[FFmpeg-user] How to copy all of 5 audio streams to mp4 file

Ulf Zibis Ulf.Zibis at gmx.de
Tue Feb 19 20:26:15 EET 2019


Hi Ted,

Am 19.02.19 um 17:14 schrieb Ted Park:
>> Yes, I know, and for monochrome images it compresses much better. But
>> I'm wondering about the scrambled content and the huge amout of PNG
>> files in the ripped data, so it seems to me, that each subtitle pic
>> becomes somehow decrypted from many PNG fragments which unfortunately
>> fill 10 % of the entire stream.
> JPEG isn’t a bitmap format. And those png’s aren’t encrypted, they are regular (more or less) black and white bitmaps. On a commercial pressed dvd it would have been kind a wireframe that the subtitles are built on using color info from the program it’s in. So ~10% isn’t too bad when it’s pulled out as something independently decodable. You can ocr the text, but if you are wanting to archive the original bitmap based subs, not much that can be done. There’s no easy way to weave it back in as subpicture either, not a reversible process.

Much thanks for your detailed teaching.

It turns out, that I was in error with the ~10 %. A 6 min. chunk of the
video with subtitles in mkv container consumes 100.6 MB, without
subtitles in mp4 it's 98.1 MB. It was late in the night when I
erroneously read 90.1 MB. So the footprint is only 2.5 % which is ok for
me. Please sorry for the noise and your efforts.

But with the OCR idea I have a problem. Have you seen the pics I
attached 18.02.19, 23:15 MEZ ?

I'm still wondering, why ffmpeg doesn't recognize the language codes of
the subtitle streams. They must be there, as one can see from the
attached dvd::rip screenshot. Is it possible to add them manually?

-Ulf
-------------- next part --------------
A non-text attachment was scrubbed...
Name: dvdrip language codes.png
Type: image/png
Size: 165464 bytes
Desc: not available
URL: <http://ffmpeg.org/pipermail/ffmpeg-user/attachments/20190219/0de2614a/attachment.png>


More information about the ffmpeg-user mailing list