little less detailed. which is why i think they may have just renumbered their quantizers starting with 1 instead of 2.

but because of 503 experience, i'm going to stick with 502 until there's good reason (quality/size) and people generally find it stable/bug-free. (or maybe i'll switch when xvid releases a "stable" labeled build)

post your thoughts and share with all.

[valnar]

I'm with you. Multiple Nth passes doesn't entice me. I'm very happy with my 5.02 results. To get them better, I'm more waiting for the final .28 GKnot release.

Is 5.04 actually better in any way?

Robert

[LordIntruder]

Waiting for Gknot will not produce any better quality unless you are a beginner in the rip scene and don't know what settings to choose, what filter to use, etc... In that case I agree. :)

Some days ago I encoded a whole movie in 5.03 (an old one with some scenes pretty noisy and others not). I've tried the exact SAME settings with the new 5.04 last night. Differences may sometimes be seen but you have to be very closed to your screen. Watching both 5.03 or 5.04 version at correct distance does not allow you to distinguish any differences.

Of course my experience on 5.04 is only on ONE movie at the time I write this message. I can't conclude on this.

Anyway, I tested the EKG and this is the force of the new 5.04. Sometimes a scene has not been encoded correctly, you see some artefacts or the wall is "moving" or the quality is simple not good when the rest of the movie is pretty fine. Here you just have to select the bad encoded scene within EKG, you increase the quality by increasing the modulation as describe in the 5.04 tutorial guide available on divx.com (support section). So you do another pass and magically the scene looks now good. :)

It was impossible before to increase the quality of a specific scene but now you can and in a few seconds with EKG. I was septical, I gave it a try and it works good. :)

Try on a sample of one minute. Increase the modulation of the first 30 sec by 30% and see what happens to the first and the last 30 sec. You should be definitevely convinced. :)

[valnar]

Originally posted by LordIntruder
Waiting for Gknot will not produce any better quality unless you are a beginner in the rip scene and don't know what settings to choose, what filter to use, etc... In that case I agree. :)

Not a newbie, just lazy. :)

The new GKnot will have better resize filters built-in so that at least could make it sharper. Sure, I could edit my AVS files now in GKnot .27, but that requires extra effort. As stated before, I'm lazy.
:)

Robert

[madoka]

Originally posted by ang3l
With the new Codec 5.04 final Version

In "Multipass, 1st pass" mode... if I choose " write Log file" and "write MV file" ... it is not created the file MV to make to go swifter the second pass, but alone the file LOG.


It happens also you?
Now that you mentioned it, the MV file is also missing in my encodes. So now we know there are at least 2 instances of this behavior :-)

[BiaTch 5.0]

DivX still has problems with high motion scenes

DivX 5.04
-B Frames
-3 Pass
-2 CD, High Comp test

The low motion scenes have no blocks what so ever.

Edit: waiting on attachment

[niUmex]

madoka

So now we know there are at least 2 instances of this behavior

At me the same problems. Why last versions DivX so are filled mistakes? In any way at them it is impossible to let out the codec without bugs. A team of developers as though have changed. Since has left 5.0.2, versions with which leave is impossible to work...

[Jeffster]

Originally posted by ang3l
With the new Codec 5.04 final Version

In "Multipass, 1st pass" mode... if I choose " write Log file" and "write MV file" ... it is not created the file MV to make to go swifter the second pass, but alone the file LOG.


It happens also you?

There is a thread (http://forums.divx.com/viewtopic.php?topic=48044&forum=23) on the DivX forums where the MV file was identified as causing periods of low quality during a movie.
Now perhaps they haven't fixed it, so decided to release 5.04 with the MV file option disabled instead? This is just speculation.

Jeff

[draggoon01]

keyframes aren't being placed properly. i read this from someone else, and then noticed it's true for my testing also.

well, that's enough of divx 504 for me. i'll be waiting for divx505 now. so sad that divx costs money.

[Gej]

Keyframe issue is fixed:

http://forum.doom9.org/showthread.php?s=&threadid=51602

[Gej]

BiaTch 5.0: I’m very surprised with your results, is the pictures are actually taken from the final encode or the compression test ?
What are the settings, bitrate, modulation and Psy ?

[sapo]

Normal Version [Completely free]
http://download.divx.com/divx/DivX504Bundle.exe

Pro Version [Needs a valid serial number to make it work]
https://secure.divx.com/downloads/DivXPro504Bundle.exe

Pro Version [This one is free but is also bundled with GAIN spyware]
http://download.divx.com/divx/DivXPro504GAINBundle.exe

[BiaTch 5.0]

Originally posted by Gej
BiaTch 5.0: I’m very surprised with your results, is the pictures are actually taken from the final encode or the compression test ?
What are the settings, bitrate, modulation and Psy ? No, Final
3 passes with –B frames, NO PE, GMC or Qpel.

I always use GKnot with a 70-80 comp test (DivX 5 Enc v1.11 by jonny) & so far 4:4 different DVD movie encodes have had this issue with DivX 5.04

I’m not sure if this could be the –B frames issue people are claiming I will later try without –B frames.

Is anyone having this problem?

[BiaTch 5.0]

Here are samples of my other encodes (still with -B frames, same settings)

P.S I'm a big fan of DivX & it's quality so I hope this gets fixed soon.

[BiaTch 5.0]

Well Snatch was the last movie I encoded so it was the easiest to re-encode, same settings, no B- frames.
VideoSize: 609957 kbyte.
Audio1Size: 92406 kbyte.
Audio2Size: 0 kbyte.
Overhead: 9316 kbyte.
AviSize: 711680 kbyte (695 mbyte).
FilesSize: 0 kbyte.
TotalSize: 711680 kbyte (695 mbyte).
Interleaving: vbrmp3
Bitrate: 845 k(=1000)Bits/s
Frames: 147856
FPS: 25.000
Duration: 1:38:34

The result was the same without B-frames.

[BiaTch 5.0]

Here is another sample, this time DivX vs. XivD (To prove it’s a DivX bug)

P.S I don’t know how to capture single frames so…

[Ewi]

Is this RC bug gone in the new 5.05 that BiaTch 5.0 describes above? I don't think so, cause the only thing that is mentioned in the changelog does not belong to VirtualDub encodes...

[Acaila]

Two comments about your attachment:

1) You can make a screenshot from within VirtualDub by pressing CTRL+1. Open any image editor and paste the image there. Prefereably save as PNG since it's lossless, but that depends on how big the filesize becomes. That way you'll have the image you want without the VirtualDub frame surrounding it.
2) It's a lot easier to compare pictures when they have the same resolution...

[N_F]

... And 3), are of the same frame frame. You probably used B-frames in XviD, this shifts the frames a bit compared to DivX and XviD non B-frames.

Personally I started using XviD a few days ago when I realised what I had considered as noise in XviD wasn't noise. It was detail/texture. It is most noticeble on walls and faces.

[BiaTch 5.0]

O.K will do. I changed the res because I got a lower comp test with XivD but I understand why. I’m surprised I’m the only person (as far as I know) who has encountered this issue. I can encode with 5.02 if necessary although I’m not sure it will accomplish much for I have already shown several samples, I’m just waiting for a 5.06 beta testing thread or at least Gej to acknowledge the issue.