The GetDPI Photography Forum

Great to see you here. Join our insightful photographic forum today and start tapping into a huge wealth of photographic knowledge. Completing our simple registration process will allow you to gain access to exclusive content, add your own topics and posts, share your work and connect with other members through your own private inbox! And don’t forget to say hi!

Focus-stacking MF landscape images

This weekend I've been experimenting with focus stacking on GFX landscape files. I'm a bit surprised at the results and would appreciate hearing about different experiences.

It's about 3 months since I switched from A7r2 to GFX, and I've been concerned about DOF and about diffraction with lenses stopped down to compensate for longer lenses/larger sensor. I almost bought a used Hartblei tilt-shift lens to avoid stopping down, but I realized that my landscape shots too often have overhanging trees, so tilting the lens to increase DOF wouldn't be successful. Hence my interest in focus stacking.

I'm not posting samples for pixel-peeping, because the things I've noticed so far were pretty easy to see:

1. Using the 32-64 lens, diffraction isn't as much a problem as I thought it would be, given my customary print size (17x22 paper). F16 is acceptable, but I did my single-file comparison shots at f22 to create a worse-case scenario.
2. Focus stacking 3 to 5 files shot at f8 with Photoshop produced quite good results from the standpoint of overall sharpness. The drawback of PS's focus stacking was that it sometimes missed very small areas. That is, sometimes I could see at 100% that in a midrange-focused area, there was a little gap where it should have picked up some leaves from a farther-focused file but didn't do so. The faults seemed, however, to be so minor that I didn't think they'd need manual retouching.
3. Helicon Focus disappointed me, and here especially is where I'd like to know about others' experience. While it seemed to make the stacking selections more precisely, the resulting image was never as sharp as a Photoshop-stacked comparison image. It altered tones and contrast. I tried both Method A and Method B, and tried turning down the Smoothing setting to see if that made an improvement, but it didn't. In some instances my f22 unstacked comparison image looked sharper than the Helicon versions, except with lots of foreground. The f22 and PS versions were better than Helicon in preserving color balance.
4. One advantage when using a larger aperture and focus stacking is simply the shorter exposure! At f22, leaves and branches were more likely to blow during the necessarily longer exposure. With shorter exposures at f8, I could paint in things that had moved from another layer.
5. I've ended up believing focus stacking for my kind of work (landscape, not macro, photography) is less often useful or necessary that I'd thought it would be. The 32-64 lens is excellent at f11, and loses contrast, more than it goes soft from diffraction, at smaller apertures. When using the 64mm end, or when there's a long foreground with important subject matter, I'll make several files for stacking. But I expect to need them less often than I'd anticipated.
6. I don't plan to experiment with other stacking software, because in the few instances I'll be needing it, either the less-perfect PS stacks or the apparently lower resolution Helicons give me sufficient options.

I don't think I've made any big mistakes in my experiments, and perhaps the main variable in other folks' experience is that they make larger prints than I do, and are thus more concerned about diffraction.

Question: I'm a beginner at this; so has your mileage varied significantly from mine?

Kirk

PS, I posted one of the images here – 3 files with Photoshop stacking: https://www.getdpi.com/forum/medium...-backs/4730-fun-mf-images-447.html#post749527
 
Last edited:

drunkenspyder

New member
Kirk:

I am no expert, so please take this all with a grain of salt. I have been focus stacking for about two years now. Have tried Helicon, Zerene, and PS. I always come back to Helicon.Here is why:

1. Overall, I agree with your observation that stacking can be hit-or-miss [more the latter than the former] when shooting landscapes. Any movement, and I do mean any movement of almost any part of the stack, in any image of the stack, can have a negative effect on the outcome. I also use it far less for landscape than I thought I would. BTW, Lloyd Chambers moved into the stacking zealotry camp this past year, thanks in large part to the Nikon D850 stacking capability. Given where Lloyd shoots [high altitude California Sierras], this surprised me. But if all one shoots is tree trunks, rocks, scree, and snow, it makes sense [that is not a negative observation; Lloyd uses those subjects effectively to assess equipment]. But because I love long exposure work, stacking for my landscape work is just not effective.

2. I began stacking with my Nikon D810 and CamRanger, and then moved to Case Air controller software. I liked the results enough in my pseudo-macro work [bugs and plants] to realize that I would want this capability when I made the [return] move to MF. The PhaseOne AF lenses and stacking capability were an important selling point for me.

3. Phase One opened my eyes to the importance of getting enough images in a stack. I used to shoot 5-9 images, but in my experience, that's just not enough. A "short" stack for me now is 20-30 images. A typical stack is 80-150. That's why Phase One's automation of stacking is so handy. I could not do enough images manually and precisely. However, out in the field, you take what Mother Nature gives you, and so you might be able to snag only 3-5.

4. I have never found PS stacking to be superior to Helicon. However, I do sort through my stacks in Helicon, look for obvious problem images, and also try stacking with all three methods and different settings. It's laborious, but it can be worthwhile.I also hate PS, so that may be a factor. :rolleyes:

5. I have not had a color balance problem with Helicon [with one exception, and I believe that was due to passing clouds during my exposure], but your eye may be more critical than mine.

That's just my experience so far. Everyone's mileage may vary.

Greg

P.S. I have a three-lens Hartblei TS set in Nikon mount that I like a lot. But I am likely to sell that soon. However, your comment about hanging branches did make me think of something. I recently did a stack of some blooming plants in our garden. The distance between some of the blossoms and some of the background greenery was sufficient, even at f/8, to cause enough [bokeh?] blurriness in some of the frames that it wrecked any hope of stacking. Efforts to focus on your border branches may be compromising some of the stacking algorithm's effectiveness.

This weekend I've been experimenting with focus stacking on GFX landscape files. I'm a bit surprised at the results and would appreciate hearing about different experiences.

It's about 3 months since I switched from A7r2 to GFX, and I've been concerned about DOF and about diffraction with lenses stopped down to compensate for longer lenses/larger sensor. I almost bought a used Hartblei tilt-shift lens to avoid stopping down, but I realized that my landscape shots too often have overhanging trees, so tilting the lens to increase DOF wouldn't be successful. Hence my interest in focus stacking.

I'm not posting samples for pixel-peeping, because the things I've noticed so far were pretty easy to see:

1. Using the 32-64 lens, diffraction isn't as much a problem as I thought it would be, given my customary print size (17x22 paper). F16 is acceptable, but I did my single-file comparison shots at f22 to create a worse-case scenario.
2. Focus stacking 3 to 5 files shot at f8 with Photoshop produced quite good results from the standpoint of overall sharpness. The drawback of PS's focus stacking was that it sometimes missed very small areas. That is, sometimes I could see at 100% that in a midrange-focused area, there was a little gap where it should have picked up some leaves from a farther-focused file but didn't do so. The faults seemed, however, to be so minor that I didn't think they'd need manual retouching.
3. Helicon Focus disappointed me, and here especially is where I'd like to know about others' experience. While it seemed to make the stacking selections more precisely, the resulting image was never as sharp as a Photoshop-stacked comparison image. It altered tones and contrast. I tried both Method A and Method B, and tried turning down the Smoothing setting to see if that made an improvement, but it didn't. In some instances my f22 unstacked comparison image looked sharper than the Helicon versions, except with lots of foreground. The f22 and PS versions were better than Helicon in preserving color balance.
4. One advantage when using a larger aperture and focus stacking is simply the shorter exposure! At f22, leaves and branches were more likely to blow during the necessarily longer exposure. With shorter exposures at f8, I could paint in things that had moved from another layer.
5. I've ended up believing focus stacking for my kind of work (landscape, not macro, photography) is less often useful or necessary that I'd thought it would be. The 32-64 lens is excellent at f11, and loses contrast, more than it goes soft from diffraction, at smaller apertures. When using the 64mm end, or when there's a long foreground with important subject matter, I'll make several files for stacking. But I expect to need them less often than I'd anticipated.
6. I don't plan to experiment with other stacking software, because in the few instances I'll be needing it, either the less-perfect PS stacks or the apparently lower resolution Helicons give me sufficient options.

I don't think I've made any big mistakes in my experiments, and perhaps the main variable in other folks' experience is that they make larger prints than I do, and are thus more concerned about diffraction.

Question: I'm a beginner at this; so has your mileage varied significantly from mine?

Kirk
 
Hi Greg,

Thanks for this reply! Perhaps my results were due to sloppiness, because I'd read that 3 to 5 files would be plenty for landscape work. I'll have to do another experiment with the size of stack you're recommending – though it's hard to ask Mother Nature to hold still for so long!

I can see where deep stacks are needed for macro work, but I'm surprised that so many are advantageous for landscapes. (My landscapes are shot with fairly short lenses, never venturing beyond 63mm for GFX / 50mm for FF; so at f8 each file has appreciable DOF.)

Another bit of sloppiness on my part: I started with primes but found (to my surprise) that I liked the 32-64 zoom better, because resolution was excellent and the files were a little less contrasty. But this lens' MF distance shows up only in the EVF/LCD, which doesn't allow very precise adjustments – so I'd just been setting for 3, 5, and 10 meters, and sometimes 2m and/or infinity.

And finally, I never imagined checking each file for imperfections – I just looked at the final output to see if anything was amiss and in need of retouching.

For these reasons, it looks like I should spend another weekend – sometime – working on this sort of technical stuff. :(

I'm still in the trial month for Helicon Focus and will go ahead and purchase it – you seem to have tried 'em all and have come to an informed conclusion. But I suggest trying this just once: put the same stack through the Photoshop routine and through Helicon Focus, and look at 100% to see if there's not a noticeable difference in sharpness? I use files with even the default sharpening off, and I don't think PS slips in any sharpening via the Auto-Align and Auto-Blend functions. To me the difference was obvious and surprising at 100%, though I doubt it makes a visible difference in my prints on 17x22 paper.

Thanks again,

Kirk
 

drunkenspyder

New member
Kirk:

I will try that comparison you suggest.

I tend to shoot wider open when I stack, but even when I don’t, I have found he additional images make a difference in the final result. I have also found that, if each image occupies a shallower slice of the DOF, it can be easier to dispense with any single substandard image. It also helps make sure I don’t overshoot my desired focus field, near or far. It’s funny, because the number one thing I watch out for (after movement) is making sure that the parts of the image I don’t want in focus stay that way.

Hi Greg,

Thanks for this reply! Perhaps my results were due to sloppiness, because I'd read that 3 to 5 files would be plenty for landscape work. I'll have to do another experiment with the size of stack you're recommending – though it's hard to ask Mother Nature to hold still for so long!

I can see where deep stacks are needed for macro work, but I'm surprised that so many are advantageous for landscapes. (My landscapes are shot with fairly short lenses, never venturing beyond 63mm for GFX / 50mm for FF; so at f8 each file has appreciable DOF.)

Another bit of sloppiness on my part: I started with primes but found (to my surprise) that I liked the 32-64 zoom better, because resolution was excellent and the files were a little less contrasty. But this lens' MF distance shows up only in the EVF/LCD, which doesn't allow very precise adjustments – so I'd just been setting for 3, 5, and 10 meters, and sometimes 2m and/or infinity.

And finally, I never imagined checking each file for imperfections – I just looked at the final output to see if anything was amiss and in need of retouching.

For these reasons, it looks like I should spend another weekend – sometime – working on this sort of technical stuff. :(

I'm still in the trial month for Helicon Focus and will go ahead and purchase it – you seem to have tried 'em all and have come to an informed conclusion. But I suggest trying this just once: put the same stack through the Photoshop routine and through Helicon Focus, and look at 100% to see if there's not a noticeable difference in sharpness? I use files with even the default sharpening off, and I don't think PS slips in any sharpening via the Auto-Align and Auto-Blend functions. To me the difference was obvious and surprising at 100%, though I doubt it makes a visible difference in my prints on 17x22 paper.

Thanks again,

Kirk
 

JohnBrew

Member
Hi Kirk, #2, so you've noticed the most obvious problem with PS focus stacking. I love Helicon focus and have been using it for over a year. I import RAW images, then adjust the resulting tif file in ACR. Works very well for me and the prints @ 24" are excellent. I do this with Nikon and CFV-50c files.

The only problem with focus stacking for me is wind. Helicon won't tolerate anything but perfectly aligned files.
 
Hi John,

I took a look at your Nikon and tech cam landscapes and saw you’re getting fine results - so I want to ask: How many files have you been creating (on average) for them?

TIA for advice,

Kirk
 

JohnBrew

Member
Hi Kirk, I use between three and five frames for landscape images. I know some macro enthusiasts use forty and beyond!
Much of this is lens and aperture dependent. The wider the lens the least shots, the more stopped-down you are, also a lesser amount. However, with all the resolution available to us today I like to use an aperture which is as wide open as possible yet keeping the corners sharp. While I have a Zeiss 21, it is usually reserved for single shot images. My go-to lens for stacking is a Zeiss Otus 55 for the Nikon. I find I prefer the Rodenstock 70 for stacking with the tech-cam, but I'm growing fonder of the 180.
Maybe tmi here for you as I'm well aware of your prowess as a photographer.
 

dchew

Well-known member
Well Kirk, this thread got me thinking I should check to see if I find the same thing you did. Believe it or not, I have never before stacked in PS; it as always been in HF. So I tried a few different things on this image I posted in the technical camera images thread:



Workflow:
(4) Images: sk35xl, Alpa STC, IQ3100. All images at f/13. Camera ~2ft off the ice. Approximate focus distances (ft): 4, 8, 20, inf.
Process raw files in C1, export tiffs to LR. Open in LR, perform some additional minor adjustments in LR. Sync changes to all four images.
Open tiffs w/ LR adjustments to HF. Saved two different methods, B & C. Did not notice much difference between the two. Chose method B (depth map) as "slightly" better.
Open tiffs w/LR adjustments to PS. Load files into stack, align layers, blend layers. Export as new tiff.

I don't know if you can see it in this web jpeg, but here are screenshots in LR "compare" mode (PS left, HF right):




The PS image is better on my screen (PA241w), at least for this image. Not dramatically, but I can definitely see it. Hmf; you have me scratching my head. I am not sure what to make of this other than I will proceed with both methods for a while.

Interesting.

Dave
 

stngoldberg

Well-known member
I find this thread quite interesting; as I have been using Helicon Focus for quite a few years.
I find Helicon most useful outdoors photographing static images such as rocks, architecture, lighthouses etc; but I find that images involving objects that move such as leaves on trees and images that include clouds are best served by employing a technical camera with tilting the lens.
In most landscape images using my tech camera with a Rodenstork 32mm lens at f8 with 1 degree of tilt, the focus results are very good from the front of the image to the rear without the blur from objects moving.
I wonder what Dave’s image might have looked like with my setup? Usually, if I had focused my lens just beyond the log in the foreground of his image, the result would be quite satisfactory but may have required f11 or f16.
Stanley
 

dchew

Well-known member
You may very well be right Stanley. the camera was tilted about 10 degrees down, 2 feet off the ground and the closest distance in the frame was about 3 feet away. As you know, I cannot tilt with the 35xl. However, I ran this scenario through Torger's Lumariver DoF calculator, which I have set up in a way that matches my DoF expectations. Basically ~2.5 stops narrower than the traditional lens scales (f/13 corresponds with a bit less than f/5.6 DoF printed on the lens). Using those constraints, hyperfocal distance @ f/16 is 23.3ft with a close limit of 11.7 feet. So I couldn't do this in one shot without tilt.

Running the tilt scenario through Lumariver gives a "maybe" result, but 1 degree wouldn't do it because the ground is so close. 2 degrees would have rendered the foreground, but the top of the cliffs might have gone out of focus. That may not matter though, since in a shot like this I think the distant parts can be a bit soft without degrading the image.

The problem I have with tilt in this scenario (even if I could do it with this lens), is that it was dark. I am still not thrilled with the IQ's live view in the dark, and I find it quite difficult to determine sharp focus when tilting in low light. I often find myself throwing in the towel and just focus bracketing. That's annoying though, because I have to take an LCC for each one of those brackets!

Dave

 

stngoldberg

Well-known member
Dave,
You bring up another very interesting question...is an LCC necessary if it is dark...I really don’t know
Stanley
 

WildRover

Member
I use Helicon Focus, Zerene Stacker, and at times Photoshop.

I have noticed the same softening and image loss that you have discovered with Helicon being the worst. I suspect it has to do with the resizing of each image that the program needs to do as part of the stacking process. It is similar to the degradation from using transforming tools. For images that would require small stacks of two or possibly three images I have concluded that it is better to instead, if possible, use one image and a smaller f/stop with its accompanying diffraction loss than to stack. I found that Zerene Stacker does not become as soft and does a better job, but is harder to do retouching. Perhaps its because I haven't figured out how to display the mask while retouching in Zerene. In Helicon it is easy to go back and forth between the individual images and the mask while retouching and this helps considerably. Stacking with Photoshop does a great job, but is most useful I think with images where the composition lies on a progressive near - far image plane. Similar to compositions where tilt would be utilized. Dave, your beautiful image from your morning hike is such an example. Photoshop isn't as good at finding edges and delineating fine detail such as branches and leaves or other protrusions against a far background. With Photoshop and certain images, you can get a general idea where the focus zones are and then manually correct and paint the mask into the proper focus zones.

Also with Helicon I try to use the smallest radius and smoothing settings as possible so as not to introduce more blur between images than necessary. With deep stacks, there is also the blurry halos that occur that become a PITA to try and clone away. At smaller print sizes though, I doubt they would be of concern.

Helicon and other stacking software has its place, but also isn't a perfect solution either.

Rick
 
Thank you, Dave - Stanley - Rick, for confirming what I thought I saw and reminding me of the T-S option, though I too often work under overhanging branches that ‘outlaw’ tilting. From your discussion I’ve concluded there’s no one answer. So for landscapes I’ll regularly:

(a) try a stopped-down exposure, then
(b) expose an f8 stack, and then
(c) in post-processing, compare PS and HF versions to see which works best.

Fortunately SD cards offer room for many exposures, and few of my images are successful enough to make me go through this full range of options in PP!

One further reflection: Looking at the workprints a few days after my experiments, I reacted negatively to near-perfection. :) I felt that in highly detailed landscapes, especially if there’s no sky and details fill the entire print, ‘perfect DOF’ looks fidgety and almost annoying; and so option (a) above, the stopped-down exposure, might as Rick suggested make the more pleasing print. Or I might be happiest With 2, 3, 5, and 10 meter stacks, leaving out the excruciating little details attained at infinity.

Kirk
 
Last edited:

stngoldberg

Well-known member
Attached find an example employing tilt IMHO successfully.
This was a three shot panorama using more than one but less than two degrees of tilt on a Rodenstork 32mm lens. I focused on a position in the middle of the circular pattern which allowed a sharp foreground; with as Dave mentioned an acceptably slightly soft background.
Again as Dave mentioned (if I had chosen to focus stack) the task of juggling three or more images at different focus points and than a corresponding number of LCC’S (doing that three times for the panorama); then compiling those eighteen images (remember-nine LCC’S) in Capture One and next in Helicon....lots of effort....we can all do it but it’s a PITA
For me, TILT is a major reason I enjoy owning a technical camera!
Stanley
 
Last edited:

dchew

Well-known member
Dave,
You bring up another very interesting question...is an LCC necessary if it is dark...I really don’t know
Stanley
Stanley,
It is, but someday I want to investigate how much of a difference focus distance makes with this lens. At least from 1m to inf the lens isn't moving all that much. Here are the four LCC's for those images:



Dave
 
Attached find an example employing tilt IMHO successfully.
This was a three shot panorama using more than one but less than two degrees of tilt on a Rodenstork 32mm lens. I focused on a position in the middle of the circular pattern which allowed a sharp foreground; with as Dave mentioned an acceptably slightly soft background.
Again as Dave mentioned (if I had chosen to focus stack) the task of juggling three or more images at different focus points and than a corresponding number of LCC’S (doing that three times for the panorama); then compiling those eighteen images (remember-nine LCC’S) in Capture One and next in Helicon....lots of effort....we can all do it but it’s a PITA
For me, TILT is a major reason I enjoy owning a technical camera!
Stanley
Agreed, Stanley, that's a great example of tiltin'! But at the beginning of the thread, I posed the problem as photographing landscapes with nearby top-of-frame content, such as overhanging branches, which rule out the tilt option.

PumphouseFujiBW.jpg

Kirk

(GFX 63mm, 8 files stitched – and somehow oversharpened when I retrieved a small JPG version)
 

vjbelle

Well-known member
Kirk..... your image is the perfect example as to why tilt has limited use. Just as stacking has its limitations as nothing can move.

Victor
 

stngoldberg

Well-known member
Kirk,
Without having the opportunity of capturing your attached image, I don’t know whether TILT would be sufficient to get the overhanging branches in focus
Stanley
 
A week later: Some more focus stacking experiments with landscapes using GFX and 32-64 mm lens - comparing a single shot with lens at 44mm, focus at 5m, and f16; and four shots at f8 @ 3, 5, & 10m, plus infinity – the latter stacked in all of the Helicon Methods, and also in Photoshop. This time I wasn’t interested in pixel peeping at 100%; I only wanted to know what differences would matter in 15x20 prints.

I made files of this very ugly subject that extended from 3.5m to the horizon, had a very long tonal range, no big objects near the camera, and potentially excruciating detail:

BGTreePSStack.jpg

Of the Helicon Focus Methods, B was best for this image – A was softer, and C surprised me by producing ghosting.

As in my first experiment, PS stacking was sharper on the monitor at 100% than any of the Helicon Methods – but this made virtually no difference in 15x20 prints, with differences hard to pick out even with a loupe. (If you print larger, you might see more difference – but remember that the normal viewing distance will be greater too.)

The most interesting result was how little diffraction occurs with the 32-64 lens at f16. The file is a little flatter, but both foreground and horizon detail are acceptable in the print. (On the monitor at 100%, the f16 image is actually sharper than the Helicon-stacked version, except at nearest foreground and horizon – but this isn't of interest, except with a loupe, in the prints.)

I tried one more experiment with more attractive subject matter: Without wind, I was able to both focus-stack and then stitch 2 files into a square. The detail would be sufficient for an XXL print:

ScabbyBark.jpg

Overall conclusion: Focus stacking isn’t particularly useful for landscapes shot with ‘normal’ focal length lenses, when there isn’t an especially extended foreground. But you probably already knew that.

Kirk
 
Top