"Dave Martindale" <davem_at_cs.ubc.ca> wrote in message
news:bv80lq$a2d$2_at_mughi.cs.ubc.ca...
> Alfred Molon <alfred_molon_at_yahooDELETE.com> writes:
>
> >What frequency does the AA filter cut off ? The pixel density or half
> >the pixel density (to prevent colour aliasing) ? I suspect it's the
> >pixel density, which means that colour aliasing still takes place and
> >reduces the effective resolution of a Bayer CCD.
>
> The AA filter would cut off at 0.5 cycles/pixel, which is the frequency
> necessary to prevent luminance aliasing. Yes, that still allows some
> colour aliasing, so fine colour detail can be wrong. Plus you can
> construct or find worst-case patterns that cause the camera to mistake
> luminance for colour or vice versa.
I've made a Webpage with some Bayer CFA experiments:
http://www.xs4all.nl/~bvdwolf/main/foto/bayer/bayer_cfa.htm .
Maybe it is of use for others.
> But despite the theoretical problems, real Bayer cameras take quite
> good images, particularly if you look at them at print size, not greatly
> enlarged to 100% on screen.
For Web publishing on a 1024x768 screen, anything over a 2MP image is
overkill. But down-scaling is also an art...
Bart
| 
