First insect of 2017 that wasn't an ordinary fly or bee...

[Beatsy]

...it was Bombylius major, a bee fly

Pretty big too, FoV is about 28mm here. I had to use the MP-E 65 at 1:1 for this stack (111 images), which was practically wide-angle compared to the stuff I usually shoot with Mitties. Made me proper agoraphobic it did

Back to familiar territory to shoot some close-up bits next...

[Beatsy]

Zooming in on the proboscis. One of those subjects that shows more structures and detail as magnification increases. Neat!

[Saul]

Steve, very nice. Black background rocks. Will you post bigger versions on Flickr ?

[Beatsy]

Steve, very nice. Black background rocks. Will you post bigger versions on Flickr ?

Thanks. Slightly bigger versions here

https://flic.kr/p/SK3qAn - bee fly
https://flic.kr/p/SGCPw3 - head and proboscis
https://flic.kr/p/SmwhTj - proboscis tip

[Beatsy]

A last one to complete the set - the top of the proboscis where it joins the head. 10x Mitty stack, sensor in APS-C mode. I rather like the "close-crop portrait" look of this one.

[MacroLab3D]

"blurred halo effect" returned on the last shot, right?

[Beatsy]

A touch just above the top of the proboscis, but generally not I thought. I stopped the stack short and left the background blurred on purpose so the front hairs stood out against it. Is that the part you meant?

[MacroLab3D]

A touch just above the top of the proboscis, but generally not I thought. I stopped the stack short and left the background blurred on purpose so the front hairs stood out against it. Is that the part you meant?

Yes.
So, solution for sharp hair is:
to do full stack, avoid stopping until full subject in focus to prevent blurry fuzziness at the end
+ smallest step size with massive overlap, i.e. 177 shots
+ widest aperture i.e. 5.6 on MP -e65
=success?

Or am i missing something?

[Beatsy]

Not that straightforward IMO


>to do full stack, avoid stopping until full subject in focus to prevent blurry fuzziness at the end
It will prevent the background blur (also known as bokeh) but doesn't guarantee no blur in supposedly in-focus hairs

>smallest step size with massive overlap, i.e. 177 shots
Not necessary. In fact, all the shots above (except the whole fly) had no overlap at all. I shot wider spacing for speed then reduced image size 50% before stacking - which *barely* restored overlap of in-focus parts.

>widest aperture i.e. 5.6 on MP -e65
Yes, for resolution, but it doesn't make any difference when hairs "go bad". I use my MP-E at f/4.0 up to 2x, and f/2.8 up to 5x (though I usually switch to the 5x mitty for 3.5x and above if corners aren't that important on full frame, it's sharper)

Key things I *know* to improve the situation (most of the time)...
1. Composition. Make sure the subject is posed so none (or few) of the known problem situations crop up. e.g. Don't have bright clumps of hair in front of darker areas deeper in the stack, or if you do - then reduce stacking depth to avoid the problem (as in last image above).
2. Turn on x, y shift and scaling when stacking in Zerene. Especially for deep stacks. This definitely improves the rendering of hairy bits, even if the stacking rig is 100% true and stable and doesn't need alignment corrections. I wonder if parallax is the issue with the hairs being so close together - which scaling and shifting can fix locally?
3. Careful lighting. If things must overlap, make sure one isn't significantly brighter or darker than the other. May involve little tabs of card inside the diffuser to shade tiny areas (gobos).
4. ? Not sure what else...

Cheers

[MacroLab3D]

Thank you very much, Beatsy, for sharing your settings and advice. This is gold. I appreciate.

[rjlittlefield]

Turn on x, y shift and scaling when stacking in Zerene. Especially for deep stacks. This definitely improves the rendering of hairy bits, even if the stacking rig is 100% true and stable and doesn't need alignment corrections.

Scaling should definitely be on, unless you're using optics that are telecentric or have such shallow DOF that you can pretend they're telecentric.

The issue is that with most optics, the entrance cones fan outward from the center of the lens to the edges of the subject. This results in a scale change from front to back of the in-focus slab. When you step focus, you have to rescale the new image so that subject features line up in adjacent images.

The following formula may be helpful in thinking about when you need to rescale:

Code: Select all

DeltaSubjectWidthPixels = SubjectWidthPixels * SliceThickness / EntrancePupilDistance

where
  SubjectWidthPixels is the subject width in pixels,
  SliceThickness is the effective DOF in a single image,
  EntrancePupilDistance is the distance from entrance pupil to the subject, and
  DeltaSubjectWidthPixels is the amount that the subject will change size from one slice to the next.

If DeltaImageWidthPixels is greater than 1, then you should allow rescaling because otherwise some parts of the subject will be misaligned by 1/2 pixel or more from one slice to the next.

With the MP-E 65, the entrance pupil position is about 25 mm behind the front of the lens. At 5X and f/2.8, this gives roughly SliceThickness ~ 0.025 mm and EntrancePupilDistance ~ 70 mm. If you're interested in pixel-peeping a 5000-pixels image, then DeltaSubjectWidthPixels = 5000 * 0.025 / 70 = 1.78. This not hugely greater than 1, but it's still big enough that I would definitely turn on scaling if I wanted to retain all the image quality that the lens can deliver.

--Rik

[Beatsy]

Thanks Rik - that's clear.

I further assume that scaling factor across the "thickness" of the focal plane is the same as, or similar to, perspective (further away = smaller). Explains why turning scaling on also gives an overall more pleasing perspective to deep stacks of larger insects. Without scaling, the "rear end" of some insects can look out of proportion with the head (too big).

I don't know the entrance pupil distances for mitties. I found a thread explaining ways to find out - but I think I'll just leave scaling on. The A7Rii can produce images almost 8k pixels wide - so chances are "DeltaImageWidthPixels" will exceed 1 in most cases anyway.

Thanks again for the details.

[rjlittlefield]

I further assume that scaling factor across the "thickness" of the focal plane is the same as, or similar to, perspective (further away = smaller). Explains why turning scaling on also gives an overall more pleasing perspective to deep stacks of larger insects. Without scaling, the "rear end" of some insects can look out of proportion with the head (too big).

Correct. Another way to think about scaling is that it tries to resize each slice to fit however big that focus plane appeared to be in the first frame processed. The actual operation is incremental, each frame against the previous, but in concept it's the same idea.

I don't know the entrance pupil distances for mitties.

An alternate strategy is to estimate DeltaSubjectWidthPixels by aligning two frames that are one DOF apart, then looking in Zerene Stacker's log or saved project file to see what scale factor was used, and computing as

Code: Select all

DeltaSubjectWidthPixels = (1.0 - RegistrationParameters_1_ScaleValue/RegistrationParameters_2_ScaleValue) * SubjectWidthPixels

where 
  RegistrationParameters_n_ScaleValue comes from the n'th <RegistrationParameters> block of the saved project file.

This approach is simpler and more accurate when the entrance pupil is far away, as with many Mitty's.

It turns out that you can also use this formula, in combination with the other one, to calculate where the entrance pupil must be. That can be a useful cross-check against other methods of finding it.

--Rik

[microman]

Interesting to se the proboscis up close.