Hover fly's etc

[DaveW]

Hover fly:=




Is this a bee or what?






Nikon D200, 70-180 Micro Nikkor, ISO 100, Available light.

DaveW

[rjlittlefield]

Dave, these are sure looking sharp!

The subjects are all flies. The first one looks like Heliophilus, like on our front page at the moment. The bottom one also looks like some syrphid, and I'd be inclined to say the middle one also. On the other hand, British flies are not exactly my strong point, and I will happily yield to anybody who is actually familiar with these beasts.

--Rik

[DaveW]

Rik,

I used the "Adjust Sharpness" function under "Enhance" in Photoshop Elements 5 instead of "Unsharp Mask". I think you can get them a bit sharper using this than with "Unsharp Mask" and without the halo's.

Do you think these are OK, or have I over-sharpened, because I was not able to detect much in the way of sharpening halo's on my screen?

They were all taken with the camera hand held at 1/125th second at between f9-f11, from the Exif data.

DaveW

[georgedingwall]

Hi Dave,

The second and third are both flies. They are usually known as Bee Mimic Flies for obvious reasons. I don't know the proper names for them, but they look similar to some I photographed back in July.

Very nice images Dave, and the sharpening looks fine to me.

Bye for now.

[DaveW]

Hi George,

You managed to diffuse the light on yours as you were using flash. Unfortunately I could not use a diffuser with available light as for one thing it would have spooked the fly and I only have two hands so have to use them for hand holding the camera. Therefore I got more reflections off the wings from direct sunlight

The settings I used in "Adjust Sharpness" were:-

Amount 94%
Radius 1.0 pixels
Gaussian Blur
Angle 0

I presume these were the default settings?

DaveW

[rjlittlefield]

Do you think these are OK, or have I over-sharpened, because I was not able to detect much in the way of sharpening halo's on my screen?

They look great to me. I don't see any halos either, and all of the fine textures like hair look completely believable. (Extreme sharpening can drastically change textures even without introducing noticeable halos, see this post, image 3, for example.)

Some people decry sharpening as being somehow unfair or as indicating poor technique in taking the picture in the first place. I strongly disagree. I think of sharpening as "approximate deconvolution". Its job is to compensate for the overall effect of many individual blurs ranging from lens aberrations to diffraction to monitor blur. An exact deconvolution would be better, but I'll take what I can get at reasonable cost.

Go forth and sharpen with a clear conscience. If the result looks realistic, you've done well.

--Rik

[rjlittlefield]

They are usually known as Bee Mimic Flies for obvious reasons. I don't know the proper names for them, but they look similar to some I photographed back in July.

Consider genus Eristalis, in family Syrphidae. The preceding link goes to BugGuide, which has quite a few pages of pictures. Again, I don't know the British syrphids. Some poking around on http://www.syrphidae.com might narrow it down. Eristalis tenax, the drone fly, comes to mind.

--Rik

[DaveW]

This was the thing I was disappointed with when I moved to digital Rik. With a slide if it was not sharp at the taking stage I could see straight away. However with digital because the camera deliberately blurs the image from the sensor to hide jaggies and avoid moire pattens, the image has to be "pseudo-sharpened" again.

I say "pseudo-sharpened" as you can never put back what has been removed, only fake it. The camera will do this automatically if taking JPEG's, but if using RAW you have to do it in your post processing software.

These people who shoot RAW or JPEG's but say their images are never sharpened are fooling themselves, because unless you switch it off the camera itself or most post processing software automatically default sharpens the image, although you can modify the result.

I don't think you can ever get anything sharper than the original anyway. The only problem in digital is sharpening halo's and artifacts if you turn up the artificial sharpening too much. Digital sharpening is akin to the old film photographers using acutance developers where when an edge occurs they increased the brightness of the lighter part and darkens that of the darker part, so apparently increasing sharpness. But of course it is only an optical illusion.

However that is all that matters because we are viewing the image by eye, so if it looks OK it is OK!

An interesting article on digital sensors here:-

http://www.extremetech.com/article2/0,3973,15465,00.asp

DaveW

[Planapo]

Nice and clear images there, Dave.

Like Rik, I´m not engaged in flies either, but your #2 and #3 look like belonging to the genus Eristalis.

Cheers,
Betty

edit: Meanwhile Rik has already digged out the genus name... I shouldn´t do other things while I´m writing an answer....

[rjlittlefield]

Dave,

The images that you're posting -- are those actual-pixel crops (no resizing), or full-frame reduced to forum size, or something in between?

And when you say

Amount 94%
Radius 1.0 pixels
Gaussian Blur
Angle 0

are you operating on the image as it came from the camera or raw converter, or on an image that has already been reduced to forum size?

--Rik

[DaveW]

Rik,

Original image was put through Adobe RAW converter, slight exposure correction, opened into Editor, cropped. re-sized to 800 pixels wide then sharpened as stated and then saved for web at just under 150k.

Here is one of the originals uncropped and just put straight through the Raw Converter into Editor and re-sized for web with no sharpening other than default sharpening which Elements may have applied.



DaveW

[rjlittlefield]

Dave, let me think out loud...

Assuming that your original image was the full 3872x2592 that I see in a D200 review, I measure that your posted image would have been about 1436x1332 before resizing. So, in resizing it to 800 pixels on the long axis, you reduced it to about 56% of original size (on each axis).

For my Canon 300D, that amount of reduction is almost exactly what it takes to eliminate any effects from the anti-aliasing filter and whatever other sensor blurs there might be. That is, if I take image 3 from this post, and shrink it to 56%, there becomes essentially no difference between the "simulated perfect sensor" and the actual sensor.

From there, you've sharpened by 94% at 1 pixel radius.

Yes, that does seem like quite a bit.

I usually figure that if my image is sharp to start with, and I downsize it for posting, then sharpening the downsized version by 35% at 0.7 pixels is about right to restore the appearance.

So I would guess, based on this model, that what we're seeing in the post is somewhat sharper than the image that your lens actually cast onto your sensor.

"Sharper" here means higher contrast for fine spatial detail -- essentially pushing up the end of your optics' Modulation Transfer Function. This sort of sharpening is actually a fair bit different from that done by acutance developers. If I recall correctly, they relied on local depletion of the developer chemistry, so they enhanced strong edges but didn't do much for weak ones. The digital filter affects strong and weak alike. No wet developer I ever saw could have done what USM did with the rabbit ears in this old post.

The word "sharper" is fuzzy, of course. Informally, I'm happy to say that one image is sharper than another if a typical viewer will notice finer detail in it. That's certainly true for the rabbit ears. It's also true, of course, that the fine detail was present in the original image, albeit at lower contrast, so whether the filtered version is "sharper" than the original is a matter of language, not technology.

BTW, I think it's interesting that you were disappointed when you switched to digital. I had exactly the opposite experience. The reasons are documented here. The figure at the top of that page comparing digital and film makes a nice summary.

--Rik

[DaveW]

I was disappointed in switching to digital initially, not in the final image Rik but in the sense that I could not immediately assess if I had got the image sharp at the taking stage, as with a slide, because it required "pseudo sharpening afterwards".

With a slide what you see is what you get. With digital unless you are taking JPEG's and don't touch them again, using RAW it is only the first stage of the process. Also it would be nice if they would perfect a sensor like the Foveon where what the sensor actually sees is what you get with no demosaicing etc.

I am getting a bit better now at assessing the initial image for sharpness at the taking stage.

Do you think I have oversharpened the original images then?

DaveW

[rjlittlefield]

it would be nice if they would perfect a sensor like the Foveon where what the sensor actually sees is what you get with no demosaicing etc.

Yes, true RGB like Foveon would be very nice, assuming we can get it without paying some other tradeoff such as lower resolution, higher noise, or smaller gamut. I don't have a clue how long we're going to have to wait. The Foveon technology has been difficult to perfect, and meanwhile the mosaic folks have kept raising the bar with higher pixel densities.

Do you think I have oversharpened the original images then?

Yes, in the sense that 1) I suspect these images show more contrast for fine detail than the real subjects had, and 2) I suspect the same procedure applied to other more contrasty subjects and/or different lens setups would produce visible artifacts such as halo.

However, as I wrote earlier I don't see any artifacts in these images, and the textures look quite believable. The pictures are pleasant to look at. When I wrote "these are looking sharp", I was speaking in comparison to my memories of some past posts such as this one, which you noted at the time seemed to have some problems. Whatever it is that you're doing, this round of images seems definitely improved.

--Rik