[MPlayer-users] Question about MPlayer, Theora and YCbCr444 format

[Giorgio]

Hello,

lately I've been experimenting with Theora 1.1 and I've written a
small program that reads some frames from ppm (or pgm or pgmyuv) files
and encode them in a Theora/Ogg stream. Now, Theora encoder requests
uncompressed YCbCr frames and lets you choose the pixel format among
three format:

1) 444 (no chroma decimation, the Cb and Cr chroma planes are full
width and full height)
2) 422 (chroma decimation by 2 in the X direction, the Cb and Cr
chroma planes are half the width of the luma plane, but full height)
3) 420 (chroma decimation by 2 in both the X and Y directions, the Cb
and Cr chroma planes are half the width and half the height of the
luma plane)

I have encoded black&white images (using only Y component, Cb=Cr=128)
and color images (using all three components) in both 444 and 420
format, and then I have tested the output file with MPlayer. These are
the results:

1) 420     B&W     OK
2) 420     COLOR  OK
3) 444     B&W     OK
4) 444     COLOR  NO

So I'm having problems when Mplayer plays color 444 files. It may be
my fault, a bug somewhere in my program, but before I make other tests
I was wondering if MPlayer can read Theora-encoded YCbCr444 files.

Here are two samples:
http://mywing.altervista.org/mplayer/otheo444.ogg
http://mywing.altervista.org/mplayer/otheo420.ogg

(basically, what I did is: read the frame, converted it from RGB 24bit
to YCbCr444, passed it to the encoder, and repeated that 100 times.)
Thank you.

Giorgio Vazzana