[vlc-devel] Problem with an acces_demux module

Christophe Démaret vlc at cdemaret.com
Wed Jun 14 23:05:01 CEST 2006


Hi there.

Since my last mail got no answer, i will try a new one, cleaner.

I'm trying to build a module doing the following:

1) read data from an FPGA using some well tested functions
2) date them and put them in a packet

That's all. Uncompressed RGB 24 ( RGBR GBRG BRGB ... ). Should be easy huh ?
I would like this data to be streamed and before that, I need it to be 
stored locally using standard_sout (acces=file). The most logical choice 
for the local tests seemed to be AVI.

So the right FOURCC is RV24 isn't it ?

I've tried to use the screen.c as a base but it looks laike the date 
management is buggy (may have sth to do with my frame rate 1/10 fps ?)


Right now, i have an AVI with the GOOD data (checked with an hexadecimal 
editor, old style) so I must be very close ! But the AVI is not read by 
VLC: I only have middle grey 0x808080.

The main key is probably the message given by ts_mux:
 > [00000159] mux_ts private warning: packet with too strange dts
 > (dts=12603000240,old=12583001107,pcr=12583001107)

could someone give me a hint about this ? I just didn't understood what 
the code generating this meant :-/ what is the PCR ?

Anyway, the code is on ftp://ftp2.cdemaret.com/cdemaret/ml along with 
three -vvv logs and the resulting files. (MASSIVE_DEBUG enabled in the 
first one. by the way MASSIVE debug doesn't display all the data, only 
most of it. the data in the file ARE correct, trust me.)

The source data read are:
00 00 00 00 01 00 00 00 02 00 00 00 ... 00 00 00 FF 00 00 00 00 01 00 00 
01 01 00 00 ... (endianess warning: format is GH EF CD AB , A being the 
first digit of the 32 bit counter) This is for the full 800 px x 600 px 
buffer.

I would be very thankfully for any help I could get.

Christophe.

-- 
This is the vlc-devel mailing-list, see http://www.videolan.org/vlc/
To unsubscribe, please read http://developers.videolan.org/lists.html



More information about the vlc-devel mailing list