BDSup2Sub - convert and tweak bitmap subtitle streams (VobSub,BD-SUP,BDN XML,HD-SUP)

[Pawpcorn]

Originally Posted by Chetwood

VobSub Shuffle can merge several languages into one VobSub, if that's what you mean.

Thanks, Chetwood! I'll try it out, tonight! At first glance it looks like what I was looking for!
Thanks again,
Pawpcorn

[Pawpcorn]

Well.... I tried the VobSub Shuffle program tonight, combining an English idx/sub subtitle, and a Thai idx/sub subtitle. Both had been processed by BDSub2sub, with identical sizes and palettes... and it LOOKED like it worked: (Command screen
-----------------------------------------------------------
C:\Users\Gerald\Downloads\VobSub.Shuffle.1.0.Win32\VobSub Shuffle>vsshuffle combined eng thai

VobSub Shuffle version 1.0 - (C) 2010 by jeanl

Collecting all streams
eng.idx: 1 stream(s)
thai.idx: 1 stream(s)

Creating output sub file: combined.sub
eng.sub
thai.sub

Creating output idx file: combined.idx

Checking for streams in combined.idx not present in combined.sub
combined.idx: 2 stream(s)
combined.sub: 2 stream(s)

Done!
----------------------------------------------------------------
in that it created a new idx/sub subtitle "combined" that was twice as large as the two originals...
BUT...
When I opened the resultant file "combined" with BDSub2sub, only the English subs showed.

When I opened the resultant file with Subtitle Creator, it asked me "Choose Language Stream" with the choices of English or Thai. Only ONE language subtitle would load; not a combined merged subtitle display, which had been my hope.

When I created a DVD (AVStoDVD) using this combined idx/sub subtitle file... upon viewing the resultant DVD, select to display subtitles menu (in VLC) there was only ONE subtitle choice available to display (good!) but... unfortunately.... only the ENGLISH language subtitles appeared!

Any other ideas? I want to merge the two graphic images of two separate languages subtitles... to display both subtitle languages simultaneously (as ONE subtitle), when requested by the viewer.

I'm happy to supply the source and output data subtitle files, if this is of value to anyone...

Any ideas, anyone?
Thanks,
Pawpcorn

That's because the Java BDSup2Sub doesn't handle IDX/SUB with multiple languages properly. It only loads the language specified by langidx in the IDX file and ignores all the others. The ported version I've created can handle streams with multiple streams and lets you switch between them but still contains the limitation of the original code that only one language is exported. It's something I mean to rectify but it's part of an every growing list of enhancements that I want to make.

[Pawpcorn]

Thanks, Sassbot, for your reply... You have to know that I use your program almost EVERY day, and it's just a Godsend to me! It was the final solution to sooooo many problems I used to have, working with vob subs...
Regarding the issue above, I'm just NOT clear on why these vob muxing programs have been written, if all they do is put two subtitle streams into one file... And in the various programs that input idx/sub subtitles, u have to select which stream to work with in further steps, and you're back to seeing only a single subtitle once you've done that.... Confusing to me.
WHAT program, or what situation can take advantage of multiple subtitle "streams"?
Thanks again for your terrific program!
Pawpcorn

[Chetwood]

Vobsub Shuffle works for me. Anytime I combine separate languages with it, I can later cycle through them while playing back the file. Subtitle Creator also can combine subs streams: open the second sub stream and save it under the name of the first stream. It first asks you to overwrite it but after selecting yes it asks again if you want to add this stream. In case this also doens't work, upload the files to Mediafire or something so we can have a look.

[r0lZ]

Pawpcorn, I don't think what you want is possible with a one-click program. Merging the individual images of two different subtitles into single images, to build a single multi-language stream is technically difficult, as that would require that both streams have all subtitles perfectly synchronized. For example, if the first subtitle of your English stream starts at 0:00:01.000 and the corresponding subtitle of the Thai stream starts at 0:00:01.500, the program will have to decide what time code it must keep, or create a sub with English only from 0:00:01.000 to 0:00:01.500, and then add the multilingual subtitle at 0:00:01.500. Also, the program will have to move one subtitle up or down if both are at the same place in the original streams. AFAIK, no program can do that.

Perhaps you can try to convert both subs to BMP files, and merge them yourself with a batch image processing program like ImageMagick, then rebuild the IDX/SUB steam. If there are the same numbers of subtitles in both streams, and if you can assume that the timings of one of them is suitable for the other one, that should work.

Anyway, I don't understand why you want to display two languages at the same time. What's the benefit? Is it for a multi-language audience?

[Pawpcorn]

Hi, Chetwood, thanks for your message.

I think there's some understandable confusion regarding my terminology about what I'm trying to do...

I tried your method (SubtitleCreator) and this action is appending subtitle files.

What I'm trying to do is an actual merge of two graphic subtitle images... into one, for each and every subtitle image.

I've included a picture of what the result would look like here: http://www.4shared.com/photo/rU-_1xI...subtitles.html

This is a screen dump of the procedure I've been using, utilizing VirturalDub, to create my desired effect (The Avengers).

Unfortunately, the subs are burnt into the video, using this method, and can't be turned on or off, as such.

Do you know of any software which can merge two graphic images, subtitle by subtitle, into one, to produce this effect?

I can see it as a near-impossible software task, considering that the timecodes for the images wouldn't match, exactly; and it's not always an image for image match.

In the "Catwoman" project I'm working on at the moment, for example, the English subs include 807 images, and the Thai subs include 821 images. And even in the cases of "matching" images... they don't really pop on and off at exactly the same frame...

I've included the subtitle files here, if this can be of some use:

eng.sub http://www.4shared.com/file/c11HlUkg/Eng.html

eng.idx http://www.4shared.com/file/vComAq1P/eng.html

thai.sub http://www.4shared.com/file/sLgI13QG/Thai.html

thai.idx http://www.4shared.com/file/QQM8iCrL/thai.html

Maybe this is just wishful thinking...
Thanks for your advice and help!
Take care,
Pawpcorn

[Pawpcorn]

Originally Posted by r0lZ

Pawpcorn, I don't think what you want is possible with a one-click program. Merging the individual images of two different subtitles into single images, to build a single multi-language stream is technically difficult, as that would require that both streams have all subtitles perfectly synchronized. For example, if the first subtitle of your English stream starts at 0:00:01.000 and the corresponding subtitle of the Thai stream starts at 0:00:01.500, the program will have to decide what time code it must keep, or create a sub with English only from 0:00:01.000 to 0:00:01.500, and then add the multilingual subtitle at 0:00:01.500. Also, the program will have to move one subtitle up or down if both are at the same place in the original streams. AFAIK, no program can do that.

Perhaps you can try to convert both subs to BMP files, and merge them yourself with a batch image processing program like ImageMagick, then rebuild the IDX/SUB steam. If there are the same numbers of subtitles in both streams, and if you can assume that the timings of one of them is suitable for the other one, that should work.

Anyway, I don't understand why you want to display two languages at the same time. What's the benefit? Is it for a multi-language audience?

Hi, ROlz! Thanks for your reply! I saw it just after posting my "additional info" post.
hahahaa.... yes!!! They are my Thai students, in my English class, and this method is seen as somewhat helpful in helping them learn English! (I'm a full-time English teacher, at an in-mall English (business) school, here in Bangkok)

Good query, I'm sure this need and application would seem illogical, if it weren't for my situation here... thanks for the input!

I've thought about what you suggested doing, and... as you can imagine, (I'm sure) it would be a rather labor-intensive task, especially since I'm trying to create one of these DVD movies PER week for my adult students (in my free time, NON-paid hahhahaha) and my plate is fairly full, as is

Oh... I could take care of the text placement (shift English down, shift Thai up), ahead of time, using BDSup2Sub, so there wouldn't be any visual conflict if the images were just "mixed" in an additive fashion.

Thanks again for your interest and advice!
Take care,
Pawpcorn

[r0lZ]

Indeed, with BDSup2Sub, you can "prepare" manually the two original streams so that the subtitles do not overlap. If that problem is solved, it should be possible to write a simple program or script to analyse the IDX files, and build the output IDX and BMPs accordingly with ImageMagick. The program could assume that, if a particular subpic is present in both streams, it must use the time code that comes first for the beginning, and the last one for the end. That should work in most cases. But there is an additional difficulty. For example, a long sentence can be split in two subs in one language, but kept in one single sub in the other stream. Or a new Thai sub can begin before the current English sub pops out. As you can see, that cases are more difficult to solve, but not impossible. Anyway, writing that program requires some programming skills. If I have some time, I may try to do it, but don't count on that!

[Pawpcorn]

Dear rOIZ, indeed, would be a challenging programming task!
At 63 now... my programming days seem behind me at this point in life; trying to teach English to Thais, whose native language provides a distinct disadvantage to learning English, is pretty challenging, in it's own right.

I USED to write programs, in the past, and if I were going to approach this, I'd just use brute force; create a table that duplicated reality of what got shown, and separate graphic images for every unique moment. Thus... this could result in 2 frames of Thai-only sub graphic, then 60 frames of mixed English and Thai, and maybe 3 more frames of English only sub graphic. I'd make NO effort in trying to "clean up" any of this, allow that there could be an abundance of sub images, and a large resultant file, but really, see no practical problem in huge idx/sub files, since we're really looking at a specialized situation here, and just being able to provide a working solution is success itself; it wouldn't need to be clever nor pretty

Hey, question for you, SassBot and rOIZ, I have a frequent task where I need to locate every subtitle that has been shifted to screen TOP (to avoid picture content at bottom) and manually shift it to screen BOTTOM (where I've provided a nice black area for my subtitles)

Currently, using BDSup2Sub, I have to click through EVERY SINGLE subtitle in the movie, to spot which subtitles are at screen TOP, and move them to screen BOTTOM, manually. For "The Avengers" I had to click the [>] button over 1600 times to look at every single subtitle, and... you know... how this really tests my patience and carpal tunnels...

Is there anyway to ask, and be told, which frames are at screen TOP, to avoid this manual process?

Thanks! Take care,
Pawpcorn

[r0lZ]

I don't think there is a way to automatically locate the top subtitles. But I have had to do a similar job. I've used "Move all captions". You should set the Crop Offset Y and enable the Crop Bars, then move all captions by a large Y value, for example 3000. All captions will be moved to the bottom of the screen (including the top ones), but not below the bottom crop bar. Then, accept the dialog, and use Edit -> Reset Crop Offset to freeze the subtitles at their current positions. Save.
That works fine, but with some limitations. It's not the baseline of the text that is taken into account when the caption is moved down and constrained by the crop bar, but the bottom-most non-transparent pixel. That means that the lines of the different captions are not perfectly aligned any more, but it's not visually important. Also, a caption with a single line of text is moved to the bottom, exactly like a caption with two lines. I would have preferred to move the captions with a single line less, so that it is aligned with the first line of the double-line captions. That should be possible by moving first all captions up, then down by a specific offset, but unfortunately, due to the bug I've reported above, you cannot move the captions up, and therefore it's currently impossible.
So, currently, it is possible to do more or less what you want, but not in a precise way.

[Chetwood]

Now I'm getting it too! If it wasn't for the Thai characters I'd said another way would be to convert both subs to ASS, change the vertical position of one sub and merge them manuelly. Or something...

[Pawpcorn]

Thanks, r0IZ, I'll give that method a try!
This always bugs me, because I could SWEAR, a couple years ago... using some forgotten piece of software on my laptop... I remember seeing statistics for how many (or which?) subtitles were TOP subtitles... Damned if I can remember what program that was...
Thanks again for the advice!
Take care,
Pawpcorn

[Pawpcorn]

Originally Posted by Chetwood

Now I'm getting it too! If it wasn't for the Thai characters I'd said another way would be to convert both subs to ASS, change the vertical position of one sub and merge them manuelly. Or something...

Yes! The Thai font makes things rather compicated, especially since I can't OCR the darned thing down to raw text...
We have so many tools to manipulate subtitles... But no way to truely mix 2 idx/sub subtitle images together, keeping the original timecodes of both...

[jd213]

Would it be possible to separately convert each idx/sub file to xml/png format with BDSup2Sub, put all of the png files into one directory, and then edit the xml file to include all of the png files/timecodes? I've done this a couple times when merging multiple subtitle tracks from anime DVDs into one subtitle track (for reauthoring the same title on Blu-ray). It seemed to work fine, although it only contained English subs and I didn't try converting back to idx/sub.

[Chetwood]

Originally Posted by Pawpcorn

Yes! The Thai font makes things rather compicated, especially since I can't OCR the darned thing down to raw text...

I wanted to give SubExtractor a try, however, I'm not signing up on 4shared for this. Also, why not compress all files into one rar, it should be small enough.

Originally Posted by jd213

I've done this a couple times when merging multiple subtitle tracks from anime DVDs

Yeah, I was a going to suggest checking anime fansubber forums who might know some tricks.

[Pawpcorn]

Originally Posted by Chetwood

I wanted to give SubExtractor a try, however, I'm not signing up on 4shared for this. Also, why not compress all files into one rar, it should be small enough.


Yeah, I was a going to suggest checking anime fansubber forums who might know some tricks.

I appreciate the idea about anime fansubber forums...

Oh, I tried to upload (attachments) the subs files as an .rar, but it was over 2MB, (the Thai sub alone was 1.2MB), and I'm limited to 200 kb here...

I looked at subextractor, but I'm pretty certain it doesn't include OCR software for the Thai Language.

The 4shared site is free and without a catch, AFAIK...

I could email them to you, if you PM your email address...

Take care,
Pawpcorn

[Pawpcorn]

Originally Posted by jd213

Would it be possible to separately convert each idx/sub file to xml/png format with BDSup2Sub, put all of the png files into one directory, and then edit the xml file to include all of the png files/timecodes? I've done this a couple times when merging multiple subtitle tracks from anime DVDs into one subtitle track (for reauthoring the same title on Blu-ray). It seemed to work fine, although it only contained English subs and I didn't try converting back to idx/sub.

Thanks for the advice, jd213! This method would learning a number of new (for me) procedures... I may give it a try, as time permits;
Thanks for the advice!
Take care,
Pawpcorn

[Tappen]

I'm the author of SubExtractor, and I did add support for Thai OCR in the most recent release (1028). I worked for several months with a Thai speaker to get the diacriticals and other combo characters OCR'd and ordered correctly.

[Chetwood]

Mh, so since I don't have any Thai font installed on my system, Subextractor does not show these chars to enter below the current OCR'ed char? Pawpcorn's Thai sub is hard to OCR, apparently whoever created it, messed with the palette.