A DCP VF with a image layer is possible???

Anything and everything to do with DCP-o-matic.
Post Reply
Silas Alexandre
Posts: 1
Joined: Mon Nov 14, 2016 10:35 pm

A DCP VF with a image layer is possible???

Post by Silas Alexandre » Thu Apr 13, 2017 9:27 pm

Hi, everyone. First of all, sorry my english.

I'm looking for a solution to show a video file (with sign language, you know for deaf people enjoy the movie) into the dcp package. I wondering if its possible to create a kind of dcp package VF with a layer of another video, instead of the subtitle of the movie.

And if that is possible ... i'll could choose if the exibition will happen with the "sign language layer" like a DCP VF, or no. Like is possible with DCP VF subtitle. And if this happen i'll be able to create the same DCP package to "Hearing impaired sessions" and "Normal sessions". That's my final goal.

Anybody can give me a light here? I atached a frame, just in case my message is difficult to understand.
Attachments
dcp vf.jpg
dcp vf.jpg (254.52 KiB) Viewed 3747 times

carl
Site Admin
Posts: 1050
Joined: Thu Nov 14, 2013 2:53 pm

Re: A DCP VF with a image layer is possible???

Post by carl » Thu Apr 13, 2017 9:31 pm

As far as I can see this might be possible with PNG subtitles... I haven't looked at them much yet, though, and DCP-o-matic can't create them (yet).

scorpio81
Posts: 91
Joined: Mon May 09, 2016 7:19 am

Re: A DCP VF with a image layer is possible???

Post by scorpio81 » Fri Apr 14, 2017 7:19 am

I have also thought of PNG subtitles. But for the goal that topic starter wants to achieve it is needed that PNG subtitle files must change with the same fps as the main movie.
According to the current DCI specs it might be possible. The PNG files should be the same resolution as DCP and rendered in the same color space as DCP. Again, according to the DCI specs, the navigation XML file will be rather big because it has to specify all the frames, and there should be hundreds of thousands of them for sign language.

liloneum
Posts: 7
Joined: Fri Apr 14, 2017 9:00 am

Re: A DCP VF with a image layer is possible???

Post by liloneum » Fri Apr 14, 2017 9:12 am

Hi,

I already thought about it as I am working on accessibility for deaf people too.
As you said, the way to go is png.
You can use png files at the size you need for the sign language translator and position them at the bottom left or right.
You'll have to convert your video with alpha channel to png files, create a special subtitle file referencing all the images every frame.
It will be a very big file, but you can script it.
I think the only way to go is SMPTE and wrap a mxf subtitle (I already have troubles on some servers with a classic subtitle and about only 1000 png files for INTEROP).
About the colour space, it has to be RGB, but some users reported troubles on some Doremi systems that need wrongly XYZ.
The only big problem will be: is a server able to display every frame a jpeg2000 and a png without latency?
Good luck with it, and I'm very interested with your results!
I can give a hand for the subtitle scripting and wrapping.

Lilian

Carsten
Posts: 1397
Joined: Tue Apr 15, 2014 9:11 pm
Location: Germany

Re: A DCP VF with a image layer is possible???

Post by Carsten » Fri Apr 14, 2017 11:27 am

There has been some testing done a while ago on PNG performance, and it does occur that systems will not have the capability to display 'live' video in the form of PNGs. The outcome will be very unpredictable. Unlike the video mxf, PNG is not a realtime stream from server to projector, but more a web-server like communication. To my knowledge, this will not change with SMPTE DCPs, as the PNGs will be unwrapped server side and then sent to the projector using the same methods as with current XML/Interop DCPs.

The ISDCF brought this issue up recently, as Brazil issued new laws towards the use of Brazilian sign language in cinemas.
However, they developed these laws without considering the technical aspects/capabilities of the current digital cinema environment. So, currently, this seems to be limited to brazil, and for cinemas, no one has a real solution, aside from setting up a second projector or other display device and sync mechanism. I would love to have PNG support in DOM for various reasons, but I don't think it will solve this specific issue.

Is has been covered in the recent ISDCF meeting/notes. The outcome currently is - the deadline for implementation is too short, the technical experts in DCI doubt there will be a solution based on standards in time. The new law will become effective in September 2017 to my knowledge. There are some technology demonstrators, and some of them may be installed in brazilian cinemas by the time. They do seem to attach to subtitle or closed captions tracks for content and sync, so, the best one could do by then is to supply subtitles and/or closed caption with the DCP.

Wondering wether a 'faked' 3D DCP could be used instead...

https://seer.ufmg.br/index.php/jidm/art ... /1034/1011

With the digital cinema standards development advancing very slowly, I think the only realistic outcome is to render sign language on external equipment, like local smartphones, probably synced from the server or by audio track, like it's done for subtitles or closed captions.
Another solution, of course is to overlay/burn in the video into the main video stream - wether this will be an OV or VF, but for a 'limited market', these special DCPs may be the only solution to obey these laws that work on every system.
People without handicaps will be annoyed by these overlays more than by subtitles already for sure, as there will be only few 'dedicated' handicapped audience presentations with them. So I think, personal display devices like smartphones or glasses, will be the only viable solution in the long run.

The question is, how much standards development can be expected for a limited market/application like BSL?

- Carsten

liloneum
Posts: 7
Joined: Fri Apr 14, 2017 9:00 am

Re: A DCP VF with a image layer is possible???

Post by liloneum » Mon Apr 17, 2017 1:19 pm

OK, it seems that PNG is not the way to go...
I will follow your intuition for fake 3D DCP and try some test.

Lilian

liloneum
Posts: 7
Joined: Fri Apr 14, 2017 9:00 am

Re: A DCP VF with a image layer is possible???

Post by liloneum » Sun Mar 04, 2018 11:07 pm


binba
Posts: 11
Joined: Thu Jul 04, 2019 3:56 am

Re: A DCP VF with a image layer is possible???

Post by binba » Sat Sep 07, 2019 2:14 am

So the way they pulled it off was to encode 1Mbps video into one of the audio channels, and sneak it out to a decoder via AES/EBU! :o
Sneaky... kinda like the 21st century equivalent of LTC...

Here is a very detailed video (by a Brazilian company) showing the process:
https://www.youtube.com/watch?v=9cqHU7REDJI
The part about sign language starts at 12:57.

Carsten
Posts: 1397
Joined: Tue Apr 15, 2014 9:11 pm
Location: Germany

Re: A DCP VF with a image layer is possible???

Post by Carsten » Sat Sep 07, 2019 10:52 am

Christian, the guy who made that video, turned up here a while ago:

viewtopic.php?f=2&t=1329&sid=0c8650dfc3 ... eb5f2ce6f7

- Carsten

carl
Site Admin
Posts: 1050
Joined: Thu Nov 14, 2013 2:53 pm

Re: A DCP VF with a image layer is possible???

Post by carl » Mon Sep 09, 2019 12:07 am

It's mind-boggling but also kind of reassuring that hacks like this still exist :)

Anyway, I hope DCP-o-matic can support these tracks directly at some point. I added a feature request to track progress.

Post Reply