Closed captions not being preserved

[boulder]

I am just trying to take away the part before and after a recorded TV show.
When I do that, closed captions (PAL) are no longer being shown on the output (Media Center or Windows Media Player).
On the input this was not a problem.
I am using the latest build available.
In Tools, Options, I do have the option set to "Preserve closed captions".

I thought closed captions were not a problem if you stayed within DVR-MS ?

[DanR]

I don't think that PAL recordings have closed captions in the same sense. Instead I believe that they have teletext or subtitles. Can you record a 1 min. video with CC and upload it to our FTP site. We would like to take a look at it.

[boulder]

I don't think that PAL recordings have closed captions in the same sense. Instead I believe that they have teletext or subtitles. Can you record a 1 min. video with CC and upload it to our FTP site. We would like to take a look at it.

So is this simply ftp.videoredo.net ?
And to which folder should I upload ?

[phd]

ftp://upload:upload@videoredo.net/

Create a folder with a unique name and place the file in it.

Send an email to Support@VideoReDo.com and let us know when it has been successfully uploaded.

Also include a reference to this topic number: 4703

[boulder]

I don't think that PAL recordings have closed captions in the same sense. Instead I believe that they have teletext or subtitles. Can you record a 1 min. video with CC and upload it to our FTP site. We would like to take a look at it.
I checked, and yes, here (PAL) Teletext (page 888 most of the times) is being used.
What I don't understand is why some other free utilities (Cutter, for example) don't seem to have a problem with leaving that data in, when cutting the video.
Is it really that hard to simply leave the VBI data intact when cutting or joining DVR's ?

[DanR]

If the VBI data is placed in the MPEG stream then we preseve it in all cases, except DVR-MS. DVR-MS has unique and undocumented formats. We do handle NTSC CC with DVR-MS. Not sure about TeleText.

Did you upload the file as requested by Pat? We can tell a lot more after looking at the file.

[boulder]

If the VBI data is placed in the MPEG stream then we preseve it in all cases, except DVR-MS. DVR-MS has unique and undocumented formats. We do handle NTSC CC with DVR-MS. Not sure about TeleText.

Did you upload the file as requested by Pat? We can tell a lot more after looking at the file.
I have uploaded the file, it is in "Closed captions or subtitles issue in PAL region" folder.
It 's about 66 megabytes.

[boulder]

If the VBI data is placed in the MPEG stream then we preseve it in all cases, except DVR-MS. DVR-MS has unique and undocumented formats. We do handle NTSC CC with DVR-MS. Not sure about TeleText.

Did you upload the file as requested by Pat? We can tell a lot more after looking at the file.
Has anyone had a look at that file yet ?
I have been looking into this myself a bit.
What I have found is that when I use Graphedit, and choose to render a DVR-MS file, it says it cannot render all streams.
Now, I see that pin 1 and pin 3 are connected for audio and video.
When I connect pin 2 to the decoder tag, then do a dump, I get the teletext data.
That is to say : ALL data, not just the subtitles.
So I guess what happens is that Mediacenter when it plays a DVR-MS file filters out the correct page for Teletext subtitles (there is an entry in the setup of MCE to indicate which page contains subtitles), and displays the lines on screen during playback of the program.
The graph is indeed different when MCE is opening the file, compared to when you open it with Windows Mediaplayer or any other player.

So I think the issue with Videoredo is simply that it does not copy the second stream, in which case the Teletext data, and for that matter, the subtitles would be preserved.

I don't understand what is so hard about preserving that data ?


BTW : I have been doing some more research, and found that I can take this huge dump file, and decipher the characters.
They seem to be in some strange form of ASCII.
I have got to the point where I can see the subtitles in plain text.
I wonder if I could create a utility that would convert them into SRT files.

Problems I will probably encounter :
* codepages, I assume several languages have different ways of representing their special characters.
I found that in the file I saw : mais, whereas on screen MCE presented it correctly as : "maïs", and "zeeen" was correctly represented as "zeeën", but I could not find where in the file the ¨was indicated.
* finding out where page 888 is, or the subtitles in general, and then filtering them out.
* colours : subtitles often use different colours when one line is spoken by one person, and another by a different person.
How are these encoded ?

Anyway, Teletext takes up large amounts of space in the DVR-MS file.
From a 5Gb file, I got >300 megabytes of Teletext data, but of course you can easily compress it enormously.

[boulder]

If the VBI data is placed in the MPEG stream then we preseve it in all cases, except DVR-MS. DVR-MS has unique and undocumented formats. We do handle NTSC CC with DVR-MS. Not sure about TeleText.

Did you upload the file as requested by Pat? We can tell a lot more after looking at the file.

So, did anyone take a look at it ?
I found more info.
It seems that if you open a DVR-MS that contains PAL Teletext subtitles, that this info is all contained in the second stream of the file (the other two are audio and video).
If you connect pin 2 in Graphedit to the Dump Directshow filter, you will get a file that contains mostly illegible text, but you can decypher part of it.
If you convert the characters you get all teletext data (not just the page with subtitles).
So why is stream 2 simply not being copied over ?
This does not strike me as terribly difficult ?