DivX Pro 5.1.1 Beta 2 is now available

[FreakNGoat]

What I'm trying to get at is although under some circumstances QPel is a nice feature, both with very low bitrate encoding or for DivX Certified compatibility its a no-no. This is where Slow/Slowest modes come in extremely handy - don't be so quick to put a feature down without thinking about its benefits

I realize they have benefits, but I don't think that should make them immune to criticism. I'm saying they do not produce the highest quality encodes, and at the expense of twice the encoding time. I'm glad that Divx is working to improve quality on hardware devices and very low bitrate encoding.


May I ask one question:If you are only playing on Computer and need a quick encoding,will enabling Qpel,GMC,and Quarter pixel in standard mode(May be all together?) produce better quality than going for the slowest/slow mode in 2-pass fashion?[High birate encoding=>1100]


Yes, without a doubt if you're using higher resolutions as well (>512 horizontal), but I would turn off GMC based on what I see is a mild general consensus around here. Apparently it wastes bits the majority of the time. I haven't really done any of my own testing on it. Anyway, if your goal is a quick encode, definitely turn it off.

[allynm]

Originally posted by Tuning
May I ask one question:If you are only playing on Computer and need a quick encoding,will enabling Qpel,GMC,and Quarter pixel in standard mode(May be all together?) produce better quality than going for the slowest/slow mode in 2-pass fashion?[High birate encoding=>1100]

Thanks

-Tuning

does 5.1 / 5.1.1 beta even let you set gmc/qpel? if so, how?

[Tuning]

@All
ill enabling Qpel,GMC,and Quarter pixel in standard mode(May be all together?)
First of all sorry for this mistake,please read as Qpel,GMC and BiDirectional Encoding......

@FreakNGoat
Thanks for the info,I will do my own tests.

@allynm
You can enable/disable these features in divx config => Select profile Wizard => disable profiles =>Click Next.

:)-Tuning

[DigitAl56K]

Because various people have been reporting some bizarre timings relating to what options are enabled (some say turning off b-frames increases encoding times, likewise with psycho-vis), I did some lengthy tests this morning with the big Matrix Reloaded fight scene (I never want to see another Agent Smith so long as I live..).

I testest Standard, Slow and Slowest modes with:
1) No PV or Bi-directionsal encoding
2) PV but no Bi-directional encoding
3) Bi-directional encoding but no PV
4) PV and Bi-directional encoding
The test clip was 8421 frames (5:36:840) long. Feedback window was disabled.

Taking the no PV/no bi-directional encoding time as 100% duration, results were (averaging all pq mode results):

1) No PV or Bi-directionsal encoding 100%
2) PV but no Bi-directional encoding ~105%
3) Bi-directional encoding but no PV ~138%
4) PV and Bi-directional encoding ~141%

Which is what I had expected. Also, in case anyone was wondering there were no inconsistencies for any specific pq mode. The order in terms of ascending duration was always 1,2,3,4.

-Al

[DigitAl56K]

Is anyone still alive? ;)

[SeeMoreDigital]

Originally posted by DigitAl56K
Is anyone still alive? ;) Well I'm still alive!

I can't speak for everybody else but I bet some of them have been testing VP6 like me.

I've also been nagging nagui at Sigma for some up to date Xcard drivers!

Cheers

[colordog]

I've done a ton of encodes now with beta 5.1.1 - very happy with it, except that it's not 100 times faster. Still, I'll take the 50%.

Wondering if there's been any word on the non-beta release?

One thing I'd like to see is a way to pause the encoding while being able to disable the feedback window. I think it was Gej that said that to be faster, the feedback window had to be disabled, but then you lose the "pause" button.

[SeeMoreDigital]

Hi colordog,

I been testing the DivX beta with MPEGmediator. And never ever have the feedback window enabled!

And with MPEGmediator you also get a pause key!

Cheers

[Fr4nz]

Hi guys. I encounter a problem when I use XMPEG to encode in Divx: when I want to encode a dvd using 2 or 3 passes, selecting the IFO or the first vob of the film, xmpeg does the first pass correctly, then crashes but the encoding keeps going until it finishes correctly the n-passes.

So it's not a serious problem but it should be better to correct it.

The problem doesn't appear when I select and encode a small part of a vob (for example if I want to encode 2 minutes of a film).

Anyone else had this problem?

[Soulhunter]

Finally I found the time to install DivX 5.1.1 !!!


Was not able to install it before...

Because I did a very loooong enc. session with Matrix Reloaded.... ;)

It was pain and fun in one... No matter !

DivX 5.1 was not working for me, but 5.1.1 works now... :)


Here are the results of some test encodes I did !

Just to give something to discuss... ;)


Source:

Gladiator Trailer (PAL DVD) / 1:04 min. / 25 fps / 640x480


Results:

Quantizer2 / Bidi Enc. / Fast Psy. / Slowest
Time: 24:55 min. / Size: 37.871.616 Bytes / Avg. PSNR: 43.53

Quantizer2 / Bidi Enc. / Fast Psy. / Slow
Time: 20:21 min. / Size: 38.242.304 Bytes / Avg. PSNR: 43.51

Quantizer2 / Bidi Enc. / Fast Psy. / Standard
Time: 05:10 min. / Size: 38.064.128 Bytes / Avg. PSNR: 43.35


Quantizer2 / Bidi Enc. / Slow Psy. / Slowest
Time: 23:45 min. / Size: 90.476.544 Bytes / Avg. PSNR: 41.88

Quantizer2 / Bidi Enc. / Slow Psy. / Slow
Time: 21:55 min. / Size: 90.552.320 Bytes / Avg. PSNR: 41.91

Quantizer2 / Bidi Enc. / Slow Psy. / Standard
Time: 06:30 min. / Size: 46.712.832 Bytes / Avg. PSNR: 42.19


Note:

When Fast Psy. mode was used, DivX seems to use more B-Frames...

Think that causes a lower file size and also a higher PSNR result !!!

Bye

[LordIntruder]

Well thanks for the encodes but I deduce you made an error when reporting the filesize for the last test because if this one is correct, I think we are all gonna use those settings! :D Half less in filesize and even better PSNR result. :p

I also did a full movie encode and I really really REALLY appreciate the increased speed gain. Thanks for the work. :)

[Soulhunter]

Well thanks for the encodes but I deduce you made an error when reporting the filesize for the last test because if this one is correct, I think we are all gonna use those settings! Half less in filesize and even better PSNR result.
Hmmm... No, I'm 90% sure I did not an error !!! :D

But to be sure, Ive redone this test !!! ;)


The only change now is in Time... :rolleyes:

Its now 08:00 min. !!! :p

Because I did the encode while writing this here !!! :D

Don't know what goes wrong here... :(

Maybe its source related !?! :confused:

Or the last test was right, but 4&5 was wrong... ???

Ill do a second test with a other source !!! ;)

Bye

[Soulhunter]

Hi again !!!

As I said, here is a second test with an other source... ;)

Just to be sure, and see what was wrong with my first tests !!!


Source:

Matrix - Lobby chapter (PAL DVD) / 1:00 min. (last 1500 frames) / 25 fps / 640x256


Results:

Quantizer2 / Bidi Enc. / Fast Psy. / Slowest
Time: 11:38 min. / Size: 23.070.720 Bytes / Avg. PSNR: 44.18

Quantizer2 / Bidi Enc. / Fast Psy. / Standard
Time: 02:57 min. / Size: 20.875.264 Bytes / Avg. PSNR: 43.75


Quantizer2 / Bidi Enc. / Slow Psy. / Slowest
Time: 12:14 min. / Size: 29.556.736 Bytes / Avg. PSNR: 43.36

Quantizer2 / Bidi Enc. / Slow Psy. / Standard
Time: 03:32 min. / Size: 23.470.080 Bytes / Avg. PSNR: 43.19


This results looking a bit more logical... ;)


Bye

[LordIntruder]

Hi Soulhunter.

Thanks to redone some tests. :) Sorry I didn't got first you were working at constant quantizer 2 and the difference must come from here. I only work with Nth passes and in this area there are small differences between Standard, Slow and Slowest.

However I'm very surprised to see such a difference. On your first test you got 90Mb in Slowest and 45Mb in Standard for the exact same encode with Slow Psy. And same with the next test, there is a 5Mb difference in size between Standard and Slowest with Slow Psy.

Did anyone else find such differences with constant quantizer? I wasn't aware of this.

[Soulhunter]

Hi Soulhunter.

Thanks to redone some tests. Sorry I didn't got first you were working at constant quantizer 2 and the difference must come from here. I only work with Nth passes and in this area there are small differences between Standard, Slow and Slowest.
No problemo... !!! ;)

I did the Quantizer2 tests just because I do mainly 3 or 4 CD and 1/2 DVD-R encodes...

Yeah, I know I'm a HQ Freak !!! :D

So, for this size I aim for, Quantizer2 is mainly used... :p

This way I could check DivX 5.1.1 ability's for hi-bitrate encodes !!!


But MultiPass encodes are in progress... ;)

Just wait a little bit more !!!

Think they are done in 2-3 hours... :rolleyes:

[EDIT: Was a good guess... Finally 02:07h !!! :D]

Bye

[Soulhunter]

Today is Sunday... Sunday is testing day !!! :D

So, here are my MultiPass results... ;)


Source:

Chaser Trailer - MPEG1 (CG) / 1:19 min / 25fps / 384x288


Results:


MultiPass 250 kbps / No Psy. / Bidi Enc. / Standard

Pass1:
Time: 03:02 min. / Size: 14.090.240 Bytes / Avg. PSNR: 40.24

Pass2:
Time: 02:52 min. / Size: 16.564.224 Bytes / Avg. PSNR: 40.46

Pass3:
Time: 02:51 min. / Size: 16.564.224 Bytes / Avg. PSNR: 40.55



MultiPass 250 kbps / Fast Psy. / Bidi Enc. / Standard

Pass1:
Time: 03:02 min. / Size: 14.090.240 Bytes / Avg. PSNR: 40.24

Pass2:
Time: 03:28 min. / Size: 16.566.272 Bytes / Avg. PSNR: 40.31

Pass3:
Time: 03:26 min. / Size: 16.562.176 Bytes / Avg. PSNR: 40.22



MultiPass 250 kbps / Fast Psy. / Bidi Enc. / Slowest

Pass1:
Time: 15:22 min. / Size: 14.090.240 Bytes / Avg. PSNR: 40.54

Pass2:
Time: 15:29 min. / Size: 16.562.176 Bytes / Avg. PSNR: 42.02

Pass3:
Time: 15:43 min. / Size: 16.560.128 Bytes / Avg. PSNR: 41.17



MultiPass 250 kbps / Slow Psy. / Bidi Enc. / Standard

Pass1:
Time: 03:20 min. / Size: 14.090.240 Bytes / Avg. PSNR: 40.24

Pass2:
Time: 04:18 min. / Size: 16.568.320 Bytes / Avg. PSNR: 40.19

Pass3:
Time: 04:28 min. / Size: 16.560.128 Bytes / Avg. PSNR: 40.10



MultiPass 250 kbps / Slow Psy. / Bidi Enc. / Slowest

Pass1:
Time: 15:32 min. / Size: 14.090.240 Bytes / Avg. PSNR: 40.53

Pass2:
Time: 15:50 min. / Size: 16.562.176 Bytes / Avg. PSNR: 41.93

Pass3:
Time: 16:51 min. / Size: 16.560.128 Bytes / Avg. PSNR: 41.09


Thats all so far !!!

But more will follow...

Next Sunday is not far away ! ;)

Bye

[jos]

Hello,

It's my first post here. I hope I won't say anything particularly stupid.

I've done some compressibility tests with DivX 5.1.1 beta 1 on American Beauty at 800 kbps. The focus of these tests was on dependency between psychovisual enhancement modes and compressibility. The results are quite interesting for me:

PVE off: 71%
PVE fast: 61%
PVE slow: 51%

(HomeTheater profile, B-frames on.)

I used to think previously that PVE increases compressibility at the expense of some invisible details, giving more bits to the more visible parts of the movie. So now I have a question: does it still make any sense to use PVE regardless of the above results? What so good it does to balance the negative impact on bitrate?

Another topic -- I've read some discussions on MV-file usage. In one of my tests I found that when using GMC together with MV-file I get a choppy playback in some scenes.

Thank you all for very interesting posts :-)

J.

[Gej]

For theses that do not read the Subject !: DivX Pro 5.1.1 Beta 2 is now available


Hello folks,

This is the second and (hopefully) last Beta for DivX Pro 5.1.1. In this version you’ll find again some speed optimization and some bugs fixes.
Please enjoy…

http://labs.divx.com/archives/000021.html

What's new in DivX pro 5.1.1 Beta 2
-----------------------------------

Encoder fixes:
- Slow mode 5-10% faster than 5.1.1 Beta 1
- Slowest mode 5-10% faster than 5.1.1 Beta 1
- Issues with Psychovisual and DirectShow Encoder
- Resize issues with DirectShow Encoder
- Fix issues with capture applications

Decoder fixes:
- Issue with n B frames and Deblocking fixed

Changes:
- Psychovisual Disabled by default
- Feedback mode Disabled by default

NOTE: How we measure speed of the Codec ?
We want to measure the speed of the encoding, not of the Encoding application , of a MPEG2 decoder or some preprocessing tools. So we create some tests files that are already pre processed and resized and that are encoded using HuffYUV (and predict left option) and we use theses to actually measure the Codec speed using VirtualDub. This reduces the influence of the actual reading of the source file. People with SMP system or HyperThreading CPU can also create a simple AVS script that will just load and color converts the HuffYUV file before sending it to the codec, allowing the AVS script to run on a separate CPU/Thread than the codec itself.

If you already own DivX Pro, a DivX Beta release will install over your existing DivX Pro version unless you install it to a different directory. (You may want to keep a copy of your officially released DivX Pro codec in case Beta Prototype releases on a 15-day trial, making it easy for everyone to help improve the codec. If we push out yet another beta release after your 15 day trial is up , we will create an additional 15-day trial so you can continue using the prototype version.

The DivX Pro codec released on DivX Labs should be relatively stable. Remember, however, that these releases have a Beta or prototype release status so, uh….you could possibly run into a few bugs from time to time.

[colordog]

Howdy -

I just installed DivX 5.1.1 beta 2 over beta 1.

Under the "Options" tab, the DivX 5 "First Pass", "Nth Pass", etc., buttons no longer open up the codec settings window.

Not sure if this is a complete DivX bug, or if it's related to the version change. I do know for sure that I could bring them up in beta 1 though.

Um... since I'm an idiot, how can I access the codec settings now, since my GKnot buttons don't work?

[colordog]

Also -

Uninstalling beta 2 and reinstalling beta 1 restores functionality to the buttons.

I think beta 1 will expire in a day or two though.

[Angelus]

I just tried installing the new beta 2 codec but Gordian Knot is unable to see it. I did hte "bgregister.exe" after I installed it but that didn't help at all either. I re-installed beta 1 over it and beta 1 works normally. Any ideas?

[colordog]

@Angelus

I encountered the exact same thing.

I posted a question about it in the GKnot developer's forum, but no bites yet.

[Druizk]

@Angelus & @Gej

These error in DivX Pro 5.1.1 beta 2, no register en drivers of Windows the fourCC divx.

I installed manually in register of Windows with "regedit" utility.
In the key:

HKEY_LOCAL_MACHINE\SOFTWARE\Microsoft\Windows NT\CurrentVersion\Drivers32\

append New Value AlfaNumber with key "vidc.divx" whitout quotes and Value: "divx.dll" again whitout quotes....

Sorry my bad english.

[Gej]

Installer is back online

http://labs.divx.com/archives/000021.html

Sorry again for the delay

Gej

[Soulhunter]

Hi again...

Ive just finished my first "full" encode with DivX 5.1.1 !

It was again Matrix Reloaded... :cool:

IIRC, Its the 7th encode now !!!


My settings:

Res. 960x384 / @ 25fps

1st Pass: Standard / 2nd Pass: Standard

Bitrate: 2592 kbps

Bidi. Encoding = on / No Psy Enh.


Ive done exactly the same encode with DivX 5.05 some days ago... :rolleyes:

Now I have to say that the 5.05 encode looks better than the 5.1.1 one !

With 5.1.1, fine details are blurry and flat areas are smoothed...

Its not much difference, but its visible !!!

Sure, this was with beta 1...

And before you ask, PP was not enabled... ;)

Has someone else the same impression ???


Bye

[SeeMoreDigital]

Hi Soulhunter,

Personally I don't think it so strange that you've found that your DivX 5.0.5 encodes look better than your 5.1.1(b1) encodes, especially without post-processing. As I am of the opinion that DivX is relying more heavily on post-processing techniques to make encodes with it's newer codecs look better!

May I ask why you used an image pixel frame size of 960x384. As this equates to an aspect ratio of 2.50:1! 912x384 would be nearer to 2.35:1.

However, if you want encode the same quantity of vertical image pixels as your 2.35:1 PAL DVD source, you could also try 1024x432!

Shame about DivX 5.1.1(b2). Looking forward to tomorrow!

Cheers

[Soulhunter]

May I ask why you used an image pixel frame size of 960x384. As this equates to an aspect ratio of 2.50:1! 912x384 would be nearer to 2.35:1.
Uhm... Has it not a AR of 2.5:1 when you crop the black boarders ???

Think the original AR is 2.35:1, but with small boarders... !?! :confused:

Thought this just because the 16:9 AR with cropped boarders results in a 2.5:1 AR... !!!

Hmm... Now I'm really confused !!! :p

As I am of the opinion that DivX is relying more heavily on post-processing techniques to make encodes with it's newer codecs look better!
I remember that the DivX 5.05 encode had a avg. quant. of 2.1 !!!

So I think there should not be much need for PP... ;)

Or is DivX 5.1.1's PP more than de-blocking ???


PS: I'm also looking forward... ;)


Bye

[SeeMoreDigital]

Originally posted by Soulhunter
Uhm... Has it not a AR of 2.5:1 when you crop the black boarders ???

Think the original AR is 2.35:1, but with small boarders... !?! :confused:

Thought this just because the 16:9 AR with cropped boarders results in a 2.5:1 AR... !!!

Hmm... Now I'm really confused !!! :p Don't be too confused. I take it you just wanted to encode the image only and not the matte (black bars)!

Here's some 'Info about... Pixel Ratios and Pixel Totals' you may find useful when encoding from PAL sources
PAL DVD Player & DV Camcorders
Frame Aspect Ratio = 1.25:1 'Anamorphic Frame'
Image To Nearest To Nearest Matte Image Matte %
AR 16th Pixel Pixel Pixel Qty Pixel Qty Of Frame
1.33:1 720 x 576 720 x 576 000,000 414,720 00.00
1.77:1 720 x 576 720 x 576 000,000 414,720 00.00
1.85:1 720 x 560 720 x 554 15,840 398,880 4.04
2.35:1 720 x 432 720 x 436 100,800 313,920 32.17
2.40:1 720 x 432 720 x 427 107,280 307,440 34.98
--------------------------------------------------------------------------------
PAL Standard (4:3) TV/PC Monitor
Frame Aspect Ratio = 1.33:1 'True Frame'
Image To Nearest To Nearest Matte Image Matte %
AR 16th Pixel Pixel Pixel Qty Pixel Qty Of Frame
1.33:1 768 x 576 768 x 576 000,000 442,368 00.00
1.77:1 768 x 432 768 x 432 110,592 331,776 33.33
1.85:1 768 x 416 768 x 415 123,648 318,720 38.79
2.35:1 768 x 320 768 x 327 191,323 251,136 76.14
2.40:1 768 x 320 768 x 320 196,608 245,760 80.00
--------------------------------------------------------------------------------
PAL Widescreen (16:9) TV/PC Monitor - 'Not a Recognized Standard'
Frame Aspect Ratio = 1.77:1 'True Frame'
Image To Nearest To Nearest Matte Image Matte %
AR 16th Pixel Pixel Pixel Qty Pixel Qty Of Frame
1.77:1 1024 x 576 1024 x 576 000,000 589,824 00.00
1.85:1 1024 x 560 1024 x 554 22,528 567,296 4.04
2.35:1 1024 x 432 1024 x 436 143,360 446,464 32.17
2.40:1 1024 x 432 1024 x 427 152,576 437,248 34.98

Originally posted by Soulhunter
I remember that the DivX 5.05 encode had a avg. quant. of 2.1 !!!

So I think there should not be much need for PP...

Or is DivX 5.1.1's PP more than de-blocking ???I too think that it should be the case that PP should not be required when encoding at a bitrate of 2590kbps. Personally I would not bother using any of the DivX tools such as BiDi, GMC, PSY etc, at all!

I take it you are using GK to generate your encodes. What resizing tool are you using?

Cheers

EDIT: 2.40:1 aspect ratio's added and 'To Nearest Pixel' instead of ' To Nearest Even Pixel'

[Soulhunter]

Don't be too confused. I take it you just wanted to encode the image only and not the matte (black bars)!
Yes, I always crop the black boarders away... !!! :D
Here's some 'Info about... Pixel Ratios and Pixel Totals' you may find useful when encoding from PAL sources
Nice n' usefully this list... THX for it !!! ;)
I take it you are using GK to generate your encodes. What resizing tool are you using?
I use GK only to calculate bitrate and making my AVS... :p

PS: For resizing I used Lancoz...


Bye

[Angelus]

Gej said last night that there was an error in the new beta DivX installer, and thus took the link for the download offline. He should fix it soon, and that's why the buttons dont work. For more info, look at the Divx Codec section

[CruNcher]

@SoulHunter and SeeMoreDigital
Matrix Reloaded Pal is not 2.35:1 its 2.40:1 ;)

[SeeMoreDigital]

Originally posted by CruNcher
@SoulHunter and SeeMoreDigital
Matrix Reloaded Pal is not 2.35:1 its 2.40:1 ;) Well that's good to know!

So now if Soulhunter wants to keep his original 384 vertical pixel size, he must now use 928 horizontal pixels!

Hey CruNcher,
I've been doing some more tests with VP6 on an 3Ghz P4 PC with a clean install of WinXP O/S.

My VP6 tests so far have just involved CBR encodes and I have to say they look a lot better than my original batch of tests!

I'll get back to you with regard to sending you some of my test encodes!

Cheers

[jggimi]

His comments have now been "merged" from the GK development forum into this thread, just to keep everything in one place for Gej and his team to track.

Since replies are by date/time, they're on the previous page of this thread, and it can look just a little out of sequence there due to comments from two threads that are now in one.

[CruNcher]

Originally posted by SeeMoreDigital
Well that's good to know!

Hey CruNcher,
I've been doing some more tests with VP6 on an 3Ghz P4 PC with a clean install of WinXP O/S.

My VP6 tests so far have just involved CBR encodes and I have to say they look a lot better than my original batch of tests!

I'll get back to you with regard to sending you some of my test encodes!

Cheers

huh hehe i think you confound me with C0mPr355 ;)

[Fr4nz]

XMpeg 5 keeps crashing between passes also with 5.1.1 beta 2 when you encode the ENTIRE dvd. So all is normal during the first pass, then xmpeg crash but the encoding keeps going for all the n-passes. You have to keep the window error if you don't want to terminate the encoding. Very annoying. Plz Gej, can you fix it?

I noticed that this bug doesn't occur if you select to encode only a small portion of the film, but only if you encode it ENTIRELY.

EDIT: I'm sure this is not an xmpeg bug because this doesn't occur with 5.0.5 and with 5.1.

EDIT2: I'm doing further tests...stay tuned!

[colordog]

Just as with beta 1, I'm getting the further speed increase claimed by DivX with beta 2.

Nice.

[Custom_VCD]

When I go to encode the capture in the brand new Divx 5.1.1 beta 2, virtualdudmod gives me error saying

Can't Start Video Compression

The Source Format is invalid
error code -2

It always occurs during the second pass? Thanks in advance

[Angelus]

check the GKnot FAQ.

If VdubMod fails to encode the first pass with "videosourceavi error: the sourceimage format is not acceptable. (error code -2)" you may have to make a manual change in your Windows registry. See TelemachusMH's comment in http://forum.doom9.org/showthread.php?s=&threadid=51906.

Also, what version of GK and VDubMod are you using?

[fedge]

Maybe since this is beta2 it should have its own string.. just a suggestion


Also, I ve had no problms with beta 1(but i never do any capturing)

[Gej]

No problem capturing here, the Wrong video format can come from two things:
- The capture card is sending a color format not recognized by DivX
- You trial/Registration is wrong, the DivX codec in this case throw a "wrong video format" exception to exit the encoding app.

[Fr4nz]

UPDATE ABOUT DIVX BUG WITH XMPEG:

There are no problems if you decrypt the DVD on the HD and then encode it.

Very strange bug.

[Soulhunter]

About my first 5.1.1 "Reloaded" encode...

Wow !!! :eek:

Seems that Ive enabled psy. mode in my first DivX 5.1.1 Reloaded encode, without knowing it... :D

Because Ive done another encode (now with a 928x384 res. ;)) and now (without psy. mode) It looks much better !!!

Another thing...

Can somebody confirm that this new "post DivX 5.05" versions, look better on still scenes ??? I think with 5.1.x flat/uniform background looks more static now... (less texture movement) !!!



PS:

While using this new DivX versions, I have some feature requests...

1.

Could DivX write some sort of "analyze" log-file ???

This would save up all info you can see in the feedback window...

Like the min/max/avg of Bitrate, Quants and PSNR !!!

Maybe the percentage of I,P,B frames and all other info too ?!?


2.

Why is the video preview in the feedback window not resize able ???

Its disturbing when you work with high res. video... :angry:

Because you cant reach the bottom of the feedback window then !!!

A preview resize function like in VDub would be useful... ;)

Like 25%, 50%, 75%, 100%, 125%, 150%, 175%, 200% !!!


Bye

[Fr4nz]

Gej I've found another problems with XMPEG/Divx 5.1.1 combo.

As you can see from this image:

http://users.libero.it/i3ltt/Cazzate%20varie/movie.JPG

there are blocks which appears after a change scene (so they come from previous scenes, before an I-Block occurs).

This problem appears sometimes and without a reason (sometimes encoding is good, sometimes has this "block problem".

With Xmpeg/5.0.5 I've never had this problem.

Gej I'm guessing now if there's an incompatibility between these two programs... :confused: :confused:

Thanks for any answer!

PS: The encoding paramenters I used for the film in that pictures were: 1650kbps, b-frames activated, psy off.

[Torian]

@SeeMoreDigital (6th November 2003 18:46):
"I too think that it should be the case that PP should not be required when encoding at a bitrate of 2590kbps."
Definitely, it deteriorates the image quality at DVD-Quality encodes.

"Personally I would not bother using any of the DivX tools such as BiDi, GMC, PSY etc, at all!"
I disagre. Not using BiDi is a waste of Bitrate, but most importantly you should use PSY as Soulhunter seems to have discovered too. It is completely rewritten (since 5.1) and does not blur the image as it uses to with 5.0.5. If you want to preserve the image quality of a DVD (I do) this is mandatory and you can discern every pore of the actors faces. (I have a sample of a dvb-s capture w/ and w/o Psy which shows a dramatic difference).
As I intend to keep my movies somewhat compatible to standalone-players I do not use GMC.

[SeeMoreDigital]

Hello Torian,

Welcome to the forum

Originally posted by Torian

Ref to post by SeeMoreDigital (6th November 2003 18:46):
"Personally I would not bother using any of the DivX tools such as BiDi, GMC, PSY etc, at all!"

I disagre. Not using BiDi is a waste of Bitrate, but most importantly you should use PSY as Soulhunter seems to have discovered too. It is completely rewritten (since 5.1) and does not blur the image as it uses to with 5.0.5. If you want to preserve the image quality of a DVD (I do) this is mandatory and you can discern every pore of the actors faces. (I have a sample of a dvb-s capture w/ and w/o Psy which shows a dramatic difference).
As I intend to keep my movies somewhat compatible to standalone-players I do not use GMC. There's no doubt using BiDi can make your encodes look better!

I've used BiDi with many low bitrate 2pass encodes and it does make a difference for the better. But I've not seen any difference using BiDi when geneating 2pass encodes above 2000kbps. And this certainly seems to be the case when I view my encodes on my TV via Xcard!

Personally I can't say how well DivX performs when capturing or how well it performs when generating high bitrate CBR encodes, with or without BiDi + PSY!

However, I've dug out some old 600~700kpbs CBR encodes I did with DivX, WMV9, RV9, XviD, etc and compared them agianst the new VP6 codec. And VP6 is astonishing!

Can you confirm whether your are generating 1pass or 2passes encodes when using BiDi + PSY?

Cheers

[vio]

Just did a little test.

DVB source -> DVD2AVI -> Gordian Knot -> Avinsynth -> Virtualdub

704x400, 25fps, 500 frames.

B-frames, no pse, no mv. 1-pass quant 1.

Speed time (s) size (kb)

5.1
slowest 349 16,610
slow 271 16,590
standard 40 18,336
fastest 33 28,148

5.1.1
slowest 186 16,702
slow 176 16,720
standard 30 18,350
fastest 21 29,562

These are the settings I use to temporary backup small clips. Didnt realise the negligable difference in size between slow and slowest.

So half percent increase in size and about half the time to encode, not bad. Visually no difference and obviously (quant 1) great quality.

[Narsus]

Originally posted by Fr4nz
Gej I've found another problems with XMPEG/Divx 5.1.1 combo.


I had same problem with Divx 5.1.1b1.
Divx 5.1.1beta2 fixed this.

[Soulhunter]

@Torian
Not using BiDi is a waste of Bitrate, but most importantly you should use PSY as Soulhunter seems to have discovered too.
Uhm... What ??? :confused:

Ive nothing like this discovered... !!! :p

Please, read my posts again !!! ;)

I wrote that my first 5.1.1 "Reloaded" encode looked worse than the 5.05 one (smeared, vague, blurry) !!! Thought it was without psy.... But seems it was not !!! Because later I did a second encode (same source) without psy. mode (checked this time) that looked much better than the first 5.1.1 one !!! So the only thing I can think of, is that my first 5.1.1 encode (blurry one) was done with psy. !!! Maybe psy. is nice for lower bitrates, like RV9's in-loop filtering, but for high bitrates not...

But for the thing with BiDi, I think its also very useful for high bitrates... Seems to drop the image quality a bit... (Is this because B-Frames using lower Quants. ???) But for my "Reloaded" example, not using BiDi would drop the Avg. Quant. from 2.1 to something like 3 or so... (Not sure, haven't tested this yet ! ;)) No matter... Having every 2nd frame a bit less detailed does not hurt the overall quality when watching the movie !!!


PS:

Can now someone confirm now, that the new versions of DivX (5.1.x) look more static/solid and have less texture movement on plain background than with DivX 5.05 ???


Bye

[Snippit]

Quick question: I started encoding Indiana Jones The Last Crusade using DivX5.1.1b2, GKnot 0.28.6.2 at a rate of about 1600 kbps. GMC and BiDi turned on, PV off, slow setting, 2-pass session on a P4@2GHz. The encoding time appeared a bit ridiculous, so what I'm simply interested in is whether I've used the absolute wrong settings or if this is a normal encoding time using those settings.

Enc time: 9 hours for the first pass and an estimated 12.67 hours for second pass(terminated).

/S

[Angelus]

For most DVD encodes I use a 2-pass encode, using "Standard" speed with GMC and BiDi on the first pass and use "Slow" speed with the same GMC and BiDi on the second pass and it usually takes me a little over 7 hours for the whole encode and the muxing of the audio. So something may be wrong with your system. I have a P4 1.6 Ghz processor with 512mb ram. I've recently upgraded to a 7,200 RPM hard drive with 8 mb cache and i've seen encoding speeds increase 5-6 fps, sometimes even a lot more than that. It also depends on the length of the dvd you are trying to encode...

I think i read somewhere that it wasn't worth using slow on the first pass b/c it didn't really amount to any visible improved quality. And usually for me slow doubles the encoding time so standard speed to me is the best choice.