Re: Any thoughts /news on Foveon sensors?

In message <dqfgg1$r9j$1@xxxxxxxxxxxxxxx>, Dave Martindale wrote:
> Edmund <nomail@xxxxxxxxxxx> writes:
>
> >I have to admit I don't know what that Nyquist frequency means in
> >pictures. I do know the about Shannon/Nyquist in digital audio.
>
> It's almost the same. In digital audio, the independent variable is
> time, the sampling rate is in samples/second, and the Nyquist limit is
> the point at which the signal frequency is half the sample frequency.
> In other words, you need more than two samples per cycle to resolve
> a signal without aliasing.

At least exactly 2 samples per cycle in fact.
>
> In imaging, the independent variable is space, and the signal is 2D
> instead of 1D. So the sample rate is in samples per mm or samples per
> inch, and the Nyquist limit is the point at which the signal frequency
> is half the sample frequency. In other words, the wavelength is twice
> the pixel spacing at the Nyquist limit, or you need more than two pixels
> per cycle to resolve a signal without aliasing.

Thank for explaining, can I find more info on this on the Internet?
I like to know what kind of errors are made without these anti-aliasing.
In audio e.g. the SACD they just sample very high and don't use anti
aliasing anymore.
>
>
> >As we expected identical results to the SD9, and with good reason
> considering
> >the design of the sensor and the fact that the pixel count hasn't changed.
> >That's to a lack of a color filter the X3 sensor doesn't suffer from color
> >moire at the limits of resolution.
>
> The X3 sensor doesn't suffer from colour moire on B&W images, and that is
> a real advantage. But it *does* suffer from B&W moire on fine B&W patterns
> because it lacks an anti-aliasing filter, and that's just incorrect.
>
> >Because the X3 sensor doesn't employ an anti-alias (low pass / blur)
> filter it
> >continues to deliver detail past Nyquist. Taking the vertical resolution
> bar as
> >an example we can count all nine lines up to our 'absolute resolution'
> >measurement of around 1550 LPH, after this point lines begin to merge and
> by
> >2000 LPH we can count five obviously combined lines. In a real image this
> could
> >be the detail of leaves on a distant tree or bricks on a distant wall. At
> the
> >time of my SD9 review there was much debate over the 'validity' of this
> extra
> >detail. My opinion on this matter is that this detail is at least of
> >photographic merit, it is the correct color and represents detail that the
> >human eye although not able to distinguish perfectly would also see as a
> broken
> >texture, it's certainly better than the blurred area we would get from the
> >anti-alias filter of a Bayer sensor camera.
>
> That's a matter of taste. I, at least, would prefer that a camera
> render detail too fine to resolve as the average colour, not as some
> different detail that the camera would have been able to resolve had the
> image contained that. I never want 9 pickets in my picket fence imaged
> as 5 boards - ever. The eye blurs detail it can't resolve, and so does
> film. The aliased images from the Sigma camera just look unnatural to
> me.
>
> >I can understand the frequencies relative to the pixel spacing part,
> >but how about the bayer interpolation that has to guess the color of
> >the pixels?
> >Like here http://www.foveon.com/article.php?a=70
>
> Calling it "estimating" would be more accurate than "guessing".

You are right, estimating is a much better word, but hey
I am Dutch :-)
>
> But
> Bayer colour estimation works remarkably well in practice for most
> subjects on a modern camera; it requires some effort to find images
> where it will fail.
>
> >I just wonder if there is a theoretical advantage, the whole
> >concept make sense to me. Maybe it needs to be refined a bit.
>
> There is a theoretical advantage, if the 3 layers of the sensor have
> similar colour response and sensitivity to the filtered cells in a Bayer
> sensor. And in fact some video cameras use 3 CCDs and a prism colour
> separation block, and Foveon themselves once sold such a camera. This
> design is expensive but works very well. The Foveon X3 is an attempt to
> get the same advantages from a single flat sensor - but so far it has
> not worked very well.

That popped to my mind too, to use a prism for this purpose.
But that is already done, OK.
>
> Dave


.

Relevant Pages

  • Re: Any thoughts /news on Foveon sensors?
    ... >the design of the sensor and the fact that the pixel count hasn't changed. ... The X3 sensor doesn't suffer from colour moire on B&W images, ... >anti-alias filter of a Bayer sensor camera. ...
    (rec.photo.digital)
  • Re: Nikon D80/200 - Canon 30d
    ... than an el cheapo Point&Shoot camera with a tiny teeny-weeny CCD ... from a manufaturer that its CMOS sensor has less noise than a CCD ... just look at the images (not the ...
    (rec.photo.digital.slr-systems)
  • Re: 10MP compacts. I am not impressed
    ... sensor isn't a myth. ... course it will--but the measured difference in S/N between my camera ... 1200 electrons/sq micron, about average. ... the S/N in the allmost saturated blue sky of some 5D images I found ...
    (rec.photo.digital)
  • Re: lens/camera cleaning
    ... Can anyone recommend any kits/products for keeping the lenses and sensor ... free...I have heard someone mentioned air blowers? ... also to clean a filter than a lens because it is cheaper to replace etc. ... camera sensor. ...
    (uk.rec.photo.misc)
  • Re: First pics with the E500
    ... you cannot draw any conclusions about an 8 MP camera ... from these tiny 0.5 MP images. ... The noise at ISO 800 is severe. ... Thirds sensor size. ...
    (uk.rec.photo.misc)
Loading