Possible OPV Bug

[archaeo]

OK, here's the problem - when I choose a custom matrix to run in OPV mode, it appears DVDRB v1.08.1 is not sending the custom matrix to CCE (2.70.2.6) - it is defaulting to the CCE standard matrix.

The first time I tried it, I had AQME enabled, and I thought this might be the problem. So I disabled that feature (CCEAQM=0), and ran it again with the same custom matrix, the Fox Home Entertainment Matrix... Result: same problem- when I extracted the matrix with RME, it was the CCE standard matrix. Checking the item. ecl showed the command 'qmat_idx=0', w/no no custom matrix.

As a control, I decided to try another OPV encode, this time using RB Opt v21, again with CCEAQM disabled. When I
finished the encode and extracted the matrix, I found that it did in fact apply the Fox Home Entertainment matrix. And the item.ecl does contain 'qmat_idx=1' along w/ the Fox matrix.

Anyway, I'm not sure what the problem is and why DVDRB OPV is not updating the ecl with the custom matrix before sending to CCE.

[jdobbs]

Hmmm... I seem to recall disabling matrices in OPV several versions ago... but I'll check.

[archaeo]

yep, you did jdobbs. Didn't realize that. I did a search and found that you disabled that feature in v.89:

Changed OPV encoding so it doesn't use custom
matrices. Since OPV uses constant quality
(quantizer) encoding, the precalculated bitrate
shouldn't be used as a method of determining
quantizer matrix as it really doesn't apply.
This should make sizing predictions more
accurate.


It seems that sizing accuracy was the main concern for it. Were there that many problems with this feature? Any chance you might re-enable it to use ***at one's own risk***?? I have to say I do like being able to choose, even for OPV... and be self responsible for any over/under sizing. The above mentioned OPV that I did w/RB Opt using the Fox custom matrix (w/ 4% sample size) came in right at 4.33 gb.