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

in debug builds log all the support DXGI output formats
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: Francois Cartegnie <fcvlcdev@free.fr>

in the case of direct rendering picture_CopyPixels is used to transfer from
the decoder to the display and requires the same surface properties

the planes now have the same lines/pixels as the decoder, not just the visible ones

fixes #14531
Signed-off-by: Jean-Baptiste Kempf <jb@videolan.org>

do use the data in memory that wasn't actually read

Close #14783
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>

Since the bw adaptation can go really fast now,
we need to be sure to send at least 1 data segment
before switching to another.
Otherwise we'll have multiple moov bug and can
also never get any data outside of init segments.

Could be skipped as we stop parsing on moov
and index can live between moov and moof

avoids late failures

When needed (fmt->b_packetized == false), create a packetizer before creating a
decoder. The fmt_in of the decoder is copied either from the fmt_out of the
packetizer, or from the fmt of CreateDecoder.

This will allow a decoder module to be loaded/unloaded more than one time.

Use an es_format_t.

Replace it with a mutex per picture_sys_t. When doing direct rendering, the
picture_sys_t is refcounted between the vout and the decoder.