This all brings to mind the often-quoted: The difference between theory and practice is greater in practice than in theory.
I've pixel-peeped all the K-5IIs hi-rez images I could find, compared them to the same scene shot on fully-AA'ed machines where available and have come to the conclusion that the K-5IIs is entirely made of win. For my purposes shots from the K-5IIs exceed the useful resolution of the Canon 5d Mk II, a standard in the portrait and fashion shooting biz. Moire was also a non-issue. I expect the K-3 (?) 24 Mpx body to have much additional win poured into it. Martin Dopplebauer's painfully twisted knickers notwithstanding. On Thu, Feb 21, 2013 at 6:45 PM, Bryan Jacoby <[email protected]> wrote: > I think we're thinking about slightly different things. I agree with > what you say about visible moire, but I'm also thinking about > non-moire aliasing artifacts. These still represent (in my mind at > least) a corruption of the image, but they are not easily visible in > the image (without comparison to reality which people aren't usually > able to do when looking at a photo) and therefore not aesthetically > objectionable. One might say that if they aren't visible they don't > really matter, and that's fine. I'm just pointing out that some of > the perceived detail increase you get from omitting the AA filter > isn't accurate/real, but I realize that some may be willing to pay > that price for sharper looking images. > > If the image projected on the sensor (including lens imperfections, > diffraction, camera shake during the exposure, etc.) has the right > amount of blur to be Nyquist sampled by the pixel pitch then you won't > get aliasing, and adding an AA filter might unnecessarily decrease the > resolution by something like sqrt(2). So in that case there would be > no advantage to an AA filter and for all I know that may be a common > situation. Personally, I think I would rather give up a little > resolution to know that I won't be adding artifacts to the image (but > I've never used an AA filter-free camera so this is all theory), so I > hope I'll continue to have that choice as with the K-5 II and IIs. > > Nothing in here will be news to you, but I found this recently (via > PetaPixel) and thought it was pretty good (though admittedly it > approaches from the "all artifacts are bad even if you can't see them" > point of view): > http://www.martin-doppelbauer.de/foto/tippstricks/aliasfilter/index.html > > P.S. For the record, if anybody wants to give me a Leica with no AA > filter I will not turn it down! > > > On Thu, Feb 21, 2013 at 5:48 PM, Matthew Hunt <[email protected]> wrote: >> On Thu, Feb 21, 2013 at 5:30 PM, Bryan Jacoby <[email protected]> wrote: >> >>> I agree that it will be a practical problem very rarely given the >>> pixel pitch. But I think that is another way of saying that the >>> sensor is over-resolving what the lens, etc. can do. Which I think is >>> another way of saying what you're mostly getting is bigger files, not >>> more real detail in the images. >> >> Well, I don't think that's quite right... with good lenses at a sharp >> aperture and careful technique, I think you can make use of the sensor >> resolution and achieve high detail. But to provoke moire, you need a >> pattern with just the right spatial frequency in the same part of the >> image where you're achieving that sharpness. I think it's the >> combination of those two factors that makes it so rare in practice, >> rather than it just being a matter of the sensor always over-resolving >> the optics. >> >> -- >> PDML Pentax-Discuss Mail List >> [email protected] >> http://pdml.net/mailman/listinfo/pdml_pdml.net >> to UNSUBSCRIBE from the PDML, please visit the link directly above and >> follow the directions. > > -- > PDML Pentax-Discuss Mail List > [email protected] > http://pdml.net/mailman/listinfo/pdml_pdml.net > to UNSUBSCRIBE from the PDML, please visit the link directly above and follow > the directions. -- -bmw -- PDML Pentax-Discuss Mail List [email protected] http://pdml.net/mailman/listinfo/pdml_pdml.net to UNSUBSCRIBE from the PDML, please visit the link directly above and follow the directions.
