[MPlayer-users] speaking of vo... (was How I take a screenshot ??)

[Roberto Togni]

On 2004.08.21 20:36, ivanova wrote:
> On Sat, 21 Aug 2004 21:09:31 +0300
> Jan Knutar <jknutar at nic.fi> wrote:
> 
> > On Saturday 21 August 2004 19:51, The Wanderer wrote:
> >
> > > As a separate note, the one potential problem which I knew about
> but
> > > forgot to mention in the previous suggestion is that the
> > > grab-screenshot-with-separate-program method will not work with
> some -vo
> > > methods, because pausing MPlayer and switching window focus
> repeatedly
> > > will soon result in the video window blanking to blue (or, at
> least, it
> > > does for me).
> >
> > Even if it didn't, all you'd see on the screenshot would be the  
> blue
> anyways,
> > atleast with xvidix, xmga, xv, xover and most likely xvmc. Or some
> other
> > colour ;-)
> 
> While we are on the subject: What is the difference between the
> different vo
> modes? I got -vo x11 to work with scrot. What makes it different to
> all the
> other modes?
some vo (like x11) draws the image on the screen, while other vo (like  
xv) use an overlay.
What's the difference?

With x11 the image is converted to the display colorspace (usually  
rgb24) by MPlayer, and then it's drawn on the screen like any other  
window. So the image is really in screen memory area, and a snapshot  
program can grab it.

With xv MPlayer paint a solid colour rectangle on the screen (blue or  
purple, i don't remember exactly but that's not so important), and tell  
the video card to treat that color as a colorkey. Then MPlayer put the  
video frames to a special memory area in the video card, and the video  
card takes care of converting the image to the correct colorspace and  
to replace the solid colored rectangle with the image.
In this case, a screen grabber program gets the rectangle (that's  
phisicalli painted in video memory) and not the image (that does not  
exist, it's created on the fly while sending data to the monitor).

Advantages:
- source image is usually YV12 (like mpeg video), that's 12bit/pixel  
compared to 24bit/pixel fior RGB, so you need less memory bandwidth
- colorspace conversion (computationally expensive) is performed in  
hardware.

Basically is the same trick they use on tv for weather forecast: the  
speaker moves in front of a solid colored backgroung (usually blue),  
and then they replace the background with the weather map before  
broadcasting the image.

Ciao,
 Roberto