B-frames?

[ed_co]

I don't understand, the reason that autogk re-running first pass and switch B-frames off.
Had the first pass with 98.20 of quality an now enconding in 73.46.
Any1 helps me?
Thank you.

[14/08/2006 16:11:17] Running compressibility test.
[14/08/2006 16:17:28] Duration was: 6 minutes 10 seconds
[14/08/2006 16:17:28] Speed was: 30,47 fps.
[14/08/2006 16:17:28] Compressibility percentage is: 96,31
[14/08/2006 16:17:28] Chosen resolution is: 640x272 ( AR: 2,35 )
[14/08/2006 16:17:28] Predicted comptest value is: 96,31%
[14/08/2006 16:17:28] Running first pass.
[14/08/2006 17:29:38] Duration was: 1 hour, 12 minutes 10 seconds
[14/08/2006 17:29:38] Speed was: 52,18 fps.
[14/08/2006 17:29:39] Expected quality of first pass size: 98,20%
[14/08/2006 17:29:39] Trying to adjust settings.
[14/08/2006 17:29:39] Switching b-frames off
[14/08/2006 17:29:39] Chosen resolution is: 640x272 ( AR: 2,35 )
[14/08/2006 17:29:39] Predicted comptest value is: 73,46
[14/08/2006 17:29:39] Re-running first pass.

[BigDid]

I don't understand, the reason that autogk re-running first pass and switch B-frames off.
Had the first pass with 98.20 of quality an now enconding in 73.46...
Hi,

This topic is covered 3 or 4 times a year. I made a title to get it retrieved easily in the future.

Two statements:
1. B frames are of lesser quality than P frames but much more compressible and visually Ok
2. Above 75% in quality is kind of wasted; and above 100% will get undersized.

You put this 2 statements together and you have the explanation for AGK behavior, specially if fixed width is choosen (prefer auto width)
- this first pass is near (or above) 100%; are B frames present?
no -> nothing changed
yes -> redo 1st pass without B frames

Some older threads about it:
http://forum.doom9.org/showthread.php?t=109626&highlight=king+kong
http://forum.doom9.org/showthread.php?t=114003&highlight=king+kong

Did

[ed_co]

Then, we can say, that enconding without B-Frames give more quality? (96%->71%).
I don't have fixed the width (I put max width to 640) and size of video 2gb (harry potter have 140 min and I choose AC3 audio).
Thank you.


Hi,

This topic is covered 3 or 4 times a year. I made a title to get it retrieved easily in the future.

Two statements:
1. B frames are of lesser quality than P frames but much more compressible and visually Ok
2. Above 75% in quality is kind of wasted; and above 100% will get undersized.

You put this 2 statements together and you have the explanation for AGK behavior, specially if fixed width is choosen (prefer auto width)
- this first pass is near (or above) 100%; are B frames present?
no -> nothing changed
yes -> redo 1st pass without B frames

Some older threads about it:
http://forum.doom9.org/showthread.php?t=109626&highlight=king+kong
http://forum.doom9.org/showthread.php?t=114003&highlight=king+kong

Did

[BigDid]

Then, we can say, that enconding without B-Frames give more quality? (96%->71%).
Yes, specially if comptest is above 100% and file would get undersized.
I don't have fixed the width (I put max width to 640) and size of video 2gb (harry potter have 140 min and I choose AC3 audio).
Thank you.
For AGK fixed width or max width is same regarding adjusting settings -> not possible.
This being said if you play on a Standalaone (SAP) with ESS chipset, 640 max is a good choice to stay out of trouble, if MTK or newer no need.

Did

[ed_co]

For AGK fixed width or max width is same regarding adjusting settings -> not possible.
This being said if you play on a Standalaone (SAP) with ESS chipset, 640 max is a good choice to stay out of trouble, if MTK or newer no need.

I'm sorry but I don't understand you very well (my english is very bad).
Not possible what?
I didn't say anything about Standalone, but I play sometimes in a Woxter X-div 600 with a Mediatek chipset (but I don't say anything about this, and I don't understand that you say, sorry :( )

Thank you for your reply and sorry for the inconveniences.

NOTE:
When I said:
I don't have fixed the width (I put max width to 640) and size of video 2gb (harry potter have 140 min and I choose AC3 audio).
I would want to say:
I didn't have fixed the width (always put max width to 640) and size of video 2gb (because harry potter have 140 min and I choose AC3 audio).

[BigDid]

I'm sorry but I don't understand you very well (my english is very bad).
Not possible what?..
Ok, I'll rephrase.

When you use fixed width OR max width 640( in your case) it is not possible for AGK to change the width (upsize) accordingly to the result of the first pass, in relation with the size of the output file and B-frames or not B-frames (:) bad joke)

You can redo your HP encode with
- Same size: 2gb
- auto width

and see what happens... to get a bigger resolution with B-frames and compare visually the 2 results to see which one suits you better, knowing that either of the two will play fine (if encoded with the MTK option) in your SAP.

Did

[ed_co]

Ok, I'll rephrase.

When you use fixed width OR max width 640( in your case) it is not possible for AGK to change the width (upsize) accordingly to the result of the first pass, in relation with the size of the output file and B-frames or not B-frames (:) bad joke)

You can redo your HP encode with
- Same size: 2gb
- auto width

and see what happens... to get a bigger resolution with B-frames and compare visually the 2 results to see which one suits you better, knowing that either of the two will play fine (if encoded with the MTK option) in your SAP.

Did

Thank you very much for your rephrase (as you can see, my english is really bad).
What do you think that is the best options by your experience, 704x3XX with B-frames or 640x272 without B-frames?

I had put this (width) because, I always rip with GK (not autoGK), with divx and always choose a resolution 640 or less width, always with pro capabilities (my Woxter Mediatek chipset supports this), and using the lanczos resize.
I change to autoGK because I don't have any time (work), and I prove them and not same quality, but very good, I verymuch time save. I NEVER NEED TO PUT COMPATIBILITY WITH MTK, really need with xvid and autoGK?

Thank you. ;)

[BigDid]

... What do you think that is the best options by your experience, 704x3XX with B-frames or 640x272 without B-frames?
...
I had put this (width) because, I always rip with GK (not autoGK), with divx and ...
Ok,

1. I forgot you use Divx as I always use Xvid; so as long as the home theatre profile is used for divx you should'nt have SAP compatibility problems.

2. Already answered: and see what happens... to get a bigger resolution with B-frames and compare visually the 2 results to see which one suits you better
We're turning in circles, I advise you to try and test and you ask me what will be the result or YOUR tests...
To get my mood on the subject,see related post: http://forum.doom9.org/showthread.php?p=863219#post863219

No need to re-post until you DID try by yourself, thank you. At that time I will be happy to exchange feedback with you, not before.

Did

[ed_co]

I encode in both methods and the results are both crappy... (because the film have VERY darks scenes... and the behavior of the codec is very bad, with the settings and the autogk).
Can I make any improvement to get more quality, with the same space?
Thank you.

P.D. I think that I'll finish to make this in Divx 3.11 with Nandub (I hope that don't be needed).

[BigDid]

I encode in both methods and the results are both crappy... (because the film have VERY darks scenes... and the behavior of the codec is very bad, with the settings and the autogk).

Could you upload pictures of 1 same frame: 1 original, 1 first encode, 1 2nd encode (P-frame)?
http://forum.doom9.org/showthread.php?t=96362

Can I make any improvement to get more quality, with the same space? Regarding darkness... afraid not, at least not inside AGK.

P.D. I think that I'll finish to make this in Divx 3.11 with Nandub (I hope that don't be needed). Your choice, even if I think it is more a provocation than a real will :D

Did

[kis2005]

I'm not sure with DivX (I use XviD) but for better quality than 640 width, try 720 (as long as it will work for your player). I have gotten great results with 720 fixed resolution with this movie and at 1/3 DVD size with VBR192. So with 2Gb size, it should be good. It's all a matter of taste.

Good luck.

[BigDid]

... (I use XviD) but for better quality than 640 width, try 720 (as long as it will work for your player)... It's all a matter of taste...
Hi,

If you look at the log, you'll see it's xvid, and it's 2 logs in one:
-one at 640 fixed -> no B-frames
-one auto-width -> with B-frames

hence my demand for frames comparatives pics. Your post is interesting because, if you did your encode with AGK and do not have problems with dark scenes, question is: is it a matter of taste...

Did

[kis2005]

Did,

Well it is all a matter of taste, but I find that I'm very picky. Some movies are better than most, but I have found that 720 fixed res. w/ 192VBR settings with appr. 450Mb / 45min, I get DVD or near DVD quality. With very long movies escpeically with dark areas, the quality will be reduced a bit, but I haven't had any come out realy bad except if the source was bad ...and none of the HP movies have been authored poorly. So with 2Gb at 720/192VBF, one should be able to get a very good encode ...even on the 4th one that is very dark.

But as stated, it's all a matter of taste.

[ed_co]

Did,

Well it is all a matter of taste, but I find that I'm very picky. Some movies are better than most, but I have found that 720 fixed res. w/ 192VBR settings with appr. 450Mb / 45min, I get DVD or near DVD quality. With very long movies escpeically with dark areas, the quality will be reduced a bit, but I haven't had any come out realy bad except if the source was bad ...and none of the HP movies have been authored poorly. So with 2Gb at 720/192VBF, one should be able to get a very good encode ...even on the 4th one that is very dark.

But as stated, it's all a matter of taste.

Hi, I encode again, but this time with Gordian Knot (not auto), with Divx 6.2.5 (in insane quality settings, during 8 hours and a half, the encode) and the results are better but, not much.
The source is not too bad (not perfect), but the difference is high with the 3 encodes... (and the differences between the encodes are smaller).

If anyone wants to prove them, the worst part are 1:57:00 to 2:05:00 aprox. in Harry Potter Goblet of Fire.

Your choice, even if I think it is more a provocation than a real will
Divx 3.11 with nandub with propper configurations (and many encodes), made better results than divx5 and xvid during a long time... (in the codec match in doom9.org you can see...).

If I have time today, can be that I prove to encode with the better codec (that doom9 said in the last codec match), the x264. But I don't know if I'll can.

Thank you.

[BigDid]

Hi, I encode again, but this time with Gordian Knot (not auto), with Divx 6.2.5 (in insane quality settings, during 8 hours and a half, the encode) and the results are better but, not much.
Hi,

A little historic. We are in the AGK section, initial post was about re-doing 1st pass and B-frames. The re-doing is explained, the proposal was to compare with and without B-frames. Kis2005 also gave his opinion from his experience: max width with B-frames.
The source is not too bad (not perfect), but the difference is high with the 3 encodes... (and the differences between the encodes are smaller)...
If anyone wants to prove them, the worst part are 1:57:00 to 2:05:00 aprox. in Harry Potter Goblet of Fire.
Nothing to prove here, just exchange, you cannot or don't want to post frames to make a comparison for the benefits of other posters or readers... never mind.
Divx 3.11 with nandub with propper configurations (and many encodes), made better results than divx5 and xvid during a long time... (in the codec match in doom9.org you can see...).
If I have time today, can be that I prove to encode with the better codec (that doom9 said in the last codec match), the x264. But I don't know if I'll can.
Thank you.
Great but outside the scope of AGK.
My approach would be to tweak some settings with AutoGKtweaker or AGKPal IF these dark scenes were really overwhelming; I don't have any element to go that way ATM, so just be it ( or leave it). Bye.

Did

[ed_co]

Nothing to prove here, just exchange, you cannot or don't want to post frames to make a comparison for the benefits of other posters or readers... never mind.

Relax.

I prefer don't share images, because you can't view the images in fullscreen.

I prefer share a portion of avi. (Remember avi 2gb with 1 ac3 and 1 mp3). (Each file size is aprox. 1,2 mb).

Without B-frames.
http://rapidshare.de/files/29647004/01withoutBframes.avi.html

With B-frames.
http://rapidshare.de/files/29647294/02withBframes.avi.html

With GK divx 6.2.5 insane quality. (8 hours and half, with a amd64 3800+...)
http://rapidshare.de/files/29647473/03divx6.2.5insanequality.avi.html



... IF these dark scenes were really overwhelming; I don't have any element to go that way ATM, so just be it ( or leave it). Bye.

Did

Enjoy.

DID.

[BigDid]

Thanks for the effort.

Cannot DL atm, will see (much) later on.

Did

[BigDid]

... I prefer don't share images, because you can't view the images in fullscreen...
Hi,

Well, I can do it (view images fullscreen) with MPC.

I finally had a litle time ( a few minutes) at home to DL the files and have a quick look at it. I concentrated on the P-frame 291, a close up on Harry enclosed by the death statue.

Details:
1. 640w without B-frames AGK/Xvid
Some ringing/haloing around the glasses, seems of lesser quality
2. 704x304(or 720?) with B-frames AGK/Xvid
No ringing, good details, a little sharper but darker than the divx encode
3. 704x288(or 720?) No B-frames GK/Divx
Good details, less sharper but a little brighter than the xvid encode

I will concentrate on the N°2 and 3, n°1 (640 without B-frames) being of lesser visual quality.

Overall:
I made some quick exam with DRFanalyser. The birate is really low: between 520 to 570 kbp/s.
1 question: did you use any degrain pre-processing for the divx encode with GK? (AGK uses removegrain(mode=2) by default).
I also quickly compared with my encode which is a 624w with B-frames for 1480Mb file size with AC3 so less overall bitrate. I made some brightness, denoise, degrain, sharpening pre-processing (with AGK-PAL); results: brighter, more bitrate for that sequence, but a less "pure" picture.
I will try to identify the same sequence to cut it and upload later if interested.

Did

[ed_co]

...2. 704x304(or 720?) with B-frames AGK/Xvid
No ringing, good details, a little sharper but darker than the divx encode

720


Overall:
I made some quick exam with DRFanalyser. The birate is really low: between 520 to 570 kbp/s.

Each file has 2gb size, the average size has 1338 kbp/s, this bitrate low, is for the VBR algorithm and for the dark scenes are very compressible (but very crappy).


1 question: did you use any degrain pre-processing for the divx encode with GK? (AGK uses removegrain(mode=2) by default).

I think that no, but you can view the .log that I have attached below.


I also quickly compared with my encode which is a 624w with B-frames for 1480Mb file size with AC3 so less overall bitrate. I made some brightness, denoise, degrain, sharpening pre-processing (with AGK-PAL); results: brighter, more bitrate for that sequence, but a less "pure" picture.
I will try to identify the same sequence to cut it and upload later if interested.
Did
I don't understand anything, I don't use any special features...
Your encode features are...:
- Program: AGK
- Codec: ????
- Number of audio: 1?
- Audio format: AC3
- B-frames: yes
- Resolution: 624x?
- Any feature to add: ?????

Thank you.

[BigDid]

Each file has 2gb size, the average size has 1338 kbp/s, this bitrate low, is for the VBR algorithm and for the dark scenes are very compressible (but very crappy).
I know that, as soon as I will be able to cut and upload, you'll be able to compare. In the meanwhile 1 pic:
http://img53.imageshack.us/img53/9041/pframe291nd9.th.jpg (http://img53.imageshack.us/my.php?image=pframe291nd9.jpg)
I think that no, but you can view the .log that I have attached below. Ok, so no degrain in GK, this is why there is a little sharpness difference between:
GK -> H263 no degrain
AGK -> Sharp matrix but removegrain=2

- Program: AGK
- Codec: Xvid
- Number of audio: 1
- Audio format:re-encoded to Ac3 320 CBR
- B-frames: yes
- Resolution: 624x256
- Any feature to add: all filtering done via AGK-PAL

Edit: link for video: http://www.badongo.com/vid/185590
Your frame 291 should be +- my frame 280 or 281

Did

[ed_co]

sharpness difference between:
GK -> H263 no degrain
AGK -> Sharp matrix but removegrain=2

Where is this option in both AGK?


- Audio format:re-encoded to Ac3 320 CBR

I know that, here isn't the place to comment this, but, why 320 CBR (I suppose to 5.1), should be better VBR. (I don't have any experience in encode AC3->AC3, so sorry if I make a mistake) What program do you use to encode?


- Any feature to add: all filtering done via AGK-PAL

I don't know who is that... EDIT: I found that is a program (version 0.11), exist any other version? (Will try to use tomorrow...)

Thank you.

[BigDid]

Where is this option in both AGK?
:confused: :cool:
I was comparing the sharpness with GK and the H263 CQM vs AGK and the sharpmatrix -> conclusion the H263 encode should have been softer (= less sharp). It is not because AGK uses internally removegrain in mode=2 and you did not use any degrain in your GK encode.
I prefer (with AGK PAL) to use removegrain( mode=5) or 6 not mode=2 too soft :devil: :cool:

I know that, here isn't the place to comment this, but, why 320 CBR (I suppose to 5.1)
Yes
should be better VBR.
Not possible ATM, not Dolby specs compliant; see aften thread in audio section (already linked) for more infos
(I don't have any experience in encode AC3->AC3, so sorry if I make a mistake) What program do you use to encode?
BeHappy (already linked); input AC3 5.1 -> DRC-> Normalize -> ouput 384(or 320) AC3 5.1 with aften
Edit: Shon3i just made a new Behappy package with aften available see here: http://forum.doom9.org/showpost.php?p=829754&postcount=259

I don't know who is that... EDIT: I found that is a program (version 0.11), exist any other version? (Will try to use tomorrow...)
No new version.

More on the HP Goblet test encode. With DRFanalyser same bitrate distribution between your xvid encode (n°2) and mine. Difference is mine has a lower overall bitrate +-430 vs 520 for yours. Had some difficulties to cut the AVI, I believe it is related to VDMod version, I got success with the 1.5.10.2 version.

Did

[ed_co]

Hi,
First, sorry for the delay. I couldn't post until now.
When I try to install AGK-PAL, an error window appears: Installation is corrupter. Please, reinstall the aplication.
I try to reinstall the aplication, and no change, the window appears again.
I have the autogk 2.32b.
Thank you.

[BigDid]

Hi,
First, sorry for the delay. I couldn't post until now.
When I try to install AGK-PAL, an error window appears: Installation is corrupter. Please, reinstall the aplication.
I try to reinstall the aplication, and no change, the window appears again.
I have the autogk 2.32b.
Thank you.
Hi,

1. AGKPal requires the Microsoft .NET Framework 2.0

2. As far as I recall (i have been using it for a long time now), I didn't install it, just dezipped, run AGK, start the agktweaker from the agkpal folder, etc ...?

Did

[ed_co]

1. AGKPal requires the Microsoft .NET Framework 2.0


I know, and it were installad a long time ago for another aplications.

2. As far as I recall (i have been using it for a long time now), I didn't install it, just dezipped, run AGK, start the agktweaker from the agkpal folder, etc ...?
Did
Yes I know, but if the program fails, corrupt the original .exe and when I reinstall, fails too.
NOW: when I reinstall only reinstall the beta version, and now reinstalling the stable and then the beta, works properly again.
But the program AGK-PAL don't works...
Steps:
-Run AutoGK.
-Run AGK-PAL.
-After configuring a compilation, add Job & Run.
And nothing at all.

Thank you.

[BigDid]

...NOW: when I reinstall only reinstall the beta version, and now reinstalling the stable and then the beta, works properly again.
But the program AGK-PAL don't works...

Hi,

I have other concerns ATM and may be not posting for some time.
It works for me and worked for other users also. So it has to do with your configuration or installs. Suggestions:
-Deinstall all, install 2.27 stable install AGKPAL, make it work and install 2.32 after: Len0x stated he changed installer in the 2.32
- or Post in the AGKPAL thread
- or Create a new thread
- or ...

For the time being, this will be my last post in this matter. Bye.

Did