Welcome to Doom9's Forum, THE in-place to be for everyone interested in DVD conversion.

Before you start posting please read the forum rules. By posting to this forum you agree to abide by the rules.

 

Go Back   Doom9's Forum > General > Subtitles

Reply
 
Thread Tools Search this Thread Display Modes
Old 30th March 2006, 06:26   #381  |  Link
ai4spam
Programmer
 
ai4spam's Avatar
 
Join Date: Sep 2003
Posts: 378
Well, conceivably you could apply the filter "in real-time", while viewing the movie. You'd need a pretty powerful machine, as right now the interpolation is really slow, or saving the results of some sort of precomputation as a separate file with removal information, and apply that in real-time. But until then no, it's impossible without recompression.
ai4spam is offline   Reply With Quote
Old 1st April 2006, 10:52   #382  |  Link
Teebeeke
Registered User
 
Join Date: Jun 2005
Posts: 36
Dunno if this has been asked before, but is it possible to combine all my character matrixes into 1 big file ?
Teebeeke is offline   Reply With Quote
Old 1st April 2006, 18:57   #383  |  Link
Egh
Registered User
 
Join Date: Jun 2005
Posts: 630
Quote:
Originally Posted by FarCry
so this will be possible in the future? because like you said this filter requires to uncompress the video and compress it again. is there any other way to do it? apart from this filter. thanks
well if you want to re-distro that file then you'll need that. What is more, with VDub-only filter you will need to recompress to RGB32, adding more time for processing. If you recompress, it will be then converted into YV12 again.

But even that filter is very useful for e.g. removing hardsubbed subs from transports. So I'm looking forward to test it :P
Egh is offline   Reply With Quote
Old 2nd April 2006, 05:51   #384  |  Link
ai4spam
Programmer
 
ai4spam's Avatar
 
Join Date: Sep 2003
Posts: 378
@Teebeeke: Not really, but you can use the "search matrix" button every time you get an unknown char, if it finds it in an older matrix, it adds it to the current one.

@Egh: Believe me when I say it's VERY slow. Now, if someone figured out how to make VobSub remove subs instead of putting them in... . The way this filter works is to generate some information on where the missing parts (what's under the subtitles) should come from, then to apply that to remove the subs. If one had that info precomputed, it could be applied in real-time, while watching. I'll put up the filter soon, maybe someone wants to pick it up and make some software to output the info quickly from a .sub/.idx file or a bunch of bitmaps, and a filter to remove them in real-time while playing. Or... if other removal methods are faster (like gimp's fill texture filter), those would work as well.
Alas, I have no time to do any of this...
ai4spam is offline   Reply With Quote
Old 6th April 2006, 20:09   #385  |  Link
molitar
Registered User
 
Join Date: Jul 2003
Posts: 177
I would like to see Subrip be able to use ffdshow if installed. I am trying to rip a hard sub video but it has trouble identifying yellow subs on say blonde hair. If ffdshow was able to be used than I could adjust brightness and contrast making the subtitles have more contrast allowing the OCR to be able to detect them better. It would be alot of work I think to build the various hues, saturations, brightness, and contrast settings into the program itself.
molitar is offline   Reply With Quote
Old 6th April 2006, 22:05   #386  |  Link
ai4spam
Programmer
 
ai4spam's Avatar
 
Join Date: Sep 2003
Posts: 378
@molitar: Hmm, I never thought of that. Well, playing with the colors may help, but you'd have to tweak them mid-way, while processing the movie.
If you know what values you want, try using AviSynth filters inside your own .avs script.
ai4spam is offline   Reply With Quote
Old 7th April 2006, 00:30   #387  |  Link
Egh
Registered User
 
Join Date: Jun 2005
Posts: 630
Quote:
Originally Posted by molitar
I would like to see Subrip be able to use ffdshow if installed. I am trying to rip a hard sub video but it has trouble identifying yellow subs on say blonde hair.
You can actually use any filters from ffdshow in video by calling it thru avs wrapper plugin.
Egh is offline   Reply With Quote
Old 7th April 2006, 02:12   #388  |  Link
molitar
Registered User
 
Join Date: Jul 2003
Posts: 177
How would I call the filter to start ffdshow through an avs wrapper plugin? If It starts ffdshow than I should be able to tweak to my hearts content throught it's own interface.
molitar is offline   Reply With Quote
Old 7th April 2006, 02:57   #389  |  Link
Egh
Registered User
 
Join Date: Jun 2005
Posts: 630
Quote:
Originally Posted by molitar
How would I call the filter to start ffdshow through an avs wrapper plugin? If It starts ffdshow than I should be able to tweak to my hearts content throught it's own interface.
it's not working that way you'll be able to save presets in ffdshow and use then in avs (with tweaking of the settings).

I wonder, btw, how subrip itself plays the video. Does it use Directshow for that? Cause if it does then implementing ffdshow in the processing chain will be pretty straightforward.
Egh is offline   Reply With Quote
Old 7th April 2006, 15:03   #390  |  Link
ai4spam
Programmer
 
ai4spam's Avatar
 
Join Date: Sep 2003
Posts: 378
@molitar: I was suggesting some sort of separate AviSynth filter that changes the brightness/contrast/gamma. I'm sure there are plenty. The problem is, you'd have to use something else to figure out the values, then put them in.

@Egh: SubRip uses AviSynth and avifil32.dll, and does not care what's underneath (sometimes it's DirectX, sometimes it's not). It only gets the frames it needs and does processing on them. I guess some simple brightness/contrast/gamma sliders wouldn't be too bad to implement, I just don't have the time these days.
ai4spam is offline   Reply With Quote
Old 11th April 2006, 17:31   #391  |  Link
ericf
Registered User
 
Join Date: Sep 2005
Posts: 63
i becomes j in french sub. Why?

Hello. I have problems with i becoming j in the saved srt sub.
Example:
Ma¡s ¡l nous faudra¡t plus de temps.
Hmm... It doesn't show on the board. It does in Google and worldlingo.

I used both 1.40b and 1.50b.

I used your matrix but when I discovered the problem I edited out all i and j entries. No change.

If I do it manually (put in every character myself) the i and j are correctly displayed.

Where is the problem?

I should add that in the sub file box the i's and j's look correct when they have been ripped. it's somewhere in the saving process or... No, can't be. where in the matrix could the problem be?

Thank you.

ericf

Last edited by ericf; 11th April 2006 at 17:36.
ericf is offline   Reply With Quote
Old 12th April 2006, 15:01   #392  |  Link
ai4spam
Programmer
 
ai4spam's Avatar
 
Join Date: Sep 2003
Posts: 378
That looks more like an inverted exclamation sign to me, not a j. Try setting the OCR sensitivity to 1000.
ai4spam is offline   Reply With Quote
Old 14th April 2006, 09:33   #393  |  Link
ericf
Registered User
 
Join Date: Sep 2005
Posts: 63
Thanks. Ill try that.
ericf
ericf is offline   Reply With Quote
Old 14th April 2006, 10:44   #394  |  Link
ericf
Registered User
 
Join Date: Sep 2005
Posts: 63
Sorry. Sensitivity didn't work.
Also, this is what some of my subtitles from the french DVD of Dr Wai looks like:

275
00:24:48,314 --> 00:24:50,782
//s <i>s'rntrr- tuèrrnt dans</i> /r <i>désrrt.</i>

276
00:24:51,674 --> 00:24:54,507
<i>0n racnntr</i> qu'après /r massacrr,

277
00:24:54,714 --> 00:24:55,988
/r sab/r <i>drv/nt rnugr.</i>

278
00:24:56,434 --> 00:24:57,787
<i>D0nne- le-</i> m0¡ 0u meurs.

279
00:24:57,994 --> 00:25:00,713
Jama¡s!
S¡ tu l'0uvres, c'est la f¡n.

280
00:25:00,914 --> 00:25:02,063
Ta¡s-t0¡!

281
00:25:15,434 --> 00:25:16,469
<i>Le c0ffret</i> est là-bas!

So the sensitivity didn't work.
And r's are coming up instead of e's.

Is there any way of removing the cursive tags from the program? This time only some of the subs were interpreted as cursive while others in the same word wasn't.

I can't find that kind of options in the Advanced OCR options window.

Thank you.
ericf

EDIT:
Is there a way I can delete several characters in the character matrix? If not, could you make it work? It seems a lot of characters are read the wrong way and if I remove the bad characters in the matix I can put in good ones. I deleted the upside down esclamation marks and put in an i when it asked for it. No more problems with those.
Thank you.

Last edited by ericf; 14th April 2006 at 11:36.
ericf is offline   Reply With Quote
Old 14th April 2006, 16:21   #395  |  Link
ai4spam
Programmer
 
ai4spam's Avatar
 
Join Date: Sep 2003
Posts: 378
For r's instead of e's there must be a character that's wrong in the matrix.
About the italics, there is now a post-processing option that italicizes a whole word if more than half of its letters are italicized. There is no way to remove the tags other than search and replace in NotePad.
About removing multiple characters, I'll give it a thought as a feature for the next version.
If you still get this many errors with the sensitivity all the way up to 1000, you're better off starting a new char matrix for this particular subtitle.
ai4spam is offline   Reply With Quote
Old 15th April 2006, 19:56   #396  |  Link
ericf
Registered User
 
Join Date: Sep 2005
Posts: 63
Hello again. I removed all the upside down i (spanish I guess) and removed a character from the r's that said e. Now I have a working rip without the bad i's. The tags can be removed with notepad or Wordpad. I didn't understand the italicized or not option in the corrections window. I'll have a look at that. If you intend to use these files for a new DVD the tags are in the way though. They will sho up in the sub.

Thanks for thinking over the suggestion of being able to remove several characters at once in the character matrix.
ericf
ericf is offline   Reply With Quote
Old 15th April 2006, 21:34   #397  |  Link
Dreassica
Registered User
 
Join Date: May 2002
Posts: 384
When it comes to italizing I foudn that subrip tends to fuck up on ... 's, making them italized in sentences that obviously arent Italized.
Dreassica is offline   Reply With Quote
Old 16th April 2006, 02:29   #398  |  Link
ai4spam
Programmer
 
ai4spam's Avatar
 
Join Date: Sep 2003
Posts: 378
The option is "Format whole words only" in the post-processing window. It looks at each word, and if more than half its letters are italics, it makes the whole word italic, otherwise it removes the tags. Same with bold and underline, although nobody seems to use those nowadays.
ai4spam is offline   Reply With Quote
Old 19th April 2006, 11:19   #399  |  Link
jesus2099
Registered User
 
jesus2099's Avatar
 
Join Date: Oct 2005
Location: Paris
Posts: 29
The hard coded subripavi is great!
I needed it so much and used it.
It's very interesting.
As it is from VHS, the letters move often and I have to type many times the sames letters in the matrix
After having typed many times the same letters I see that the best guess is often correct so I put the auto use best-guess after one second global option (fill in after ...). And I correct the small errors with subtitle workshop afterwards. It saves lots of time!

I had a 2 problems though.
- I reached the maximum matrix size quickly thus being unable to add more letters to the matrix. What I don't understand is that your Matrix has 9345 characters. Mine only has 3546 but subrip tells me it's full. (file attached)
- I was unable to open a simple DV AVI (tried MS codec, Canopus readonly Codec, Mainconcept Demo Readonly Codec). I had to make a DivX out of it and I lost some time.

Oh yes and it would be useful to be able to put 0 second in
fill in after best guess option (currently, the minimum is 1 second)

Thanks
Attached Files
File Type: rar mexpress.rar (71.6 KB, 348 views)

Last edited by jesus2099; 19th April 2006 at 18:05.
jesus2099 is offline   Reply With Quote
Old 20th April 2006, 05:13   #400  |  Link
ai4spam
Programmer
 
ai4spam's Avatar
 
Join Date: Sep 2003
Posts: 378
@jesus2099:
The matrix constraint is something I'd have to look into, it's a bug in Delphi's memory management. Meanwhile, uncheck "add best guesses to chat matrix". Your letters will still be recognized correctly most of the time, yet the matrix won't fill that easily.
The 1 second minimum is there for a reason: in case you happen to be in front of the computer and notice a mistake, you can quickly type in the correct answer .
ai4spam is offline   Reply With Quote
Reply

Thread Tools Search this Thread
Search this Thread:

Advanced Search
Display Modes

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off

Forum Jump


All times are GMT +1. The time now is 15:15.


Powered by vBulletin® Version 3.8.11
Copyright ©2000 - 2019, vBulletin Solutions Inc.