Luminous Landscape Forum

Equipment & Techniques => Digital Cameras & Shooting Techniques => Topic started by: kgross on June 15, 2008, 02:42:19 pm

Title: anti-aliasing filters?
Post by: kgross on June 15, 2008, 02:42:19 pm

Hi,

I've seen some discussions on this forum about anti-aliasing, but I haven't found an answer to my question and I'd appreciate any input.

Does anybody know at what frequency anti-aliasing filters are designed to cut at? Is the cut frequency just below or above the grey-scale Nyquist frequency or is it just above or below the bayer Nyquist frequency?

Thanks
Kevin

Title: anti-aliasing filters?
Post by: AJSJones on June 19, 2008, 04:20:51 pm

Quote

Hi,

I've seen some discussions on this forum about anti-aliasing, but I haven't found an answer to my question and I'd appreciate any input.

Does anybody know at what frequency anti-aliasing filters are designed to cut at? Is the cut frequency just below or above the grey-scale Nyquist frequency or is it just above or below the bayer Nyquist frequency?

Thanks
Kevin
[{POST_SNAPBACK}][/a] (http://index.php?act=findpost&pid=201766\")

Can't answer your question as posed, but you might be able to do the optics/maths.

 The following image came from a Canon document (my link expired) where the shift for each LPF was stated as " one pixel up and one pixel left".  The 4 point images would now each be at ~25% of the original photon flux and each point would go to four sensor elements, presumably ensuring that a sensor element of each color recorded some information about the original point source, as required to minimize color moire.  

(http://www.fototime.com/D4606C91BADB2C6/orig.jpg)

[a href=\"http://www.sensor-film.com/filter.html]Actual images[/url] of AA filters are consistent with Canon changing the thickness of the lithium niobate layers in the AA filter assembly in proportion to the pixel spacing to maintain the 1 pixel shift, but whether it is actually 1.0 pixel is unknown!  

Andy

Title: anti-aliasing filters?
Post by: kgross on June 20, 2008, 02:37:41 am

Andy,

Assuming that the birefringent crystal is the same material in the 5D and the 20D it might be possible to figure it out. A math problem for tomorrow when I'm not so sleepy.

Thanks for your input.
Kevin

Quote

Can't answer your question as posed, but you might be able to do the optics/maths.

 The following image came from a Canon document (my link expired) where the shift for each LPF was stated as " one pixel up and one pixel left".  The 4 point images would now each be at ~25% of the original photon flux and each point would go to four sensor elements, presumably ensuring that a sensor element of each color recorded some information about the original point source, as required to minimize color moire. 

(http://www.fototime.com/D4606C91BADB2C6/orig.jpg)

Actual images (http://www.sensor-film.com/filter.html) of AA filters are consistent with Canon changing the thickness of the lithium niobate layers in the AA filter assembly in proportion to the pixel spacing to maintain the 1 pixel shift, but whether it is actually 1.0 pixel is unknown! 

Andy
[a href=\"index.php?act=findpost&pid=202464\"][{POST_SNAPBACK}][/a]