Greetings!

I have pre-ordered the D800E and am now wondering if that is really the best option. I appreciate the thoughts and comments in this forum and have viewed other sites and watched numerous videos. (I won't share how many times I have enjoyed "Joyride"--more than once!!)

Honestly, I had never heard of Moire until a week ago and am now learning lots.

What I have not seen addressed are repetitive patterns in nature--in particular, Death Valley...I understand that Moire is caused by lines and dots and in fabric. One individual mentioned the possibility in "bug eyes" and flower centers with macro.

My preference is nature, but I shoot people too...all my glass is suited for FX as I have been waiting long for the D800 and planning accordingly.

So, who has dealt with moire in nature?

Thanks much!

Laura

Visit my Nikonians gallery.

Visit my Nikonians gallery.

Hi Laura,

It's worth noting that removing the AA filter is nothing new. Medium format cameras generally don't have one, and there are companies who will remove it from your current camera (sometimes called hot-rodding) as well.

The issue isn't just repeating patterns, it's repeating patterns near the same frequency as the repeating pattern of pixels on your sensor. Part of the thinking on the D800E (and medium format cameras in general) is that the pixel pitch is fine enough, that the repeating patterns which cause moiré would be so small the moiré would not be very visible.

Because companies have been modifying cameras like this for a while, you can find opinions on the modification around the internet. Thom Hogan has some good information on his site. I have seen reports of moiré showing up in nature shots in grass, bird feathers, and rocks. For that reason, I never opted to "hot rod" any of my cameras. You should be able to find information with Google, or go to http://www.maxmax.com/hot_rod_visible.htm and look at some images before and after AA filter removal.

That said, because Nikon itself is offering this configuration, and including software for moiré correction, that puts a different spin on it for me. It says that Nikon engineers have spent some time studying this and believe there's enough benefit to offer a specific model. I've heard it said that moiré is nearly impossible to remove in post-processing, but if a blur filter in your camera can do it, I don't understand why a blur filter applied to pre-bayer interpolation raw data can't do the same thing. OTOH it's debatable if a round of capture sharpening doesn't render the softer, AA-filtered files basically the same as the very sharp looking hot rod files. You can grab example files, play around, and make the call yourself.

Anyway, I'm a nature shooter and decided to pre-order the D800E, although for all the reason outlined about, I'm not sure it's a clear-cut decision. I'm honestly mostly worried that the E will be in shorter supply and therefore a much longer backorder.

Larry - a Bay Area Nikonian
My Nikonians gallery

www.tempered-light.com

Larry,

Thanks much for the reference. Good info on that site.

Think I am sticking with my first instinct and staying with the D800E.

Where I am now in my learning curve--matches post processing and I just want the best possible options with my new camera.

Big decisions, but such an exciting time!

We'll have to compare arrival time

Best wishes to you,

Laura

Visit my Nikonians gallery.

Visit my Nikonians gallery.

Larry,

>The issue isn't just repeating patterns, it's repeating patterns near the same frequency as the repeating pattern of pixels on your sensor.<

It also is visible colour speckles, i.e. false colour that is not present in the subject being shot.

>I've heard it said that moiré is nearly impossible to remove in post-processing, but if a blur filter in your camera can do it, I don't understand why a blur filter applied to pre-bayer interpolation raw data can't do the same thing.<

The filter in the camera blurs light before it is recorded by the sensor. Applying a digital filter to the RAW data in a camera without an AA filter will not help - the RAW data already represents moire.

Rob
www.robbuckle.co.uk

> I don't understand why a blur filter applied to pre-bayer
>interpolation raw data can't do the same thing.

I'll give this a shot and hopefully I won't mess up too badly.

I will start with the one dimensional case in time as the principles are the same but somewhat less complicated.
Suppose I had a sensor that took 10 evenly timed samples over a one second period (10 Hertz). Now suppose I sample a sine wave that oscillates at 2 cycles per second or 2 Hertz. That is, every half second, the signal repeats itself. At 10 samples per second, I would get 5 samples before the signal repeats itself. And so, the next 5 samples would just be a repeat of the first 5. If I were to sample a constant signal (one that does not change) then all 10 samples I take would be the same. Now, consider what happens when I sample a signal that repeats itself 10 times per second. If I am sampling at 10 times per second and the signal repeats 10 times per second, then I am sampling the signal at the same value every time and every sample would be the same. So, if I just look at the samples, there is no way to tell if they came from a 10 Hertz signal or a constant signal. Both signals produce 10 samples that are all equal to each other. In signal processing, we say the signal has been "aliased down".
It turns out that any signal with a frequency higher than half of the sampling rate will be "aliased down" to another frequency. For example, in our 10 Hertz sampling system, an 8 Hertz sine wave would produce the same samples as a 2 Hertz sine wave. (Wiki has a nice graphic under Aliasing which shows two sine waves that produce the same data when sampled in time).
In order to combat this, we put an analog Anti-Aliasing filter in front of the system. This filter blocks frequencies higher than half the sampling rate (5 Hertz in our example). Unfortunatley, we cannot build a perfect filter which would block everything above 5 Hertz and leave everything less than 5 Hertz untouched. Instead, a practical filter will begin to effect frequencies around say 4 Hertz in order to fully reject frequencies above 5.
In the spatial case, we don't really deal with nice sine waves but generally speaking we can associate higher detail or rapid variations with higher frequncies. As such, the AA filter is causing some loss of detail (blur) by rejecting frequencies (detail) that the sensor could otherwise record. With the AA filter removed, any rapid spatial variations will be aliased down and produce artifacts in the image. Moire is the result of high frequncy detail being aliased down to a lower frequency.

So, if you are still with me, the answer to your question.
Moire patterns result when high frequency spatial variations are "aliased down" to lower frequencies. Since a blur filter in post processing is just a filter that rejects high frequencies, it cannot necessarily get rid of the Moire because the Moire is now a lower frequency problem. It had to be done before the sampling. Back when you could tell the difference between 0 Hertz and 10 Hertz (or 8 Hertz and 2 hertz).

Hopefully I have not confused more than helped.

Brian

Thank you Brian, that is very interesting!

Laura, the only examples of Moire I've personally seen are all on man-made objects - fabric, building surfaces, roof tiles.

Also, the moire examples I've seen mostly were a pretty small amount of the frame by area (with one exception).

For travel, my subjects might include a fair amount of man-made objects, so for the moment I am leaning away from a D800E even though I'd like one. I've got about a year to decide before I get a D800, it will be very interesting to see how all this works out!

Best regards, SteveK

'A camera is an instrument that teaches people how to see without a camera.' -- Dorothea Lange
My Nikonians gallery
My Nikonians Blog

>> I don't understand why a blur filter applied to
>pre-bayer
>>interpolation raw data can't do the same thing.
>
>I'll give this a shot and hopefully I won't mess up too
>badly.

<snip>

This is of course the correct interpretation.

I believe the moiré correction in NX2 is applied to the color channel only, which isn't that difficult a thing to do. If the moiré exists in the luminosity channel then cloning it out is probably the only approach.


Larry - a Bay Area Nikonian
My Nikonians gallery

www.tempered-light.com

...

"Anyway, I'm a nature shooter and decided to pre-order the D800E, although for all the reason outlined about, I'm not sure it's a clear-cut decision. I'm honestly mostly worried that the E will be in shorter supply and therefore a much longer backorder."

Exactly my position. Also, I expect to keep my D700 for the occasional moire-prone subjects in my studio shoots of my wife's jewelry and as a 2nd body for wildlife and bird shoots with a less-than-pro-level Sigma Bigmos 500mm zoom.


Gator Bob in Gainesville FL
D700 & SB800 * D800E on order
Nikkors: *14-24 * 28-300 * PC-E 85mm *50mm 1.8
Tamron 90mm Macro

Visit my Nikonians gallery.

Laura,
I had the same concern and went through several hundred of my images to see what "worst case" natural patterns I could find. The following three were taken in Death Valley last year and I have posted them on some other forums to get opinions as to whether they might cause moire' with the D800E. No one thinks this type of pattern will cause problems with moire'.
Dave Jolley

Visit my Nikonians gallery.

Attachment #1, (jpg file)
Attachment #2, (jpg file)
Attachment #3, (jpg file)

Dave, you have shown some good candidates from your collection but I think the spacial frequency of the patterns in these is higher than the D800E sensor's problem areas with the exception of the upper left area in the first shot. If taken with a higher resolution sensor I think that area could possibly cause some moire patterns in that small part of the image, although it is difficult to say with the small size of the images posted here.

I have observed some moire patterns in nature with my D40 when shooting fields of saguaro cacti. Saguaros close to the camera have a spacial frequency of the ribs much lower than the threshold of the sensor and cacti in the distance have a frequency much higher than the sensor can resolve but where the spacial frequency of the pattern is about equal to the sensor I can see moire patterns. But a 6MP sensor has an entirely different threshold than the 36MP D800E would have. I have never seen any problem with my D7000 so far.

Dean
Phoenix, Arizona USA
Nikonians Team Member
Website: The Splendid Silence of Light

Recent Trips: Grand Canyon 2012 Glen Canyon 2012 West Clear Creek

Visit my Nikonians gallery.