Tariq Gibran wrote:
This color moire does not really concern me too much overall. It's the luminous moire such as shown with the Nikon Kimono example which would be a real nightmare.
I'm not sure if I would call that moiré. I thought moiré is what you see when a pattern of bands is created. I'd call this false color -- but this is present on every camera with or without AA filter, and it doesn't require a very sharp lens. It's a result of the way the Bayer pattern works. I always see it emerge as cyan and orange localized fringes in places where the spatial image detail is smaller than what the sensor can resolve.
AhamB wrote:
I'm not sure if I would call that moiré. I thought moiré is what you see when a pattern of bands is created. I'd call this false color -- but this is present on every camera with or without AA filter, and it doesn't require a very sharp lens. It's a result of the way the Bayer pattern works. I always see it emerge as cyan and orange localized fringes in places where the spatial image detail is smaller than what the sensor can resolve.
I saw same thing from 1Ds (that first one), either 1Ds had very weak AA filter (could be) or its simply as you said, bayer issue.
Moiré is caused by repeating paterns, those branches are most definitely not pattern.
Btw. D800E still has AA filter, just different kind (Nikon designer or whoever that was explained that somewhere).
Mescalamba wrote:
Btw. D800E still has AA filter, just different kind (Nikon designer or whoever that was explained that somewhere).
It effectively does not have an AA filter. One of the components which would make it an AA filter has been changed so that it has no blurring property thus technically it does not have a working AA filter.
AhamB wrote:
I'm not sure if I would call that moiré. I thought moiré is what you see when a pattern of bands is created. I'd call this false color -- but this is present on every camera with or without AA filter, and it doesn't require a very sharp lens. It's a result of the way the Bayer pattern works. I always see it emerge as cyan and orange localized fringes in places where the spatial image detail is smaller than what the sensor can resolve.
I believe the colored specks are a form of moire caused by the high frequency detail and the inability of the bayer pattern to predict the correct color, thus the false color. It can show up with a camera with an AA filter depending upon how weak the filter is as well as how sharp the lens is. You can see in that example that the D800 AA filter is fairly weak and seems to be pretty well matched. I expect with a very sharp/ high micro contrast lens such as some zeiss lenses, even the D800 would show some moire in certain situations. Lloyd discusses this type of moire here: http://diglloyd.com/articles/MiscReviews/HasselbladH3D.html
Now, he could be technically wrong about what it's called but it's definitely something which is either eliminated or dramatically reduced when an AA filter is used.
The patterns in the pattern bands you are referring to is luminous moire isn't it?
Tariq Gibran wrote:
It effectively does not have an AA filter. One of the components which would make it an AA filter has been changed so that it has no blurring property thus technically it does not have a working AA filter.
Its created by two layers, one splits rays, second one puts them together. When its normal AA, its two layers that both split rays, creating pretty much gaussian uniform kind of blur (which is why its so easy to regain sharpness).
Theres slight problem, that those two layers are still glass (or plastic) and they still manipulate light, I dont believe it will work same as true AA-less camera (with just thicker IR filter instead of AA) and samples show that they dont work same as AA-less sensor.
Problem is that proving this would need volunteer and MaxMax. Tho would be nice to know what would MaxMax say on this topic.
My understanding is that it has a "blur" / "re-blur" that offset one direction, while retaining the other. Not sure @ horizontal vs. vertical for retain vs. blur/deblur.
Similar in concept (i.e. not a true non-AA filter) to the HotRod conversion that MaxMax does on some models to remove one direction ( @ h vs. v) of the filter, but the 90 degree counterpart remains.
I doubt the conversion would be worth it. The benefit would be so much smaller than with the normal AA filter. But what does a MaxMax conversion cost? Is it worth buying a D800 and removing the AA filter?
I'm not suggesting an HR of a D800 ... definitely not. I doubt it is even a service they offer.
I was only using it to psuedo-parallel that neither are truly non-AA filter. They are more of a "1/2 AA filter" and 1/2 something else (either removed or blur/deblur offset) ... iirc.
As to cost of HR @ MaxMax, I think the 5D will run $450.
Hi Tariq, is it those little areas of magenta near some very fine branch ends, I see one for example at the top left, and a few other spots also...it looks like fine detail got replaced with the same content shape, but in colour?
I occasionally see something similar to this but not so pronounced for similar content (nature photography) with the Distagon 21mm on A900, and always thought it was some minor CA. But it seems to surround the fine detail, whereas here it seems to have replaced it - pretty ugly IMO! Whatever, a good catch by them and you.
But my word, the rendition is in another league from the AA camera in these two images. Even fine detail in the rock. I guess, once seen however, you always are going to be looking for it, the 'effect'.
Then again, when is it going to intrude on output apart from occasional glitches - it's not in the same ballpark as the TV reporter moire effect as his shirt seems to ripple with light...will it show in large prints? Will s/w vendors have a (good) fix - for the mainstream market, that is?
Mescalamba wrote:
Problem is that proving this would need volunteer and MaxMax. Tho would be nice to know what would MaxMax say on this topic.
At the risk of getting it wrong ... he would of course need a donor camera to test convert and from my dialogue today ... it would likely be very, very low on his priority list.
My takeaway was basically "suggesting" that if you aren't getting moire', it's because you're glass could be sharper. I guess there will always be a trade-off / balance / sacrifice involved between the two, and of course it is peculiar to certain instances more than general practice. So, it would seem that if your quest is for maximum sharpness ... moire' just comes with the territory ... and you learn to be aware of it, watch for it & deal with it.
He went on to say something about using ACDSee (iirc) for deinterlacing RAW files ... but it went over my head for now, so I told him "google (and FM'ers) is my friend" rather than ask him to further explain.
philip_pj wrote:
Hi Tariq, is it those little areas of magenta near some very fine branch ends, I see one for example at the top left, and a few other spots also...it looks like fine detail got replaced with the same content shape, but in colour?
I occasionally see something similar to this but not so pronounced for similar content (nature photography) with the Distagon 21mm on A900, and always thought it was some minor CA. But it seems to surround the fine detail, whereas here it seems to have replaced it - pretty ugly IMO! Whatever, a good catch by them and you.
But my word, the rendition is in another league from the AA camera in these two images. Even fine detail in the rock. I guess, once seen however, you always are going to be looking for it, the 'effect'.
Then again, when is it going to intrude on output apart from occasional glitches - it's not in the same ballpark as the TV reporter moire effect as his shirt seems to ripple with light...will it show in large prints? Will s/w vendors have a (good) fix - for the mainstream market, that is? ...Show more →
Hi Philip, Yes, there are magenta and cyan false color specks and it would appear that artificial artifacts/ shapes have been created as well. This is actually an issue if one needs to make really large prints because, ironically, the D800 image is better suited for enlargement as compared to the D800E - probably the opposite of what many might think and the reason some may go with the D800E erroneously. The image artifact you note is a good example why.
I have opened both raw images and interpolated them up in size to approx. 44"x66" at 240ppi using Alien Skin Blow Up II. The D800 image was then sharpened. Sharpening the D800E image would be a mistake imo as it would further accentuate the issue with the false information. The side by side crops below show them at this size viewed at 50% - pretty close to what the actual print would look like detail wise. Which one do you prefer? To me, the false data of the D800E has ruined the image - not just due to the color specks which could be removed but due to the false shapes which you noticed and can't easily be dealt with. Also note how nicely the D800 has reacted to sharpening. Apologies for the large side by side for those using low res screens. http://www.gibranstudio.com/moirec.jpg
carstenw wrote:
I doubt the conversion would be worth it. The benefit would be so much smaller than with the normal AA filter. But what does a MaxMax conversion cost? Is it worth buying a D800 and removing the AA filter?
Actually - you would gain some, but also lose quite a lot. Removing the top layer from a D800E would make the camera physically the same as a D800 (normal) with the top layer removed.
The problem with the D800E sensor is that the bottom layer is actually physically attached to the sensor surface. It's very hard to remove without damaging the microlens arrays. And if you remove the top layer in a D800E - you would actually get a more blurred image.
Removing the top layer from a normal D800 would leave one layer, most probably the "Y" layer. You'd gain some, but also lose some. They use a very high quality matched IR filter that's laminated together with the AA filter, that's hard to replace with something as thin without losing filtering effect (and a LOT of red/orange/yellow color accuracy).
Smridevan wrote:
This comparison got me all worried about that D800e so I took a look at the files using Nikon NX2 and to my eyes the false color really isn't there –- or it's not as noticeable. So it could be the raw conversion that is exacerbating the false color. It should be noted that the D800e image was taken at 1/250s and the D800 was taken at 1/320s. Perhaps that could contribute to the difference.
NIkon NX2 certainly handles the false color much better than ACR, which is what I used. I also tried Raw Photo Processor and surprisingly the results were about the same as ACR - I don't think it has been optimized yet for the D800.
The brightness difference does impact the final contrast and look of the fine tree branches and in the conversion I did, I tried to equaljze the slight exposure difference. I don't have NX2 to play around with unfortunately. As it is in your comparison now, the brightness difference makes the D800E look better but once the exposure is equalized and the D800 sharpened, I'm curious if the difference would still be there. Guess I need to load NX2 as it clearly looks like it will be required for optimum results from the D800E.
Thank you, Tariq - I prefer the left (D800) image. It seems to have a more realistic look as well..this is pretty much the direction the LL thread was taking also - careful sharpening and/or deconvolution from a known base. I doubt, going on this shot, there is much reason for the E, for this subject matter which should really favour it.
Smridevan, this is another concern - conversion, of course we can expect NX2 to be 'tuned' to the output. This E image appears to have a kind of hypernatural look to the edges - almost too much contrast, not being critical here, just staring at the content can do that sometimes. And it's all new to us.
It's just one set of images, a single step on the way. Thanks.
I find that ACR in CS5, which I think is the same engine used in Lightroom, does a very poor job in converting Nikon raw files. It's like night and day difference. The dynamic range, microcontrast, and colors are all muted using ACR conversion compared to NX2. It defeats the purpose of using Zeiss lenses when converting NEF files using ACR. So I try to do as much processing as possible with NX2 for my NEF files. If I need to work on the file in CS5 or Lightroom, I convert it to Tif. However, it's a pain to deal with such a large file. But I haven't found a better solution...
Btw.for those wondering, moiré reducing in Capture One can remove false colors too (I know it cause 1Ds files, tho its bit easier to simply stick to either DPP or NX2). Just one need to be careful and not overdo it.
Yea and Im seeing false colors in both samples, D800 and D800E, tho in first case its much less..
There is false color in both D800 and D800E examples. I just spent some time with Capture NX2 and it does a much better job with false color and the D800E (it's still there though, even if one turns the moire correction to high). After running both files through Capture NX2 and then processing as I did above for the enlargement, I still don't see any advantage to the D800E processed through NX2 versus the D800 when processed through ACR and sharpened. I'm actually somewhat surprised because the ACR D800 file looks more detailed with better color and contrast versus either file processed through Capture NX2, both at default settings (beyond bosting the D800 file about a third of a stop to match the D800E exposure). Anyway, my take away from this is that the D800 has a really weak and well matched AA filter and is less critical of which raw developer one uses. The D800E is going to be more picky - ACR looks like it will be the worst choice for any image which has moire/ false color issues.
Smridevan wrote:
So I increased the exposure compensation and added a little bit of sharpening for the D800 image using NX2. The D800E image was untouched. The images are 100% crop. As an experiment, can you tell which one is from the D800 and D800E?
I would suspect Sample A is the D800E - it shows a little more false color artifacts.
After further scrutinizing both files in great detail and looking at the sharpened image from the D800 vs D800E, I'm starting to think that the D800E's resolution advantage over the D800 can be negated by simply sharpening the D800's image. Since the AA filter is pretty weak on the D800 as you said, it wouldn't require that much sharpening to match the resolution. Any adverse effects from sharpening would not be an issue since sharpening is not heavily applied. Is this analogy valid? Or am I missing something?
Tho would be nice to know what would MaxMax say on this topic.
@ h vs. v) of the filter, but the 90 degree counterpart remains.