[Kdenlive-devel] Kdenlive now handles <setSceneList>

Rolf Dubitzky dubitzky at pktw06.phy.tu-dresden.de
Sun Dec 1 16:40:27 UTC 2002


On Sunday 01 December 2002 04:48 pm, Jason Wood wrote:
> That behaviour is fine for Piave, in which case the task of setting up
> overlays will be left on the GUI side.
>
> By overlaying, I am talking about the task of placing one video on top of
> another, as would be done when performing alpha-blend, pic in pic, or any
> other transition-type effect that you can think of :-)

The simplest thing one can think o f is: 
 two video tracks and one effect track
 whenever there is a clip in both tracks you have to define an effect in the 
 effect track

Of course we don't want to limit ourselves to two tracks. But even with more 
tracks, the principle is the same. You always connect two inputs to form an 
output. (unary effects, i.e. effects which just operate on one input track 
are easy to display anyway)
I think one way to represent this in the graphical representation of the 
timeline is grouping. If you select a number of clips, where no more than two 
clips overlap each other at a time, you can call them a group and define 
effects between the two tracks. Then you can collaps the view and form a 
single 'groupclip' from this arangement of clips (or don't collaps it, if 
your screen is big enough). The resulting 'clip' can then be decorated with 
effects and/or connected via transitions to other clips. 
So the baseline of what I want to say is, whenever you want to define an 
operation to connect two clips (other than sequentially in time), you should 
make the two inputs and the operation form a "group". This could be indicated 
with a colored frame or so. I think this mandatory grouping will very much 
help to keep track of which track is connected via which operation to which 
other track. With the ability to collaps a group (which will maybe cover 
multiple tracks) to a single track this will also help to keep the timeline 
clean and not get crowded.

> But to be back on topic of when I was talking about 0xDEADBEEF - as soon as
> we start messing around with alpha channels, performing blue screen effects
> and the like, it would be useful if there was a way to tell that we have
> "covered" the entire picture or not with video. I think a test screen that
> would effectively be the bottom-most layer of every scene would be useful
> for this purpose.

That's exactly what I meant. And in piave it doesn't even add overhead, 
because effectively it is not the "bottommost layer" which would add an 
operation every time, but it would be the top node, connected via an 
OverlayOp to the rest of the tree where the OverlayOp by default does not 
move a single byte if the second input doesn't have transparent alpha channel 
(or any alpha at all)  ;-)

-- 
Cheers,
Rolf

***************************************************************
 Rolf Dubitzky  
 e-mail: Rolf.Dubitzky at Physik.TU-Dresden.de
 s-mail see http://hep.phy.tu-dresden.de/~dubitzky/
***************************************************************






More information about the Kdenlive mailing list