[cpia] yuv to rgb conversion in kernel-space?

[Nicolas Roussel]

> Actually IMHO this is something teh v4l2 people should think
> about. They should provide such a library as part of a minimal v4l
> API.

I totally agree.

I think the kernel drivers should reflect exactly what the camera or
video board can do, and not what v4l expects from them. This means:

  - support only the native pixel encodings of the hardware (i.e. no
    kernel space encoding conversion), even if they are vendor
    specific ones (e.g. compressed modes)

  - provide hooks for all the get/set functions available

Then, a second level (user space) driver should be written to map v4l
or v4l2 features to the kernel driver. 

This approach would lead to more stable kernel drivers. In addition,
it would allow developers to use all the features of a particular
hardware and to experiment with various generic APIs (such as v4l,
v4l2, or others).

Nicolas

-- 
________________________________________________________________________

Nicolas Roussel                            mailto:nicolas.roussel@gmd.de
GMD - FIT.HCI, Schloss Birlinghoven      http://www-ihm.lri.fr/~roussel/
53754 St. Augustin, Germany       phone: +49 2241 14-2154 / fax: 14-2065
________________________________________________________________________

_______________________________________________
cpia mailing list  -  cpia@risc.uni-linz.ac.at
http://mailman.risc.uni-linz.ac.at/mailman/cgi-bin/listinfo/cpia