PDA

View Full Version : Anti Aliasing Filters (LPF) removal thereof



Kym
25-09-2013, 8:50am
http://image-sensors-world.blogspot.com.au/2013/09/ir-on-aliasing-artifacts.html


Imaging Resource (http://www.imaging-resource.com/PRODS/olympus-e-m1/olympus-e-m1A.HTM) posted an interesting note on the recent trend to remove optical low pass filters over the sensors:

"There's been a strong move in the camera industry lately to remove low-pass filters (aka anti-aliasing filters or LPFs) from cameras, in pursuit of greater image sharpness... At IR, we feel strongly that eliminating low-pass filters is a bad idea, and a mistake for the industry. While the vast majority of natural subjects aren't subject to aliasing and moiré issues, many man-made objects have the sort of regular patterns that trigger the problem."


http://www.imaging-resource.com/PRODS/olympus-e-m1/olympus-e-m1A.HTM

So lets debate the issue of the K-5ii vs the K-5iis or Nikon D800 vs D800e etc.

Is moire as big an issue or is the extra sharpness worth it?

Kym
25-09-2013, 8:53am
Also the science of it all https://en.wikipedia.org/wiki/Nyquist%E2%80%93Shannon_sampling_theorem#Application_to_multivariable_signals_and_images

arthurking83
25-09-2013, 9:57pm
I don't know about the K5 vs K5iis, but so far from what's been seen as actual results, moire is not as big an issue as it used to be.

From the chatter going around, the major difference between what we used to see(in the form of moire) and why it's not such an issue with the D7200 vs D5200 and D800 vs D800E .. is simply pixel pitch.

With the higher pixel density sensors we're now getting .... 24Mp D7200 and 36Mp D800E where diffraction is becoming an issue with smaller pixels, moire from weaker/none LPFs is also reducing as well.

I used to get moire from the D70s(being 6Mp) even tho it has a LPF(although a weak one).

I think it's something we will see more of in the future as more camera makers adopt higher pixel numbers on a given sensor size.

In 4000 exposures with the D800E, I'm yet to see a single instance of moire. I know it's possible, I've seen other people post images where you would expect to see moire from the scene.


Something I'm sure we will probably see more of in future is moire from older cameras(for those that still use them) where they probably never or rarely saw moire previously.
Lenses are just as much to blame for moire as is the removal or weakening of the LPF.
Afterall, it's the sharp rendering of finely pitched repeating patterns that causes it in the first place.
As lenses seem to be getting sharper with each new incarnation, as more people acquire these new sharper lenses, used on their older camera bodies, they'll probably discover higher instances of moire.

For me, I ummed and ahhed a lot about the choice of D800 or D800E, and it was a post by Rob Galbraith that convinced me that the moire issue that the D800E had where the D800 didn't have it, lead me to believe that moire was really a non issue.
He allowed downloading of images from both cameras to allow the individual to assess for themselves if the problem was in fact going to be a problem.
The moire on the D800E image was easily accounted for with minimal moire removal tool(a simple single click) .. and bingo ... I wasn't as apprehensive of the D800E.
Finally the choice between E or non E was made easy for me .. the D800E was in stock, the D800 wasn't! :p


As for the K5iis, vs the K5 of old ... the major difference between this pair and the two newer Nikons is simple. The K5iis still uses the same sensor as the K5 does, which obviously isn't optimised for use without an anti alias filter.
The D70s with it's occasional but annoying moire issues was also a case of inappropriate sensor for the anti alias filter setting Nikon used.
The D100(same sensor) didn't suffer the same moire issues because it used a stronger AA filter, and the D50 and D40 after the D70 were also built with the D70 issue in mind(stronger AA filter).


I don't think its a matter of do we - don't we or is it better - is it worse .. it's a simple matter of is it appropriate for the rest of the tech within the system.

Obviously Nikon thought it appropriate for the D800/D800E to turn them into twins, but not for the D4(which has no twin. D4 has a much lower pixel pitch at 16Mp.

An AA filter is simply a blurring attachment. The idea behind it is to just blur the image slightly so as to reduce the acute rendering of this repeatedly fine detail that causes moire on digital sensors.

Of course Fuji's response to the problem of moire is to make the sensor pixel random instead. They're X-Trans sensors have no AA filter either, but the problem is 'solved' because the sensors pixel aren't set in the same manner as a bayer sensor.
Sigmas Foveon sensor also has no AA filter as the system is different again(three separate layers of dyes over the same pixels).

Like I said, I think as the Megaring of pixels in new cameras increases, more manufacturers will use a similar route for the advantage it offers.
I can see this removal of the LPF/AA in Canon's future as well, if their long rumoured D800 killer is going to come to light, and they maintain the bayer type sensor.

Funnily the extra sharpness is not really as much as an advantage is you would initially think it would be.
Without any extra sharpening, yes the non AA image will look more sharp. Well it does on the D800E vs D800, and I haven't really seen much of the K5 vs K5iis to comment on that.
But, the D800 images(from those I've downloaded) can be sharpened up more(ie. more high pass or USM) to result in an image just as sharply rendered as the D800E image.
But the more you sharpen up the D800E image, it doesn't necessarily render any more useful detail in the image.

Basically the D800 image catches up with the same D800E image with careful use of PP!

The only advantage in the D800E(vs D800) comparison, that I've noted, is that where the D800 image may find an advantage with some USM/high pass, the D800E doesn't need it.
Saves some PP time.

- - - Updated - - -

Damned.. I should have read the Nyquist-Shannon article first(before posting).

Here's the quote that will almost certainly open the floodgates to many more cameras without LPF ...


Similar to one-dimensional discrete-time signals, images can also suffer from aliasing if the sampling resolution, or pixel density, is inadequate. For example, a digital photograph of a striped shirt with high frequencies (in other words, the distance between the stripes is small), can cause aliasing of the shirt when it is sampled by the camera's image sensor. The aliasing appears as a moiré pattern. The "solution" to higher sampling in the spatial domain for this case would be to move closer to the shirt, use a higher resolution sensor, or to optically blur the image before acquiring it with the sensor.

(I underlined the important part!)
As sensors become more high resolution, the AA filter is less necessary!
An invitation to join the party if ever there were one, I reckon.

Note that high resolution doesn't simply mean more pixels .. it means more pixels for a given sensor size.

your typical P&S or camera phone sensor is usually more high res than the D800/D800E is.

I don't think it's unrealistic to believe that phone cams don't have AA filters! .... don't know(actually don't care :p) .. does anybody know?