Subtitle Edit

[Emulgator]

A writeup of SE's proxy file depencies/conventions, maybe helpful for any manual pruning:
https://forum.doom9.org/showthread.php?p=2010776

[markfilipak]

Originally Posted by Emulgator

"None" here as well, never touched it. Hm.

It's a mystery... Thanks for trying. ...Umm, look at your SRT in list mode and make sure that "are" is followed by "<br />" with nothing in between?

PS: I now think this bug has nothing to do with punctuation (see https://forum.doom9.org/showthread.p...77#post2010877).

[Emulgator]

(Retyped from list window)
- PICARD: Situations like these are<br />-What?
renders correctly as 2 lines in Edit window.
Without video, without audio, naked SE 4.0.8 standalone, unpacked from .zip, no common paths.
Conflicting/shared Settings.xml ?

[markfilipak]

Originally Posted by Emulgator

A writeup of SE proxy...

I believe you are misusing the word "proxy". I think you mean "specification".

[markfilipak]

Originally Posted by Emulgator

(Retyped from list window)
- PICARD: Situations like these are<br />-What?
renders correctly as 2 lines in Edit window.
Without video, without audio, naked SE 4.0.8 standalone, unpacked from .zip, no common paths.
Conflicting/shared Settings.xml ?

Yours:

- PICARD: Situations like these are<br />-What?

Mine:

- PICARD: Situations like these are<br />- What?

PS: I've noticed that SE has a lot of off-by-one bugs. This may be one of them.

PPS: The dialog is from "Star Trek: The Next Generation", season 5, episode 123, "The Next Phase". I just reloaded it and ran a test. I could not replicate the problem. That's disturbing. It means the problem has something to do with timecodes and/or shotchanges, which were not reloaded even though they still exist. I continue fixing ST:TNG subtitles and will keep my eyes open to this problem and see if I can replicate it in the future.

Originally Posted by markfilipak

Yes. Use SE to rip the subs.

OK, thanks, I will it a shot

[leoenc]

Is there a way to set more parameters for VobSub export via the command line? (E.g. font family, size, simple rendering...)
Perhaps there's a way to use a template file with all of the settings that are available in the GUI VobSub export window?

[markfilipak]

Originally Posted by leoenc

Is there a way to set more parameters for VobSub export via the command line? (E.g. font family, size, simple rendering...) ...

Subtitles on DVDs and Blu-rays are pictures of text. You can use FFmpeg to split out the pictures and save them as graphics. That's what SE does. To get letters with fonts and sizes you need to optically convert the pictures to strings of textual characters. That's what optical character recognition (OCR) does.

[leoenc]

@markfilipak, maybe I wasn't clear. I'm converting SRT to VobSub, not the other way around.
Here is my current CLI:

SubtitleEdit.exe /convert subtitle.srt VobSub /fps:25 /resolution:3840x2160 /targetfps:25

[markfilipak]

@leoenc, if you are making an MP4, I've not seen any computer-based player that will not render an embedded (muxed-in) SRT stream. So, I assume you're making a DVD and need vobsubs. I don't do that. I can't help. Sorry. I suggest you try FFmpeg.

[leoenc]

Actually I need to create MPEG-TS with DVB subtitles in ffmpeg, which requires bitmap based subtitles as input.

[markfilipak]

Originally Posted by leoenc

Actually I need to create MPEG-TS with DVB subtitles in ffmpeg, which requires bitmap based subtitles as input.

I've not done that, but I know that others have. FFmpeg has the text renders, with font control, and the codecs to get what you want. If you figure out the FFmpeg commands that do it, you may be able to use the very same commands in SE since SE uses FFmpeg... or you can just do it in FFmpeg and skip SE entirely.

[leoenc]

I haven't been able to convert textual subtitles to bitmap based subtitles with ffmpeg. It is limited to text->text or bitmap->bitmap workflows.

BUT, I managed to get much better results with SubtitleEdit /Convert when using ASS instead of SRT as I can edit the font/size etc. in the ASS file itself before conversion.

[Emulgator]

A resolution 3840x2160 is is out of vobsub specs anyway.
In .ts 1920x1080 SUP would work.
Then your workflow would be <your file> SE -> .srt -> tsMuxeR -> .ts or .m2ts
In tsMuxeRGUI you can specify font, color and alignment for the first SUP stream.
If you need more SUP streams then you would handedit the other streams into the tsMuxeR .meta file from first attempt, then run the tsMuxeR CLI from a batch file pointing to that .meta file.

[markfilipak]

Originally Posted by leoenc

I haven't been able to convert textual subtitles to bitmap based subtitles with ffmpeg. It is limited to text->text or bitmap->bitmap workflows.

You are presuming. It is not true. I know people have made bitmaps from text using FFmpeg and then made subtitles from the bitmaps. I've not needed to do that myself.

[leoenc]

Originally Posted by Emulgator

A resolution 3840x2160 is is out of vobsub specs anyway.
In .ts 1920x1080 SUP would work.
Then your workflow would be <your file> SE -> .srt -> tsMuxeR -> .ts or .m2ts
In tsMuxeRGUI you can specify font, color and alignment for the first SUP stream.
If you need more SUP streams then you would handedit the other streams into the tsMuxeR .meta file from first attempt, then run the tsMuxeR CLI from a batch file pointing to that .meta file.

I deliberately use double the resolution of full HD, as then the subtitles render with antialiasing when scaled down.
When using SUP as source, the subtitles render incorrectly by ffmpeg (half of the subtitles is cut off).
I cannot use GUI tools, I'm building an automated workflow.

[leoenc]

Originally Posted by markfilipak

You are presuming. It is not true. I know people have made bitmaps from text using FFmpeg and then made subtitles from the bitmaps. I've not needed to do that myself.

I'm not presuming. Here is the output straight from ffmpeg:

[sost#0:2/dvbsub @ 0000024b03ac9cc0] Subtitle encoding currently only possible from text to text or bitmap to bitmap

Trust me, I've posted this after days of researching this online. I couldn't find anyone that made this work otherwise.

[Emulgator]

3840x2160 is is out of SUP specs as well, this plane is limited to 1920x1080.
No GUI ? see tsMuxeR CLI.

[markfilipak]

Originally Posted by leoenc

Trust me, I've posted this after days of researching this online. I couldn't find anyone that made this work otherwise.

FFmpeg can make static pictures out of any text. Once they are pictures, the pictures can be converted to various graphical subtitle formats. I've not done that sort of thing, but others have. If the graphical subtitles get cut off, then the pictures need to be dimensioned to oppose the cut off. Again, I've not done that, but others have.

There is an FFmpeg user's mailing list by which you can ask questions.

PS: When you render the text as pictures, you don't have to save them as files with serialized names. The pictures can be streamed into the coder that makes the subtitle stream, then the streams can be muxed -- all in one process, one command line.

PPS: You see, FFmpeg is not meant to be evoked by end-users. It's meant as a go-between between a script or GUI and the coders. That's why so many FFmpeg error messages say "Fix your code". That's why the FFmpeg documentation is unsuited to novices.

[markfilipak]

I've noticed that this: "- This is one speaker<br />- This is another speaker" splits to make this:

This is one speaker

This is another speaker

But this: "- This is one speaker--<br />- This is another speaker" splits to make this:

- This is one speaker--

- This is another speaker

The "^- " (regexp) substrings are not being removed. It is a hard fault. It has nothing to do with punctuation.

Again, I suspect an off-by-one bug.