Focus shift on D850, Z6 etc

[Edwin Barkdoll]

I've fiddled around some with focus shift on the D850 and although I like the idea I'm puzzled about some of its implementation. I also use Congnisys's StackShot as well as simply manually changing focus with the lens or advancing the camera on a rail. So far my main use of FS has been in the field where the rail setup isn't practical, I don't have it with me or I'm shooting a landscape.

 

For me an advantage of Cognisys's method is that you can a) specify the start and endpoints of your stack, the number of shots to take and the software calculates the stepsize or b) specify the starting point, step size and number of shots to take. With each method the actual step size is known and can be roughly matched to the depth of field given lens parameters so that a smooth stack can be constructed.

 

With Nikon's FS the step size is a dimensionless number 1-10 and starting point and number of shots are user specified. Through some trial and error, this yields a series of photos that produces a good stack.

 

I don't see the advantage of Nikon's FS and don't see how it would have been difficult to implement a similar focus shift method where the step size is actually known.

 

Anyone else use Nikon's Focus Shift?

 

[ilkka_nissila]

I use it and find determining the right step size difficult. I tend to error on the side of caution and use a small step size. Nikon have given some tables to help the users in this document:

 

http://download.nikonimglib.com/archive3/olWKv00OIiiw03YkdPO64B8kqA89/D850_TG_Tips_(En)03.pdf

 

It would seem based on the tables that the focus step size is somehow related to the depth of field and a smaller number of shots using a given step size parameter setting is needed at f/8 than f/5.6.

 

One has to wonder why the step size units are not specified so that we could do our own calculations.

[Edwin Barkdoll]

I use it and find determining the right step size difficult. I tend to error on the side of caution and use a small step size. Nikon have given some tables to help the users in this document:

 

http://download.nikonimglib.com/archive3/olWKv00OIiiw03YkdPO64B8kqA89/D850_TG_Tips_(En)03.pdf

 

It would seem based on the tables that the focus step size is somehow related to the depth of field and a smaller number of shots using a given step size parameter setting is needed at f/8 than f/5.6.

 

One has to wonder why the step size units are not specified so that we could do our own calculations.

 

Thanks; the tables are consistent with what one would expect qualitatively. It would be nice to know the numbers.

[mike_halliwell]

One has to wonder why the step size units are not specified so that we could do our own calculations.

I've always wanted a very simple system of...

 

1 Press here to START

2 Press here to STOP

3 40 slices please HAL.

4 GO!

 

Sorry Mike, I can't do that...!

 

I mean, what's a touch screen for? Press it....:p

[ilkka_nissila]

What I would like to see is autofocus tilt-shift Nikkors with automatic focusing and tilt that would be used in the following way: user turns on LV, touches on three points in the image after which the lens would adjust the plane of focus through all there points automatically and recenter the image using automatic shift. I do more or less this manually and it can sometimes take me 15 min to be satisfied.

 

Of course, additional depth of field could be achieved by applying focus shift (and post-capture focus stacking) using the AF AT PC Nikkor images.

 

But, the camera manufacturers don't seem interested in my needs in the slightest degree. ;-) That's okay, I can work the lens manually.

[mike_halliwell]

But, the camera manufacturers don't seem interested in my needs in the slightest degree.

Typical selfish manufacturers....:p

[mike_halliwell]

One has to wonder why the step size units are not specified so that we could do our own calculations.

I guess it's 'increments' are related to degrees of rotation or smaller subsets thereof, rather than actual linear dimensions in the real world.

[ilkka_nissila]

I guess it's 'increments' are related to degrees of rotation or smaller subsets thereof, rather than actual linear dimensions in the real world.

 

but why then stopping down the lens from f/5.6 to f/8 with subject distance range unchanged results in a smaller number of focus steps required to cover the subject without changing the step size? These tables would suggest that Nikon have attempted to compensate for the increased depth of field at smaller apertures without the user having to dial in larger step sizes to account for more depth of field.

 

I can think of a few reasons why they might not want to specify the units. Older motors may not give completely reproducible steps and so Nikon don't want to claim that the step size is exactly such and such.

 

But it would really help me if they could at least give an approximate formula of how the step size relates to depth of field, or whatever they attempted to achieve with this approach.

[mike_halliwell]

but why then stopping down the lens from f/5.6 to f/8 with subject distance range unchanged results in a smaller number of focus steps required to cover the subject without changing the step size? These tables would suggest that Nikon have attempted to compensate for the increased depth of field at smaller apertures without the user having to dial in larger step sizes to account for more depth of field.

I suppose that they have a certain CoC in mind and account for it regarding aperture?

 

I still think this is cart-before-the-horse thinking.

[roland_vink]

What I would like to see is autofocus tilt-shift Nikkors with automatic focusing and tilt that would be used in the following way: user turns on LV, touches on three points in the image after which the lens would adjust the plane of focus through all there points automatically and recenter the image using automatic shift. I do more or less this manually and it can sometimes take me 15 min to be satisfied.

It would be even better if a camera was designed so the sensor can tilt, so it would work with every lens not just expensive tilt/shift lenses. Sensors already vibrate to shake off dust and move to stabilise the image, why not add tilting to the mix? By tilting the sensor only there is no need to re-center the image, the framing stays the same, only the plane of focus is tilted so it fits best over the subject. The camera could have a landscape mode where the sensor tilts to maximise sharpness across the entire field.

 

And while they are at it, add forward/backward movement of the sensor also. This would allow limited AF with manual lenses - the user focuses approximately and the camera moves the sensor forward or back to refine the focus, for example in portraiture to ensure the eyes are in perfect focus. It would be useful for focus stacking also. Just a couple of mm would be enough to cover most cases, especially for wide-standard lenses.

[mike_halliwell]

It would be even better if a camera was designed so the sensor can tilt, so it would work with every lens not just expensive tilt/shift lenses. Sensors already vibrate to shake off dust and move to stabilise the image, why not add tilting to the mix? By tilting the sensor only there is no need to re-center the image, the framing stays the same, only the plane of focus is tilted so it fits best over the subject. The camera could have a landscape mode where the sensor tilts to maximise sharpness across the entire field.

 

And while they are at it, add forward/backward movement of the sensor also. This would allow limited AF with manual lenses - the user focuses approximately and the camera moves the sensor forward or back to refine the focus, for example in portraiture to ensure the eyes are in perfect focus. It would be useful for focus stacking also. Just a couple of mm would be enough to cover most cases, especially for wide-standard lenses.

I sincerely hope someone has you on retainer because all of these ideas are relatively easily doable....now....:cool:

[roland_vink]

The "landscape mode" I mentioned is really a self-aligning sensor. Whether the image plane is tilted or perpendicular, it can eliminate problems where the lens/adaptor/TC/camera/sensor mount is out of alignment due to manufacturing tolerances or knocks, which is increasingly a problem as camera resolution gets higher. It will ensure the sensor is always properly aligned to the image plane when the picture is taken.

[mike_halliwell]

Auto Scheimpflug?

[Mary Doo]

Auto Scheimpflug?

Unlike some other cameras, such as Olympus M43, Nikon's focus-stacking procedure stops short of combining all the shots in the sequence into one final image. It is "Auto Scheimpflug" enough; and it actually provides the photographer more control on which images to include or exclude into the final composite. I think this works better. I had tried this feature with Olympus M43 casually and it did not work out very well and I did not care to pursue it further. To be fair it was most likely user-error :eek:.

[ilkka_nissila]

If the sensor is tilted, the object shapes in the image change. I am not sure I would want the camera to make such changes automatically. But, I would not object to more freedoms of movements if technically feasible. Edited March 17, 2021 by ilkka_nissila

[mike_halliwell]

Unlike some other cameras, such as Olympus M43, Nikon's focus-stacking procedure stops short of combining all the shots in the sequence into one final image. It is "Auto Scheimpflug" enough; and it actually provides the photographer more control on which images to include or exclude into the final composite.

 

That's got nothing to do with...Scheimpflug....;)

 

Principle

 

The basics of the idea is that the plain of focus is manipulated by angling the lens plain and the sensor plain and the subject plain respective to eachother.... usually to meet at a theoretical point. This is often done to maximise the DoF, but can be the opposite if needs be to produce a very, very thin band of sharpness.

[Mary Doo]

That's got nothing to do with...Scheimpflug....;)

Not directly but it does. It is a big word, yes. ;)

The basics of the idea is that the plain of focus is manipulated by angling the lens plain and the sensor plain and the subject plain respective to eachother.... usually to meet at a theoretical point. This is often done to maximise the DoF, but can be the opposite if needs be to produce a very, very thin band of sharpness.

At the end of the day it is much simpler, more cost-effective, more practical, more pixels, time-saving, to use Focus-stacking to achieve the target DOF one desires than to buy and deal with expensive fast tilt-shift lenses. More importantly, it is unlikely what you described would become reality in non-medical and non-engineering camera designs such as the Nikons we use.

Edited March 17, 2021 by Mary Doo

[mike_halliwell]

Focus-stacking to achieve the target DOF one desires

Try that with a scene where stuff moves! Think crashing waves and scudding clouds with everything from the pebbles at your feet to the horizon....sharp...;)

 

But, yes it is niche. However, that's why digital backs are made to fit on monorail cameras with full movements, usually 5" x 4". But it would work for MF and 35mm too!

[Mary Doo]

Try that with a scene where stuff moves! Think crashing waves and scudding clouds with everything from the pebbles at your feet to the horizon....sharp...;)

True. Fortunately perfect depth of field of these subjects are generally not necessary nor desirable. At most it is a just novelty which soon gets boring. However, I can see why it is needed in the medical field; say, an ophthalmologist dealing with a twitching patient with rapid blinking eyes.

Edited March 17, 2021 by Mary Doo

[mike_halliwell]

Fortunately perfect depth of field of these subjects are generally not necessary nor desirable

I think all 3 of Ilkka's great 'Frozen' shots in last week's Nikon Wednesday were with Shift or Tilt'n'Shift lenses for this very purpose.

[q.g._de_bakker]

That's got nothing to do with...Scheimpflug....;)

[...]

The basics of the idea is that the plain of focus is manipulated by angling the lens plain and the sensor plain and the subject plain respective to eachother.... usually to meet at a theoretical point. This is often done to maximise the DoF, but can be the opposite if needs be to produce a very, very thin band of sharpness.

Not maximizing DoF at all!

 

The Scheimpflug principle positions the image plane such that all of the subject plane (which we have to select) is brought into focus. It changes the conjugate distances across the image, as required for a sharp image of every part of a subject plane that is not parallel to the image plane.

DoF has nothing to do with that. DoF depends on (aperture and) those conjugate distances, so is not equal across the image (smaller in the part of the subject closest to the camera. Larger in the part of the subject furthest away). But It is not maximized at all. As always, to maximize DoF, you have to stop down as far as you can (not a good idea, by the way).

Yes, you can turn the principle around, and position image and subject planes such that only a very small part of the subject plane is in focus. But again, that is not about DoF, but about actual focus. You defocus so much that even DoF can't create the illusion of sharpness.

Edited March 17, 2021 by q.g._de_bakker

[mike_halliwell]

But It is not maximized at all. As always, to maximize DoF, you have to stop down as far as you can

To MAXIMISE the DoF, at a set aperture, in the plane you select/require, you utilise the Principle.

 

The Field, in the context in DoF, still extends both sides (not always equally) of the nominal plane and is often wedge shaped.

 

Scheimpflug dictates the plane, aperture decides on the planes apparent depth....they are inextricably connected and decided upon by the final choice of CofC.

 

Oh, and if we want to split hairs, there is no such thing as DoF, just an allowable degree of softness, both sides of the plain.

 

Oh, and I did use the terms 'the basics'... and... 'is often done to maximise the DoF'...,.;)

 

Anyway, Focus Stacking has very little to do with Scheimpflug...!

[Edwin Barkdoll]

Try that with a scene where stuff moves! Think crashing waves and scudding clouds with everything from the pebbles at your feet to the horizon....sharp...;)

 

But, yes it is niche. However, that's why digital backs are made to fit on monorail cameras with full movements, usually 5" x 4". But it would work for MF and 35mm too!

 

Well, I was all set to sell my 24mm PC-E Nikkor but you've almost convinced me to haul it out and dust it off! I'll set myself a challenge of shooting something with it by next week's Nikon Wednesday.

[mike_halliwell]

Last time I tried this was a low angle shot of spring crocuses on a slightly breezy day with my 85mm T&S PC, (non E version)

 

Use a nice tripod and have fun...:D

[q.g._de_bakker]

To MAXIMISE the DoF, at a set aperture, in the plane you select/require, you utilise the Principle.

 

The Field, in the context in DoF, still extends both sides (not always equally) of the nominal plane and is often wedge shaped.

 

Scheimpflug dictates the plane, aperture decides on the planes apparent depth....they are inextricably connected and decided upon by the final choice of CofC.

 

Oh, and if we want to split hairs, there is no such thing as DoF, just an allowable degree of softness, both sides of the plain.

 

Oh, and I did use the terms 'the basics'... and... 'is often done to maximise the DoF'...,.;)

 

Anyway, Focus Stacking has very little to do with Scheimpflug...!

 

No, Mike. All that the Principle does is tilt the plane of focus.

It does very little to DoF (i explained what). And what it does is in no way maximize DoF.

(Just try to explain how tilting the plane maximizes DoF, at a set aperture.)

 

And that's not a matter of splitting hairs. It's in principle wrong to even associate Scheimpflug with anything related to DoF.

 

To MAXIMIZE DoF, the only thing you can do is stop a lens down.

Edited March 17, 2021 by q.g._de_bakker