Advertisement
SNAPCLOUDBack up every photoUnlimited RAW storage, 30 days freeTry freeSample ad
Advertisement
Advertisement

Sony FF: Remove or not remove the Anti Aliasing filter?

On left the Nikon D800. On right the Nikon D800E without the Anti Aliasing filter (Click on the image to enlarge)

Let’s not talk about the usual stuff like AF speed or high ISO tests. There are some other aspects the camera manufacturers are now competing for. Example, In the mirrorless segment we already saw the “miniaturization battle“. When it comes to video quality you often read about the “Rolling Shutter jello effect reduction“. But there is also another pretty recent evolution: The removal of the AA (Anti Aliasing) filter that usually protects you from the effect called “Moire”. Within 2012 we already had two new cameras getting rid of the AA filter: The Fuji X PRO 1 (Click here) and Nikon D800E (Click here). Here is a list of links you can check to learn more about it:

Aliasing explanation at Wikipedia (Click here).
Nikon D800 versus Nikon D800 image quality comparison at Pocket Lint (Click here).
Canon 5DmarkIII versus a modified markIII without AA filter at CanonWatch (Click here).
Fuji X PRO 1 versus the “older” Canon 5DmarkII at DSL-check (Click here).

From what I have seen so far the Sony 36 Megapixel sensor really gets a jump of resolution when there is no AA filter. But the Moire effect is really annoying in pictures with patterns. And this cannot be removed in post-production. So for me the Nikon D800E is a no way to go for normal use photography. But the Fuji X PRO 1 tells us another story. By modifying the RGB pattern on the sensor you can almost completely remove the moire issue. And I think that would a way to go for Sony. So here is my question for you:

[poll id=”98″]

—
Links to the mentioned cameras:
Nikon D800E at Amazon (Click here).
Fuji X PRO 1 at Amazon (Click here).
The Nikon D800 remains on top of the Amazon ranking (Click here).
While the NEX-7 surpassed the Fuji X PRO 1 (Click here to see the ranking).

Share this story
Community

Join the discussion

67 Responses

  1. There is a difference between can’t remove moire in post production and can’t “completely” remove it. You should get your fact straight Andrea. People have been doing just fine with C1 moire removal. And the new lightroom 4 has the moire brush now. And I don’t know how you define normal use when I have yet seen a real complain from Leica users.

    1. I’m sorry but I have to say this:

      After 15 years in this industry, I can unequivocally say that Leica users don’t complain because to do so would be blasphemy. It’s simply not allowed. Also, mentioning Leitz and Minolta in the same sentence may cause molting.

      1. Tell them that both the Hexar RF and Zeiss Ikon are better cameras than any of the M series, then break out the popcorn.

        1. I’m sorry, this is relevant to the article how? After 15 years in the industry, you should know that making “unequivocal” generalizations about how people select, use and describe their photographic equipment is a foolish thing to do.

          Back to topic: yes, moire can be removed to a large extent during post-production. Whether or not it is an issue for you depends on what you shoot. Nice that Nikon gives users the option.

          **This was intended for the comment above Carl’s.

          1. I am sorry, but could you finish what you started and demand relevance from every second comment on this site?

            After (insert number here) years in live, you should know about such words as “hyperbole”. If not, please consult dictionary. I can also recommend checking “irony” and “sarcasm”.

            If you do not have in your possession so called “sense of humor”, just check box here [ ].

          2. @ Robdel: I like making specific generalizations. It brings out the oxymorons.

            Kind of like anti-alias filters. We need to subtly blur your image in order to keep the quality at a consistent level. We hope you don’t mind!

            I love our industry.

          3. Make sure you follow this to its natural conclusion and criticise every other post that strikes you as being not relevant to the article or a generalisation. Don’t hold back now.

  2. I would like to see the mirror/pellicle gone most of all.

    Had the liveview of the 800/800E been up to the m4/3rds or NEX standards, I would have bought them already.

    Still very tempting with the 800 bombardment on this site. 🙂

  3. Well Sony plans to release multiple FF bodies, why not just offer some with, some without. They 24mp camera that is suppose to arrive in September might benefit from leaving the AA filter in place, as i’d imagine this camera is intended to be a more versatile camera in the lineup. With a camera like that in place, I think it would be smart for Sony to follow with a super high res model with no AA filter for applications where ultimate detail and resolution are required. It would be more of a special purpose camera likely used with specific intention…hopefully by users that understand the benefits/trade offs of ommiting the AA filter.

      1. Yes, you should only have ND and polar filters on a lens when its needed to capture the image planned.
        You should take them off.

        1. ND filter, agreed. Polarizer,not so much. Unless you see “walking around a seaside” as “planning”

          1. I hope the AA filter gets removed so we can see those sweet high iso cat pics that SonyA77 is just waiting to show us.

            Bahahaha

            Meatheads!

    1. Which you can do if you wish. It is removable you know, but don’t bother as it has fuck all effect on IQ.

  4. Following this blog and on occasion others, I can’t help but to think that the sensor manufactures including Sony are stuck between several older technologies and looking for a breakthrough of sorts.

    The prism, the mirror, and whatever this thing that Sony has placed between the device to capture the image and the image, is begging for change, as all have their problems it seems. I can’t help but believe that in part this is why nobody but Sony has jumped at replacing the standard DSLR. Everyone is trying to figure-out how to make the next generation of cameras actually better than the one before (Canon 5II vs. 5III). You would think that a camera made in 2012 would be much better than a camera made in 2008, right?

    I want the best image possible so I don’t want a mirror or a filter in my FF when I replace my a850’s and I like the ideal of a square sensor to eliminate the portrait to landscape silliness and I want it to fit in my hands as nicely as my a850 currently does.
    Can this happen?

    1. Yep. Remove your pellicle mirror and manual focus.

      Inconvenient? Yes. Am I going to do it when the need arises? Yes.

      I bet if I took a picture with and without a pellicle mirror and adjusted the exposure in Lightroom to match, nobody would notice the difference. Even fewer would care.

      1. I agree with the Lightroom post processing comment, but many don’t want to shoot RAW or do post.
        Having stated that, I don’t know why someone willing to pay the price in money and in the size and weight of a top of the line full-frame camera would want to do anything less than shoot RAW and spend the required time doing a little post work.

      2. Notice, no, but this is the internet, of course there are people who care. If a problem doesn’t exist, someone will make one.

      3. C, don’t bother. When the A55 was released somebody tried it with and without the mirror. It has also now been done on the A77. There is NO DIFFERENCE IN IQ.

        1. How about 5%?

          You should not shout about absence of difference so loud. Do not be such a fanboy.

          If there is some minor difference, then it is real. For me it is negligible. And absence of mirror flap vibrations should compensate it.

  5. This is a kind of bullshit topic. Firstly, when Sony is going to make a high-end full-frame DSLR/DSLT, of course it must offer bodies with and without AA filter. Secondly, Sony must have a FF camera on the market before even talking about yes/no AA filter. Thirdly, we just have to sit back and wait wat Sony does. Fourthly, the difference with or without AA will only be visible when printing huge sizes, or cropping and enlarging tiny bits of the image.

  6. No, cause Sony is using very weak AA filter on most cameras and if they keep doing it same way, their 36 mpix will have most likely almost as much details as D800E.

    Nikon and especially Canon are using very heavy AA filters, but Sony is using much more “light” handed. Its quite visible in for example A77.

    1. SONY PLEASE PUT 1080p OVERSAMPLING/DOWNSAMPLING in BIONZ like the panasonic GH2 / venus engine chip, PLEASE! please, please, please do this! YOU’RE FALLING BEHIND AGAIN.

  7. The SLT mirror already reduces resolution.

    The question therefore is: is the loss of resolution due to SLT mirror and AA filter cumulative? Or does the SLT mirror already reduce resolution so much that the removal of AA filter will not have any significant benefit?

    1. THE SLT ACTUALLY INCREASES LIGHT GATHERING ABILITIES OF THE CAMERA BY 400%. THIS WAS QUOTED IN SOME STUDY THAT I CAN’T SEEM TO FIND ANYMORE!!!

      IT WILL ALSO ELIMINATE CAMERA SHAKE AND ALLOW ISO OF UP TO AND BEYOND 1 MILLION.

        1. I’m confused, what does ISO have to do with SLT? Behind the mirror (translucent or otherwise) is a sensor. That is where ISO happens. Aside from the fixed 1/2 stop or whatever of lost light by the translucent mirror, SLT and ISO are unrelated technologies.

          1. You’re confused because you didn’t get the sarcasm.

            You’re quite right that ISO is merely the electrical gain the sensor runs at, but there are lots of misguided comments on the internet suggesting that a semi silvered mirror is somehow going to result in a precipitous drop in high ISO performance when the laws of physics don’t allow that possibility.

    1. > The D800E incorporates an optical filter with all the anti-aliasing properties removed in order to facilitate the sharpest images possible.

      Yes, it is a filter. No, it does not anti-alias. It most certainly is not an AA filter with no softening effect. I don’t know what else the filter does, perhaps it is integral to the Bayer color filter or it helps redirect the inter-pixel light onto pixels? Regardless, ‘softening’ is how you anti-alias. There are certainly other ways to avoid particular visible effects (moire by ‘random’ color mosaics), but…

      1. I am not technology expert, so please correct me if I am wrong. According to wikipedia: “the anti-aliasing filter is also known as an optical lowpass filter or blur filter or AA filter.”

        http://en.wikipedia.org/wiki/Anti-aliasing_filter

        And D800E has 2 lowpass filters. I think that still an AA filter, but is different from traditional one.

        http://www.nikonusa.com/en_US/IMG/Images/Learn-And-Explore/2012/Camera-Technology/D-SLR-Series/Moire-D800-D800E/Media/OLPF_schematic.pdf

        1. There’s two filters 180 degrees out of phase designed to cancel each other out. I assume they did it that way so they didn’t need to change the design of their quarter wave plate.

        2. Their second ‘low pass’ filter doesn’t make any sense. A spacial low-pass filter would not take two light paths and line them back up. I call marketing gimmick lingo misrepresenting the engineering work. As to the low-pass filter being an AA filter, sure. You avoid aliasing by filtering out frequencies that are too high for your sampling method. The low-pass filter does this by passing all the lower frequencies. Spatially we often refer to low-pass filtering as blur.

          Regardless, my statement was based on interpretation of their text.

          > The D800E incorporates an optical filter with all the anti-aliasing properties removed in order to facilitate the sharpest images possible.

          They specifically say that the anti-aliasing properties have been removed. In other words, aliasing can occur…

  8. For me, the crave is for higher image quality. Sony please do a FF with film colour gamut & true native 16 bit per colour channel.

    How about 2 ver of the FF? One uses an EXMOR (normal 12 bit) & the other a SuperHAD CCD2 (film gamut with 16bit)?

    Exmor sensor is 30MP & SuperHAD CCD2 is 22MP like 5D3 😀

  9. And what about the microlenses technology ? It seems that even with the Fuji X-Pro 1, the M-mount lenses don’t deliver good results as the Nex-5n… Then, we need microlenses technology of this FF is a mirorless camera !

  10. As more cameras with no AA filter come out, the anti moire applications in post processing with likely increase in effectiveness (heck I wonder if putting a bit of gaussian blur or lowering clarity would help). You can always take away most of what’s there, but you can’t add what wasn’t captured. Also, I wonder if a transluscet mirror might act as a very weak AA filter anyhow. At any rate, count me among those who would prefer having a no AA filter option.

    1. > You can always take away most of what’s there, but you can’t add what wasn’t captured.

      That’s the purpose of an AA filter. If you sample at 12khz you can’t capture a 20khz signal… heck, even if you have a 3khz signal present, the 20khz signal will destroy it. You must use a system with a 40khz bandwidth to filter the signal down to 6khz max frequency content before you can accurately sample at 12khz.

      The AA filter in cameras does this spacially. Go have a read about Nyquist. Sure, you might get a ‘sharp’ image without an AA filter… but it may or may not be accurate, it’s impossible to know once you’ve sampled it. :[

      I really don’t understand why the Foveon sensors aren’t catching on. Is it just an annoying patent issue? They would have more or less no need for any AA filter due to the fact that the space between pixels of any given color is very small compared to pixel size (obviously unlike Bayer filters that have minimum 1 entire pixel (or more) of space between two pixels of a given color). The pixels themselves are the AA filter in a Foveon.

    1. Do you find that people don’t listen to you because you deliberately misconsture events so you can put them down? It must be deliberate, because nobody is stupid enough to do that by accident.

        1. You like to sound like you know everything but that is not the case. If you have a valid source of information to back up your claims then state them. If not, then you just like to come off like some bigshot when you’re a know nothing. You must be a joy in real life – do you have any friends? Bye!

          1. My best friend in the world is Rob, I call him Robbie… And I know nothing ’bout cameras or anything else, unlike Robbie who really knows…

  11. My understanding is that moire can be removed in PP – happens with current medium format backs… I also understand that moire problems decrease as the pixel count increases – so might resolution sensors have less need of AA filters.

    1. It’s all about sampling rate vs. analog frequency content… certainly a higher resolution sensor will avoid moire (caused by the aliasing referenced by the name anti-aliasing filter) up to higher analog frequencies. That said, if the high frequency subject (say, someone with a striped shirt) is as much farther away as the increase in sensor resolution, you will be right back where you started.

      As to removing moire in PP, can’t images also be sharpened in PP? It’s the same thing either way. Creating data (possibly an educated guess, but still a guess none the less) to give the image a feel that we prefer. If you argue against stuff like HDR based on it being a ‘lie’, then you should also be arguing for AA filters.

  12. There will be no “random” matrix at all. Just take a look at fuji shots: colors are pretty bad, midtones are poor. It happens because their pattern significantly decreases color resolution: you can see 2×2 areas filled only with green pixels. It’s bad, because at these areas camera loses information about red and blue. You can see why it’s bad. Sensor resolution should be decreased 4 times to restore the similar to bayern filter color resolution in these areas. In other words, fuji sensor should have 4 times more pixels to be able to catch the same color info in issued areas ⇒ 64 megapixels. BUT: the same 64Mp APS-C with bayern filter will have even greater color res and the moire problem will disappear due to diffraction.

    Another fuji pattern quirks:
    1) Demosaic for fuji filter is complicated and thus it increases the lag in EVF and LV
    2) Sony and Samsung are about to introduce RGBW sensors in order to resolve DR issues with EVF and LV, as well as in video + increases DR for still photo. This isn’t for free, the cost is another color resolution drop. Just imagine: fuji pattern color resolution drop + W pixel caused drop. This is too much, and having W is a must in order to have EVF that is as good as OVF. Of course, the resolution will be increased again to compensate this.

    So, you see: fuji pattern is a temporary solution and unlikely to be repeated in future developments.

    1. Thanks for the informative discussion Denis – SO MUCH BETTER than hearing from know-nothing loudmouths like Luis.

  13. just remove the AA filter and when you start to see moire, just stop down until you get a little bit diffraction, this will also eliminate the moire-effect

  14. One just needs to know the aperture range to avoid with a given lens so that moire never appears. Really simple.
    Then for that avoided aperture range one simply needs a different body to use.

  15. Yes!!!

    Keep fighting like kids over tech stuff you want in a camera.

    The rest of us will be using what we have now to enjoy our photography and make money doing it.

    Typical Sony user base, always left in dust talking about tech stuff instead of perfecting their skills and actually taking pics of something other than their cats.

    Haha

  16. If they uses modified a RGB pattern like the Fuji X-Pro1, maybe OK ~ but otherwise keep an AA filter, especially if the camera is going to do video.

  17. As a fashion photographer, I’m not sure removal of AA filter is workable for me. Yes, I know it can be addressed in PP. But I certainly don’t want to use software to correct for thousands of images per day shoots.

    _____

    I just downloaded the Nikon rock sample and applied 1 Sharpen in PS. It made the AA shot sharper than the original non AA shot. I see little point in removing the AA at resolutions so high as 36mp.

  18. Non of it will matter if Sony really does have a foveon like sensor in the works. that sigma sd1 is CRAZY amazing

Leave a Reply

Your email address will not be published. Required fields are marked *

Search

Search
Advertisement
SNAPCLOUDBack up everyphotoTry freeSample ad

Subscribe to Our Newsletter

*Please also check your spam folder if you don't get the confirmation mail. And mark the mail as "non-spam" to see the future newsletter.

Advertisement
SNAPCLOUDBack up everyphotoTry freeSample ad
Advertisement
Advertisement