< ! --Digital window verification 001 -->

Why remove the anti-aliasing (AA) filter in the Nikon D800E?

Nikon 1 concept lens: moire example (taken with Fuji X100)

Moire example (taken with Fuji X100)

Why would Nikon release the D800E with the anti-aliasing (AA) filter removed? The simple answer is to produce sharper images with more details and better resolution. The AA filter removes information that cannot be recorded "correctly" by the sensor. It's basically an extra layer on top of the sensor that reduces the image quality in order to remove certain undesired artifact. If you remove the AA filter, the drawback is that you can get moire patterns in certain situations (see example above). Since several readers have asked me about this topic, here are few related links about anti-aliasing (AA) filters and moire (feel free to add more links in the comment section):

Leica M9, Fuji X100, X-pro1 and many medium format cameras have sensors without AA filters (or with very "weak" AA filters). With this move I believe Nikon is directly targeting medium format users by offering a 36MP sensor without the AA filter in a much smaller and affordable package.

Moire can be removed in post processing. Capture NX also has a moire reduction tool. Here is the above image processed in Lightroom:

Moire removed in post-processing

This entry was posted in Nikon D800 and tagged . Bookmark the permalink. Trackbacks are closed, but you can post a comment.
  • Andre

    Here is a screenshot of the Moire adjustment in Lightroom 4 Beta if anyone’s interested. Just put it on a old, hidden page on my site here: http://www.moonencreative.com.au/photo-showcase/

    I’m more happy about the added Highlights and Shadows sliders in the brush tool. But this thread isn’t about Lightroom 4…

    Andre

  • http://www.soundstagereview.com Fraggle

    I cannot wait for this announcement where we will all find out all the rumors are false and the d800 camera bodies are actually lunch boxes for kids to take to school!

    • BartyL

      Or simple plastic shells with Fuji disposable 35mm cameras inside.

  • Moe Jacknally

    Seems like I’ll be selling my Nikon equipment and finally getting a real camera:
    http://www.amazon.com/Fun-Saver-One-Time-Use-Camera-Flash/dp/B00083CZCK/ref=sr_1_1?ie=UTF8&qid=1328436910&sr=8-1

    Who’s with me?

    • Douglas Adams

      I guess some guys from Canon’s camp! They will spare some serious money and improve on auto-focus considerably!

  • kiran
    • PeterO

      Now that’s funny, I don’t care who you are.

    • Douglas Adams

      You’re alright Kiran, you’re alright….

  • Wolff

    So after reading all this, what would you say I should get -1 I never shoot video 2 I shoot mostly portraits under studio conditions 3 I have a d7000 to shoot with now.

    • Ray

      Are the people in your portraits wearing fabrics of any kind? D800.
      Are the people in your portraits naked or otherwise fabric not visible? D800E.

  • http://ronscubadiver.wordpress.com Ron Scubadiver

    Why does it cost more without the AA filter?

    • Larry Gray

      Great question
      And the follow-on question after reading the articles…

      Why is the projected cost difference greater than the price of a D3100 that presumably has it.

      In reality, though, it may he a good thing. Get the cheaper one with the AA filter and “suffer” it’s benefits.

    • Jason

      Wish D800 had higher ISO… then we can get large, clean images at 6400, which is noisy in d700.

      Besides higher camera price, minus-AA filter creates more post processing work, and not sure if the post process removal of AA moire, renders the image quality back to what it would have been with the AA filter.

    • craig

      Typically these cameras are manufactured in a line of ‘cells’ where steps are carried out. At some point in one of the cells, an AA filter is added. By creating an option to not have the AA filters requires altering that cell’s process to tell the assembly person to not add it, and then there must be a system to keep track of which cameras have the filter and which do not. Or, likely there will be some entire run of cameras that will not include the filter. These modifications to the manufacturing process have a real cost to them.

      Then there is also the cost of inventory. These cameras will sell slower than the others, meaning they’ll sit in inventory longer. This costs extra in overhead and things like taxes on inventory.

      End result: The specialty camera costs more even though it has 1 less part.

      Assuming this version of the camera exists… we’ll find out soon.

    • Q

      It’s better. Therefore Nikon can take a higher price. And at for now there is no competition.

    • Nikehabei

      Despite technically having “less” hardward (anti-aliasing removed) it is at a higher price point for marketing reasons. The 800E is aimed to grab some market share from medium format photographers (Hasselbad for example) and they are accustomed to much more expensive equipment. So, it is Nikon’s attempt at making higher profit margins on the back of medium format users’ buying habits.

  • Greg

    I think a few people have made the case that if you don’t know whether it’s better to remove or keep the AA filter, you’re better off keeping it. That’s probably right. I’m starting to think that’s also why Nikon put a premium on removing it– to force people to do enough research to justify the expense rather than pick the wrong camera and complain about the image quality. The take away from that is to learn what the AA filter does for you so you can make an informed choice. I find it strange that Nikon would offer a non-filtered version and I’ve been thinking through why they might do that. There have been a couple posts trying to explain what the filter’s doing– let me try to add another.

    Spatial frequency– you might see it specified in line pairs per mm or some such, but most photographers think of it as “detail” or “sharpness”. Every optical system suffers from distortion and loss of sharpness, but digital has a unique weakness that can result from excessive detail that originates from the spatial sampling process and in particular from the fact that it happens on a regular grid. Film has distortions too, but we’ve all noticed how the eye is much more sensitive to digital artifacts, mainly because we’re sensitive to lines.

    First, forget about color and bayer patterns, think in black and white. There is a limit to the sensor’s ability to resolve detail because you only have so many pixels across. The highest resolvable frequency is the Nyquist frequency and is usually said to be around half the sampling frequency (it gets a bit nuanced with a 2D sampling grid, but I don’t care about exact numbers here anyway). The easy way to think of it is to think about bright lines– if they’re far enough apart you’ll see two. If they fall within a pixel width you’ll see one.

    Graphically you can think of a plot where low frequencies (no detail) are on the left and higher frequencies (intricate detail) show up to the right. Zero frequency is basically the common illumination across the image (if you take a picture of a flat grey wall that is evenly illuminated, everything is at zero frequency– no detail). At some point, your sensor can’t resolve:

    | (N)
    | |
    | (A) | (B)
    L______________________|___________

    The region marked A is resolvable, the region marked B is not. The dividing line is the Nyquist limit.

    The problem though is that the light with very fine details in region B doesn’t disappear in the sensor, they *alias*. Just like a car tire on TV can appear to slow down because the frame rate is too slow, high spatial frequencies will start to appear as low frequencies.

    | (N)
    | |
    | (A) | (B)
    L_____________z__y__x__|__X__Y__Z_

    Spatial frequencies at the point marked X actually appear in the image as though they were at the point marked x. Same for Y and y and Z and z. If there is any true detail at points x, y or z it is forever corrupted by the addition of false detail from X, Y and Z.

    Thinking again of the bright lines– if they’re spaced half a pixel apart, you’ll get two per pixel and all the pixels will be the same brightness. If they’re a little less than half a pixel apart (more detailed) then every now and then you’ll get 3 lines inside a pixel instead of two and that pixel will be 50% brighter than the ones around it.

    An entry level engineering question is: “Why don’t we save money by sampling the signal first and filtering it digitally?” The answer is that there is no real way to untangle the real signals from the aliased signals just like there’s no real way to correct defocus. Once that third bright line is thrown in the pixel bucket, there’s no way to know if there was a third line, or if the usual two lines were brighter. You can post-process and create an image that matches what you think the real subject looked like but you, or the computer, are doing nothing more than making an educated guess. The sensor believes to the best of it’s ability to understand the world that the lens is painted with rainbow colors. The software is just painting them away because you tell it to, not because it has any information to know the truth.

    This aliasing will be happening in every picture, not just regular patterns, but artistically it may not matter as much in natural scenes. Moire stands out because the eye is sensitive to lines and curves. The tree trunks are going to be the wrong shape in a landscape shot too, but who would notice?

    So what does an anti-alias filter really do? It gets rid of the detail that’s too intricate to resolve. It eliminates, or at least highly attenuates, the B region above. No more alias distortion. Why would anyone think that’s a bad thing? Well, the filter isn’t perfect. No filter is. What the filter is doing is more like this:

    |100%______________
    | \ (C)|
    | (A) \ | (B)
    L____________________\_|___________

    Where as much of the detail in region A is passed through at 100% brightness as possible, and as much of the detail in region B is eliminated as possible. Physics prevents you from going straight from 100% to 0%, and forces you to make that transition gradually (steeply, for sure, but not vertical). The detail in region C is detail that would be resolvable with your fancy new 36MP sensor, but isn’t because of the filter, leading people in the forums to complain that Nikon knows they can get away with charging for a 36MP while only providing 30MP of detail because we’re all suckers.

    And the truth is if you’re taking pictures of subjects where the eye isn’t sensitive to distortion, you could probably use that extra detail without anyone knowing any different. The detail you would get by removing that filter probably isn’t *accurate* detail, but assuming you’re an artist rather than a scientist, accuracy to that level isn’t really the objective anyway (reference the “portrait touch-up” ads all over this site…)

    There are a few other balancing factors that make the AA filter either less desirable or irrelevant in some cases. It’s not desirable because it’s another element in the optical path, and anything we put there is undesirable unless it’s necessary. I’m sure the filter isn’t flat and perfect in passing 100% of region A detail, and will probably add some distortion there too. It probably steals a little light away too.

    What leads to it being irrelevant? Well, everything else between the subject and the sensor is filtering too. Most of the components are designed to resolve as much as possible but, you know, physics.

    For the people saying “36MP doesn’t need an anti-alias filter because it out-resolves the lens”, I think you’re pushing that argument a bit far. This is still a 5 micron pixel pitch. The Pentax Q is spaced at 1.5 microns– I’m not saying they’re resolving to that, but they’re certainly not getting 4MP images from their 12MP camera.

    That said, the lens does filter down the higher spatial frequencies. You might be getting something like this in the center:

    |100%_____________________
    | | \
    | (A) | (B)\
    L________________y__x__|__X__Y\_Z_

    So you aren’t getting aliasing from every hair on every fly. In the corners you might lose enough sharpness to avoid aliasing all together and the AA filter would then be making a bad thing worse– and if the subject you’re shooting isn’t bothered by a little aliasing in the center then it might be nice to reduce the excessive softness in the corners.

    |100%_________________
    | \ |
    | (A) \ (B)
    L______________________|\_X__Y__Z_

    The other thing we already know will kill detail is diffraction. Most DX cameras start to soften at f/8, and given that this 36MP FX sensor has the same pixel pitch as about a 16MP DX sensor, I suspect it’ll start to soften around there too. If you’re going to stop down to f/8 or smaller, that AA filter is probably not doing you much good and you’re losing detail for no good reason.

    I think the argument that landscape photography could do without the anti-alias filter is the right one– organic shapes, stopped down to the edge of diffraction for depth of field. I”d probably wait for someone else to try it first though– software is going to have a heck of a time correcting moire in a natural scene like that if it does show up.

    For anyone happy enough with a 30MP camera, I’d say keep the filter on. If you’re one of the people complaining that there’s too many pixels on this thing, and you’re going to take dark noisy images and then downsample them to kill the noise, keep the filter on. Downsampling is throwing away far more detail than the antialias filter is.

    Nikon isn’t stupid, if they’re offering two versions they’ve seen the value. Hopefully, in the coming weeks, we will too.

    • Greg

      Interesting… It threw away the extra whitespace from my ASCII art!

      • http://micahmedia.com Micah

        Thanks for taking the time to write this, but here’s a couple issues, most important: “…the AA filter would then be making a bad thing worse…”

        If the AA filter is throwing out (mostly) details that are beyond the sensor’s ability to accurately resolve–what’s the big deal? What is “resolved” is actually harmonics of real detail, and not the detail itself. The math to turn that into useable information just isn’t there yet, and it will never be 100% accurate anyway.

        As you point out, an AA filter doesn’t even completely remove the possibility of moire. I know–I use sharp lenses and often end up with moire!

        Diffraction isn’t a hard limit either, and has curves unique to each design. Shooting beyond diffraction limits isn’t a good argument for working sans AA filter, it’s an argument for different/better lens design!

    • Martin rock

      Thank you for this long and tedious padding waffle. Wikipedia offers a better analysis. Post a comment and not an article boring and erroneous …

      • Markus

        Martin rock, No Wikipedia doesn’t it in relation to the D800. If you’re not interested just don’t read and and if you are old enough to have saved enough money for the D800 read it, and try to understand it in relation to why Nikon decides to deliver 2 versions.

        • Martin rock

          ………….thx but Keep your advice I know exactly the difference between the two AA/WAA(see my comment on this) and yes wikipedia is an excellent article on the subject so well that dear Admin Peter mentioned in his post.

          • Ryan_T

            Greg, thank a lot. Your explanation is brillant.

      • http://codebeta.net Luis Murillo

        Boring? I hardly think so, quite informative actually and I’m glad I saw and read that comment

    • Neil Davies

      I think Greg’s comment was a valuable insight into the filter and how it effects our images. Not boring or erroneous and in fact nice to read something from someone who appears to know some science behind what we do, rather than moaning waffle which is all too common.

      • Martin Rock

        Do you like waffles? me especially with Nutella!
        Seriously, you have an interesting topic, a personal analysis then you publish it to your blog or forum but not as a comment …

    • http://www.maths-rattrapage.com Duncan Dimanche

      I’m really worried for the videos though… postprocessing moire seems a Bi°°°°

      • Martin Rock

        Like me…and even if Nikon has integrated a postprocessing system to prevent moiré, it will not be equivalent to a real AA filter. I already have the problem of moire with my D7000 that has a filter, I dare not imagine without it.

      • Jon Waisman

        I want the D800E for product shots, but if I want to use it for video, am I going to have a problem with the anti Antialiasing not being in the camera. Im getting into video more and more, but main job is products. Is the moire that debilitating.

    • David

      Thanks, Greg. Very helpful.

  • Greg

    Same post, hopefully with intact (if difficult to view) graphics:

    I think a few people have made the case that if you don’t know whether it’s better to remove or keep the AA filter, you’re better off keeping it. That’s probably right. I’m starting to think that’s also why Nikon put a premium on removing it– to force people to do enough research to justify the expense rather than pick the wrong camera and complain about the image quality. The take away from that is to learn what the AA filter does for you so you can make an informed choice. I find it strange that Nikon would offer a non-filtered version and I’ve been thinking through why they might do that. There have been a couple posts trying to explain what the filter’s doing– let me try to add another.

    Spatial frequency– you might see it specified in line pairs per mm or some such, but most photographers think of it as “detail” or “sharpness”. Every optical system suffers from distortion and loss of sharpness, but digital has a unique weakness that can result from excessive detail that originates from the spatial sampling process and in particular from the fact that it happens on a regular grid. Film has distortions too, but we’ve all noticed how the eye is much more sensitive to digital artifacts, mainly because we’re sensitive to lines.

    First, forget about color and bayer patterns, think in black and white. There is a limit to the sensor’s ability to resolve detail because you only have so many pixels across. The highest resolvable frequency is the Nyquist frequency and is usually said to be around half the sampling frequency (it gets a bit nuanced with a 2D sampling grid, but I don’t care about exact numbers here anyway). The easy way to think of it is to think about bright lines– if they’re far enough apart you’ll see two. If they fall within a pixel width you’ll see one.

    Graphically you can think of a plot where low frequencies (no detail) are on the left and higher frequencies (intricate detail) show up to the right. Zero frequency is basically the common illumination across the image (if you take a picture of a flat grey wall that is evenly illuminated, everything is at zero frequency– no detail). At some point, your sensor can’t resolve:

    | ……………………………………(N)………………….
    | ……………………………………..|………………………
    | …………………(A)………………|………… (B)……….
    L______________________|___________

    The region marked A is resolvable, the region marked B is not. The dividing line is the Nyquist limit.

    The problem though is that the light with very fine details in region B doesn’t disappear in the sensor, they *alias*. Just like a car tire on TV can appear to slow down because the frame rate is too slow, high spatial frequencies will start to appear as low frequencies.

    | …………………………………..(N)………………
    | …………………………………….|……………….
    | …………….(A)………………….|…… (B)……….
    L_____________z__y__x__|__X__Y__Z_

    Spatial frequencies at the point marked X actually appear in the image as though they were at the point marked x. Same for Y and y and Z and z. If there is any true detail at points x, y or z it is forever corrupted by the addition of false detail from X, Y and Z.

    Thinking again of the bright lines– if they’re spaced half a pixel apart, you’ll get two per pixel and all the pixels will be the same brightness. If they’re a little less than half a pixel apart (more detailed) then every now and then you’ll get 3 lines inside a pixel instead of two and that pixel will be 50% brighter than the ones around it.

    An entry level engineering question is: “Why don’t we save money by sampling the signal first and filtering it digitally?” The answer is that there is no real way to untangle the real signals from the aliased signals just like there’s no real way to correct defocus. Once that third bright line is thrown in the pixel bucket, there’s no way to know if there was a third line, or if the usual two lines were brighter. You can post-process and create an image that matches what you think the real subject looked like but you, or the computer, are doing nothing more than making an educated guess. The sensor believes to the best of it’s ability to understand the world that the lens is painted with rainbow colors. The software is just painting them away because you tell it to, not because it has any information to know the truth.

    This aliasing will be happening in every picture, not just regular patterns, but artistically it may not matter as much in natural scenes. Moire stands out because the eye is sensitive to lines and curves. The tree trunks are going to be the wrong shape in a landscape shot too, but who would notice?

    So what does an anti-alias filter really do? It gets rid of the detail that’s too intricate to resolve. It eliminates, or at least highly attenuates, the B region above. No more alias distortion. Why would anyone think that’s a bad thing? Well, the filter isn’t perfect. No filter is. What the filter is doing is more like this:

    |100%______________…………………………..
    | ………………………………\.(C)|…………………..
    | …………..(A)……………….\….|…….. (B)……….
    L____________________\_|___________

    Where as much of the detail in region A is passed through at 100% brightness as possible, and as much of the detail in region B is eliminated as possible. Physics prevents you from going straight from 100% to 0%, and forces you to make that transition gradually (steeply, for sure, but not vertical). The detail in region C is detail that would be resolvable with your fancy new 36MP sensor, but isn’t because of the filter, leading people in the forums to complain that Nikon knows they can get away with charging for a 36MP while only providing 30MP of detail because we’re all suckers.

    And the truth is if you’re taking pictures of subjects where the eye isn’t sensitive to distortion, you could probably use that extra detail without anyone knowing any different. The detail you would get by removing that filter probably isn’t *accurate* detail, but assuming you’re an artist rather than a scientist, accuracy to that level isn’t really the objective anyway (reference the “portrait touch-up” ads all over this site…)

    There are a few other balancing factors that make the AA filter either less desirable or irrelevant in some cases. It’s not desirable because it’s another element in the optical path, and anything we put there is undesirable unless it’s necessary. I’m sure the filter isn’t flat and perfect in passing 100% of region A detail, and will probably add some distortion there too. It probably steals a little light away too.

    What leads to it being irrelevant? Well, everything else between the subject and the sensor is filtering too. Most of the components are designed to resolve as much as possible but, you know, physics.

    For the people saying “36MP doesn’t need an anti-alias filter because it out-resolves the lens”, I think you’re pushing that argument a bit far. This is still a 5 micron pixel pitch. The Pentax Q is spaced at 1.5 microns– I’m not saying they’re resolving to that, but they’re certainly not getting 4MP images from their 12MP camera.

    That said, the lens does filter down the higher spatial frequencies. You might be getting something like this in the center:

    |100%_______________________……….
    | …………………………………….|…….. \………
    | ……………(A)………………….. |… (B)..\……..
    L________________y__x__|__X__Y\_Z_

    So you aren’t getting aliasing from every hair on every fly. In the corners you might lose enough sharpness to avoid aliasing all together and the AA filter would then be making a bad thing worse– and if the subject you’re shooting isn’t bothered by a little aliasing in the center then it might be nice to reduce the excessive softness in the corners.

    |100%_________________ …………………
    | ……………………………………\.|………………..
    |…………………….. (A)………….\….. (B)……….
    L______________________|\_X__Y__Z_

    The other thing we already know will kill detail is diffraction. Most DX cameras start to soften at f/8, and given that this 36MP FX sensor has the same pixel pitch as about a 16MP DX sensor, I suspect it’ll start to soften around there too. If you’re going to stop down to f/8 or smaller, that AA filter is probably not doing you much good and you’re losing detail for no good reason.

    I think the argument that landscape photography could do without the anti-alias filter is the right one– organic shapes, stopped down to the edge of diffraction for depth of field. I”d probably wait for someone else to try it first though– software is going to have a heck of a time correcting moire in a natural scene like that if it does show up.

    For anyone happy enough with a 30MP camera, I’d say keep the filter on. If you’re one of the people complaining that there’s too many pixels on this thing, and you’re going to take dark noisy images and then downsample them to kill the noise, keep the filter on. Downsampling is throwing away far more detail than the antialias filter is.

    Nikon isn’t stupid, if they’re offering two versions they’ve seen the value. Hopefully, in the coming weeks, we will too.

    • Thomas

      Greg, I hope you’ve saved your article for posterity somewhere (else).

      • KnightPhoto

        I have certainly saved Greg’s article.

        Thank you for taking the time to write this Greg, the x y z and X Y Z bit and everything else in your article combined with a quick re-read of Thom’s D3X review has very much stepped the topic to a new level of understanding for me.

    • Douglas Adams

      Greg,

      Thank you very much for this article. It has been long time since someone has put on such a worthy contribution. It explains just everything….and it was brilliantly put together. Thank you for it!

      Please ignore people like this Martin rock character. They’re still confused over fact that on Tuesday we won’t get D4x with 128 MP, D800 with 36 MP and ISO up to 1 million, and of course D400 with OFF (Over Full Frame) with 48 MP and sensor that is twice in size compared to FF.

      • Martin rock

        You do not know the new system and you based on the explanations of someone who does not know any more and that although most can be understood in a few words, Nikon certainly has other innovations to overcome the inconvenience associated with no filter …

    • QQMoar

      Thank god someone posted something useful on this site!

  • jodjac

    Ilike your explanation. Thanks for taking the time to filter out the noise.

    • http://www.maths-rattrapage.com Duncan Dimanche

      That gives you a good idea of why you should or should not keep the AA filter.
      http://www.maxmax.com/hot_rod_visible.htm

      Hope that helps a few…. it kind of help me… but still not convinced

      • Martin rock

        here’s a great explanation but sometimes that is based on systems that have not been designed to operate without AA filter. What I mean is it likely that Nikon could develop something to reduce the moiré effect.

      • Martin Rock

        What’s http://www.maths-rattrapage.com/ ?? french math ?

  • pabs

    Well, after anticipating the release of the D800 for quite a while, the option between the two versions is taking on ridiculous proportions. I admit that I’m getting frustrated with Nikons move toward niche cameras. I doubt that even professionals focusing on one type of photography use their camera exclusively in one area. ie, wedding or studio, landscape but not nights) Now the same seems to apply to the AA filter. Comments such as implying that if you don’t take pictures of repetitive patterns you likely don’t need it seem to miss the point. With the present arguments, one should have the D4 for low light and sports, the D800 for landsccape (but not landscape at night) and the D800E for that added sharpness unless your shooting feathers, tweed jackets, fences, etc. Obviously the real world tests are what will matter most but where is the high end, general purpose reasonably priced DSLR that offers a modest improvement in resolution, ISO, dynamic range and can still snap at a shutter spead of at least 6fps for sports or HDR so that a balanced advancement is achieved?

  • http://www.classecreativa.com Davide

    Thanks Greg for the nice explanation!

    That’s it, I think people that wished a 5dMII resolution and video capabilities for years and was bound to 12Mp now will have acces to a machine that will allow them to focus more and more on creativity, on light, on subjects and not on new equipment for a long time.
    I do remember photographers that were giving a little punch to their large format cameras to add little vibration and soul on a picture, I do remember people (like Newton) lighting the scene with car light… That’s the mark up to photography, the soul behind that. In other cases people use sinar and spend days composing their still life scene and that means they will know if they need D800E or just Sinar to reach the perfect reproduction. That’s another mark up, being a master of light as few in the world. That’s people that will seduce and net a relationship wit every subject and pull out amazing portraits, or see stories and tell them naturally… Another markup… All of them will live without the fear of AA or no AA. D800 or not… I only wish to become something like this and I’ll try to find that markup.

    A part for that point I really enjoyed Greg explanation because I know now what’s that and that I need AA filter and don’t have a new barrier to creativity because nobody will notice incredible resolution difference on my work (but will easily notice moire unfortunately). It was very interesting and informative, thanks!

  • mensamember

    brevity is the soul of wit, greg. nonetheless, i hope those long posts made you feel better about yourself.

    • Kragom

      “mensamember” says it all…

  • http://www.ForemostFotography.com Edward Vullo

    Call me an old dog that you can’t teach new tricks too, but I have a real problem with both Nikon and Canon following nonsensical trends that are better left to the consumer and prosumer segments of their market. I shoot both still photography and videography professionally, and the last thing I see a need for is a professional quality still camera like the upcoming D800 that has placed so much emphasis on the video aspect, or included video functions at all.

    I’ve been using the D700 since it was introduced and am still amazed at its high ISO yet low noise capability. I would have much preferred the D800 to have increased from 12MP to up 18 MP for the times that I need monster-size print capability, and improve the high ISO/low noise aspect even more while keeping the retail price more reasonable as a result of not including any of the video crap. Wedding shooters want to be able to cover a no-flash ceremony at decent shutter speeds and small apretures while roaming around at will without having to use a tripod and not have to worry about filtering excessive noise in post processing. My secondary camera is a Canon 7D that is absolutely horrible at high ISOs. I bring it along as a backup on assignments but hardly ever touch it. I have reluctantly used it for HD video purposes on occassion but ended up being embarrassed at the how clumsy the manual focusing and manual zoom ended up looking, not to mention the times when the wireless microphone was failing and I wasn’t aware of it because of the lack of a headphone jack.

    I’m aware that these shortcomings are being addressed with new and upcoming versions of DSLR video cameras, but there is still nothing like the smooth and steady servo motor zoom action that’s activated by a handle-mounted controller and searching-less auto focus of even a prosumer dedicated video camera, not to mention the built-in tilt and swivel LCD monitor.

    • QQMoar

      Resize 36MP to 18MP…done? What noise? :P

      • Thomas

        Resize Bayer-quadruple to one pixel and get the best ever 9MP image you’ve seen from a DSLR :)

        • QQMoar

          But it can’t go from 9MP to 36 megapickles :P

  • Douglas Adams

    Another thing that I found extremely important, and the one that is so easily overlooked, is the fact that we don’t know yet in which volume the D800E will be available. My best guess would be that 800Es will be available in very, very limited numbers. Furthermore, those will not be generally available to the general crowd, regardless if you are pro shooter or not.
    To my opinion, D800E is a very specific utility that will be used for specific tasks. And it will probably be landscape related.
    At this point, there are still some magazines that do not accept digital scans (think of that Arizona Highways or something like that) but just film. With this sensor, which will for sure have a notch better DR, noise and color capabilities than D7K, and with 36 MP instead of 16!? With AA filter removed, any argument for not accepting its digital captures will easily die on sight!

    I guess the main question is whether you are a person that needs that sensor, and if you’re not, you just don’t think about it any more.

    For years I’ve been working on upgrading audio equipment of my customers. Pre-amps, amps, D/A converters…Mainly it was about removing things that stood in a way of audio signal. All these things were put there with a reason, but mainly it was about circuit protections. Without them, you could easily burn your gear if you didn’t know what were you doing. These elements were killing a signal also, largely degrading the final audio output. Without them, you get cleaner sound, but with the cost of being extra cautious. I guess that’s what’s it’s all about here.

    • Martin rock

      D800e perhaps a higher-end model to d800. Just a thought…
      But you’re right, it might be for a specific use then it’ll be the first time for Nikon I think.

  • D800image
  • PoD

    I wish that D800 will has in camera ‘Auto Moire Reduction’.

    • Martin Rock

      It will be and also in that its price is justified.

  • Mats

    I’m having a hard time understanding all the comments here.
    I’m WAY behind on the technical knowledge that many of you have but then I’m not so much interested in that as of my final result.

    From all the sites and forums that I’ve been to, trying to gather info about AA or not to be able to decide which version to buy for myself one thing keeps coming up, that many of you are concerned about the moire. Afaik many photo editing tools remove all this w/o any problems, or am I missing something here?
    For ex this photo (on this site) taken without AA and then PP in LR:
    http://nikonrumors.com/2012/02/04/why-remove-the-anti-aliasing-aa-filter-in-the-nikon-d800e.aspx/#more-32586
    To me it looks like PP removes all of the moire, no problems. Again, what am I missing that so many are complaining about?

    For all you tech guys.
    I take mostly wildlife macro photos and at times I have to crop heavily too. Taking macro means using great apertures a lot of times and with high ISO. Macro of insects can contain a lot of repetitive patterns but hearing that PP, especially in the new LR4, of moire in apps like Capture NX and LR4 is of no concern I can only see benefits in using the non-AA D800. Am I on the right track here? What would you say about this?

    • http://micahmedia.com Micah

      “takes care of it” [moire] = gives you a tool to manually remove it with

      There’s no automation. If you have moire over 50% of your image, you’re going to be applying an adjustment brush to 50% of your image. And from what I’ve seen, extreme settings can introduce broad color artifacts and you’ll have to be precise about it.

      If you want to see what I’m talking about, try this: download the latest Light Room 4 demo and the samples from the M9 from imaging-resource.com

      Try it out with the image with the resolution charts. It works on most of the color moire, but there is still clearly a lot of moire and aliasing.

  • Been there guy

    After read these detailed posts, I come with an agreement that Nikon starts to make cameras that will have specialized purpose. If that’s the correct direction Nikon is taking, then, WHY the hell they mix the video and still? It jacks up the cost for everyone whether you like it or NOT!
    Now I will end up with a D800 with a video function that would never even being turned on. so much for specialized directions.
    They should just make a wedding & event camera, and leave the rest of us along!

    • http://www.ForemostFotography.com Edward Vullo

      I concur 100%. That’s exactly the point I was trying to make in my previous post.

    • jodjac

      I think that is precisely the reason they have included video. Maybe you should play with the video function a bit before completely dismissing it. The use of images is evolving, look at the iPad. There are new ways of communicating on the horizon. I think the wedding industry will be evolving too, to incorporate video into albums. Instead of dismissing it, I’m going to take a class an learn what I can about DSLR video and sound.

  • Sahaja

    Maybe Nikon figure some people will buy two cameras who wouldn’t have otherwise – one with the AA filter and one without.

  • Thomas

    Many people got their AA-filter removed by specialists. Why haggle with Nikon if they see a market and offer it for themselves, full warranty included?
    I can’t see why offering alternatives is such a bad thing. You simply have to decide. and the simplest way to come to a decision is to get what you normally would get: with AA-filter.
    image what hell would break loose if Nikon would sell the specialist version D800E for less than the normal (and safe) version D800. Luring people into buying something that can make some trouble. Ha!

  • http://tylerevert.com Tyler Evert

    What a shocker, Nikon is making *Pro cameras for pro photographers. I have a D3s and this is exactly what I have been waiting for to compliment it as my Studio/Landscape camera.
    Less speed, more megapixels are exactly what people who make a living at this and have multiple systems need.

  • Royster

    The D700 ,D3s and the D3x all had the AA filter and everyone says how great the IQ is on those cameras.
    Why now the D800 is coming out is the AA such a bad thing all of a sudden.OK without the filter the IQ may be better but that does not mean that with the filter the image IQ will be bad .
    It’s like saying the IQ of the D800 isn’t any good as MF cameras have much better image IQ

    • J0rge

      Exactly! Also, looking at the example on the maxmax web site, in order to see the difference in sharpness they had to magnify very small portions of the photo.

  • knightPhoto

    The reason to get Nikon to do your D800E is you get to keep the dust-shaker.

    If I’m not mistaken, MaxMax or IR conversions, you lose the dust shaker afterwards.

    I’m still toying with the idea of IR-converting my D300… project for another day for me.

  • http://www.jimprisching.com Chicago Photographer

    I may be skeptical about Nikon making two D800′s when they may have this technology available to them as discussed on this forum..http://nikonrumors.com/2011/05/21/nikon-patent-for-onoff-optical-low-pass-filter-olpf.aspx/..Seems to me if they apply this patent to the D800 , they can have it both ways and since they have the patent , may stop Canon from using the same technology. If in fact they have developed that technology, it makes no sense for them to announce two cameras, one with and one without an AA filter.

    • http://rkstephensphotography.smugmug.com Randy Stephens

      To I understand correctly that the D800 version with the AA Filter might have a stronger than desired AA Filter for still photography to insure the video if good? Therefore photos may be softer than we would like or are used to?

      • QQMoar

        It’s like saying 5Dmk II has soft image because it has video lol.

  • Martin Rock

    Wait for another 2 days and Nikon will answer your questions but I think the AA filter helps prevent moiré, as on a video taken with my D7000, the moiré effect is accentuated. I fear that without AntiAliasing filter the videos will be difficult to use but wait patiently Nikon doesn’t venture into compromises. I saw the difference with a Nikon d200 with and without AA filter. The image quality without AA filter is fantastic (Sharpness, contrast, …) but the moiré generated compromises my enthusiasm. I’m sure Nikon will surprise

  • tigrebleu

    Indeed, Nikon has some strong AA filters. Canon got it just right IMHO.

    Owning both a Pentax K10D and a K-7, with rather weak vertical AA filters, I’m used to rather sharp images when using decent lenses (at least when shooting RAW that is: Pentax JPEG files are on the soft side, both in default and with in-camera sharpness raised to a high level). I must say I love the very high detail level I can get out of my .PEF files at ISO 100 and 200.

    But on the other hand, I get more demosaicing artifacts, especially at high ISO. While most of these artifacts are hard or impossible to spot at ISO 200 and below unless you go for large size prints, these artifacts become visible and get quite annoying at ISO 800 and above. That’s when I’d rather have a stronger AA filter.

    But with 36 Mpix, I guess it will be hard to see these artifacts at all but the highest ISO or the larger print available today. So a D800 without AA filter makes sense, especially if Nikon aims to compete against the entry-level medium format digital backs and cameras.

    However, for people shooting at high ISO or who don’t want to remove possible moiré in PP, this could be a different situation. So the D800 with AA filter should be perfect in these occasions.

    So for maximum sharpness at all but the highest ISO, I’d go for a D800E. For high ISO performance, I’d stick to the D800.

    Just my two cents.

    Anyways, once the D800 becomes available, I will jump on the D700 while they’re still available (the D700 could benefit from good rebates, thanks to the introduction of its successor). With a D7000 for video and backup, I should have a great combo.

    • craig

      I would think any artifacts at high ISO are going to be influenced more by noise than the sharpness of the light coming through the filter. If anything, high ISO grain might actually mask apparent moire artifacts, though I have no examples to back this up. Would be interesting to see someone test these.

      In MF, backs without AA filters are the norm. Some systems that are marketed specifically at fashion shooters have included them though.

      Personally, I’ll likely buy the D800 without the filter if it exists. I’ll continue to use my D3 for high ISO work.

  • tedmok

    My D70 has a thin low-pass filter and produces moire, but it also produces the sharpest images with my 50mm f/1.4 out of all my Nikons. So I kind of like it for that.

  • http://stve stve

    I wonder if Nikon have been collaborating with Adode on the moire issue , the timing is very close to be just a coincidence.

  • Derek Matarangas

    I hate that they add video to these cameras. I will never use it either ever!! Just like many other readers have been saying.

    If you want video buy a video camera not a DSLR!

    It frustrates me when people cheap out and use 5d mrk II to make videos and not real video cameras.

    • Hsh

      Obviously you know nothing about videography and film making so… shut up.

    • neversink

      Why do you hate that people use these SLRs for video.. They are getting better. I hate war and killing. I couldn’t care less what people use to make videos…. The quality will be getting much better with video. Save your hatred and disgust for real things, not for people shooting video with these cameras.

      I think the video, which I don’t use on SLRs, might actually be feet on the D4. I can’t wait to try it, but please don’t hate me for it!!!!!

      • Edward Vullo

        I believe you’re over reacting a bit here. The point is that at the professional level of DSLR cameras, the consumer should be offered a choice of video versions and non-video versions for assumedly less money.

    • Mats

      That is a really ignorant comment.
      Why t h… would I wanna carry 2 big pieces of equipment around when I can carry one?
      Video mode adds SO much to the possibilities.
      Most of us don’t have as much money as you have so we can only buy one good equipment. A lot of times when I have my camera with me I see events where video documenting gives more than a photo. The sounds and action.
      If I’m out photographing in the wild there are situations when I want to document certain situations for all kinds of reasons, for.ex. scientific (how that species behaves in “that” situation).

      And, if you don’t know it, even top rated movie directors use dSLR’s movie functions as it gives them much more pleasing DoF than traditional movie cameras.

      I can go on and on.
      Just because you can imagine any reason doesn’t mean that the rest of us are that poor in our needs and imagination.

  • neversink

    Hmmmmm…. All this AA filter vs. anit AA filter arguments are making me think of selling my cameras and returning full time to film….

    With film one camera could do nearly everything…. You just needed to buy different films, not different camera bodies…

    As a pro photographer for more than 30 years, I was reluctant to switch to digital, and only did so in 2009 with trepidation. I yearn for the good old analog days of film……..

    • Zen-Tao

      Hurry up and purchase all Kodak stuff from your local dealer I’m sure we’ll attend to Photochemical imaging end. Have a look to the Academy Awards and rekon up how many films are been done on digital cámeras, don’t say about post. Buy a D800 as soon as it appears on the market

  • Zen-Tao

    All those figures about sampling an Niquist limits are quite interesting but I have some doubts. What happens if your sensor has more resolution or sharpness than your lense? 36 mpx exceeds the best Nikkor, Nikon or Zeiss lense capabilities so the sampling rate may be bigger than the image onto the CMOS. Thus the moire effect would be quite decreased so that users or manufacturers don’t need to attach an AA filter

  • q

    Look at the picture for the 300S. Amazing. But it is bogus. Look att the light pole. Very different exposure environment, see the shadows. Many hours between those pics! But im in to a 800E anyway…

  • D lee

    It’s completely feasible that the patent filed in may of last year is indicitive that Nikon is releasing a d800e with a switchable AA filter. Think about it. The timing of the patent in addition to the fact they are charging substantially more for the e version says to me that it is an added feature not just a variation. Pretty exciting if you ask me. So then the question becomes not if we want the 800e version, but how much more we are willing to pay to have the option of AA vs no AA filter.

  • http://www.chriswrightphotography.com Photographer Dundee

    Interesting. Watching avidly.

  • Beverly

    I shoot mainly wildlife and birds. I am wondering what the chances would be of moire showing up either in feathering or in fur … any thoughts?

  • RichardinJax

    There are a lot more issues to this than mentioned. The removal of the AA filter provides a modicum and that is one small modicum of increased detail but it wracks the camera very hard as well. The D800E is not recommended for use other than in controlled light situations and on a rock solid tripod ( see Nikon on this ) and I can assure folks that advise is spot on.

    I use the D800 and was recently on a wildlife shoot in the Amazon with a guy with a D800E. That camera was simply lousy in hand held uncontrolled lighting. You could not give me one. Morie is a trivial ( and correctable during imaging , pan the camera or after with software ) concern. The D800E is meant to cut into medium format studio guys that make slick images for magazine covers..not any other type of work.

  • Back to top