First attempt at macro focus stacking

[John Cothron]

I had been observing the growth of this magnolia seed pod for a week or so.  On the spur of the moment I decided to shoot it.  I had to go get my macro lens out because I literally haven't use it in 3 years or so.  I decided to try a focus stack, and used a macro rail for the shot.  I sort of guessed on the stepping, and used about .50mm (manually).  I think I could have used far less than the 61 images in this stack after having done some calculations later.  The stack was done in Ps.  I'd be interested in hearing your thoughts.

[Bob_B]

I like it, especially that bit of purple on the lower right part of the stem that complements the green hues on the bud itself.

[francois]

Really nice shot, even without considering that it's your first attempt. Good framing, great colors and well managed OOF areas, I like it very much.

[Paulo Bizarro]

Very nice. I would prefer it if the background were entirely homogeneous.

[Bart_van_der_Wolf]

I had been observing the growth of this magnolia seed pod for a week or so.  On the spur of the moment I decided to shoot it.  I had to go get my macro lens out because I literally haven't use it in 3 years or so.  I decided to try a focus stack, and used a macro rail for the shot.  I sort of guessed on the stepping, and used about .50mm (manually).  I think I could have used far less than the 61 images in this stack after having done some calculations later.  The stack was done in Ps.  I'd be interested in hearing your thoughts.

Hi John,

Well 'attempted'   I also like the lighting you used. Such focus-stacked shots deserve to be displayed big, because of the overwhelming amount of detail and the larger than life display size (which by itself will draw attention).

As for backgrounds, one can opt for either a fully blurred background (which may make the image look sterile/artificial), or shoot an additional shot (of a different setup) with barely enough detail to define surroundings/habitat. This can also be mounted in at a later stage.

For shooting planning, I like to use a DOF calculator that allows setting, besides aperture used, a COC that matches the sensel pitch of the sensor used, and that uses magnification factor instead of distance as input. Macro-lenses often have a scale indicating the Magnification factor. On a focus rail, the magnification factor remains constant, so the slice depth is also constant, and the magnification factor automatically deals with the effective aperture (which changes when focusing closer than infinity).

One can also pre-compute the constant per-slice DOF for a few magnification factors and make a small cheat-sheet from that with:

where N = aperture number (e.g. 5.6) and c = COC and m= magnification factor. P is the pupil factor for an asymmetrical lens, but if you don't know that for your lens, just use 1 as for symmetrical lens designs. When the COC is given in millimetres (e.g. for the EOS 5 Mark IV the sensel pitch is approx. sensel pitch = 5.36 µm or 0.00536 mm), then the DOF will also be in millimetres.

At closer focusing distances or larger magnification factors, diffraction also becomes a magnified issue, so I often select something in the f/4 - f/5.6 range for optimal sharpness when approaching 1:1 magnifications.

Cheers,
Bart

[John Cothron]

Hi John,

Well 'attempted'   I also like the lighting you used. Such focus-stacked shots deserve to be displayed big, because of the overwhelming amount of detail and the larger than life display size (which by itself will draw attention).

As for backgrounds, one can opt for either a fully blurred background (which may make the image look sterile/artificial), or shoot an additional shot (of a different setup) with barely enough detail to define surroundings/habitat. This can also be mounted in at a later stage.

For shooting planning, I like to use a DOF calculator that allows setting, besides aperture used, a COC that matches the sensel pitch of the sensor used, and that uses magnification factor instead of distance as input. Macro-lenses often have a scale indicating the Magnification factor. On a focus rail, the magnification factor remains constant, so the slice depth is also constant, and the magnification factor automatically deals with the effective aperture (which changes when focusing closer than infinity).

One can also pre-compute the constant per-slice DOF for a few magnification factors and make a small cheat-sheet from that with:

where N = aperture number (e.g. 5.6) and c = COC and m= magnification factor. P is the pupil factor for an asymmetrical lens, but if you don't know that for your lens, just use 1 as for symmetrical lens designs. When the COC is given in millimetres (e.g. for the EOS 5 Mark IV the sensel pitch is approx. sensel pitch = 5.36 µm or 0.00536 mm), then the DOF will also be in millimetres.

At closer focusing distances or larger magnification factors, diffraction also becomes a magnified issue, so I often select something in the f/4 - f/5.6 range for optimal sharpness when approaching 1:1 magnifications.

Cheers,
Bart

Hello Bart,

I appreciate the information.  Obviously I can't easily go back (at least from the office) and get precise numbers, but I have a decent idea of the magnification based on know approximately how big the pod is and how much of the frame it is taking up.  The magnification factor would be .75-.80x based on those estimates.  According to the numbers you give me at the formula provided that would slice of focus would be between .136 and .150 mm.  Does that sound right to you?  To me it sounds really shallow as based on my calculations using different methods I would have expected a 10x multiple of those numbers.  I'm wondering if I've made a decimal mistake somewhere.

I shot at f4.5 so by your formula I have (2 * 4.5 *.00536 * (1 + .75 or .80))/ .75 or .80 ^2.  In each case that gives me (.04824 * 1.75)/.5625 or (.4824 * 1.80)/.64 yielding .15mm or .136mm respectively.  I feel like I must be missing something here.

Thanks,

John

[Bart_van_der_Wolf]

Hello Bart,

I appreciate the information.  Obviously I can't easily go back (at least from the office) and get precise numbers, but I have a decent idea of the magnification based on know approximately how big the pod is and how much of the frame it is taking up.  The magnification factor would be .75-.80x based on those estimates.  According to the numbers you give me at the formula provided that would slice of focus would be between .136 and .150 mm.  Does that sound right to you?  To me it sounds really shallow as based on my calculations using different methods I would have expected a 10x multiple of those numbers.  I'm wondering if I've made a decimal mistake somewhere.

I shot at f4.5 so by your formula I have (2 * 4.5 *.00536 * (1 + .75 or .80))/ .75 or .80 ^2.  In each case that gives me (.04824 * 1.75)/.5625 or (.4824 * 1.80)/.64 yielding .15mm or .136mm respectively.  I feel like I must be missing something here.

Nope, you're not missing anything. The DOF is really that shallow, so you did well and probably needed as many slices as you used (especially at the edges of the pod where depth is changing more rapidly for each next slice than in the frontal/centered surfaces). Of course, using a CoC equal to the sensel pitch is offering a super high standard, allowing to magnify the image to huge sizes (the edges of the pods can otherwise visibly go in and out of focus, similar to e.g. insect antennae pointing towards the camera). Smaller output sizes can still work with proportionally larger CoCs and fewer slices. Also the use of narrower apertures will add such an incredible amount of magnified diffraction blur, that a larger CoC can be used (because the overall resolution will be lower anyway).

When you check the edges of the stacked pod at full file size, you might find that your rail steps were slightly on large side.

Cheers,
Bart

[John Cothron]

Nope, you're not missing anything. The DOF is really that shallow, so you did well and probably needed as many slices as you used (especially at the edges of the pod where depth is changing more rapidly for each next slice than in the frontal/centered surfaces). Of course, using a CoC equal to the sensel pitch is offering a super high standard, allowing to magnify the image to huge sizes (the edges of the pods can otherwise visibly go in and out of focus, similar to e.g. insect antennae pointing towards the camera). Smaller output sizes can still work with proportionally larger CoCs and fewer slices. Also the use of narrower apertures will add such an incredible amount of magnified diffraction blur, that a larger CoC can be used (because the overall resolution will be lower anyway).

When you check the edges of the stacked pod at full file size, you might find that your rail steps were slightly on large side.

Cheers,
Bart

There are most definitely some areas that are blurred.  I didn't think it was bad, but it is there.  I wasn't sure whether it was movement (there was some the wind was blowing) or if it was just beyond the capability of Ps to get the stacking right (which I think is true to some extent).  If I keep playing around with this technique I'll probably invest in some stacking software such as Zerene Focus.

Thanks for the input!

John

[BradSmith]

Not on the topic of macro, but I loved your pod.  Here is my formal portrait of a another member of the family.
Brad

[John Cothron]

Not on the topic of macro, but I loved your pod.  Here is my formal portrait of a more more member of the family.
Brad

Thank you, and very nice!  I'm going to keep an eye on my trees and try to get the different stages I think.  I find them very interesting looking.

[JKoerner007]

Very nice. I would prefer it if the background were entirely homogeneous.

+1

[John Cothron]

Yeah in retrospect I agree.  The edge of the leaf in the background is distracting more than I would like.  I kind of like it in the lower right to contrast with the stem though.  Still, if I were doing it again I would do it a little differently.

[Bart_van_der_Wolf]

Yeah in retrospect I agree.  The edge of the leaf in the background is distracting more than I would like.  I kind of like it in the lower right to contrast with the stem though.  Still, if I were doing it again I would do it a little differently.

It may not be necessary to do it over again, just add a blurred image of some foliage to the stack. As long as there is slightly more detail than the current background, it may automatically replace the current background.

You can also try making a 2 image stack from such a backdrop and your finished stacked image, or do a formal masking of the stacked image on a different background. Lots of creative options, as long as we're not dealing with photojournalism or forensic photography.

Cheers,
Bart

[bjanes]

I shot at f4.5 so by your formula I have (2 * 4.5 *.00536 * (1 + .75 or .80))/ .75 or .80 ^2.  In each case that gives me (.04824 * 1.75)/.5625 or (.4824 * 1.80)/.64 yielding .15mm or .136mm respectively.  I feel like I must be missing something here.

John and Bart,

Congratulations to John for his excellent work and to Bart for his useful comments. From the EXIF I see that a Canon camera (which unlike Nikon does not report the effective f/number) and a 150 mm f/2.8 lens (? Sigma) were used for the stack. With this lens, he pupillary magnification is likely less than 1 and this could affect the calculations slightly.

In the DoF equation that Bart quoted I assume that the factor (1+m/p) corrects to the effective N from the nominal N. For Nikon users I assume that the factor would simply be (m/p). Is this correct?

Bill

[Bart_van_der_Wolf]

John and Bart,

Congratulations to John for his excellent work and to Bart for his useful comments. From the EXIF I see that a Canon camera (which unlike Nikon does not report the effective f/number) and a 150 mm f/2.8 lens (? Sigma) were used for the stack. With this lens, he pupillary magnification is likely less than 1 and this could affect the calculations slightly.

Yes, but with many (especially Macro-) lenses, internal focusing groups make it hard to exactly know what's going on. It can help to 'focus'/extend the lens for a given magnification, take the lens off of the body, and look at the exit/entrance pupil diameters ratio, which is P.

Another benefit of using Magnification factor, is that the unknown exact focal length (which also changes in most Macro lenses) is taken out of the DOF equation. Magnification is relatively easy to determine for close-up and Macro photography, by just taking a ruler and see how many millimeters are projected to the 36mm frame width (assuming 'Full frame' sensors).

In the DoF equation that Bart quoted I assume that the factor (1+m/p) corrects to the effective N from the nominal N. For Nikon users I assume that the factor would simply be (m/p). Is this correct?

The term (1+m/P) is indeed the effective adjustment of the nominal aperture number. So if the camera automatically adjusts the effective aperture value in the camera EXIF, BUT KEEPS the physical aperture at the same diameter, then the formula remains unchanged. I assume that is what Nikon does, it reports the effective aperture but keeps the physical aperture diameter the same. The apparent loss of diameter is caused by projecting that same physical aperture from a larger distance to the sensor, so it seems narrower and the image loses exposure (to be compensated by a longer exposure time of factor (1+m/P)^2, which is automatic for TTL measuring).

I'd say that if that's the case, the DOF approximation formula applies universally for close up photography, regardless of Camera.

Cheers,
Bart

[bjanes]

[quote author=BartvanderWolf link=topic=119199.msg991176#msg991176 date=1501955794

The term (1+m/P) is indeed the effective adjustment of the nominal aperture number. So if the camera automatically adjusts the effective aperture value in the camera EXIF, BUT KEEPS the physical aperture at the same diameter, then the formula remains unchanged. I assume that is what Nikon does, it reports the effective aperture but keeps the physical aperture diameter the same. The apparent loss of diameter is caused by projecting that same physical aperture from a larger distance to the sensor, so it seems narrower and the image loses exposure (to be compensated by a longer exposure time of factor (1+m/P)^2, which is automatic for TTL measuring).

I'd say that if that's the case, the DOF approximation formula applies universally for close up photography, regardless of Camera.
[/quote]

Bart,

You raise an interesting point which I have not considered before, but that is not my understanding on how Nikon set and reports the aperture. Most recent Nikon lenses do not even have an aperture ring and the aperture is set on the camera. With my 105 mm f/2.8 macro I can set the lens wide open at infinity and the camera reports f/2.8. If I rack out the lens for 1:1, the aperture is reported as f/5.6 f*(m+1) and there is no way to set the lens to f/2.8.

The Photopills macro depth of field calculator has a special checkbox for Nikon which reports the effective aperture.

http://www.photopills.com/calculators/dof-macro

Also, Rik Littlefield has a separate DOF table for Nikon macro lenses.

http://zerenesystems.com/cms/stacker/docs/tables/macromicrodof

What do you think?

Thanks,

Bill

[Bart_van_der_Wolf]

Bart,

You raise an interesting point which I have not considered before, but that is not my understanding on how Nikon set and reports the aperture. Most recent Nikon lenses do not even have an aperture ring and the aperture is set on the camera.

I'm not a Nikon user, so I have to go with what people tell me. However, from the physical point of things, the 'Canon approach' and the 'Nikon approach', must both use the same physical aperture diameter (otherwise the resulting exposure time would differ, i.e. 4x longer at 1:1 magnification).

In that case, and you can only set the aperture (automatically converted to effective aperture) on camera, I'd suggest dropping the (1+m/P) term from the formula, and if the pupil factor is significantly different that 1 divide the result by P (since that seems to be not included in the reported aperture).

That would change the formula to DOF = 2 x N x c / (m^2 x P) for Nikon reported effective apertures, which should give the same result as the earlier formula for other cameras that report physical aperture.

With my 105 mm f/2.8 macro I can set the lens wide open at infinity and the camera reports f/2.8. If I rack out the lens for 1:1, the aperture is reported as f/5.6 f*(m+1) and there is no way to set the lens to f/2.8.

I can only assume that that is the reported aperture, but not the physical aperture. You can check by looking at the entrance pupil diameter, I assume that it does not significantly change as you go from infinity focus to 1:1 magnification.

The Photopills macro depth of field calculator has a special checkbox for Nikon which reports the effective aperture.

http://www.photopills.com/calculators/dof-macro

Also, Rik Littlefield has a separate DOF table for Nikon macro lenses.

http://zerenesystems.com/cms/stacker/docs/tables/macromicrodof

Yes, they also account for the differently reported Apertures.

It's interesting that Rik Littlefield arrives at larger steps than the DOF formula suggests. I have not asked him for his calculations, but since he is a very experienced micro- and macro-photographer I'd say it's usually safe to take his word.

Cheers,
Bart

[John Cothron]

John and Bart,

Congratulations to John for his excellent work and to Bart for his useful comments. From the EXIF I see that a Canon camera (which unlike Nikon does not report the effective f/number) and a 150 mm f/2.8 lens (? Sigma) were used for the stack. With this lens, he pupillary magnification is likely less than 1 and this could affect the calculations slightly.

In the DoF equation that Bart quoted I assume that the factor (1+m/p) corrects to the effective N from the nominal N. For Nikon users I assume that the factor would simply be (m/p). Is this correct?

Bill

Bill,

Thank you for the comment.  Yes I used a Sigma 150mm APO Macro for the shot.  I did a rather quick search the other day for the pupillary magnifcation on the web and didn't find a number so I used 1 for the calculation.  I definitely need to figure it out.

***Sorry for the late response to this thread, I had an operating system drive go out Saturday morning, so you might imagine what I have spent most of the weekend doing.  The upside is I didn't lose anything drastically important (like images).  Also, I'm now running Windows 10   I still have a lot of catch-up work to do but so far so good.

Thanks!

[Bart_van_der_Wolf]

I did a rather quick search the other day for the pupillary magnifcation on the web and didn't find a number so I used 1 for the calculation.  I definitely need to figure it out.

John, it's not that hard to figure out. Just hold the lens (without lens-cap and body-cap) at arm's length, and try to measure the apparent exit pupil (camera side), then the entrance pupil (subject side) diameter. The exit/entrance pupil diameter ratio is the Pupil factor. It's the apparent narrowing or widening of the physical aperture due to asymmetric optics. Modern Macro lenses often use internal moving focusing groups of elements, which can lead to different Pupil factors depending on the chosen magnification. In addition, internal focusing often changes the actual focal length, so formulas involving focal length are not as useful as those with magnification for close-up and Macro work.

Magnification factor also allows easier calculation of the 'bellows factor' (required exposure increase) for non-TTL measurements, it's just (1+m)^2 or more accurately (1+m/P)^2 , multiplied with a handheld exposure time metering.

***Sorry for the late response to this thread, I had an operating system drive go out Saturday morning, so you might imagine what I have spent most of the weekend doing.  The upside is I didn't lose anything drastically important (like images).  Also, I'm now running Windows 10   I still have a lot of catch-up work to do but so far so good.

In my experience, Win10 is an improvement over Win7 (and  8 / 8.1 which I skipped ).

The benefit of such a forced upgrade is that it gives an opportunity to redo the structure and leave a lot of obsolete stuff behind. It does take some time to reinstall lots of more urgently needed stuff. Good luck.

Cheers,
Bart


P.S. An excellent resource for such things is the website by Paul van Walree, a site that seems to have gone off-line recently. Fortunately, we still have the Web Archive's Wayback machine for access to the latest version here:
https://web.archive.org/web/20161227153359/http://toothwalker.org/optics/dofderivation.html

It explains the Pupil factor for asymmetric optics, and the Magnification based DOF formulas can be found as equation (12 and) 13 on that page.