Focus stacking together with panorama assembly

[John Fine]

I want to learn a decent workflow for doing focus stacking together with panorama assembly ("together" meaning NOT one before the other).  There are a bunch of different aspects of this for which I could use advice.

My camera doesn't do focus stacking.  With a camera that does focus stacking, the different focusses of each position of the panorama would be very easy to align well, so it would make more sense in post to merge each focus stack before assembling the pano.  Instead, I will be taking all the shots for one focus setting, then changing focus and taking them all again, which simplistically sounds like I should assemble a pano for each focus setting before merging by focus setting.  But that seems to not work right either.

I think the heart of the solution must involve the save-mask and load-mask operations in enfuse.  I haven't yet experimented with those to figure out proper use.  I want to have enfuse compute some contrast based weighting and save that.  Then I want to use other tools to examine that weighting and modify it.  Then I want to have enfuse use those weights rather than contrast etc.  Maybe I want also to get some kind of masks out of enblend as well to mix in.

I don't know whether save and load masks actually work that way, nor what other settings to use.  I don't know what tool to use to look at a grey scale masks together with the photo (make one semi-transparent to look at the other through it or what?)  I have Corel software that came with my camera for such things, but don't know how to use it, and/or I'll download whatever freeware is suggested for viewing/editing the masks.

I'm a software engineer.  For the alignment part of this workflow, there are a few small code changes I want to make to Hugin.  The code itself was easy for me to understand.  But the Windows build process is beyond cryptic.  The build instructions I found are full of necessary links all of which are broken (go to 404 page missing).  Once I start actually understanding enfuse/enblend I expect I'll want to change those as well.  Build suggestions appreciated.

I'm hoping/expecting to be able to construct masks that have 100% in certain areas (not isolated pixels) of high contrast for an individual photo (with 100% in any one forcing zero in overlapping part of each other photo).  Then have mixed weights (not 100% / 0% selection) in transitions between the selected part of one and the selected part of another as well is in the shared low contrast sections of all photos (I'm expecting to stack two or three photos with low depth of focus so objects further in the background intentionally remain blurred).

Sorry this is so long and implies questions for multiple different kinds of expert.  Thankyou for taking the time to read it and for any suggestions on parts of it.

[Bruno Postle]

Hi John, I have never done more than simple tests of focus stacking, but I expect that this technique (shooting a panorama at each focal distance) will be ok as long as each panorama has a similar layout and can be assembled by hugin into stacks, which can be fused and then blended into a final panorama.

Before you dive into saving and loading enfuse masks, or attempting entire panoramas in Hugin, I would experiment with a single stack to determine what enfuse parameters are necessary for the kind of photos you are taking.

Enfuse has extensive PDF documentation with a whole section on 'Focus Stacks - Depth-of-Field Increase':

 http://enblend.sourceforge.net/enfuse.doc/enfuse_4.2.pdf

--

Bruno

--

Bruno

[johnfi...@gmail.com]

I will keep trying things.  But from what I've tested so far, that idea doesn't work (Align within stacks and combine each stack before combining the results into a panorama.)

A "similar" layout isn't nearly good enough.  The layout would need to be near pixel perfect and I can't get close.  These are outdoors and the ground gets compressed under the feet of the tripod as you work.  That isn't the only source of error but it is a big enough source.  When you repeat a panorama with a different focus, the alignment is poor.

I have temporarily concentrated on getting a Windows build of these tools to work, because being able to modify the code can help my experiments even if I make no permanent enhancements.

I will later get back to experimenting with saving and loading masks.  Seeing a saved mask should also give me more insight into what is happening inside the tools and why that isn't doing the job for me.

I read all that documentation before trying any of this.

[Bruno Postle]

On Tue, 21 Dec 2021 at 23:40, john wrote:
>
> I will keep trying things. But from what I've tested so far, that idea doesn't work (Align within stacks and combine each stack before combining the results into a panorama.)
>
> A "similar" layout isn't nearly good enough. The layout would need to be near pixel perfect and I can't get close. These are outdoors and the ground gets compressed under the feet of the tripod as you work. That isn't the only source of error but it is a big enough source. When you repeat a panorama with a different focus, the alignment is poor.

Hugin should still be able to align these stacks, even if they are a
few degrees off, the sides can be a bit of a mess after enfuse has
merged them, but these sides get discarded when the panorama gets
blended. Another source of problems with focus stacks is that the
angle of view of the lens changes with different focal distances, so
you may need to optimise the angle of view of each layer separately.

--
Bruno