Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

If the packet starting at i_pagepos is larger than a page, i_input_position
will point to the last of those pages and we need to seek to the first one.
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

If the granulepos is negative we need to pagein the page nevertheless
because it might contain a packet that is continued on the next page. Also
we want to return the position of the first page containing the packet.
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

The edge detection filter uses the Sobel operator to detect and highlight
edges in a frame. In order to successfully detect an edge, the image must
first be converted to gray scale (because Sobel is a gray scale operator) and
then have a slight Gaussian blur applied to it. We do this because Sobel is a
noisy operator and the Gaussian blur reduces this noise.
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Close #16484
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

Implements the interface to provide the VLC pictures
from vout mapped to GstBuffers, along with per video
plane stride/offsets info, to the upstream gstreamer
video decoder. If the stride/offsets info of the vlc
pictures (created by vout) matches the video decoder's
expectations, the decoder will directly decode into
the vout allocated pictures, enabling zero-copy. Else
it will continue to use it's own bufferpool
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

The setting vbi-position does not appear on global level. Thus the regarding
variable callback is never invoked and therefore can be removed.
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

This commit adds an option to the ZVBI settings to select the Teletext
Presentation Level which is used for decoding. Thus one can switch to a lower
level in case of problems or can compare the rendering of a page at different
levels. The default value is Level 3.5, which already has been used each time
for decoding here.
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>