Re: [Bino-list] SubTitles?

[Joe]

Hi!

Dne 3/20/2011 1:58 PM, Martin Lambers napsal(a):
> On 17/03/11 01:16, Joe wrote:
> > On Tuesday 15 March 2011 21:15:23 Martin Lambers wrote:
> > > > Yes, we need a high level library that hides all those messy system
> > > > dependent details, and ideally also takes care of any character encoding
> > > > issues for us.
> > > >
> > > > I agree that we should avoid adding too many dependencies. Pango
> > > > actually does not seem to be too much of a problem in this regard: it
> > > > does depend on glib, but not on GTK, and on my system (Ubuntu 10.10)
> > > > libqt4gui also depends on glib (to allow using the glib mainloop), so
> > > > the number of additional dependencies would be limited to pango+cairo.
> > > > Still, there might be better choices than pango.
> > > >
> > > > Maybe we should just use Qt, but I really would like to have subtitles
> > > > working with Equalizer, too. From a quick look at the code, Equalizer
> > > > itself seems to only support crappy old-style X11 fonts on Linux, and
> > > > only in US-ASCII encoding.
> > > On second thought, there is probably no real advantage in using Pango
> > > over Qt. It looks like one would need to use QPainter on a QImage to
> > > draw the text, and then upload the QImage as a texture to the GPU. No
> > > additional libraries would be required.
> > >
> > > If we keep the subtitle rendering code a little bit separated from the
> > > rest, we can probably figure out a way to use this from within Equalizer
> > > later.
> > >
> > > Martin
> > I got rid of FTGL and reimplemented font rendering in QT. Everything looks
> > much cleaner now - system font selection dialog, all the fancy options and now
> > there is subtitles text encoding selection.
> >
> > I'm drawing subtitles into argb buffer and I managed to print it directly into
> > left/right textures, so finally I can implement depth of subtitles (paralax is
> > correct name?). However I need help with alpha blending of two textures. I'm
> > using  glTexSubImage right now to copy the subtitle image directly into
> > left/right texture, but that does not do alpha blending. I hope you will find
> > the time to implement proper blending/overlaying of those textures.
> >
> > https://github.com/cuchac/bino has fresh sources and the code should be ready
> > for testing.
> Many thanks for your work, it is highly appreciated!
>
> Your code does not yet work for me: I get OpenGL errors and video
> playback stops before the first subtitle is shown.
Can you please debug it a bit? I added only one opengl call 
"glTexSubImage2D" to copy subtitle bitmap to left/right texture. Is the 
opengl call causing a problem or there is a bug in subtitle 
processing/rendering? For me the example videos I posted are working 
well. Have you tried them?

> Nevertheless, I'm beginning to merge your work into Bino, since I'm sure
> the OpenGL problems can be fixed later.
>
> I would like to make a few conceptual changes if that's ok with you:
> - Keep subtitles and video frames separate in the media data. Just
> provide simple individual subtitle boxes from the media object, and do
> all buffering of subsequent subtitles (that may be stored in the same
> packet) in the media object. (This is similar to the way audio data is
> handled now.)
I hope I keep subtitles and video separated. I just added one pointer to 
video frame that says whether video frame has or has not a subtitle 
attached. Where do you see the mixture?
I agree with breaking subtitle list into subtitle_box that will hold 
only one subtitle and will be buffered in the same way as video or audio 
frames/blobs.
> - The subtitle reading/decoding is separated into threads just like
> video and audio reading/decoding.
I will redo your patch on top of my branch and unify processing of 
audio, vide and subtitles. I will also move font rendering from 
video_output to subtitles decoding thread. Is it what you want?
> - I have changed the decoding of subtitles based on my (limited)
> understanding of the FFmpeg subtitle API.
> See the attached patch (against current Bino in git) as a proposal. This
> code compiles, but it is *untested* (i.e. broken). Its purpose is just
> to give an idea of the proposed subtitle input and management;
> everything else is not yet included.
>
> Do you agree to these changes? Do you have other suggestions?
Is it possible to point me to same example how to alpha-blend two 
textures? I dont know how to replace glTexSubImage2D - see 
https://github.com/cuchac/bino/blob/master/src/video_output.cpp#L375. I 
have ARGB memory buffer with subtitles and I need to blend it with the 
video frame texture. glTexSubImage2D does just simple replacing, not 
blending.

Thanks,
Joe.