Re: [xine-devel] New Video "Decoders"

[Michael R. <mr...@us...>]

Hi List,

> 	Anyway, the YUV decoder just does raw YVU9 right now until I find
> samples that are encoded with other formats. And the RGB decoder will
> be adapted to decode raw RGB from QT files as soon as I find some of
> those samples.

I was just wondering about all this colourspace conversion. I am using 
XShm video out (Xv unsupported). When I play a file with a video format 
that decodes to RGB, the video decoder will now convert it to some YUV 
format, send it to the video out and there, XShm will convert the YUV 
back to RGB. Right?

Wouldn't it be a cleaner approach, if we move the colourspace 
conversion out of the individual video decoder / video out plugins into 
the generic video out loop? I have something like this in mind:
* decoders would output frames in their native colourspace
  (RGB decoders would use IMGFMT_RGB, it's already there in video_out.h,
  only commented out)
* video out plugins would have a function for asking them the format
  they want
* video_out.c would run the conversion, if formats don't match
This is similar to what we are currently doing in audio out with sample 
rate and mono-stereo conversion.

Unfortunately I don't have enough knowledge about decoders and frame 
formats to implement this myself, but I think we should at least 
discuss that.

Michael

-- 
"When you say "I wrote a program that crashed Windows", people just
 stare at you blankly and say "Hey, I got those with the system, *for
 free*."
	-Linus Torvalds