By Date: <-- -->
By Thread: <-- -->

cx88 default contrast,volume,balance

Dnia 2006-03-09 15:52, Uzytkownik Mauro Carvalho Chehab napisal:
> Ian, Marcin and Michael,
> 
> We've postponed those patches for quite a long time! Due to lots of
> other stuff, those silly fixes were just staying. To return this
> subject, let me point the pending issues:
> 
> 1) Saturation
>    0xff seems to be oversaturated in most of the cases.
> 
>    Proposed solutions were:
> a) Limit maximum saturation at half of maximum chipset scale
> 
> if we limit it to 0x7f, most softwares will show this as 50% and seems
> ok, but we lose a hardware capability of improving saturation with bad
> signals.
> 
> b) Creating a non-linear control
> 
> This breaks applications like tvtime that reads the value before
> incrementing/decrementing it
> 
> I would add these two other alternatives:
> 
> c) App knows minimum, maximum and default values. So, application may
> show default as 50% even when default is less than it, and do
> interpolation at userspace
> 
> d) We can create a shadow register for it, ranging, let's say, from 0 to
> 512, where 0-0x3f is interpolated to 0-256 at writing the real register.
> On reading, we can just return the sreg value.
> 
> I am more to (c)
> 
> 2) Audio balancing
> 
> The way Marcin proposed the fix will produce real weird results if the
> user changes balance and volume, then changes back. The better is also
> to use shadow register for volume and balance and calculate db
> attenuation when writing the real registers.
> 
> To be practical on that and not postponing those patches, I suggest to
> threat (1) and (2) with separate patches. Maybe Marcin may propose such
> patches.

1) Current v4l-dvb code represents c) and I don't mind it. Anyway maybe
in few days I'll try to adopt my earlier nonlinear implementation of b+d
if someone thinks different.
Meantime, I'm finishing general cx88_fix_controls without my earlier
balance stuff, which fixes rest of controls.

2) I'm also just finishing testing fixed balance solution using shadow
register.

Regards,
Marcin Rudowski

--
video4linux-list mailing list
Unsubscribe mailto:video4linux-list-request (at) redhat.com?subject=unsubscribe
https://www.redhat.com/mailman/listinfo/video4linux-list