Hi again,
It seems that when resolving the
intersection between two element with a capfilter in between, the caps on the
first element get caught from the peer pad of the other pad of the element
itself. So, in the example stated:
GST_REGISTRY_FORK=no ./gst-launch videotestsrc pattern=1 !
ffmpegcolorspace zorder=56 ! video/x-raw-rgb-zorder,
framerate=(fraction)30/1, width=100,height=100 ! sdlvideosink
The resulting capslist to be linked to
sdlvidesink0:sink will be taken from ffmpegcolorspace0.sink ---> =
peer_pad=videotestsrc0.src
that obviously does not have a
‘zorder’; capability. It seems strange to me cause this way every
element connected to the ffmpegcolorspace should implement that capability,
that is in contrast to the usefulness of ffmpegcolorspace itself. Implementing
‘zorder’; cap on videotesrc too seems to work. Maybe I’m
missing something, could anyone shed some light on it?
Many thanks,
Andrew
|