[FFmpeg-soc] BFI Decoder

[Sisir Koppaka]

I sent this message earlier, but it hasn't appeared in the archives so
far...I think the near- 1MB size of the video is disqualifying it. The video
is now here: http://www.youtube.com/watch?v=Bl1b4NL06DU

---------- Forwarded message ----------
From: Sisir Koppaka <sisir.koppaka at gmail.com>
Date: Tue, Apr 1, 2008 at 3:48 PM
Subject: BFI Decoder
To: FFmpeg Google SoC list <ffmpeg-soc at mplayerhq.hu>


Hi,
I posted the demuxer and the log of shell output on the FFmpeg-devel mailing
list here  :
http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/2008-April/044944.html
The decoder is not yet ready for FFmpeg-devel but I think code suggestions
might come in handy at this stage. I've attached the decoder patch + a low
fps(sorry about that!) screencast of the video playing - maybe someone will
be better able to identify the problem here.
I think the problem might be because of this particular part of the
decompression:

2 : skip chain
      if length is zero, read two more bytes of the length
      if length still zero, finish the decompression

      leave (length) bytes of the output unchanged from the last frame

Right now, that's what the decoder does: just skip those bytes. But it may
not be remembering the previous frame.  Reimar recommended me to do what the
nuv.c decoder was doing, so, I'm going through that code now. This is one
issue.

Another issue is regarding 'data'...In Eli's patch, I saw that he was using
data[0], data[1] and data[2] for Y, U and V but the decoders I was referring
used only data[0]. So, that is another confusion.

Please have a look at the screencast, I hope you might be able to recognize
whether there is any issue other than these that I should be looking into.
-----------------
Sisir Koppaka
-------------- next part --------------
A non-text attachment was scrubbed...
Name: bfi_decoder_00.patch
Type: text/x-diff
Size: 8963 bytes
Desc: not available
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-soc/attachments/20080401/513f84db/attachment.patch>