[AG-TECH] WebCam

[Douglas Kosovic]

Hi Piers,

> The UCL version of vic already supports V4l2 see:
> https://mediatools.cs.ucl.ac.uk/mbone/mmedia/wiki/VicDownload
> https://mediatools.cs.ucl.ac.uk/mbone/mmedia/browser/vic/trunk/video/grabber-v4l2.cpp

My code is more or less based on video/grabber-v4l2.cpp from SVN UCL
vic. The main issue the SVN code is that it only supports the YUYV pixel
format and not YUV420 common with webcams. Ideally I want to get it to
support all the pixel formats that the V4L version did.  

There were other issues with incorrectly initialised ioctl arguments
with the PWC V4L2 driver and capture size.

> There are still a number of drivers in the kernel (including some for
> webcams) that are still video4linux version one only - Now all video
> capture devices are listed as /dev/video - one has to open the devices
> and see which ones can be opened with V4L1 or V4L2 - thus the UCL
> version compiles both drivers.

I'm currently compiling both, but the AGTk sets the VIC_DEVICE env
variable to let AG vic know which capture device it should use.

So in grabber-v4l2.cpp I currently have:
  #ifdef HAVE_V4L2
          const char *myDev = getenv("VIC_DEVICE");

and grabber-video4linux.cpp I have a ifndef version of the above.

Another approach may need to be used instead of the VIC_DEVICE env
variable, e.g. using the so called vic device nickname.

> We are in the process having the AG features ported into the UCL
> version of vic - either by ANL or by UCL, we also plan to integrate
> work from NCHC vic - under our SUMOVER project - as mentioned by me at
> the recent AG retreat.
> 
> We would welcome any developer work on vic - we are keen to bring on
> board any active developers and potentially provide commit access to
> the SVN repository.

I was actually going to contact you guys about merging my Mac TCL/Tk
Aqua AG vic port (in ANL's CVS) to UCL vic, but haven't gotten around to
it yet.

Cheers,
Doug.