Re: Michael Reichmann reasoning for AA filters?

Kennedy McEwen wrote:
In article <blYRi.33588$c_1.17190@xxxxxxxxxxxxxxxxxxxxxxxxxx>, David J
Taylor <david-taylor@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx>
writes

I agree with you. However, the actual AA filter characteristics
differ between audio and digital cameras. In audio, brick-wall
characteristics are the aim, whereas in photography certain cameras
(Sigma, in particular) have a very weak AA filter, or even omit it,
"as the results look sharper"!

Not only do the filters differ between audio (or any bipolar signal)
and digital cameras they are required for different reasons - usually
there is only one cause of aliasing in digital audio, but there are
several (at least two) types of aliasing going on in digital cameras,
creating different levels of problem in the image. Some are
acceptable, others are not.

The "standard" aliasing issue occurs with digital cameras when the
image projected onto the detector contains spatial frequencies which
are higher than 1/(2p), where p is the pixel pitch. This results in
fine regular patterns being reproduced by the sensor as coarse
patterns - luminance aliasing.

However, in sensors based on the Bayer filter array, only half of the
sensors are green, while a quarter are red or blue. This means that
each colour has a much lower Nyquist limit than would occur without
the Bayer filter, 1.4x 2x and 2x for green, red an blue respectively.
This gives rise to chroma aliasing, where the different alias
frequency creates false colours. It also means that the AA filter
has to start cutting its MTF at a much lower spatial frequency than
it would if luminance aliasing was the only problem.

Take a look at David Littleboy's example of aliasing on the Canon 5D.
It isn't luminance aliasing, even though it is present, that makes the
effect objectionable, it is the presence of false colours created by
chroma aliasing that causes the problem. If only luminance aliasing
was present in that image, and chroma aliasing can't be eliminated
just be reducing it to grey after the event, I doubt that you or
David would ever notice it.

If the sensor doesn't produce chroma aliasing, and the Foveon approach
shouldn't, then there is much less need for an anti-aliasing filter
and you get sharper images as a side benefit. Luminance aliasing is
much less significant and certainly much less objectionable, even on
fabric, given the image resolutions we are working with these days.

Guess which I won't be buying!

A decision made for completely the wrong reason isn't a wrong
decision, just a bad decision, indistinguishable from (aliased as?) a
guess. ;-)

Thanks for the elaboration, Kennedy. I was only considering monochrome
sensors. We don't [yet] have colour in audio!

Cheers,
David


.

Relevant Pages

  • Re: Michael Reichmann reasoning for AA filters?
    ... In audio, brick-wall characteristics ... Not only do the filters differ between audio and digital cameras they are required for different reasons - usually there is only one cause of aliasing in digital audio, but there are several types of aliasing going on in digital cameras, creating different levels of problem in the image. ... However, in sensors based on the Bayer filter array, only half of the sensors are green, while a quarter are red or blue. ...
    (rec.photo.digital)
  • Re: Pentax K10D beats (sharpness, detail) Canon 40D?
    ... To get rid of all possible aliasing ... is the lens MTF around and above the Nyquist frequency). ... Perhaps it's easier to think in audio terms. ...
    (rec.photo.digital.slr-systems)
  • Re: Pentax K10D beats (sharpness, detail) Canon 40D?
    ... To get rid of all possible aliasing ... is the lens MTF around and above the Nyquist frequency). ... an audio system will have a very good low-pass filter ...
    (rec.photo.digital.slr-systems)
  • Re: Pentax K10D beats (sharpness, detail) Canon 40D?
    ... moire etc - and sensel grid aligment aliasing. ... filter will produce false detail at some stage. ... lesser quality lenses don't produce as much aliasing. ...
    (rec.photo.digital.slr-systems)
  • Re: The Korg MR-1 portable DSD recorder
    ... reduction requires a pre-reduction LPF to avoid aliasing. ... DSD doesn't get an exception. ... for a converter with PCM output, this filter has to be ... the PCM format is generated, which is in most cases the ...
    (rec.audio.pro)
Loading