The Nikon D850 and Focus Stacking

[Michael Erlewine]

Most folks who know my work know I am interested in focus stacking and have been for years. In the course of the last year and a half I have been through (IMO) a little photography hell, trying to find a camera to work with (or replace) my Nikon D810. This involved ordering and waiting for months and months for a copy of the Hasselblad X1D and then the Fuji GFX, both of which did not work out (for my purposes) as I had hoped they would and they cost a bundle, once you started adding lenses to the mix.

And of course, Nikon was conspicuous by its absence all that time, so much so that I kind of began to give up on them, although I am a confirmed Nikon user and have way too many lenses for that mount.

Then the Nikon D850 arrived and was delivered relatively quickly. But the question remained, aside from its many new features, most of which I will not use much (like high ISOs and fast autofocus), how does it work for what I really care about, which is working with LiveView and low ISOs... and especially focus stacking?

Well, the verdict is in (for me anyway); it works well for everything I need it to. Well, for starters the new LiveView screen is considerably better than the D810. And marvel of marvels, I was surprised to find how much I love the ability to run silent with Electronic-First-Curtain and no mirror slap. Wow!

As they say, “Silence is golden,” and it really is when stacking 100 layers for one photo. And it is very much faster because I’m no longer doing mirror-up and waiting for the vibration to die down. Without the sound of shutter activation, I just watch the LiveView screen update the screen with each press of the remote. And everything proceeds so quickly and easily.

I have not tried the automatic focus-stacking feature because I like to march around objects (especially round or spherical subjects) when focus stacking, and even increments defeat that. I find the whole process of stacking focus something I mostly enjoy doing by myself.

So, anyway, for those who wonder how the D850 takes to focus-stacking, my answer is better than I could have ever expected. I’m still working on the processing in post and, although the color of the D850 is different (to my eyes) than the D810, with not too much adjustment in my process, the results are what I am used to. I can’t say (or yet tell) if they are better.

I can say that Nikon D850 has killed my interest in medium-format cameras, especially in terms of computer-post. I have a very fast (and expensive) computer, custom made. Even so, I can feel the difference in processing moving to 47 Mpx as compared to 36. It’s OK, but I could see that moving to 75m or 100 Mpx would seriously impact my patience. At least for now, 47 Mpx does all I need, and I very much notice that small difference in sensor size between the D810 and the D850. For me it is just enough to push me over the edge into what I have been looking for.

Photos with the D850 and the APO El Nikkor 105mm on the Cambo Actus.

[Bart_van_der_Wolf]

Hi Michael,

Thanks for sharing your impressions. You being an experienced Focus-stacking user, makes the comments more valuable to others.

Cheers,
Bart

[Rory]

I have not tried the automatic focus-stacking feature because I like to march around objects (especially round or spherical subjects) when focus stacking, and even increments defeat that. I find the whole process of stacking focus something I mostly enjoy doing by myself.

So, anyway, for those who wonder how the D850 takes to focus-stacking, my answer is better than I could have ever expected. I’m still working on the processing in post and, although the color of the D850 is different (to my eyes) than the D810, with not too much adjustment in my process, the results are what I am used to. I can’t say (or yet tell) if they are better.

Can you please clarify Michael as I'm not sure I understand what you mean.  Are you saying that you have not used the D850 focus shift feature?  Are you manually changing the focus between shots?  What do you mean by "I like to march around subjects"?

[Michael Erlewine]

Can you please clarify Michael as I'm not sure I understand what you mean.  Are you saying that you have not used the D850 focus shift feature?  Are you manually changing the focus between shots?  What do you mean by "I like to march around subjects"?

I don't use the D850 focus-feature, which is stepping through even increments, etc. And this because our subject requires finer or courser increments if we a photographing a sphere or globe. A sphere requires very tiny increments to capture its rapidly changing surface as it appears to our eyes. Try it. So, I prefer focusing by hand for this reason.

[bjanes]

I don't use the D850 focus-feature, which is stepping through even increments, etc. And this because our subject requires finer or courser increments if we a photographing a sphere or globe. A sphere requires very tiny increments to capture its rapidly changing surface as it appears to our eyes. Try it. So, I prefer focusing by hand for this reason.

Michael,

The D850 focus stack step size is not even but varies with magnification. Here are some preliminary results with the D850 and the 105 mm AFS lens. I set the focus step to 5. The aperture was set at f/8. As the graph shows, the step size increases with decreasing magnification. This makes sense, since the DoF increases as the subject distance increases and the magnification decreases.

Regards,

Bill

[Bart_van_der_Wolf]

The D850 focus stack step size is not even but varies with magnification. Here are some preliminary results with the D850 and the 105 mm AFS lens. I set the focus step to 5. The aperture was set at f/8. As the graph shows, the step size increases with decreasing magnification. This makes sense, since the DoF increases as the subject distance increases and the magnification decreases.

Hi Bill,

I think Michael is not only referring to the shallower DOF at higher magnification factor (closer manual focusing) but also to the need to use smaller step increments when the surface normal of the subject is more tilted away, like at the perimeter of a sphere. So what may be an adequate step interval for relatively flat surfaces, may be too large for such subject's edges.

When shooting with an automatic focus rail, I tend to use the more strict approach, by calculating the DoF for the slices that all have the same Magnification factor and using that for the whole run. One can always toss out the superfluous slices before stacking. Better safe than sorry.

Cheers,
Bart

[32BT]

But purely for entertainment purposes it would be nice to test and see the camera's internal result, no?

[Bart_van_der_Wolf]

But purely for entertainment purposes it would be nice to test and see the camera's internal result, no?

Yes, it looks like a useful feature with AF lenses.

Cheers,
Bart

[bjanes]

When shooting with an automatic focus rail, I tend to use the more strict approach, by calculating the DoF for the slices that all have the same Magnification factor and using that for the whole run. One can always toss out the superfluous slices before stacking. Better safe than sorry.

Bart,

That is an interesting approach. The magnification changes with the object distance and what do you mean for equal magnification?

Also, equal steps is ambiguous. With an automatic focus rail such as the Stackshot the step size is constant in most cases. When one is stacking using the focusing ring using steps equal in terms of angular rotation of the focusing ring using a lens with a linear focusing helicoid, the lens moves in equal steps according to the pitch of the helicoid, but the focus distance steps are nonlinear, increasing with greater object distances.

For interested readers, Marianne Oelund gives the applicable formula in this thread on the DPReview forum.

Regards,

Bill

[Bart_van_der_Wolf]

Bart,

That is an interesting approach. The magnification changes with the object distance and what do you mean for equal magnification?

On an automatic rail, one normally uses a lens at a fixed distance or magnification setting. Manual focus lenses cannot be Auto-Focused anyway. So, given a relatively large magnification factor, one can easily pre-calculate the DoF with a simplified formula that's accurate at short distances (much shorter than the hyperfocal distance). That Dof is then used for the most critical 'edge' features.

Also, equal steps is ambiguous. With an automatic focus rail such as the Stackshot the step size is constant in most cases. When one is stacking using the focusing ring using steps equal in terms of angular rotation of the focusing ring using a lens with a linear focusing helicoid, the lens moves in equal steps according to the pitch of the helicoid, but the focus distance steps are nonlinear, increasing with greater object distances.

Yes, it depends on lens design how the focus distance (and thus magnification factor) changes with equal rotation steps. This may differ between brands. That also makes a built-in-camera focus-stepper interesting, because it can incorporate knowledge of this relationship, coupled with feedback from the AF-lens' distance setting.

For interested readers, Marianne Oelund gives the applicable formula in this thread on the DPReview forum.

Yes, Marianne is very knowledgeable, about Nikon equipment in particular.

Cheers,
Bart

[bjanes]

On an automatic rail, one normally uses a lens at a fixed distance or magnification setting. Manual focus lenses cannot be Auto-Focused anyway. So, given a relatively large magnification factor, one can easily pre-calculate the DoF with a simplified formula that's accurate at short distances (much shorter than the hyperfocal distance). That Dof is then used for the most critical 'edge' features.

Quite true. Sorry, I was not thinking clearly and should have known better. Here is an experiment I did some time ago, stacking a set of rulers first focusing by ring (focusing ring) and then with a focusing rail. With the rail, magnification is constant from front to rear. When the images are stacked with the focus by rail, the stacking algorithm has to adjust magnification resulting in some cropping of the image. The stack by ring is more pleasing. Helicon Focus was used here. This confirms Rik Littlefield's recommendation that for a bouquet of flowers size image, one should focus by ring. At high magnification, as with insect photography, the rail comes into its own.

Regards,

Bill

[Bart_van_der_Wolf]

Quite true. Sorry, I was not thinking clearly and should have known better. Here is an experiment I did some time ago, stacking a set of rulers first focusing by ring (focusing ring) and then with a focusing rail. With the rail, magnification is constant from front to rear. When the images are stacked with the focus by rail, the stacking algorithm has to adjust magnification resulting in some cropping of the image. The stack by ring is more pleasing. Helicon Focus was used here. This confirms Rik Littlefield's recommendation that for a bouquet of flowers size image, one should focus by ring. At high magnification, as with insect photography, the rail comes into its own.

Yes, although there's another aspect to it, perspective change.

With a rail, each shot is taken from a different position, and the shifting entrance pupil thus produces a slightly different perspective for each slice (most of it out of focus). This can create issues with occlusions.
With lens focusing on the other hand, if internal focusing is used (instead of simply adding an extension), the shift of the entrance pupil position is likely to be less, and thus perspective will also change less for each slice.

Cheers,
Bart

[fdisilvestro]

With a rail, each shot is taken from a different position, and the shifting entrance pupil thus produces a slightly different perspective for each slice (most of it out of focus). This can create issues with occlusions.

With a rail you may also move only the rear standard (camera) and leave the entrance pupil at the same position to avoid perspective changes

[bjanes]

With a rail you may also move only the rear standard (camera) and leave the entrance pupil at the same position to avoid perspective changes

Moving the rear standard is ideal except for high magnification, but AFAIK the Stackshot and other automated rails can not move the rear standard and the rack and pinion focusing found on many rails is too coarse for high magnification work.

Bill

[HSakols]

I'm curious if anyone is using focus stacking for non macro landscapes.  And how large are these files after they are processed?

[leuallen]

Using mf3. Tried EM-1 II focus stacking with various setting. Unsatisfactory results: the best result had 3 usable frames and 8 or so out of focus, next best was similar but not as sharp, the remaining 4 or 5 tries were all out of focus except for the first shot. It is easier to just do it manually: take close focus, one or two at middle distance, and final at infinity.

Most of my landscapes are with a telephoto lens. Flat farm country. Results may be different with wide or normal lenses.

Normal image size is 18mb, flattened stack is 115, layered stack 300+mb. This is when I was using more frames, up to 9 or so.

Larry

[MeCruncher]

Using mf3. Tried EM-1 II focus stacking with various setting. Unsatisfactory results: the best result had 3 usable frames and 8 or so out of focus, next best was similar but not as sharp, the remaining 4 or 5 tries were all out of focus except for the first shot. It is easier to just do it manually: take close focus, one or two at middle distance, and final at infinity.

Most of my landscapes are with a telephoto lens. Flat farm country. Results may be different with wide or normal lenses.

Normal image size is 18mb, flattened stack is 115, layered stack 300+mb. This is when I was using more frames, up to 9 or so.

Larry

Wonderful result

[bjanes]

Using mf3. Tried EM-1 II focus stacking with various setting. Unsatisfactory results: the best result had 3 usable frames and 8 or so out of focus, next best was similar but not as sharp, the remaining 4 or 5 tries were all out of focus except for the first shot. It is easier to just do it manually: take close focus, one or two at middle distance, and final at infinity.

Most of my landscapes are with a telephoto lens. Flat farm country. Results may be different with wide or normal lenses.

Normal image size is 18mb, flattened stack is 115, layered stack 300+mb. This is when I was using more frames, up to 9 or so.

Larry

Look at this post by Diglloyd. He finds the focus stacking provision of the D850 to be invaluable in landscape work.

Regards,

Bill

[HSakols]

Thanks Larry for the example.  Bill, thanks for the link regarding focus stacking. Diglloyd provides a very good write up on the subject. 

[leuallen]

Works for the 850 but not the EM-2 II. I tried it on more than one occasion with no luck.

Larry