[FFmpeg-soc] libavfilter audio work - qualification task

[S.N. Hemanth Meenakshisundaram]

On 04/09/2010 06:07 PM, S.N. Hemanth Meenakshisundaram wrote:
> On 04/09/2010 04:18 PM, Stefano Sabatini wrote:
>> On date Friday 2010-04-09 08:02:50 -0700, S.N. Hemanth 
>> Meenakshisundaram encoded:
>>> On 04/08/2010 09:54 PM, S.N. Hemanth Meenakshisundaram wrote:
> [...]
>>> Index: vf_drawtext.c
>>> ===================================================================
>>> --- vf_drawtext.c    (revision 0)
>>> +++ vf_drawtext.c    (revision 0)
>>> @@ -0,0 +1,589 @@
>>> +/*
>>> + * vf_drawtext.c: print text over the screen
>>> + 
>>> ****************************************************************************** 
>>>
>>> + * Options:
>>> + * -f<filename>     font filename (MANDATORY!!!)
>>> + * -s<pixel_size>   font size in pixels [default 16]
>>> + * -b               print background
>>> + * -o               outline glyphs (use the bg color)
>>> + * -x<pos>          x position (>= 0) [default 0]
>>> + * -y<pos>          y position (>= 0) [default 0]
>>> + * -t<text>         text to print (will be passed to strftime())
>>> + *                  MANDATORY: will be used even when -T is used.
>>> + *                  in this case, -t will be used if some error
>>> + *                  occurs
>>> + * -T<filename>     file with the text (re-read every frame)
>>> + * -c<#RRGGBB>      foreground color ('internet' way) [default 
>>> #ffffff]
>>> + * -C<#RRGGBB>      background color ('internet' way) [default 
>>> #000000]
>>> + *
>>> + 
>>> ****************************************************************************** 
>>>
>>> + * Features:
>>> + * - True Type, Type1 and others via FreeType2 library
>>> + * - Font kerning (better output)
>>> + * - Line Wrap (if the text doesn't fit, the next char go to the 
>>> next line)
>>> + * - Background box
>>> + * - Outline
>>> + 
>>> ****************************************************************************** 
>>>
>> User-documentation for filters should be put in doc/libavfilter.texi,
>> also I don't like this syntax, possibly we should try hard at giving a
>> consistent interface to the user, so we should either use
>> PARAM-1:PARAM-2:...:PARAM-N either use libavutil/parseutils.c (which
>> still depends on opt.c so I don't believe Michail won't accept this to
>> be included in the main repo, but that's not a problem for now).
> Am sorry for the confusion here. The syntax above is from the old 
> vhook drawtext file, the current syntax coded is :
>
> ffplay -vfilters 
> drawtext=f:<font_name>:t:<text>:T:<file_name>:s:<font_size>:x:<x_position>:y:<y_position>:c:<font_color>:C:<bg_color>:b:o 
> video.avi
>
> where all but font name and text are optional and order is not important.
>
> Is the syntax above acceptable? I will add all this to the texi file 
> if this is fine and start using parseutils.
>
> [...]
>
> I'll make changes based on the other comments and post the updated 
> vf_drawtext.
>
> Thanks
>
>
Some good news. I tested the filter and it works :) I will make the 
changes in coding style, argument parsing etc and then finish yadif.

ffplay -vfilters drawtext=f:FreeSerif.ttf:t:TestText invideo.mp4

Right now the font file needs to be in the ffmpeg folder. Will add 
FONTPATH to filename to fix this.