[MPlayer-users] Mencoder and transparent subtitles, again

[The Wanderer]

D Richard Felker III wrote:

> On Mon, Apr 19, 2004 at 04:03:12PM -0400, The Wanderer wrote:
> 
>> Tell me about it. I haven't AFAIR even found a Linux OCR program
>> which can tell that a tiny circle (a period) is different from a
>> large circle (the lowercase letter o), even *with* training (the
>> interaction referred to above). Admittedly the shapes are, aside
>> from size, identical - but size should not be completely ignored!
> 
> Actually they're not at all similar. They're topologically
> different.
<snip>
> One has a hole in the middle. The other does not. So this OCR
> software is just stupid...

Actually, it depends on the font. In that particular font, the period is
a tiny circle - topologically identical to the o. (Bubblegum Crisis
DVDs, if you're interested.)

The software in question is gocr - I tracked it down after a suggested
get-SRT-from-VOBsub commandline on this mailing list turned out to
require it. It's much better than nothing, but still quite a pain.

What it needs is the ability to say both "yes, even though it looks like
just one contiguous block this part is actually a separate unit" and
"no, even though you think they're identical these two blocks mean
different things". This is, drifting from the topic.

-- 
       The Wanderer

Warning: Simply because I argue an issue does not mean I agree with any
side of it.

A government exists to serve its citizens, not to control them.