XF Camera Focus Stacking, step by step, with Brad Kaye

[Steve Hendrix]

Sorry everyone that I've not been as present as usual, many of you, among others, are keeping me very busy with off-forum matters, that I'm grateful for. As we go along, all I can say is what an amazing time for camera technology! I'm sure this year will yield even more interesting developments.

This past week, Brad Kaye, our CI IEMT (Intrepid Expert of Many Things), spent some time with the Focus Stack tool from the version 2 software and published this hands on report.

https://captureintegration.com/phase-one-xf-feature-focus-stack-tool/


Also - CI has rated the firmware updates for XF and IQ cameras as *Approved for XF, *Not approved for IQ. See below link for details:

https://captureintegration.com/xf-feature-update-2-firmware-update-fw2-01-2-approved/


Steve Hendrix
CI

[Christoph B.]

Hmm some of the photos seem to have some sort of blur or rather a 'bloom' or halo... I assume it's because of the software Helicon, have you tried it with Photoshop as well?

I guess it's because the software recognises the widened shape of unsharp objects (such as flower pedals etc) and wants to keep that shape as "detail" in the final picture. Not sure if I'm expressing myself well enough to be understood.

[BernardLanguillier]

That can happen if there is too much focusing distance between shots.

Cheers,
Bernard

[Steve Hendrix]

That can happen if there is too much focusing distance between shots.

Cheers,
Bernard

Yes, in discussion with Brad, he indicated that possibility.


Steve Hendrix/CI

[Bart_van_der_Wolf]

Yes, in discussion with Brad, he indicated that possibility.

Hi Steve,

Besides setting the most near and far positions (and the camera doing the necessary interval focusing and adding the file sequence meta-data), does the user have any possibility to determine the intervals (number / COC)? I did not see that mentioned in Brad's article, but I do see a total number of exposures in the display capture.

Cheers,
Bart

[Wayne Fox]

Number of frames captured is set by the rear thumbwheel, simple math used to determine motor turns between captures through the sequence, but other than automating the process, there is no method of determining the number of captures needed to cover the focus range.  That’s still up to the user.  I believe in the article for the tire image he mentioned that 48 wasn’t enough, so he basically overshot and used 100 to make sure it was covered.

[Bart_van_der_Wolf]

Number of frames captured is set by the rear thumbwheel, simple math used to determine motor turns between captures through the sequence, but other than automating the process, there is no method of determining the number of captures needed to cover the focus range.  That’s still up to the user.

Hi Wayne,

Okay, number of focus bracketed as set by the user with thumb-wheel.

I believe in the article for the tire image he mentioned that 48 wasn’t enough, so he basically overshot and used 100 to make sure it was covered.

Yes, but it was not clear to me how that was done, and based on which criterion.

Cheers,
Bart

[Peter McLennan]

mentioned that 48 wasn’t enough, so he basically overshot and used 100 to make sure it was covered.

In my limited experience with focus stacking I've found that no matter how many images you think you need, you need more.
I manually adjust the focus by the minimum amount possible on the scale.  Like "half a line" per image.

[Bart_van_der_Wolf]

In my limited experience with focus stacking I've found that no matter how many images you think you need, you need more.
I manually adjust the focus by the minimum amount possible on the scale.  Like "half a line" per image.

Hi Peter,

I know that most people underestimate the optimum number, but it can be calculated in advance, or for a recurring scenario (front-rear distance, focal length, and aperture) it can even be empirically determined and re-used. Important is to use a CoC that equals the sensel pitch (or 1.5x pitch at most).

Cheers,
Bart

[iblink]

I am surprised there is no automatic calculation of the number of captures required. If the start and end points are fixed, as well as the f/stop, lens, and camera back, one could calculate the number of captures required for a given "sharpness quality" ( C1 would probably have to implement a high, medium and low quality setting as acceptable sharpness falloff is by definition subjective). I assume an automatic calculation has not yet been implemented simply due to the time required to prepare the tables for all the lens/back/fstop combinations (an admittedly high number).

This leads me to a question: when the number of captures are set, are the captures equidistant from each other based on the motor steps(or some linear scale), or on the depth of field? As the plane of sharp focus recedes depth of field increases, after all. There does not seem to be any documentation and it would be important to know if the software automatically takes this into account when stacking.

Finally, the start and end points may not be precisely repeatable, at least from initial reports. It would be good to know what's happening. Are there stats available that indicate the precision through the focus stacking? One could reasonably assume that at some point, under very high magnification and with small f/stops, the step motor is not sufficiently accurate to run automatic focus stacking.

[Wayne Fox]

I am surprised there is no automatic calculation of the number of captures required. If the start and end points are fixed, as well as the f/stop, lens, and camera back, one could calculate the number of captures required for a given "sharpness quality" ( C1 would probably have to implement a high, medium and low quality setting as acceptable sharpness falloff is by definition subjective). I assume an automatic calculation has not yet been implemented simply due to the time required to prepare the tables for all the lens/back/fstop combinations (an admittedly high number).

This leads me to a question: when the number of captures are set, are the captures equidistant from each other based on the motor steps(or some linear scale), or on the depth of field? As the plane of sharp focus recedes depth of field increases, after all. There does not seem to be any documentation and it would be important to know if the software automatically takes this into account when stacking.

Finally, the start and end points may not be precisely repeatable, at least from initial reports. It would be good to know what's happening. Are there stats available that indicate the precision through the focus stacking? One could reasonably assume that at some point, under very high magnification and with small f/stops, the step motor is not sufficiently accurate to run automatic focus stacking.

It may be somewhat challenging for the hardware inside the camera to calculate number of captures, although it certainly has enough information.  would be sweet if the number was set by default based on the two distances and the chosen f/stop, allowing a manual override.

As far as the distances, it appears to be pretty simple math, one shot at each point, remaining shots are even increments.  So If your points are 100 and 400 and you set it up to 4 shots you get 100,200,300,400.  I haven’t specifically tested this but in watching the increments when it shoots a stack it appears to work this way. Not a big deal, for many applications such as macro or studio work this is probably the best approach, and for landscape it seems to work fine, although you may want to add a couple of extra frames to make sure you get everything.

For me the end points seem repeatable, any focus issues are because the lens didn’t return to the correct position in my attempts to use AF.  But when i’ve tried it with manual focus the results seemed accurate.  Using AF is still challenging and reliability is somewhat sketchy.

[imagetone]

Hmm some of the photos seem to have some sort of blur or rather a 'bloom' or halo... I assume it's because of the software Helicon, have you tried it with Photoshop as well?

I guess it's because the software recognises the widened shape of unsharp objects (such as flower pedals etc) and wants to keep that shape as "detail" in the final picture. Not sure if I'm expressing myself well enough to be understood.

Helicon focus is an amazing tool, so I'm kind of surprised to see those halos.

They look like typical focus stacking artefacts. to me. In my experience different stacking methods produce different artefacts and to different degrees but those kinds of subjects with detail in different distinct and overlapping planes or surfaces rather than one continuous receding surface like the tyre, produce these artefacts that require retouching in the stacking software or afterwards in PS.