DivX 5.0.4 Beta 1 (codename: Schizo)

[timeToy]

Update2:

What's New in DivX Schizo Beta 3

DivX
Fixed:

1 Pass and Nth Pass™ undersized file should be fixed
Certain dark sequences would produce a long period of frames using a Quantizer of 31.
Psychovisual Enhancement field sometimes displayed "Custom" instead of Light when unselected
"Do not prompt with errors and warnings" would select itself when random characters were entered in the CLI after the "-key" option
When "-d 3" option was entered in the CLI field, "Source interlace" became blank.
Max bitrate would not update at times.
DirectShow decoder postprocessing slider became disabled with ATI radeon 9700 Graphics card
DirectShow decoder would "fast forward" when seeking backwards. This could cause synch issues when re-encoding DivX Files using the DivX Decoder Filter.
EKG
Fixed:

Windows 98 OS would give an Error when EKG was exited.
Clicking on FFwd (Rewind)/Left End (Right End) in "Actual Size" preview window switches the screen to Main window
Seeking for EKG was horribly slow. Updated the DivX Decoder to enable more efficent seeking.
Clicking on FFwd (Rewind)/Left End (Right End) in "Actual Size" preview window would switch the screen to Main window.
Moving "Legend" icon would create brown colored patches
EKG would not warn the user when electing a log file that does not match the video file.
Garbled (noisy and slow ..) audio heard during the seeking of preview video
Seeking to the request frame from the previous keyframe upon pressing "previous frame" button on the EKG.
"Seeking" to a previous frame" button was disabled upon pressing "seek to first frame".
"Seek to next frame" button was disabled upon pressing "seek to last frame".
Known Issues
A Frame selected for editing and the respective displayed scene video preview may be offset by a single frame.


Update:

DivX 5.0.4 Beta 2

A major registration bug has been fixed, sorry about that, all users could now use the Beta, and not only previously registered ones.




See at DivX.com forum:

DivX 5.0.4 Beta 1 (codename: Schizo)

Hi, welcome to the External Public Beta of Schizo, the next version of DivX Pro.

This version fixes lots of bugs found in DivX Pro 5.0.3, and it adds a new tool, EKG , that helps manually change the bitrate distribution for multi-pass encoding.

The Beta program is free and open to everyone.

Go grab the beta here: http://www.divx.com/divx/windows/beta/

The DivX 5.0.4 Beta 1 codec is limited to 30 days beginning on the day of installation. A final version of the codec will be released at the end of the Beta period; this release will be publicly announced on the DivX.com web site.

Beta testers who are not registered users of DivX Pro will need to update to the final version in order to keep using DivX after the beta period has expired. Beta testers who are currently registered users of DivX Pro will be able to use the Beta codec with no time limitation, but will be strongly urged to update to the final version at the end of the Beta period.

The main purpose of the External Beta Program is to gather valuable feedback for the benefit of our users. The Beta program is NOT designed to evaluate the DivX Pro Codec.

Users who discover bugs, inconsistencies or other issues should use the Bug Feedback Form to report their findings. This form is rather long and complicated, but please take the time to fill it out completely. The purpose of this form is not to annoy you, but rather to make sure we have all the valuable information necessary to understand and correct the submitted bug.

Bug feedback form: http://www.divx.com/divx/windows/beta/feedback.php

To help us process your valuable feedback, please make sure that all the hardware configuration fields, DivX Codec configuration fields and bug reports fields on the report are COMPLETED in full. The bug title and bug description must be as clear and understandable as possible. Please take your time when filling out the feedback form, especially when detailing the steps to recreate your bug. We prefer to have one properly described bug feedback per user rather than several quickly submitted but incomplete reports.

We have also created a Beta forum to allow users to share their thoughts and issues regarding the Beta release. The Forum is not a place to post bugs—all bugs MUST be submitted using the Bug report form. The Beta forum will include a sticky post that summarizes all the bugs discovered in order to avoid double submission.

Beta Forum: http://forums.divx.com/viewforum.php?forum=23

Have fun and see you on the forum.

Gej

[Morbo]

Anyone think its quite "xvid" like to use an open beta...;)

Anyways,time to see if 3 passes is better than 4 four passes,yadda yadda yadda...

Also glad to hear Gknot went open source...

Cheers!!!!

[Sirber]

EKG says that my license is invalid. How did he know? :p

[Morbo]

I just made a test...
1 pass,CQ
No bframes,GMC OR Qpel
589k

I used track two of minority report...

The slow pan up,when Tom notices the merry go round,the file kills Mplayer......

"Its a beta"

I know.....

I warned...:(

[timeToy]

Did you use the divx decoder ? Did it crash in VirtualDub ?

[Morbo]

UNinstall DIVX 5.0.2 first.........

plays fine now..

Cheers!!:angry:

[chris25frfr]

EKG says that my license is invalid. How did he know?


a lot of people have this problem . no solution at this moment ( view in the beta forum)

[BaronVlad]

Hi,

if you have problems regarding the installation of schizo, please look into my thread in the divx.com board:

http://forums.divx.com/viewtopic.php?topic=47631&forum=23

There are some hints what to do. This helped me since I had the same problem...:)

Have Fun, take care

Thanks gej, thanks DigitAl56K for your help

[Morbo]

Ah, I’ve no idea what theses applications does...

LOL....

A better answer:

Ah,Im not sure what you've done with these apps....

Im SURE he KNOWS what they are.....:D

Im fine now,but I dont like it as well as XVID or RM9 anymore...
I get Horrible blocks in still scenes(looks good in high motion BTW),and Im not content to use another program to "correct" scenes..call me lazy..but thats not a great "PRO" feature IMHO.

Cheers!!

[BaronVlad]

Originally posted by Morbo

Im SURE he KNOWS what they are.....:D


Did you mean:

Im sure HE (Vlad doesnt) know(s) what they are... ? :D

Maybe, but the proggies did the job as they always did...

I use XVid for caps and normally DivX for DVD, please dont ask me why...;)

[midiguy]

why?

[Sirber]

Use DivX, use XviD, but don't use both... :P

[chris25frfr]

somebody can explain me what ekg do please ??

thinks

[Acaila]

Just read the link timeToy gave: http://www.divx.com/divx/windows/beta/ it explains it the best way possible in my opinion.

[chris25frfr]

ok thinks. very clear but i have a simple question .

Is it possible to make sveral changes in the divx.log . for example 3 times , or we must encode each time(do a new pass and then modify the log file ?)

in fact , i want to modify in the middle of the movie and then credits.


thinks again

[Acaila]

To see the effect your changes, and adjust accordingly, have you'll have to run an encode after each change. Of course you can make changes to more than one section before encoding.

[timeToy]

From the DivX.com Website

http://www.divx.com/divx/windows/beta/

What's New in DivX Schizo Beta 3

DivX
Fixed:

1 Pass and Nth Pass™ undersized file should be fixed
Certain dark sequences would produce a long period of frames using a Quantizer of 31.
Psychovisual Enhancement field sometimes displayed "Custom" instead of Light when unselected
"Do not prompt with errors and warnings" would select itself when random characters were entered in the CLI after the "-key" option
When "-d 3" option was entered in the CLI field, "Source interlace" became blank.
Max bitrate would not update at times.
DirectShow decoder postprocessing slider became disabled with ATI radeon 9700 Graphics card
DirectShow decoder would "fast forward" when seeking backwards. This could cause synch issues when re-encoding DivX Files using the DivX Decoder Filter.
EKG
Fixed:

Windows 98 OS would give an Error when EKG was exited.
Clicking on FFwd (Rewind)/Left End (Right End) in "Actual Size" preview window switches the screen to Main window
Seeking for EKG was horribly slow. Updated the DivX Decoder to enable more efficent seeking.
Clicking on FFwd (Rewind)/Left End (Right End) in "Actual Size" preview window would switch the screen to Main window.
Moving "Legend" icon would create brown colored patches
EKG would not warn the user when electing a log file that does not match the video file.
Garbled (noisy and slow ..) audio heard during the seeking of preview video
Seeking to the request frame from the previous keyframe upon pressing "previous frame" button on the EKG.
"Seeking" to a previous frame" button was disabled upon pressing "seek to first frame".
"Seek to next frame" button was disabled upon pressing "seek to last frame".

[Victoire]

there is support of ATI 9500/9700 since 5.04 b3.
Source: http://playland.ru/forum/viewtopic.php?t=905 (in Russian, try babelfish.altavista.com or www.translate.ru for translation into your native language)

[salatec]

http://www.divx.com/divx/windows/beta/

[colordog]

I know a lot of people were saying that B-frames got screwed up in the 5.0.3 release, so I never updated to it. Any clue on whether this has been remedyed with any of the Schizo updates?

[Big_Berny]

I don't know because I never had problems with b-frames and 5.03. 5.04 works too, you should try it. And don't activate GMC.

Big_Berny

[mikeson]

@Big_Berny:
And don't activate GMC
Why?

[Big_Berny]

Look here: http://forum.doom9.org/showthread.php?s=&threadid=47559

Greets
Big_Berny

[leadman584]

Divx 5.04 Beta 4 was released coupla days ago, and wow. I've been working with this beast since Beta 1. Doing full video encodes to put this thing through it's paces. The fixes described at Divx info page are a small taste of what they have managed to repair so far. I have encountered tons of probs along the way, and DARC has managed to address them all in this latest beta. Yesterday did a 3 pass 640x480 of SWE2(from widescreen to fullscreen) with PV1 and Bits/pixel of .172, 2 disc encode. No distiguishable chroma ringing(shocked me) and from 1 foot or more from the screen virtually identical to original DVD. This most recent Beta from Divx is by far their best effort, ever. I sincerely hope that many of the skilled encoders from this site will give it a try. I have been guilty of criticizing Divx Codec in the past particularly 5.02(almost all versions of Xvid were superior). It looks like we're getting a winner this time in Divx 5.04. Good thing too, with set top players coming down the pipes for Divx in the near future(well they are kinda here already).

Anybody who can help test this monster, is definitely appreciated. The more problems we can all uncover now, the better the codec will be in it's final form. Remember these guys get paid just to do Codec developement, let's make them earn their paychecks.

[BiaTch 5.0]

Originally posted by leadman584
Divx 5.04 Beta 4 was released coupla days ago, and wow. I've been working with this beast since Beta 1. Doing full video encodes to put this thing through it's paces. The fixes described at Divx info page are a small taste of what they have managed to repair so far. I have encountered tons of probs along the way, and DARC has managed to address them all in this latest beta. Yesterday did a 3 pass 640x480 of SWE2(from widescreen to fullscreen) with PV1 and Bits/pixel of .172, 2 disc encode. No distiguishable chroma ringing(shocked me) and from 1 foot or more from the screen virtually identical to original DVD. This most recent Beta from Divx is by far their best effort, ever. I sincerely hope that many of the skilled encoders from this site will give it a try. I have been guilty of criticizing Divx Codec in the past particularly 5.02(almost all versions of Xvid were superior). It looks like we're getting a winner this time in Divx 5.04. Good thing too, with set top players coming down the pipes for Divx in the near future(well they are kinda here already).

Anybody who can help test this monster, is definitely appreciated. The more problems we can all uncover now, the better the codec will be in it's final form. Remember these guys get paid just to do Codec developement, let's make them earn their paychecks.

You sound like a sales man. I don’t think your very vague description of DivX 5.04b quality will help the users or developers much.

So far it seems to me,

B-frames need to be retested with the current beta

GMC needs work & isn’t worth using

Qpel (as with 5.02) isn’t worth mentioning

Anyone who has done any decent testing please correct me

[BoNz1]

Well, I really don't know how you can say that without backing it up with some of your own testing. leadman584 has done a lot of testing, if you look in the beta testers forum at divx.com you can see for yourself. I have done some testing myself with beta1-3, I haven't tested 4 yet though. From my tests b-frames work quite well and do not seem to blur as much as before in 5.02. I think if you did your retest you would probably find the same. Quarter pixel doesn't help PSNR much if at all, but then this is the case with most mpeg4 codecs. Of course this depends what you are encoding, whether there is a lot of high motion scenes or low motion. It does however produce a much sharper picture so I like to use it. They have completely reworked 1/4 pixel from 5.02 so you may want to take a look at it again. As a far as GMC you are right, it doesn't do much but again this is case for every mpeg4 codec. Look at xvid, using GMC might only decrease 1st pass size by like 1%. I would really like to see DXN give us back some more control in the codec. I seems that the limited control we did have from 5.02 was taken away. I understand that they have reworked the rate control and it seems to work good so there might not be any need to fuss with rc averaging settings etc. But, I would like to see an option where I could specify max and min I-frame, P-frame, and b-frame quantizers. I have other things on my wishlist as well such as mpeg quantization/custom matrices but those can wait. :)

[leadman584]

I have used B frames (virtually every encode), Q-pel, and GMC in some of the tests I have done. All of these features work fine in the current version of Divx, Though really only B frames are needed. The biggest problem I have encountered so far involves bit distribution when any level of PV is used. I posted here to try and get a few more folks with very critical eyes to take a look at this codec and help find bugs. EKG is a little strange to use, for the most part it does work, but I hope some folks will throw their 2 cents worth in to improve the layout of this tool.

As for experience with this beta testing period, Last night I passed the 15 million frames encoded mark. Hopefully I can get another 10 million in before they close the testing period.

[dTb]

Originally posted by leadman584
As for experience with this beta testing period, Last night I passed the 15 million frames encoded mark. Hopefully I can get another 10 million in before they close the testing period.

Not a bad effort there leadman584, much much more than I've done (are you counting each pass though? :sly: ).
I've used betas 2,3 and 4 and have had good results. A couple of movies I've encoded are definately the best quality I've done yet.
I never experienced any probs with b-frames in 5.0.3 and have yet to experience any with the beta releases.
There is a problem with pve while using the mvinfo.bin file so if your going to use pve it's best to disable the mv file.

[leadman584]

True PV is still screwed (I always use MVinfo.bin). You kinda got me on frame counts, but I was trying to get some videos to work ok with PV. I'd run 2 passes together, then additional passes seperately(10 passes total, on 2 different videos). I have however run more than 100 passes of full videos using this codec, and without PV they do look pretty darn good.

[BiaTch 5.0]

@leadman584

In 5.02 did you use PE? many say not to use it but I found it to produce cleaner frames. Really I found GMC not to do much B-frames helped to stop blocks. I have always found ALL DivX releases 3,4,5 to produce a kind of water mark in some darker frames & I remember testing Aliens with PE there was a scene where this would become horrible with hard PE, light wasn't bad but still PE off was best these where rare cases in my findings I found PE light to be a good tool to encode with. The PE artifact would not be present in single frames, I have a 56K connection but if anyone is interested I can provide a sample file.

[leadman584]

On previous versions of Divx, I found normal or even heavy PV were giving better results than strong filtering. I have used light PV on most past encodes, and prefer it to higher filter levels in C3D. The darkening you mention that seems to occur with PV, is shift of data to more focused area of frame. I use Lumafilter(10,1.2) in the .avs file to provide a luma gain that compensates pretty well for the effects of PV. Because I do like to use PV, I tested Beta4 with light PV on, and MVinfo.bin off. Resulting 3 pass encode looks good, no dropped frames, no noticeable panning problems, but average Q values were slightly higher. The only real sore point with PV use is that you have to actually do the encoding to see the effects.

I think we are seeing the same kind of results using Past and present versions of Divx. If you shut off the MVinfo.bin, you will will still have PV available as a decent encoding tool. It's encouraging that so many of us are indeed able to duplicate this conflict in the current Beta test. Thanks to dTb for pointing out this conflict here, and I can confirm his findings.