HCenc 0.20 OPV problem

[Sharc]

HCenc 0.20 OPV failed. Log file is attached (prepare phase).
2-pass VBR worked fine.
No problem with CCE OPV, though.

Can anyone help?

[kumi]

Attachments sometimes take a while to get approved. Perhaps in the meantime you can upload your .zip file to a free hosting service, like zshare.net, mooload.com, or sendspace.com. Or if you have just a text file, paste it to http://p.caboo.se .

[jdobbs]

Sorry about that. I read the contents of the zip file and went on to another post without approving it.

I'm currently running my second disc using HC v0.20 in CQ mode (OPV) as a test and it is working correctly. I'm not sure what's happening in this case -- except that the M2V file isn't getting created.

[Sharc]

Thank you for the response. No hurry though.

Actually HCenc starts to encode after the prepare. But terribly slow (1 .... 2fps), and bitrate is out of control (average close to 6000 kbps), probably as a result of the CQ converging to 0.1 in the prepare phase.
CQ for CCE was around 25, btw.

[jdobbs]

It's getting set to .1 because nothing is being created and the output size is zero... making the routine think the file needs to be bigger (lower Q) -- the problem is in the fact that the prediction passes are failing to create an M2V. That's something I haven't been able to repeat yet.

[Sharc]

Appears as if HCenc has a problem with the avisynth script for creating the analysis sample - SelectEvery(xxx,xxxx)

[Sharc]

I managed to capture the HC GUI during the analysis.

The info line reports:
(1) ERROR,Avisynth message:invoke environment error - pass 1

Perhaps this helps a little.

[jdobbs]

1. What version of AVISYNTH are you using?

2. Any filters?

[Sharc]

1. Avisynth 2.5.6
I uninstalled and re-installed it. No effect.

2. No filters

Edit:
....and here the analysis script:

#------------------
# AVS File Created by DVD Rebuilder
# VOBID:01, CELLID:13
#------------------
LoadPlugin("C:\Programme\Video\DVD Rebuilder\DGDecode.dll")
mpeg2source("C:\xxxxxx\xxxxxxx\TEMP\D2VAVS\V01.D2V")
SelectRangeEvery(600,12)
ConvertToYV12(interlaced=true)

[Rippraff]

I've just tested 3 PAL movies in a row with HCenc 0.20 OPV (prepare phase only). No problems here.

Cu Rippraff

[jdobbs]

I don't see anything wrong there... how big is the source? Maybe the "SelectEvery()" is returning zero??? If my memory serves me correctly, though, I don't think that should be possible.

[Sharc]

Thanks to all for your comments.
Good to know that it works on other machines.

Strange here that CCE with OPV works fine with the same analysis script being produced. Q final value = 37, reduction level about 54%.
When I change to HC the analysis.m2v is created but remains permanently at 0 bytes.
I will try with other discs later today and perhaps reinstall the RB package from scratch.

Edit: I just noted a difference between CCE and HC in the analysis.avs :
- For CCE the line AudioDub(BlankClip()) is included as last line
- For HC this line is missing
Don't know if this is relevant to the problem.

[Rippraff]

Edit: I just noted a difference between CCE and HC in the analysis.avs :
- For CCE the line AudioDub(BlankClip()) is included as last line
- For HC this line is missing
Don't know if this is relevant to the problem.
I'm pretty sure that this isn't relevant to your problem.

In fact I don't know why this is still the standard setting for CCE (probably to stay on the save side). It's a workaround for a combination of an old encoder with an old cpu: CCE 2.5 in connection with AMD Duron or Thunderbird.

Cu Rippraff

[jdobbs]

@Sharc

Try opening the AVS with Microsoft Media Player. It should playback as a group of .5 second clips.

[writersblock29]

HC OVP, testing with R1 The Land of the Dead*, worked okay--but it took something like ten prediction passes before it found the Q factor it wanted to use. Region 1 of Star Wars Episode One* (why did you keep Jar-Jar in there, Mr. Lucas? WHY!?!?!) did pretty similar, needing bunches of prediction passes. Both yielded 4.12-4.20GB file sizes in the final output. It seems to work on my dual-core machine!

*I pre-processed with DVD Shrink, creating movie-only files. I did so because of the forced subtitles in Phantom Menance, knowing (okay, assuming) that those subs would be dropped using Rebuilder's Movie-Only mode. For Land of the Dead, I can only say it was out of force of habit, since this movie has no forced subs.

[Sharc]

@Sharc

Try opening the AVS with Microsoft Media Player. It should playback as a group of .5 second clips.

That's exactly what is does.
I tried both scripts -- produced by CCE and HC.

The problem is that the analysis.m2v is always 0 length for HC.
Can I change the initial CQ factor (it is about 9 now, for HC) to a larger value that is closer to the final of 37 (as for CCE)?

Anyway, I will try with another disc now.

[jdobbs]

The two values aren't related 1:1 -- CCE's OPV value is related to computed quality. It ranges from 1 to 300 or so... the CQ value in HC is referencing quantization and only goes from 1 to 32.

That's exactly what is does. Weird that it plays back but won't encode with HC. Try opening that AVS directly with HC's GUI (if you have it) and see what happens.

[Sharc]

Try opening that AVS directly with HC's GUI (if you have it) and see what happens.

When trying to load the analysis.avs, HC GUI reports an error with line 6 of the script which is:
mpeg2source("C:\xxxxx\xxxxxx\xxxx\LOCAL SETTINGS\TEMP\D2VAVS\V01.D2V")

Edit:
Interesting though, no propblem when I open the V01.D2V directly in the HC GUI.
=> HC just seems not to like the avisynth script ......

[jdobbs]

Are those "xxx" values actually there in the AVS file, or did you edit...

[Sharc]

Are those "xxx" values actually there in the AVS file, or did you edit...

xxx = edited values.
I thought that the original path may perhaps be too long, so I made a version where I put the relevant files into a simpler path to check if this would help, but it didn't make a difference.

Still HC works fine in VBR 2-pass mode with the original path settings. The problem only comes when I change to OPV. Then there seems to be a problem with the acceptance of the *analysis.avs script (?).

[jdobbs]

Try it with another disc so we can see if it is something peculiar about your setup, or that disc.

[kumi]

I'm not sure if it's related, but I just experienced a problem with 0.20 in 2-pass mode, using NaN's DGDecode.dll. On the 2nd pass, HCenc gets to about 15% complete, then it stops encoding any further frames. I/O drops to zero, CPU remains high, the HC window is still responsive, and the VBV indicator stops moving.

[Sharc]

Try it with another disc so we can see if it is something peculiar about your setup, or that disc.

I tried with another disc. Still the same.
I am a little bit lost .....

Anyway, I can always switch to 2-pass VBR. RB and HC is a great package.

[Sharc]

I still can't use HC for redistribution, neither with DVD-RB nor with RB-Opt. HC aborts with the info:
ERROR, Avisynth message; invoke environment error - pass 1.

Reinstalling Avisynth (2.5.6a or 2.5.7) and HC did not help.

No problem with HC in standalone mode using HCgui_021.exe for 2-pass or OPV, though.

Any idea what this could be?

[Boulder]

Maybe you should PM hank315 and ask if he could create the special HC build he mentioned in the HC encoder thread.

[Sharc]

Thanks, I'll do this.

[Sharc]

hank315 has been kind enough to send me a special build which does not reload avisynth.
HC works perfectly now. :)
:thanks:

[jdobbs]

I just wonder why it was only happening in your case? Any additional information as to what was the root cause?

[Sharc]

I scratch my head as well ......
Maybe it is somehow related tho the problem reported here:
http://forum.doom9.org/showthread.php?p=1016478#post1016478
Apparently memory management of avisynth is kind of strange .... hank315 will take care of it in the coming HC release.