[popo]

Looks very good to me. As others have said, only the bit on the side of the tram stood out when looking specifically for problem areas. Nothing else in image looked bad in that sense.

[backcountryskier]

Near perfect in these photos.



Try shooting some clothing, and fabrics with a tight weave.

[popo]

[quote name='Brightcolours' timestamp='1290973344' post='4537']

Reality is: subjects are not perfect to be broken down into hard edged squares of the size determined by pixel pitch. Details in a scene do not have edges that coincide magically at borders between square pixels. At least in the computer graphics industry the last decade or two anti aliasing is universally used to increase the quality of graphics.[/quote]

Interesting you raise that, as graphic AA in general involves rendering additional sub-points at bigger contrast changes in order to provide better smoothing of that edge when presented at lower screen resolution. Back on camera sensors, I've been thinking for a long time the current pixel densities on bigger sensors (APS-C and bigger) are too low. If the pixel densities were made much higher, say 150MP in APS-C is achievable based on scaling up compact sensors, the AA filter would be a lot weaker in absolute terms. Also the higher density will help to a large extent with the colour resolution loss from bayer pattern by spreading the error. I think it would also improve (random) noise through oversampling. Of course there are many reasons why this route isn't viable for mainstream use today!

AA-filterless (no matter what kind of sensor, bayer CFA, sigma/foveon, gray scale or other) will always record false detail. And like some people love over-sharpened images, or failed HDR conversions, or other oddities, there will always be people that like AA-less cameras producing false detail. In that sense, it is a matter of taste. What is not a matter for discussion is that an AA-filter actually has a reason based on solid science, and is a valuable part in a camera to capture a scene as faithfully as possible.

There are two parts here. We've only been talking about "strength" of the AA filter, whereas the characteristic is more complex. In theory if you had a filter which cut perfectly at the Nyquist point then there would be no loss to the wanted image and no passing of unwanted aliased information. My "day job" is in sound, and the theory there is the same, but I'm not familiar with the specific characteristics of optical filters of this type. I have to assume that near-ideal filters are not cheaply realisable so there is a tradeoff between how much you cut and where.



The second part is how much of the signal will exist in the region to cause aliasing? In short, you need high contrast higher spatial frequency patterns for that risk. Back to the samples posted, there's only the side of tram which showed a significant case. The rest of it was insignificant or non existent to me. If this holds true in most cases, then why not leave out the AA filter? You can draw a kinda parallel with the SLT mirror ghosting issue here.

[Guest]

I'm with brightline here, there's good technical reason to introduce an AA filter. The "trouble" is that in real life scenes, exhibiting a great deal of "chaos," one wouldn't possibly be able to spot a lot of anti-aliasing artifacts because we have no idea what is supposed to be there and what is not in the first place. That's exactly why it's only clearly visible in places with super-regular structures like the side of the tram. I'm convinced there's a great deal of aliasing in all those pictures, but alas, how would one notice e.g. in the trees?



Interesting would be a side-by-side shot of something similar to FIG. 1 on the wikipedia page on aliasing here: http://en.wikipedia.org/wiki/Spatial_anti-aliasing

[deejjjaaaa]

[quote name='jenbenn' timestamp='1290986379' post='4547']

I guess there is reason why AA filterless cameras are only availabe as medium format pro bodies.

[/quote]

you forgot Sigma (& bayer P&S w/ high MPs on tiny sensors)

[deejjjaaaa]

[quote name='Brightcolours' timestamp='1290973344' post='4537']giving more detail (just hang around Sigma forums to notice that).

[/quote]

I never saw a person who was claiming that output from 4.6mp (not 8-10+ mp) AA'd bayer will be better that from 4.6mp AA'less foveon

[netrex]

I say yes.



Looking at those images makes me really want to remove the filter for my K-5 <img src='http://forum.photozone.de/public/style_emoticons/<#EMO_DIR#>/tongue.gif' class='bbc_emoticon' alt='' />

[sth]

I know that some of the detail is false detail, but that kind of artificial sharpness, at least judging from these few images, doesn't really look that bad. Also far less moiree artifacts than I expected (but probably worse with lenses that easily outresolve the sensor).



If I could choose, I'd probably go for a relatively weak AA filter (risking a certain degree of moiree) but wouldn't completely omit it.

[Sylvain]

I voted near perfect as I think these images are good examples, sharp and crisp.I like it. But I'm tempted to think "near perfect with this camera, this lens, ..." and I fear the generalization that all sensor/lens combo are equally well behaved when AA filter is removed.

[Tiz]

I do not know the original motif. But is this also moire? (The structure of the bricks seems strange to me.)