Hi Guennadi, Thanks for the comments. On Thu, Jul 14, 2011 at 07:56:10PM +0200, Guennadi Liakhovetski wrote: > On Thu, 14 Jul 2011, Sakari Ailus wrote: > > > Hi all, > > > > I was thinking about the sensor binning controls. > > What wrong with just doing S_FMT on the subdev pad? Binning does in fact > implement scaling. Nothing really. Supporting setting binning using S_FMT is fine. However, the interface does not express binning capabilities in any way. To effectively use binning settings one must know the capabilities. Binning is scaling but the choices are so coarse that the capabilities are a must. The capabilities could be found implicitly by trying out different formats and looking back at the result. That's still not quite trivial. If there would be a good way to enumerate the binning capabilities, combined with S_FMT it'd be close to perfect. > > I have a sensor which can do binning both horizontally and vertically, but > > the two are connected. So, the sensor supports e.g. 3x1 and 1x3 binning but > > not 3x3. > > > > However, most (I assume) sensors do not have dependencies between the two. > > The interface which would be provided to the user still should be able to > > tell what is supported, whether the two are independent or not. > > > > I have a few ideas how to achieve this. > > > > 1. Implement dependent binning as a menu control. The user will have an easy > > way to enumerate binning and select it. If horizontal and vertical binning > > factors are independent, two integer controls are provided. The downside is > > that there are two ways to do this, and integer to string and back > > conversions involved. > > > > 2. Menu control is used all the time. The benefit is that the user gets a > > single interface, but the downside is that if there are many possible > > binning factors both horizontally and vertically, the size of the menu grows > > large. Typically binning ends at 2 or 4, though. > > > > 3. Implement two integer controls. The user is responsible for selecting a > > valid configuration. A way to enumerate possible values would have to be > > implemented. One option would be an ioctl but I don't like the idea. > > > > Comments are welcome as always. > > > > Cheers, > > > > -- > > Sakari Ailus > > sakari.ailus@xxxxxx > > -- > > To unsubscribe from this list: send the line "unsubscribe linux-media" in > > the body of a message to majordomo@xxxxxxxxxxxxxxx > > More majordomo info at http://vger.kernel.org/majordomo-info.html > > > > --- > Guennadi Liakhovetski, Ph.D. > Freelance Open-Source Software Developer > http://www.open-technology.de/ -- Sakari Ailus sakari.ailus@xxxxxx -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
