It's this one:
commit 2609fb8077125b5b31f1bcc2f98c12ff1e6572d7
Author: Even Rouault <even.rouault@spatialys.com>
Date: Mon Jun 12 17:15:23 2017 +0100
Packet header writing: set empty packet header bit to 0 when appropriate (small optimization)
I'll try to work out whether it's doing something "illegal".
Dolby DSS200 and test version
-
Carsten
- Posts: 2798
- Joined: Tue Apr 15, 2014 9:11 pm
- Location: Germany
Re: Dolby DSS200 and test version
I just reconfirmed:
Sony 515/S10 plays post2.11.22 DCPs okay, so doesn't suffer from the openjpeg 'bug' introduced with 2.11.23.
At least video and audio appear okay to me when playing 2.11.31 DCPs. Of course, Sony also plays 2.11.34 DCPs.
Barco ICMP on current firmware will not play post2.11.22 DCPs, but as indicated last night has no issues with 2.11.34, this time also checked this visually and aurally.
So, waiting for Dolby DSS 200/220 testing.
- Carsten
Sony 515/S10 plays post2.11.22 DCPs okay, so doesn't suffer from the openjpeg 'bug' introduced with 2.11.23.
At least video and audio appear okay to me when playing 2.11.31 DCPs. Of course, Sony also plays 2.11.34 DCPs.
Barco ICMP on current firmware will not play post2.11.22 DCPs, but as indicated last night has no issues with 2.11.34, this time also checked this visually and aurally.
So, waiting for Dolby DSS 200/220 testing.
- Carsten
-
Eriksson
- Posts: 18
- Joined: Sat Sep 24, 2016 11:08 am
Re: Dolby DSS200 and test version
From a remote location (a train up north) I can confirm that DCP:s made with 2.11.34 work on a DSS200/Cat862. Both SMPTE and interop.
That is: I can se "running", a green progress bar moving and the counter ticking.
Win10 64bit
DSS ver 4.9.0 (96)
/M.E.
That is: I can se "running", a green progress bar moving and the counter ticking.
Win10 64bit
DSS ver 4.9.0 (96)
/M.E.
-
Carsten
- Posts: 2798
- Joined: Tue Apr 15, 2014 9:11 pm
- Location: Germany
Re: Dolby DSS200 and test version
That sounds good. I may be able to verify on another DSS200 and CAT745 combo.
- Carsten
- Carsten
-
carl
- Site Admin
- Posts: 2530
- Joined: Thu Nov 14, 2013 2:53 pm
Re: Dolby DSS200 and test version
Thanks all for the test reports.
-
Eriksson
- Posts: 18
- Joined: Sat Sep 24, 2016 11:08 am
Re: Dolby DSS200 and test version
A friend of mine just tested DSS200/Cat745 with SMPTE/interop. No problems.
DSS ver. 4.9.0 (96)
/M.E.
DSS ver. 4.9.0 (96)
/M.E.
-
Carsten
- Posts: 2798
- Joined: Tue Apr 15, 2014 9:11 pm
- Location: Germany
Re: Dolby DSS200 and test version
Excellent. Good to know 2.11.x is back on track j2c wise.
All in all, a strange occurence. There must have been quite a few changes to openjpeg over the years, but it seems none of the previous ones ever caused issues like this one. And this one, with my admittedly limited knowledge, appears to me like a rather irrelevant, or minor modification. Makes me wonder how the various manufacturers actually confirm the stream robustness of their media blocks against the various encoders. That should tell a lesson to all those who think that GPU aided j2c is a piece of cake to implement.
Every software shows different types of bugs, but to have a software that fails on a specific function is a different kind of problem than to have a software that creates a seemingly correct distribution product that actually fails (hard) in just 'some' cases of field installed target devices.
Carl - were you able to report this issue to the openjpeg team? I mean, it may be an issue on their side, but as some servers fail, while some others have no problems, it may as well be the fault of Dolby or Barco (or whoever makes their mediablocks or maintains their decoder software).
- Carsten
All in all, a strange occurence. There must have been quite a few changes to openjpeg over the years, but it seems none of the previous ones ever caused issues like this one. And this one, with my admittedly limited knowledge, appears to me like a rather irrelevant, or minor modification. Makes me wonder how the various manufacturers actually confirm the stream robustness of their media blocks against the various encoders. That should tell a lesson to all those who think that GPU aided j2c is a piece of cake to implement.
Every software shows different types of bugs, but to have a software that fails on a specific function is a different kind of problem than to have a software that creates a seemingly correct distribution product that actually fails (hard) in just 'some' cases of field installed target devices.
Carl - were you able to report this issue to the openjpeg team? I mean, it may be an issue on their side, but as some servers fail, while some others have no problems, it may as well be the fault of Dolby or Barco (or whoever makes their mediablocks or maintains their decoder software).
- Carsten
-
Carsten
- Posts: 2798
- Joined: Tue Apr 15, 2014 9:11 pm
- Location: Germany
Re: Dolby DSS200 and test version
Bringing this up again - Carl, did you ever receive feedback from the openjpeg team or Even Rouault about this issue?
- Carsten
- Carsten
-
carl
- Site Admin
- Posts: 2530
- Joined: Thu Nov 14, 2013 2:53 pm
Re: Dolby DSS200 and test version
They reverted the patch. By all accounts what they did is fine per the JPEG2000 standard, but it causes implementation complexity in decoders and some don't support it.
-
Carsten
- Posts: 2798
- Joined: Tue Apr 15, 2014 9:11 pm
- Location: Germany
Re: Dolby DSS200 and test version
Alright. I am currently looking into an issue that happened just recently during a german film festival. They created a couple of slides to be played during the festival with DCP-o-matic. They presented them within a playlist, and while playing, the Sony went bellyup. They needed to reboot, and as far as I know, where able to play some of the slides manually (outside of playlist). So far, I can only assume they may have used one of the ill fated 2.11 versions to create these slides. I have yet to receive these DCPs in order to have a look at them.
The poor guy who installed and operated the Sony did not create the DCPs himself, so he doesn't know the version. Of course, even when created with a stable release, you shouldn't play content on a high-profile event without a single test run...but it seems, as often, there wasn't enough time, and they probably also trusted DCP-o-matic to create working DCPs. Which I do multiple days each week when I create slides and trailers for our screenings and add them to playlists without prior testing on our servers. But I never use test versions for this. The cinema machine only runs stable versions, and even those I update with some delay.
I remember there used to be a warning on the test version download page not to use it for production work. No big deal, but maybe we should add it again. Maybe add something to that effect in the test version apps as well?
- Carsten
The poor guy who installed and operated the Sony did not create the DCPs himself, so he doesn't know the version. Of course, even when created with a stable release, you shouldn't play content on a high-profile event without a single test run...but it seems, as often, there wasn't enough time, and they probably also trusted DCP-o-matic to create working DCPs. Which I do multiple days each week when I create slides and trailers for our screenings and add them to playlists without prior testing on our servers. But I never use test versions for this. The cinema machine only runs stable versions, and even those I update with some delay.
I remember there used to be a warning on the test version download page not to use it for production work. No big deal, but maybe we should add it again. Maybe add something to that effect in the test version apps as well?
- Carsten