Eye of the bee

[billtils]

I shot it from this angle deliberately to concentrate on the head and eye. I'm happy enough with the outcome but curious about the pattern in the eye that I was not able to edit out (it looks like a moire pattern perhaps because of the eye structure?).

Advice and suggestions please!


Look me in the eye

[xpatUSA]

I shot it from this angle deliberately to concentrate on the head and eye. I'm happy enough with the outcome but curious about the pattern in the eye that I was not able to edit out (it looks like a moire pattern perhaps because of the eye structure?).

Advice and suggestions please!


Look me in the eye

Definitely moiré, Bill, but all is not lost - if the posted image was downsized for the web. See here for info on how to fix that:

http://kronometric.org/phot/iq/Down%...%20methods.htm

On the other hand, if the posted image is a 100% crop from a larger image, then a much longer lens and/or a much closer shot is needed.

The problem being that aliasing, e.g. moiré, will occur if the spatial frequency of the eye facets (in the image plane) is greater than your sensor's resolution (in lp/mm). The closer you get, the lower the spatial frequency of the eye facets (again, in the image plane). Ideally, you'd want to frame the whole head and little else.

A bit of googling re "Nyquist Frequency" will aid understanding of aliasing.

P.S. The link "Look me in the eye" takes me to a Yahoo log-in page which perhaps was not your intent!

[billtils]

Thanks Ted, I'll follow up your link. It was not highly cropped, taken with a 90mm maco, and captured on a 24MP FF sensor (Nikon D750).

[marlunn]

Nice image Bill, psst, its a Hoverfly not a Bee, but dont tell and most wont know

as an aside your sensor has an anti-aliasing filter so it should reduce this - however my D7100 does not have this AA filter and I very rarely get anything like that, thats seems a bit back to front to me, but what the heck.

[DanK]

As mark said, a hoverfly. Giveways: the shape of the eye, the flat paddle-shaped antennae, and the fact that there are only two wings, which you can see because the critter obligingly held them out from her body. The eyes and antennae of most bees and wasps look like the shot below, with two caveats: the antennae of some species are not bent, and native American bees don't have hairs in the eye itself (this is a honeybee).



This definitely looks like moire to me. It's a puzzle. I shoot Canon, not Nikon, all with AA filters, and I have never had this that I recall on bug eyes. Sorry I can't be more helpful.

[xpatUSA]

Thanks Ted, I'll follow up your link. It was not highly cropped, taken with a 90mm maco, and captured on a 24MP FF sensor (Nikon D750).

If the 1024x1024px image wasn't cropped much then it was most certainly down-sampled quite a bit.

Something's wrong with the processing:

A tag in the EXIF says "Optimized by JPEGmini 3.13.3.15" . . No idea what 'JPEGmini' is . . . anybody?

But worst of all is that 'JPEGsnoop.exe' tells of a large amount of JPEG compression in the posted image, hopefully not intentional:

"Compression stats:
Compression Ratio: 241979.08:1 [shock, horror]
Bits per pixel: 0.00:1"

A 'Bits per pixel' of around 6:1 would have been nice . . .

That's enough to explain it all . . . but:

[edit] I've just had a read of the Wiki and that level of compression is ridiculous and does not match the posted image quality, see here:

https://en.wikipedia.org/wiki/JPEG#C..._and_artifacts

So my suspicion passes to "JPEGmini" or some accident in post-processing. [/edit]

[billtils]

Nice image Bill, psst, its a Hoverfly not a Bee,.

Thanks Mark - yes I did know that when I took the pic but it all got lost in the downloading/indexing/editing of the approx 300 shots from the day. I was more bothered about exploring the reason for the moire than the ID () as I've not seen it in the several hoverfly shots I've taken with this body.

Yes, I shoot only RAW, and will take another look at the original later (it's on a different machine from this one).

[Saorsa]

Dean Martin had a song about that.

When a grid's misaligned
with another behind
That's a Moire

When the spacing is tight
And the difference is slight
That's a moiré

[xpatUSA]

Thanks Ted, I'll follow up your link. It was not highly cropped, taken with a 90mm maco, and captured on a 24MP FF sensor (Nikon D750).

Thanks for the info, Bill.

Looks like the image was down-sized to about 25% of the original (1024 vs approx 4000 tall), ushering in the real danger of aliasing caused by re-sampling, especially if e.g. "bicubic sharper".

Then the head ended up about only 200px square, perhaps 10% of what it could have been with a 1:1 macro close-up shot.

Based on the above, moiré was just about invevitable, I reckon.

[billtils]

Thanks for the info, Bill.

Looks like the image was down-sized to about 25% of the original (1024 vs approx 4000 tall), ushering in the real danger of aliasing caused by re-sampling, especially if e.g. "bicubic sharper".

Then the head ended up about only 200px square, perhaps 10% of what it could have been with a 1:1 macro close-up shot.

Based on the above, moiré was just about invevitable, I reckon.

Thanks Ted. I have uploaded the original of the image we are discussing here and some other fly shots for comparison, all but one as screen shots showing the EXIF for information. The "Bee" shot is the only one which shows the moire pattern in the eyes.

You can see them here: https://www.flickr.com/gp/billtils/D8Kh97

[xpatUSA]

Thanks Ted. I have uploaded the original of the image we are discussing here and some other fly shots for comparison, all but one as screen shots showing the EXIF for information. The "Bee" shot is the only one which shows the moire pattern in the eyes.

Possibly the only one with a perfectly-focused eye? By which I mean that the AA filter and even slightly out-of-focus would be enough to eliminate aliasing of the eye facets at that distance.

You can see them here: https://www.flickr.com/gp/billtils/D8Kh97

Thanks again, I looked at the shot there but couldn't deduce much from it. Would love to download the raw DSC_5451.NEF somehow for analysis - is that possible?

Or, if not, convert it to a full size PNG without post-processing esp without sharpening, crop out the bee/fly and post it here?

[billtils]

Possibly the only one with a perfectly-focused eye? By which I mean that the AA filter and even slightly out-of-focus would be enough to eliminate aliasing of the eye facets at that distance.

Thanks again, I looked at the shot there but couldn't deduce much from it. Would love to download the raw DSC_5451.NEF somehow for analysis - is that possible?

Ted

Interesting suggestion on the focus accuracy.

As far as getting the NEF, yes that can be done. However, it's 11:50 pm here and (a) I'm about to go to bed and (b) would need to fire up the iMac and upload the NEF so (a + b) = will leave until the morning (UK time). Look forward to your feedback and really appreciate your engagement in this - thank you.

[xpatUSA]

Ted

Interesting suggestion on the focus accuracy.

As far as getting the NEF, yes that can be done. However, it's 11:50 pm here and (a) I'm about to go to bed and (b) would need to fire up the iMac and upload the NEF so (a + b) = will leave until the morning (UK time). Look forward to your feedback and really appreciate your engagement in this - thank you.

Excellent - talk to you manana.

[Shadowman]

Nice effort, the moire pattern doesn't spoil it from me but as Ted stated it can be modified to good effect.

[Wavelength]

Why worry about moire, if it only adds to the beauty? it is a very nice capture....
A doubt: Is moire an outcome of diffraction?

[IzzieK]

Dean Martin had a song about that.

When a grid's misaligned
with another behind
That's a Moire

When the spacing is tight
And the difference is slight
That's a moiré

Still laughing, Brian...
That's a moire, huh??? ...close enough ...

OK..I was distracted...The moire did not distract me...

[billtils]

Dean Martin had a song about that.

When a grid's misaligned
with another behind
That's a Moire

When the spacing is tight
And the difference is slight
That's a moiré

Brian

I missed this first time around - very good! However, I'm aware of what causes a moire pattern but that wasn't the concern. I have many insect eye shots taken with the same body, same lens, and near enough same magnification but not a hint of moire.

My best bet right now is Ted's suggestion that I got the focus exactly right/? wrong on the surface of the eye.

[xpatUSA]

Bill, thanks for the NEF which I've just had a look at. I used RawDigger to view the capture with and without demosaicing. In the raw composite view (no demosaicing) we see what the sensor saw through the Bayer CFA. It looks like the spatial frequency of the pattern on the eye is less than that which would cause aliasing in the raw view . . but not much less:



In this view we can just about make out the pattern of the facets on the eye and, importantly, the lack of moire. Opened in FastStone Viewer (FSV), the same pattern is readily visible after a lot of tweaking:



Although this is a converted image, FSV seems to have suppressed the moire; no idea how it does that, but FSV does use DCraw as it's converter.

However, when RawDigger converts the NEF for the RGB review image, guess what?:



There's the Moire in all it's glory. Classic color aliasing - but caused by what?!

My view is that the process of demosiacing introduces color aliasing simply by the way it works. For any pixel, it's color is estimated from the values of itself and nearby pixels as proposed by old man Bayer hisself. Always remember that a 24MP Bayer pattern sensor is 12MP green, 6MP red and 6MP blue, so rendering of things like insect eyeballs is quite a challenge.

My view remains that, had you got quite a bit closer, the aliasing in the final image would have been less likely.

In the 67% view of the other fly, the facets in the eye don't show at all and the bright area appears obliterated by JPEG artifacts, so I'll stick with the focus theory too.

Hope this helps . . .

[xpatUSA]

Opened in FastStone Viewer (FSV), the same pattern is readily visible after a lot of tweaking:



Although this is a converted image, FSV seems to have suppressed the moire; no idea how it does that, but FSV does use DCraw as it's converter.

Out of interest, here's FastStone Viewer's conversion of your NEF with slight sharpening:



Cropped, but not re-sized. I post this as living proof that different converters do different things and I wonder what your Capture One does that brings out the color Moire so much . .

Looks like FastStone doesn't use the bad pixels list though . . see arrow in lightbox view,

[billtils]

Thanks Ted - that was very helpful indeed.