Quoting Jeroen Vreeken <[EMAIL PROTECTED]>:
>
> > So the only thing I would like to see added to the v4l spec is a way to
> > find out what sizes and formats a driver supports.
> > The library would simply ask a driver if it can scale or if it can't it
> > takes the nearest available format out a driver supplied list and scales
> > it itself.
>
> I took a quick look through the api document and would like to propose the
> following:
>
> Two new fields are to be added to VIDIOCGCAP, namely 'sizes' and 'formats'.
> If the hardware can scale between the max and min sizes it simply sets
> sizes to zero.
> (All bttv folks are happy now), if not it returns the number of sizes the
> driver supports.
Sometimes a device can deliver more than one format at different cost,
cost here being either CPU cycles or the frame rate or whatever. Maybe
we need to add this "preference" or "cost" factor too, so that the
library will be able to choose the most appropriate format.
As I mentioned earlier, sizes and formats often are mixed together - and
that is logical because the cheap DSP inside most cameras can not do
efficient processing at higher pixel counts, and then the device just
starts dumping raw RGB - sometimes even not demosaiced at all - into
the USB wire. So probably the device has to advertise not just a format
that it may support now and then, but a complete list of all formats,
image sizes, color depths, frame rates etc. (with wildcards, maybe):
enum Formats {
{ rgb24, r/g/b, 8-8-8, 3-15 fps, {176x144, 352x288}, ... },
{ rgb15, b/g/r, 5-5-5, 10-15 fps, {176x144}, ... }, ...
};
Some cameras have windowing capability where they can cut a portion
of a bigger picture and deliver it to the user space. This feature
must be accessible too.
Other cameras have the button (or several) on them. Currently V4L
has no means of using the button, short of marrying the webcam
driver with an input subsystem. This should be supported.
As I said, the new V4L interface is not an easy thing to design.
But 2.5.x is an appropriate place for that sort of hacking.
Thanks,
Dmitri
--
quit When the quit statement is read, the bc processor
is terminated, regardless of where the quit state-
ment is found. For example, "if (0 == 1) quit"
will cause bc to terminate.
(Seen in the manpage for "bc". Note the "if" statement's logic)
PGP signature
