BD Rebuilder Beta - Bug Reports Only

[PurpleMan]

Originally Posted by turbojet

PurpleMan: Would fixmpls help you debug it further?

I tried using fixmpls on the original MPLS file, but fixmpls does not fix sub-path chapter points, only main chapter points. I'm going to hex-edit the bd-rb mpls to fix this myself and see what happens.

Either way, I reckon this should be addressed by our beloved jdobbs

[Sharc]

Originally Posted by turbojet

QUICK_USE_QUALITY=1 is working for me

Good to know. Seems like those Extras which are 480i in the original and which I let BD-RB convert to 720p are still encoded 'ultrafast'. Maybe this is intentional?
I will try another disc with a simpler structure.

Added:
I found the 'bug': The correct command is QUICK_USE_QUALITY=1. I put in QUICK_USE_EXTRAS=1 as erroneously documented in the HIDDENOPTS.TXT.

[PurpleMan]

Progress report:

I've hex-editted the out_times, and now when I click play movie it indeed starts playing. So that's a winner.

However, if I try to seek or jump through chapters, powerdvd crashes. Which indicates I've either screwed up something in the MPLS while editting it, or that the timing values weren't the only thing wrong with the MPLS to begin with.

[laserfan]

Originally Posted by PurpleMan

...fixmpls does not fix sub-path chapter points, only main chapter points.

What's a "sub-path chapter point"?

Also, what is the magic incantation for downloading fixmpls--all I ever get is "Slow down, you have to wait 17 seconds..."?

[jdobbs]

Originally Posted by Sharc

BD-RB v0.30.03, 1-pass CRF mode.
I tried QUICK_USE_EXTRAS=1. It doesn't seem to have an effect. The Extras are still encoded 'ultrafast' -- as indicated by the lastcmd.txt and by analyzing the encoded file with MediaInfo.
I selected "High Quality" (default), which uses --preset medium IIRC, and enabled "Quicker" Encode for Extras. I would expect that QUICK_USE_EXTRAS=1 should now have taken the "High Quality" (i.e. --preset medium) switches.
Did I miss something?

Secondly, is it ok to use a preceding # to disable certain options in the config file for testing?

I typed it wrong in the HIDDENOOPTS.TXT file. It is "QUICK_USE_QUALITY".

[jdobbs]

Originally Posted by PurpleMan

@jdobbs:

I think I may have found a new bug with the way BD-Rebuilder handles MPLS files for pip discs.

I did Resident Evil: Extinction and left out the PiP in the full disc (BD25) backup. The result disc would play perfectly in powerdvd7 and tmt3, but would not play in PowerDVD9 (when you click 'play movie', it would show the FBI warnings then get back to the main menu, refusing to even start playing the main m2ts).

I figured this must be a PowerDVD9 bug with the disc, so I tried playing the original one. To my surprise, it played perfectly. So I started to dig - what could possibly make PowerDVD9 not play the bd-rb output? I replaced bd-rb's main-movie MPLS with the original disc MPLS (knowing it will screw up alot of stuff), just wanted to see if powerdvd will then at least start playing, and - IT DID! So I concluded that there's something in the BD-RB outputted MPLS that it did not like.

The original disc MPLS had the IN_TIME set to 00:00:11,650, and I know that BD-RB changes it in the rebuild to correspond with TSMuxer (IN_TIME = 00:10:00,000). Nothing wrong with that, and the chapters have been adjusted accordingly.

Further digging in BDEdit revealed something that is probably wrong. The sub-path chatper-points (used for the pip, I think) only had the IN_TIME adjusted to the new main m2ts IN_TIME (10mins). meaning, for each of the chapters the IN_TIME increased in ~10mins, which is good, but the OUT_TIME was not (and it should have been!). So the OUT_TIME for each of the chapters was BEFORE the IN_TIME (not good!)

I attached two screenshots of BDEdit illustrating this. One of the original disc, and one of the BD-RB output disc. You can see what I'm talking about there.

P.S.
This disc was done with 0.29.03, so if you addressed the bug since then, I apologize.

Hmmm... good catch. I'll check it out.

[jdobbs]

@PurpleMan

Thanks for the outstanding bug report. The detail was fantastic and pointed me in exactly the right direction.

I just looked at it, and this could be the cause of other problems also. When the new OUT value was rewritten, there was a character mistyped on the line of code that rewrote the value to the MPLS -- and (oddly enough) it didn't cause a compile error because here was another variable by that (mistyped) name. The value was written to the wrong location in the file.

I'm updating it and will post a fix.

[jdobbs]

Originally Posted by turbojet

Thanks a lot for the new version. 59.94 and crf extras are much more efficient. However a few things I've noticed that have changed.

- REBUILDER.INI is included in the zip again, which wipes out all settings, is that needed?
- x264 --me umh --psy-rd 1.0:0.2 don't exist any longer in the high settings, could they be used again for higher quality?

"--me umh" is set in both the "High Quality (Default)" and "Highest (Very Slow)" modes as a part of the preset. Are you saying the defaults in X264 have changed? I don't remember seeing anything in the changelogs...

"--psy-rd 1.0:0.0" should be present in all settings except "Good" and "Highest Speed (BD-25)"

[PurpleMan]

Originally Posted by jdobbs

@PurpleMan

Thanks for the outstanding bug report. The detail was fantastic and pointed me in exactly the right direction.

I just looked at it, and this could be the cause of other problems also. When the new OUT value was rewritten, there was a character mistyped on the line of code that rewrote the value to the MPLS -- and (oddly enough) it didn't cause a compile error because here was another variable by that (mistyped) name. The value was written to the wrong location in the file.

I'm updating it and will post a fix.

No, thank you!

And I find great pleasure in that I was able to help! Keep up the good work!

To continue our discussion, I wrote a small program that corrected all the OUT_TIMES on the BD-RB MPLS, and like I said, PowerDVD9 then agrees to start playing. Thing is, after only 1 minute of playback, the program crashes. Further digging revealed that it crashes EXACTLY at points where PiP would appear if I hadn't ripped it out with BD-RB. What do you think could cause this?

Maybe it's a bad idea to keep all the PiP info in the MPLS even though it doesn't actually exist on file (references to tracks 1A00 and 1B00, and the long list of sub-path chapter points).

Otherwise, there must be some reason PowerDVD9 crashes on the RB but not the original disc. We could say it's a PDVD9 bug since it's crashing instead of erroring, but it's obvious that it's seeing (or not seeing) something in the MPLS that it's not expecting to. And maybe it crashes because that thing is out-of-spec and the coders didn't think they'd need to handle such a case.

A penny for your thoughts.


Edit:
In addition to the main MPLS (00001.mpls), the disc has another playlist (00060.mpls) that is the same as the first, except it doesn't contain the subpath and pip data. I tried overwriting 00001.mpls with 00060.mpls, and viola - PowerDVD9 plays the disc perfectly, chapter selection works, seeking, and everything else. Could this, perhaps, validate my abovementioned point?

Edit 2:
I can now confirm that the OUT values calculated by 0.30.04 are correct. However, output still crashes on PDVD9 where PiP should be.

[jdobbs]

I've just updated the first post in this thread to include a link to the most recent release (v0.30.04) of BD-RB. Changes:

- Corrected an error in which the SubPlayItem OUT
  value was not being updated properly in the SubPath
  table of the MPLS.  This could be the root cause of
  several reported inconsistencies.
- Other minor corrections and cosmetic fixes.

[chudm]

hi jdobbs i kept having this problem with braveheart, region a bluray:




what can i do?

[turbojet]

OLD SETTINGS
Firstpass: --subme 7 --me hex --bframes 3 --b-adapt 2 --8x8dct --direct auto --weightb --threads auto
Very High Quality: --subme 9 --me umh --ref 4 --bframes 3 --b-adapt 2 --mixed-refs --8x8dct --direct auto --trellis 1 --weightb --threads auto --psy-rd 1.0:0.2
Description: Everything and the kitchen sink, but no insane settings.

Firstpass: --subme 2 --me dia --bframes 3 --direct auto --weightb --threads auto --partitions none
High Quality: --subme 7 --me umh--ref 3 --bframes 3 --direct auto --8x8dct --weightb --trellis 1 --threads auto --psy-rd 1.0:0.2
Description: A good balance.

Firstpass: --subme 2 --me dia --bframes 3 --direct auto --weightb --threads auto --partitions none
Medium Quality: --subme 6 --me hex --ref 2 --bframes 3 --direct auto --8x8dct --weightb --threads auto
Description: A good balance, but toned down a bit.

Firstpass: --subme 0 --me dia --bframes 3 --direct auto --weightb --threads auto --partitions none
Low Quality: --subme 3 --me hex --bframes 3 --8x8dct --weightb --threads auto
Description: Much faster, but with a significant quality penalty.

Firstpass: None
Ultra-Low Quality: --subme 1 --me dia --partitions i8x8,i4x4 --bframes 3 --8x8dct --threads auto
Description: Even faster, but with a very large quality penalty.

Firstpass: None
Ultra-Fast: --subme 0 --no-cabac --partitions none --me dia --aq-mode 0
Description: Incredibly fast, but with an enormous quality penalty.


NEW DEFAULTS
ultrafast: --ref 1 --scenecut 0 --nf --no-cabac --bframes 0 --partitions none --no-8x8dct --me dia --subme 0 --aq-mode 0
veryfast: --partitions i8x8,i4x4 --subme 1 --me dia --ref 1 --trellis 0
fast: --mixed-refs 0 --ref 2 --subme 5
default: --subme 7 --bframes 3 --weightb --8x8dct --ref 3 --mixed-refs --trellis 1 --crf 23 --threads auto


NEW EQUIVALENT (add --slices 4 -vbv-bufsize 14500 --vbv-maxrate 17500 --stats --output --level 4.1 --sar 1:1 --aud )
very high p1: --pass 1 --slow-first-pass --b-adapt 2 --direct auto
very high p2: --pass 2 --subme 9 --me umh --ref 4 --b-adapt 2 --psy-rd 1.0:0.2
high p1: --pass 1
high p2: --pass 2 --me umh --direct auto --psy-rd 1.0:0.2
medium p1: --pass 1
medium p2: --pass 2 --subme 6 --ref 2 --direct auto
low p1: --pass 1 --subme 0
low p2: --pass 2 --subme 3 --ref 2
ultralow: --preset veryfast
ultrafast: --preset ultrafast

CURRENT HIGH QUALITY: --pass 2 --preset medium (this is default, redundant) --threads auto (default/redundant) --thread-input (default/redundant)

--me umh --direct auto --psy-rd 1.0:0.2 are missing from high setting, I don't know about the others.

[chudm]

Originally Posted by chudm

hi jdobbs i kept having this problem with braveheart, region a bluray:




what can i do?

how could i rebuild all the files myself? thx

[chudm]

Originally Posted by chudm

how could i rebuild all the files myself? thx

i replaced the main movie on the original bluray (00006.m2ts) and copy the new clpi files and burnt the movie, but, instead of having the almost 3 hours movie, the movie ends about 1 hour before that, and then only gets a blank screen!

any suggestion guys?

[samtroy]

@turbojet:

Could you please explain for us "normal" users :

Does that mean that BD Rebuilders output is different now because x264's presets have changed in the newest version (r1281)?

[Sharc]

Originally Posted by turbojet

CURRENT HIGH QUALITY: --pass 2 --preset medium (this is default, redundant) --threads auto (default/redundant) --thread-input (default/redundant)

--me umh --direct auto --psy-rd 1.0:0.2 are missing from high setting, I don't know about the others.

--preset medium uses: --me hex, --psy-rd 1.0:0.0
The --psy-rd 1.0:0.2 would need to be extra set, or may be invoked by --tune film

(It may be easiest to check an encoded .m2ts with MediaInfo and read the encoder settings back to see what has eventually been set ..... not always easy to keep track of the changes..)

[jdobbs]

Originally Posted by Sharc

--preset medium uses: --me hex, --psy-rd 1.0:0.0
The --psy-rd 1.0:0.2 would need to be extra set, or may be invoked by --tune film

(It may be easiest to check an encoded .m2ts with MediaInfo and read the encoder settings back to see what has eventually been set ..... not always easy to keep track of the changes..)

All the defaults are used (from MEDIUM) unless changed by the preset.

Here are the presets:

ultrafast (used in QUICK and FASTEST by default):
ref 1, scenecut 0, no-deblock, no-cabac, bframes 0, partitions none, no-8x8dct, me dia, subme 0, aq-mode 0, no-mixed-refs, trellis 0, b-adapt 0, no-mbtree

veryfast (Used for "GOOD"):
partitions i8x8,i4x4, me dia, subme 1, ref 1, no-mixed-refs, trellis 0, no-mbtree

fast (Used for "Better, Fast"):
ref 2, subme 6, rc-lookahead 30

medium (Used for "High Quality (Default)"):
No changes from the X264 defaults.

slow (Used for "Highest (Very Slow)")
me umh, subme 8, ref 5, b-adapt 2, direct auto, rc-lookahead 50

In addition to these settings I've added limiters to assure BD compliance, including B frame limitations, GOP length, REFS, appropriate bitrate limitations, and --SLICES settings.

I was mistaken, though, "--me umh" isn't used in the "High Quality (Default)" setting. It uses the default "--me hex." I've spent a lot of time optimizing these settings and don't plan on changing them back to what they were when BD-RB first started...

[Sharc]

Originally Posted by jdobbs

All the defaults are used (from MEDIUM) unless changed by the preset.

I spent a lot of time optimizing these settings...

I have no doubt

[chudm]

any suggestion jdobbs about my problem i posted before?

[jdobbs]

Originally Posted by chudm

any suggestion jdobbs about my problem i posted before?

I can only say the the one time I've seen that error reported before was on a downloaded movie that was improperly formatted by the hacker. BD Rebuilder is meant to work with original movies and I don't support preprocessing (or hacking for that matter). That's why I didn't want to discuss it. Error 52 means "bad file name". As I recall, in the previous case the file name had illegal characters in it.

There's not really anything I can do. I get the filename from the Operating System via directory functions, then when I try to open it -- boom -- the name isn't legal. Garbage in, garbage out.

I'm not saying your copy is hacked -- but I am saying it has very probably somehow been trashed.