For beta testers : AQE 0.34.0.2 available.....

[SAPSTAR]

After a long period without anything new.....here it is:

• Slight improvements for animations
• Changes in the core to be compatible with Avisynth 2.5.7 : With this version your output MUST be YV12 if not, then AQE will stop, the next version will take care of it....
  

I didn't have time to test it much...so feel free to test it as much as you can!!! 

[danpos]

@SAPSTAR

Nice to hear news about your toy, mate!

I will give it a try asap (through wine since that I gave up Window$ - right now running Ubuntu GNU/Linux ).

Kept up,

[Marvelix]

Hello an thank you SAPSTAR for the updated AQM. 

I performed some tests using AviSynth 2.5.7 and the problem I had never appeared again.

Now the generated log file seems to be more verbose and I found this new behaviour.

When AQM works in CBR everything is OK.

When AQM works with QMatOp feature enabled and MAX_BITRATE is 8500 I often see the following lines:

Code:

[Mon Jan 29 00:24:16 2007] - AutoQMatEnc Version 0, 34, 0, 2
[Mon Jan 29 00:24:16 2007] - AutoBias : Bias selected : 43
[Mon Jan 29 00:24:16 2007] - Auto QMat Intra Max : 114
[Mon Jan 29 00:24:16 2007] - Auto QMat Inter Max : 114
Use a real OS!
[Mon Jan 29 00:24:16 2007] - Width : 720 ; Height : 576 ; Bitrate Equivalent : 4272499
[Mon Jan 29 00:24:16 2007] - QMatOp feature enabled.
[Mon Jan 29 00:24:16 2007] - Following constraints will be used for QMatOp :
[Mon Jan 29 00:24:16 2007] -   Intra Min : 8 ; Inter Min : 16 ; Intra Max : 114 ; Inter Max : 114 ; DCTune precision : 5
[Mon Jan 29 00:24:16 2007] - Automatic Multithread mode enabled : 6 threads will be used
[Mon Jan 29 00:24:16 2007] - Normal Bitrate mode engaged
[Mon Jan 29 00:24:16 2007] - Pass 1/4 - Bitrate : 5127000
[Mon Jan 29 00:24:17 2007] - Info: 'G:\DIKO_temp\movie0.avs' has audio.
[Mon Jan 29 00:24:17 2007] - G:\DIKO_temp\movie0.avs: 720x576, 25 fps, 173610 frames
[Mon Jan 29 00:24:17 2007] - converting ? -> RGB24
[Mon Jan 29 00:24:17 2007] - Assigned new RGB clip
[Mon Jan 29 00:24:17 2007] - AVS File opened successfully
[Mon Jan 29 00:24:17 2007] - (YUV) AVS File opened successfully
[Mon Jan 29 00:24:17 2007] - Autodetect number of frames mode : 173610 frames detected
[Mon Jan 29 00:24:17 2007] - Chosen multiplier for QMatop : 5
[Mon Jan 29 00:24:17 2007] - QMatOp Dark Limit : 0.160219
[Mon Jan 29 00:53:17 2007] - Normalized matrix
8 8 8 8 9 10 13 30
8 8 8 8 8 9 12 27
8 8 9 9 10 11 21 59
8 8 9 11 14 30 68 98
9 8 10 15 51 90 110 114
10 9 12 47 98 114 114 114
12 13 40 95 114 114 114 114
15 53 103 114 114 114 114 114
16 16 16 16 17 18 20 22
16 16 16 16 16 17 18 20
16 16 17 17 17 18 19 21
16 16 17 18 19 20 22 25
17 16 17 19 22 24 27 32
18 17 18 20 24 30 37 45
19 18 20 22 28 37 50 63
22 20 22 25 33 46 63 79
[Mon Jan 29 00:53:17 2007] - Darkness level : 0.163814
[Mon Jan 29 00:53:17 2007] - Ratio : 16/9
[Mon Jan 29 00:53:17 2007] - Init Buffer : 205080
[Mon Jan 29 00:53:17 2007] - Initial Complexity : 0.090196
[Mon Jan 29 00:53:17 2007] - RCmin : 1001107 ; DarkLim: 0.080109 ; Tolerance : 2248666
HBRC Constraints: 0 : 5127000 : 6800000 | 1 : 8
Warning, QMAT_SHIFT is larger then 21, overflows possible
[mpeg2video @ 10284008]rc buffer underflow
....
more than 30 lines
....
[mpeg2video @ 10284008]rc buffer underflow

and this sequence repeats with similar lines and values for each pass executed.

The final result is typically marked by error in size:

Code:

...
[mpeg2video @ 10284008]rc buffer underflow
[mpeg2video @ 10284008]rc buffer underflow
[Mon Jan 29 04:19:46 2007] - The Average FPS for this pass was : 57.000000
[Mon Jan 29 04:19:46 2007] - Done
[Mon Jan 29 04:19:46 2007] - Final Result :
[Mon Jan 29 04:19:46 2007] - Expected size : 4450492350 ; Real Size : 4103840877 ; Error : 7.8 %
[Mon Jan 29 04:19:46 2007] - Total time elapsed : 14130 seconds

Generally, if I put down MAX_BITRATE to 8000 AQM works better and the final result is matching the expected size.

I understand this behaviour is highly dependant from the source, nevertheless I would know if it is normal or it is possible to trim something other than MAX_BITRATE in order to achieve the best result as possible.

Anyway, good work and AQM is a great piece of software! 

[SAPSTAR]

@Marvelix: At high bitrates, I always "problems", so I guess this is not something new...however I may have to tweak few things to improve the rate control! Once again! Anyway thank you for the feed back and Happy Encodes!

[tamahome]

excellent build!
Its have to reconcile me with AQE

respect of final dimensions, of the bitrate  (no exceed ) (to test until 2500kbps, for the very high bitrate I do not think that it is possible to do better than the CCE which is to optimize for those)

Continue in this way

[M1ckran]

@SAPSTAR

Thanks for the update. Long time no hear from me!

I did two backups of The Matrix Reloaded (again) using the latest versions of HCEnc and AQE with AutoQMatOp, both through DVD-RB. The HCEnc version was superb and I could happily watch it through to the end.  But when I played the AQE version, I was struck by the far better picture!  I'm no expert on such things, but I think the contrast setting and the level of detail are far superior with AQE. This results in a picture that is more realistic, with a better 3D effect and more "distance" and "depth" from the foreground to the background.

Sadly, I found a problem. One, I think, that you already know about but I hope my description might help you to fix it.

During the CGI titles of The Matrix Reloaded (AQE version), the picture starts jerking very heavilly. Using the readout on my Pioneer DVD player, I noticed that the bitrate of this sequence varied from almost nothing to a peak of 11Mbps. It was the peaks above 10Mbps on the AQE version that triggered the jerkiness.  In comparison, the HCEnc version peaked at 9Mbps and had no trouble.  From memory, I think the original DVD peaked lower than HCEnc but I no longer remember how low it really was.

I hope the above helps you to isolate the problem. As I said before, I could happily wath the HCEnc version, but I really would like to use AQE for everything.

Somebody mentioned a MAX_BITRATE setting but I'm not sure what it is or how I would use it with DVD-RB.

Regards,
Michael

PS - Fix the bitrate problem and I will be very tempted to increase my previous meagre donation!! 

[Darksoul71]

PS - Fix the bitrate problem and I will be very tempted to increase my previous meagre donation!!

Hm, AFAIK the max bitrate problem is inherent to encoding library which AQE uses. QuEnc and FreeEnc have pretty much the same problems from time to time 

[artur.aragao]

AQE is dead?

[vmesquita]

Unfortunatelly we haven't heard from SAPSTAR in a long while