Antialiasing (AA) filter robs us of too much detail

There is only one thing I miss from the D70 and that is the weaker AA filter (and on a few occasioins the 1/500th flash synch speed. But the price to pay for this feature is too high). I don't mind the occasional moiré from a weaker filter, but I hate to be robbed of detail. Maybe the D70 AA filter was too weak, but the D300 filter is too strong. There must be a better compromise between these two cameras.

Do other people here feel that way? Do you wish Nikon should revise the AA filter in future dSLRs?
--
Thierry

TOF guy said:

Do other people here feel that way? Do you wish Nikon should revise the AA filter in future dSLRs?

Click to expand...

I agree.

Thom Hogan has spoken about this a few times. But he also seemed to think that Nikon would be reluctant to change. Apparently, they were bitten by moire in the past and may be loathe to undo what was seen as successful fix.

But I'm with you. I think we're paying too high a price in terms of detail to prevent a problem that is rarely a problem (at least for me). I think a weaker filter plus a little effort on some moire-removal software would be a win for everyone.

I also wonder if a weaker filter would enable camera makers to push DX-sized sensors up to 15 MP or so. That might reduce the pressure (temptation?) to go FX, at least for some users.

A D400 DX body with a weaker filter, 15 MP of real resolution, good high ISO performance etc. could be a very hot product indeed.

malch said:

A D400 DX body with a weaker filter, 15 MP of real resolution, good high ISO performance etc. could be a very hot product indeed.

Click to expand...

My wish would be for a sensor second to none in a Pro spec body i.e. built in grip. Greater MP's not necessary.

Considering the DX format does sell in large numbers, is this too much to ask for?

--
Jordan

JordanS said:

My wish would be for a sensor second to none in a Pro spec body i.e. built in grip. Greater MP's not necessary.

Considering the DX format does sell in large numbers, is this too much to ask for?

Click to expand...

I really don't think Nikon could make a dime off such a product.

Several reasons, not least of which is the fact that the prosumer market is chasing smaller bodies right now. A larger body would plop.

malch said:

JordanS said:

My wish would be for a sensor second to none in a Pro spec body i.e. built in grip. Greater MP's not necessary.

Considering the DX format does sell in large numbers, is this too much to ask for?

Click to expand...

I really don't think Nikon could make a dime off such a product.

Several reasons, not least of which is the fact that the prosumer market is chasing smaller bodies right now. A larger body would plop.

Click to expand...

I would therefore be an anomaly to the marketing hype. The current state of development is clearly on FX by Nikon as evidenced by the release of the D3s. This camera is on the cutting edge of what is available, where is the DX equal? There is none. Profitability is in the DX sales. The DX sales are a subsidy to FX development.

It is from this perspective I demand more from my chosen investment in Nikon DX.

--
Jordan

TOF guy said:

There must be a better compromise between these two cameras.

Click to expand...

though it's hardly a compromise, you can get the filter removed.

Has anyone here actually had the AA filter removed on a D300/300s? A weaker AA filter might be nice but none worries me because I haven't see any software to fix moire. I'd much rather have sharper detail and post process to correct for moire. All my images would have better detail and I could selectively fix the few that needed it.

--
-Dan Rode
http://rodephoto.com

arachnophilia said:

arachnophilia said:

There must be a better compromise between these two cameras.

Click to expand...

though it's hardly a compromise, you can get the filter removed.

Click to expand...

Yes, but that does not come cheap: ~ $400 cost or a quarter of the cost of a new D300(s). I'd rather see a weaker filter from the get go.

Besides Nikon probably refuse to fix a camera w/o the original final for a repair under warranty. Which means having the filter put back before sending the camera back etc.

Ideally, Nikon would design the camera so that the filter can be removed at will, and it would offer various AA filters of different strengths and / or no IR or UV filtration for those who want to do that kind of photography. Now that would be a nice feature.

--
Thierry

A full size DX camera would make sense for a number of users. But Nikon could have a problem to fit it in its line-up.

Say it costs $1,000 more than a D300s, that would put it price-wise against the D700. And probably most people would prefer the FF camera to the full-size DX body.

--
Thierry

TOF guy said:

Ideally, Nikon would design the camera so that the filter can be removed at will, and it would offer various AA filters of different strengths and / or no IR or UV filtration for those who want to do that kind of photography. Now that would be a nice feature.

Click to expand...

Yes, it would indeed.

I've used D70, Canon 5d - both with too weak AA filters, D80 and D300.

If you print, then aliasing from no AA filter really is a problem. Even if colour moire can be eliminated in raw processing, aliasing can't be dealt with so easily.

In my opinion, if you're looking so close on screen that you're complaining about loss of accutance from the D300 AA filter, then you're looking too close. On screen at 100% view equates for a 12mp D300 file to a print size of about 40-60 inches on the long side.

Now print an image (or a crop from it to give an equivalence of a 50 inch print), and if there's aliasing of edges, then it's likely to be easily seen as a serious image flaw, but a slight difference in accutance isn't. Even worse, at smaller print sizes (more normal sizes) aliasing stands out like appendages under a male mutt, yet any difference in accutance just isn't likely to be seen at all.

And to make matters even worse, because of poor resampling in most image viewing applications, most of us have now "mentally adjusted" to seeing aliasing of images scaled to screen size view, and as such it's very easy to completely miss aliasing on the image file - until you've wasted ink and paper.

In any case, using good quality lenses, D300 test resolution is very close to nyquist, so it's somewhat a fallacy that the AA filter loses resolution in the first place.

freddyNZ said:

In my opinion, if you're looking so close on screen that you're complaining about loss of accutance from the D300 AA filter, then you're looking too close.

Click to expand...

I fully disagree: Firstly we do sometimes crop images. Second, the D300 images are truly 12 MPs in the first place. They are 12 MP through a Bayer filter, the de-mosaicing of which involves interpolation and therefore reduces resolution, so your assumptions need correction. But the above is not that important. The main issue with the D300 filter is that one has to use larger radius during sharpening, which narrows the window between the minimum value where sharpening starts to be effective and the maximum value where negative artefacts (such as halos and noise more readily visible) start to be distracting.

freddyNZ said:

In any case, using good quality lenses, D300 test resolution is very close to nyquist,

Click to expand...

That statement can't even be close to reality. If it was close to Nyquist we would see moiré much more often than we do. Moiré is very rare in D300 images. In fact I don't even remember seeing a D300 image with that defect. What does that tell you? That the D300 AA filter is very conservative.

freddyNZ said:

so it's somewhat a fallacy that the AA filter loses resolution in the first place.

Click to expand...

I think the fallacy is to believe that the AA filter does not lose significant resolution.

--
Thierry

I think it would sell.

On almost any site that has a bird / wildlife photography forum, it's in many instances the most posted to section.

Most of us shoot DX.

I know I'd buy one and so would many that I know.
--
Jim Fenton
My Wildlife Images at:
http://www.pbase.com/soonipi1957

TOF guy said:

TOF guy said:

In any case, using good quality lenses, D300 test resolution is very close to nyquist,

Click to expand...

That statement can't even be close to reality. If it was close to Nyquist we would see moiré much more often than we do. Moiré is very rare in D300 images. In fact I don't even remember seeing a D300 image with that defect. What does that tell you? That the D300 AA filter is very conservative.

Click to expand...

No - that it's optimised very well indeed.

TOF guy said:

TOF guy said:

so it's somewhat a fallacy that the AA filter loses resolution in the first place.

Click to expand...

I think the fallacy is to believe that the AA filter does not lose significant resolution.

Click to expand...

Well even if you look on this site's reviews, there's samples of raw showing extinction resolution (without moire) of about or over 2800 (1400 lp/h). There's also lens reviews on this site using D300/s showing the same.

If you want more real resolution than the D300 offers (not the marginal and almost invisible increase shown by higher mp APS-C dslrs) then you need to go to Fx dslrs, D3x, 5DII etc, or medium format.

Even then when it comes to printing, the results of increase from 12-24 aren't large.

Removal of the AA filter results in serious image flaws. If you can live with those flaws, then good luck - I can't.

TOF guy said:

Yes, but that does not come cheap: ~ $400 cost or a quarter of the cost of a new D300(s). I'd rather see a weaker filter from the get go.

Click to expand...

isn't the d300's AA filter already weaker than the d200? i shoot a lot with my d200, and i've just learned to deal with it. extremely sharp glass helps a lot, but if you miss that focus even by like microns, it's gone.

i believe there once was a FF nikon-mount dSLR with no AA filter. made by kodak. probably very low mp count. this was years ago.

TOF guy said:

Ideally, Nikon would design the camera so that the filter can be removed at will, and it would offer various AA filters of different strengths and / or no IR or UV filtration for those who want to do that kind of photography. Now that would be a nice feature.

Click to expand...

well, that would likely be a complicated procedure, regardless. i doubt they can simplify it much more than they already have.

however, it would be nice if you could order them from nikon in different varieties. no AA filter, no UV filter, no IR filter, etc. i know fuji made some IR/UV s5's.

TOF guy said:

I fully disagree: Firstly we do sometimes crop images. Second, the D300 images are truly 12 MPs in the first place. They are 12 MP through a Bayer filter, the de-mosaicing of which involves interpolation and therefore reduces resolution, so your assumptions need correction.

Click to expand...

bayer interpolation does not decrease resolution in the way that you probably suspect. each photosite is essentially a monochromatic sensor with a colored filter over it. color resolution is about 1/4th of the actual mp value, but luminance resolution is the same . depending on the sensor and that demosaicing involved, there might be some averaging going on in the interpolation, but no resolution is lost. if you had a 12mp monochrome sensor, it would still be the same 12mp you'd see in the color version.

where resolution is lost is at the edges. sensor typically have some extra photosites surrounding the "effective" area of the sensor, which are only used to record the RGB values for using in interpolating the color of the pixels at the edges. this is why you see different actual and effective resolutions.

TOF guy said:

I think the fallacy is to believe that the AA filter does not lose significant resolution.

Click to expand...

speaking as a user of a d200, which has a VERY strong AA filter, it's generally not out of the range of some good old fashioned USM. it's just a little picky about what lenses it likes.

Please tell me which lenses your D200 prefers. I shoot with just a D200 (love it, and am waiting to update to Nikons next FF due anytime). I have noticed that a few select lenses give me incredible sharpness, detail, and an almost 3D depth where my photo looks like I am looking out the window, not at a photo--reach out and touch.
--

The day that cameras can do everything when set on automatic, will allow monkeys,

robots, and the untalented to take good pictures. Rebel now, set everything to manual!

The anti-aliasing filters on modern Nikon cameras are just about perfectly judged, in my opinion. They strike a good balance between avoiding the worst aliasing and maintaining crisp detail in typical situations.

I have a D60 (similar sensor to D200) and a D300S (similar to D300) and with both cameras, good lenses at optimum apertures easily and frequently produce luminance aliasing and colour moire. If you don't see this it's because your lenses or technique aren't good enough, or your raw converter is either not good enough to resolve the full detail in the file or automatically reduces moire.

The pixel-level softness difference between a D70 and a D300 is more attributable to higher demands on poor lenses (plenty of expensive lenses aren't terribly good, by the way) than an overly strong anti-aliasing filter in the D300. Resolving 12 million crisp pixels on a crop-sensor requires truly excellent lenses.

malch said:

I also wonder if a weaker filter would enable camera makers to push DX-sized sensors up to 15 MP or so.

Click to expand...

How would an AA filter prevent anyone from building a 15 MP sensor (apart from the fact that Canon is at 18 MP already...)?

Perhaps the lens is everything. I recently bought this old lens, and it knocks my socks off! 50mm f 1.8 AIS manual focus. Shot was handheld 1/60 sec., f 5.6.
The full size TIFF file looks even sharper.





--
'Nice photo, you must have a good camera!?'