[MEncoder-users] cropdetect cuts off bottom of subtitles

[DaveG]

On Saturday 21 January 2006 14:47, Raphael wrote:
> It occurs to me (after talking on IRC) that maybe cropdetect doesn't
> actually store the maximum lit area it ever finds. If this is the case,
> then it is a problem. In my case the subtitles are encoded into the film
> and so you never get the right settings from cropdetect if it
> recalculates per scene without storing the maximum (some subtitles have
> no "p"s and "g"s for example).  Is this what it is doing?

Yes.  The value output by cropdetect is the value for the current frame.  
That's why the man page (or docs??) specify that you test cropdetect on 
multiple parts of the video video source.

You could probably write a wrapper script to take the values produced and 
store/display the maximum values.

Alternatively, play through the video and find the parts where you do see 
lower case p and q etc and then use -ss hh:mm:ss with your -vf cropdetect to 
get the value for that segment. (or just right/up arrows keys to skip 
through)
-- 
Dave