[Feature Request] Preserve CEA-608 Closed Captions in MPEG-2 video streams

[yukichigai]

Currently MakeMKV does a fantastic job of perfectly preserving 99% of video content from DVDs with one notable exception: CEA-608 Closed Captions embedded in the MPEG-2 video stream itself. MakeMKV does use CCExtractor to convert such captions to SRT subtitles, but this conversion is not lossless; specifically, positioning information is lost as part of the conversion. Additionally, these captions are not present in the MPEG-2 video data found in the final MKV produced by the program, apparently stripped out during some portion of the processing. This means there is no way to do a 1:1 restoration of CEA-608 Closed Captions in content backed up with MakeMKV.

To address this I would like to suggest that MakeMKV's default processing be altered to leave the CEA-608 Closed Captions in the user data in MPEG-2 video streams. Failing that, CCExtractor is capable of outputting to subtitle formats which preserve all caption information, particularly SCC and CCD, though those are somewhat niche formats. SubStation Alpha, while not a perfect 1:1 match, would preserve positioning information better than SRT and is more widely supported than SCC or CCD. Having SSA/ASS as the default conversion format with an option for 1:1 capturing with SCC would at least be an improvement.

EDIT: Unfortunately, CCExtractor does not preserve positioning information when converting to SubStation Alpha or any other more conventional subtitle format. Positioning info is only preserved in SCC, CCD, and the CCExtractor-specific formats like RAW and BIN.

[yukichigai]

I normally wouldn't post a message to bump my own topic, but I made this post just as chaos was erupting over the issue with keys being unavailable. Now that the situation has been resolved I'm hoping this suggestion might get some attention.

[jgirot]

I agree that if makemkv would leave the CEA data alongside the video (as it is on the DVD), that would be helpful!

Ffmpeg can copy CEA608 data during transcoding, but cannot mux it in separately yet. So if the CEA608 data was alongside, you could carry it through. Or if you're not transcoding, then have it embedded in the video stream.

I know this isn't quite what you want to read, but I use ccextractor separately to rip to raw/bin from the DVD and then use the ccextractor spupng output option to get graphic subtitles with correct screen positioning.

[yukichigai]

I know this isn't quite what you want to read, but I use ccextractor separately to rip to raw/bin from the DVD and then use the ccextractor spupng output option to get graphic subtitles with correct screen positioning.

That's still an improvement over the current state of affairs. It does limit the ability to manipulate the text (translations and the like) but at least the positioning is preserved. Also, do you know if it parses the lesser-used Closed Caption options like Text Opacity, Background Opacity, and Background Color?

It would still be far more efficient to just preserve the stream information though.

[jgirot]

I don't know specifically about those features. I haven't seen any of those effects in my files.

Once you have the rcwt-bin file you could output to a ton of different formats. If you have software that supports it, SCC and SCC disassembled might be good for editing. I believe ccextractor can use SCC as input? I haven't tried. I believe webvtt might include screen positioning. I haven't tried lately.

I saw a GitHub project for converting to and from rcwt/bin and JSON. That could be an interesting way to edit the subtitles. I'll try to post a link later.

[yukichigai]

I actually started making my own conversion tool, or rather a modification of CCASDI, the first SCC -> SRT conversion tool (written by the guy who made the CCD format). So far I've had decent luck getting it to convert properly formatted SCC to SSA/ASS with positioning... emphasis on "properly formatted". Different versions of CCExtractor have problems extracting to SCC from different sources. On the other hand, that Grid 608 format they added seems to ALWAYS come out right, so I've shifted to converting from that. Also had luck there. Also it's way easier to edit than SCC.

I'm glad to know I'm not the only one aware of the issue here though, and working on a possible solution.

[jgirot]

That sounds like some awesome work you've done! I'll send you a PM to discuss further, so that this thread can stay focused on your feature request.

[stellar.dream]

I would love to see this added to MakeMKV! I have a very specific use case of streaming DVD video to a cable tv modulator that supports CEA-608 closed captions. As of now I have no easy way of saving video from DVDs while keeping the CEA-608 stream intact.

I've also been following the requests to add this functionality to ffmpeg. I'm so glad other people care about CEA-608 captions!

[thetoad]

I'm on the other end of the spectrum, while positioning is nice, I'd prefer a text base solution that then strips the CC out, so I'd hope this is configurable.

took me a while to figure out how to strip it out of mpeg2 with ffmpeg (-bsf:v filter_units=remove_types=178 for those that care).

[yukichigai]

I'm on the other end of the spectrum, while positioning is nice, I'd prefer a text base solution that then strips the CC out, so I'd hope this is configurable.

took me a while to figure out how to strip it out of mpeg2 with ffmpeg (-bsf:v filter_units=remove_types=178 for those that care).

More options are always preferable, and really it would fit with the rest of the way MakeMKV works. We can already select what audio, video, and subtitle streams are extracted and preserved. Closed Captions should be no different.

[dyoung]

I am on both sides of this fence.
I definitely saw plenty of bad subtitles/captions before which needed correcting, but I also enjoy the positioning/styling/coloring options of the picture formats.
Perhaps I should explore an option to edit those picture formats instead of converting them to plain text...