History of the versions

[latsilva]

AQE 0.33.0.5 - 25/05/2006

• Tweaked  the high bitrate control
• Tweaked the rate control for the embedded Pulldown application
• Improved quality at high bitrates.(Macroblock at scene changes bug)
• Worked on the Rate Control without QMatOp, a new function was implemented in order to work as if QMatOp is on.
• Corrections of minor bugs.

This release is not the one I wanted to release (I started implementing the audio encode), but some people are waiting for some included corrections and I dont have so much time to spend on the audio encoding......So I chose to release this one. But expect .6 soon, once I have implemented the audio encoding.

I will also need some tests on the high bitrates, I tweaked it in order to improve the quality, but I wonder if some spike could'nt reappear.

@Danpos: I worked on your "Lost" episode. It should be OK now. (I didn't retest it since some changes I made) but it should be OK.

FYI : I didn't update the Vmesquita mirror yet. I will do it later.

AQE 0.33.0.4 - 26/04/2006

Taaaaadaaaaa 

Here are the features/changes....many many tests to perform   

• Tuned QMatOp on very hard to encode scenes. (Ex: Pitch Black "crash down" scene)
•   Reviewed the Rate Controls for Normal and High Bitrates. (spikes hunt )
•   Added a pulldown function (very experimental) :

   
• pulldown_detect=#    (1: 23.976->29.97, 2:24.000->29.97, 3:25.000->29.97)
   
• It's applied once the movie is encoded, so for a PAL->NTSC conversion, select a PAL encode, then pulldown_detect=3

[li]  AutoBias : If vbr_bias>=0 , automatic else for a manual bias just set it as a negative value.[/li]
[li]  Bias : Behavior brought back to the CCE standard => 1 means maximum VBR, 100 means CBR.[/li]
[li]  Added a special mode correction when in high bitrates if the movie is oversized, a new faster method is called to correct it.[/li]
[li]  Added some timed log functions. Now everything is timestamped in the log file.[/li]
[li]  OPV Mode : It should simulate a little bit more the CCE behaviour :[/li]

   
• Range is now 0-300 like CCE
   
• The curve is a 1/x^2 kind of function.....a bit like CCE

[/list]

I wish you good luck on this one  Many things were added so many risks....but I want to release soon an official 0.33b to allow jdobbs to release the next DVDRB with AQE support soon....

@Danpos: Still thinking about MPEG2 CBR..... But I know you'll like this one my friend!!

AQE 0.33.0.3 - 12/04/2006

As promised....a new beta version for you to test 

• Fine tuned QMatOp for Ultra Low Bitrates
• Totally rewrote the rate control for high bitrates. (3 passes highly recommended for a good sizing)

• Constraints are respected now : No more spikes above the Max bitrate
• Sizing as good as "normal" bitrates with 3 passes (1 Vaf + 2 VBR)

[li]Adjusted the rate control for MPEG2 CBR, yet to be improved and tuned.[/li]
[/list]

[EDIT] Pay a close attention to your VBR Bias values....in this new rate control, it takes a greater importance to choose it well. I mean too small and the sizing maybe achieved faster but the quality could suffer, too big and it will be hard to achiveve a good sizing. My recommended values are 20-25........The default value in DIKO is 1, which is good for sizng but not for quality (it means the variation will be close close from nul, so close to be CBR).

Good tests....let me know your feedback...I will now start improving OPV....rate control and quality targeting....depending on your feedback of course
I may also have time to start working on other features....like audio encoding....

AQE 0.33.0.2 - 15/03/2006

Another version....I'm sorry I was really busy with my job...For the last weeks, so I didn't have much time to code, but I found few free hours :

New features/corrections :
# Tweaked QMatOp (A bit slower but better) : The default speed is a bit lower
# Tweaked the rate control...a bit better at high bitrates
# MPEG2 CBR implemented....experimental
# FPS displayed (Requested by Fox Ace)
# Log cleaning

When I'll have real free time, I'll work on implementing the audio encoding + improving the rate control at very high bitrates...I'm not yet satisfied.

AQE 0.33.0.1 - 21/02/2006

Back ....

As promised another version of AQE, for those using OPV, I urge you to use this one, the previous version had a major flaw.

Here are the changes :
# Corrected OPV, it was not working properly when Q=1 (Thank you Danpos for finding it)
# Improved OPV quality, should be better but slower!!!!
# Added new parameters for QMatOp tweakings (Requested by Archeo)

    * qmat_min_intra=### , Minimum value for intra values (default 8 )
    * qmat_max_intra=###, Maximum value for intra values (default 255)
    * qmat_min_inter=###, Minimum value for inter values (default 8 )
    * qmat_max_inter=###, Maximum value for inter values (default 255)
    * dctune_prec=##, Precision percentage (integer value from 1-100)  for DCTune (default 5)

Good tests and good encodes...

To Fox Ace : Desole je n'ai pas eu le temps de rajouter le FPS, mais le bug OPV etait trop important pour que j'attende plus longtemps.

[latsilva]

AQE 0.32.0.8 - 22/01/2006

Taaaadaaa a new version to be tested.....
If this version had no major bug, it will be the official 0.32b.

Here are the new features :
# Corrected the Multipass RC, Sizing at high bitrates should be improved.
# Added the MPEG1 encoding feature :

    * MPEG1 = 1 means MPEG1 will be encoded
    * cbr_rate : Bitrate for MPEG1 (which is always CBR)

# Chapters insertions (multipass only)

    * Same than CCE SP : chapter= ## , ## being the frame number. Only constraint for now, they must be ordered.
Example:
          o chapter=0
          o chapter=15000
          o chapter=30000


Good encodes and tests.....

AQE 0.32.0.7 - 09/01/2006

Yeeah !! Me again ! 

Here is a new release, OPV lovers, I will need your feedback on this one...the OPV RC was modified in this one...I want to know if I'm on the right way or not.

# Corrected a bug in the display of the size of the very big files (Multipass)
# Corrected a bug brought by the MTT QMatOp, the encoder was unstable.(Random deaths)
# Improved the memory consumption of the MTT QMatOp.
# Corrected the RC for OPV.

I urge everybody using the .5 or .6 to install this one. A very strange unstability introduced by the MTT QMatOp was found . It should be corrected by now. 

Good encodes and give me your feedbacks.....

[EDIT] I tested it with DIKO/DVDRB. It worked fine for both Multipass and OPV encodes.

AQE 0.32.0.6 - 09/01/2006

Hi,

As promised the quick correction is now available...you can forget the 0.32.0.5, it was really unstable....a stupid error in my mutexes prevented some part of code to be stable.
So only change for this release :
  # Corrected and Optimized the Multithreaded QMatOp
  # Smaller displays

I ran 3 tests this weekend....2 with DVDRB and 1 with DIKO, all were fine. They were multipass encodes. I will now concentrate on OPV and DIKO integration.

AQE
0.32.0.5 - 06/01/2006

Taaadaaa as promised, here is the first beta released for the new year :

    * Auto Multithreading : if threads=1 or not mentioned => Automatic, if threads = # manual number of threads, if threads = 0 Multithreading disabled

    * Mulithreaded QMatOp phase (highly experimental) (Windows threads for this part)
    * Smaller logs, only really pertinent info is kept now.
    * Corrections in the Matrix linearization functions

Now I will start working hard on the DIKO integration and the OPV mode, of course if no major issue is found in the new functionalities.

Good encodes....

AQE 0.32.0.4 - 21/12/2005

Taaaaadaaaa 

# Total time info displayed and added in the log
# Bug Corrected : Threads were not closed after each pass (their number was increasing)
# HTT/Multi Core  detection added => A warning is displayed if threads = 1 on this type of architecture

OK This will be my last release before christmas !!

@all : Enjoy your christmas time. May Santa Claus be generous with you and your family...Best wishes.

Take care !

[EDIT] I think I may use the threads parameter differently in the next version : threads = 0 would mean no MT, = 1 would mean automatic mode, = ## would mean manual value

AQE 0.32.0.3 - 19/12/2005

As promised here is a new version :
# Display version info in the log
# Create New VAF parameter taken into account now
# Multithread enabled version (Highly experimental) - Posix Threads used.

    * Parameter : threads=##   ,   ## Number of threads.

This multithread version is highly experimental, I couldn't test it on my new test PC, as it is not finished. So I dont know what will be the gain on an HTT architecture....Test it as much as you can and let me know...The default value is 1 so multithreading not enabled....
[EDIT] For now the only part which is multithread is the encoding part (pass 1 + VBR passes, OPV), but the QMatOp phase is not yet multithreaded...I have to think about a way to make it multithreaded.

AQE 0.32.0.2 - 30/11/2005

Hi there....I'm still alive ! 
Here is a new version, the RC should be improved again, it was close from perfect, but I still had some rare oversize, for example with an old french movie (UN COUP FUMANT), with the 0.32.0.1 I finished at 4.43GB, with this one I finished at 4.26G, a bit undersized but nothing major....

I still have an idea I want to implement in the RC and the Multipass RC....maybe next version No improvement in this version for OPV....no time to work on it...yet. But the test PC is close to be finished, I bought all parts except the monitor...I already received the motherboard.

Remember each time you report a pb, please share your log file(s) with rapidshare...thx.

AQE 0.32.0.1 - 22/11/2005

Back....I have a new version.....I think this time I got it (for the Multipass). If everything is fine with it (Quality/Bugs/Speed), then I may find time to work on OPV....but I still dont know how to test it as I dont want to use it for my own encodes.

You will see the new QMatOp : Faster and Better, yep I'm now using the full capabilities of the MPEG2 specs.....Some inter matrices may surprise you...but the visual quality is up to now very very good !!!

AQE 0.31.1.0/0.31.1.1 - 07/11/2005

Hi there,

A major beta release, forget the two last versions (.8 & .9), I found a major flaw in the Ultra Low engine Threshold...it was always engaged. Thats why some people experienced some blinking squares...like Gino.

# Corrected a major bug : The Ultra Low Mode was always engaged !!!! (since 0.31.0.Cool
# Added a new VAF analyzer to assess more accurately the Qmin/Qmax after the first pass.
# Added a min bitrate correction function
# Spikes should be gone at high bitrates !!!!!

I also worked hard on the spikes pb....I think I'm almost perfect now at high bitrates.

Please take note, that I had no time to test OPV. I need another computer to be able to run tests on OPV in addition to the one running for multipasses.

AQE 0.31.0.9 - 02/11/2005

Hey ! Guess what  Grin Another beta version....pleeeeeaaaaase tell me it will be the official 0.31b Wink
    *  Corrected a bug found by oxa (blinking squares due to a very low quantization)
    * Added the remaining time info for OPV
    * Corrected slightly the RC in VBR/OPV in order to reduce the number of spikes

I already encoded : batman begins, Starwars 3 and Land of the dead successfully with it and DVDRB...the visual results were very good.(Especially starwars 3, I chose to keep everything, the compression ratio was bigger than 50%)
Let me know your own results....

Something else, I have a hard time to test both OPV and VBR because it takes me too much time to run both...so I mainly test VBR (I prefer it for its quality). When I'll have gathered enough donations to buy another computer I will dedicate this new computer to run tests........
So far, my donations are : 0 $  Cheesy So a long way before the new computer