I’ve just recently started exploring using stacking in macro photography and attended a webinar last night where the presenter essentially admitted that stacked macro imagery is more art than science. I have struggled with how useful stacked images are when discovering that different devices yield different results. I did an experiment with my OM-1 and used in-camera stacking, then used the same set of images in Photoshop to create a stacked image. The Photoshop image resulted in an image where the body of the insect was about 2.5% wider when compared to the length of the body. Not much, but it was perceptible. I have found that Photoshop is more “forgiving” about using images that might be skewed more than the OM-1 finds acceptable.
Curious, I did some online research to see what guidelines they may be for using “enhanced” imagery in publications. This seems to be the most concise:
My first take-away on this is that “publishing” macro images by placing them on iNaturalist seems to create the obligation to disclose that the image has been created by stacking, and is therefore, “enhanced” and “modified”.
Notably, “focus stacking” is listed as generally acceptable, but anything that misrepresents the organism should be marked No to “Evidence accurately depicts organism/scene” in the Data Quality Assessment.
Thank you, but I did review this. It is clear that we can (and many do) use focus stacking. My concern is that we be clear so that folks who might use this platform for research (and may not be savvy to focus stacking’s limitations and methods of enhancement) are aware of what they are looking at or considering as “data”. This is especially true for some insects where measurement or proportionality of certain parts anatomical parts might relate to identification.
I’m a little off on the sidetrail, but more curious about your experience comparing in-camera vs Photoshop stacking. Myself, I found Pshop stacking to be horrendously slow. But yes, more forgiving than most other stacking-ware I tried. Until I tried Affinity Photo! It blew Pshop out of the water for speed and auto-align/transforming. Also by far the easiest to use. If you’re interested, they have a free trial. No wait, the whole Affinity package is free now, no strings attached. D’oh!
I have had success with using 4K video frames for my stacking. I just kinda move the shot in and out of the subject’s focus field for say 10 seconds, then fish for frame candidates to run through the stacker. All hand held. It actually works.
I haven’t had any issues with Photoshop being slow except when I was experimenting with 30+ images to determine how many I needed for a good stack. The answer for my insect work is somewhere around 8 frames for “tiny” (fruit fly) and about 18 for larger (bumblebee). I also find that more frames means more potential for artifacts.
What I found is that the OM-1 is really fussy about the images it uses for in-camera stacking. Essentially, it has to be done from a tripod, then using a remote shutter (phone app). Otherwise, just pressing the shutter button will create too much motion for enough images to use for a stack. Photoshop took all of them and just squashed them into a stack. But that did need a lot of cleaning up, later.
I’ll look into Affinity.
The video method is intriguing. Working on a syrphid project now and we’re trying to do it with photographs. After the first year, it is apparent that identifying them to species is not possible for about a third of the genera/tribes because some of the key features just can’t be captured on a moving subject. What camera are you using and how long does it take to sort through video segments for frames?
A lot of the photos of organisms I shoot below 1cm in size are stacks. I dont keep track of how many images are in each stack. Often I play around with different numbers, until the stack works. Of the hundreds of pics I have which are stacked, I wouldnt be sure how many images are in them. Its already a large number of hours processing raw images into stacks, its less of a motivation to track that information, than just gettings a usable image which shows multiple features of the organism in focus.
The stacked images I show are a true representation of the species, they arent adding anything but depth, which cannot really be caught on small organisms.
Images like this on a 2x4mm critter, you just wont get that detail without stacking.
But I couldnt tell you how many images were in the stack. There is possibly a way to find out. Not sure if they is kept somewhere in meta data when the stack is flattened?
That said, I only stack raws later in PS, and make sure the stacks work. I dont trust in camera stuff for this level of focus.
I use Helicon focus which works great for me, though I’ve heard Zerene stacker works great too. Photoshop is slow, and apparently does mess up sizes. :)
Would it be that much of a burden to just add a note indicating that the image is stacked? That’s really my basic point. Tell observers they’re seeing a stacked image just in case they don’t know.
My sense is that it adds to the credibility (and value) of your observation.
Possibly, take my recent trip to ecuador, that was ~13,000 images I processed. I process by week all the images, then I upload by week.
Ultimately thats 1753 observations (Most with multiple images) covering 784 species (Until more get IDed).
I could make a spreadsheet to keep track of which images, per observation, have how many images, for when I upload.
I understand if you are only processing a few images here and there it probably seems like nothing. But when I am doing hundreds of stacks, and spending a couple hundred hours processing images. Tracking 1000 images in a spreadsheet, to include that info for uploading, is more hours, when I am already trying to cut down on processing time.
If my images dont seem credible I am okay with that. For the people IDing, and those who use my ccby images, they seem easy/popular enough.
What value is lost when producing a image with more depth and features to ID?
I don’t think focus stacking is bad at all, but it is good to make a note of any processing that distorts the proportions, since little measurements can matter for insect ID. The difference between the worker forms of the ants Solenopsis carolinensis and Solenopsis molesta is the difference between the sum of antenna segments 3-8 being 26% vs 28% of the head length, for example (and they overlap around 27%).
Not that there is a clear delineation between the two, but I think it’s inaccurate to say that focus stacking is more art than science. It goes without saying that focus stacking is ubiquitous in scientific imagery.
I think it can be nice to disclose processing methods on iNat for others who are curious about macro photography.
When it comes to accurately portraying nature as we see it, I think all of the (AI) pre-processing happening behind the scenes in smartphones results in quite a few more artifacts than most focus stacking.
Every photograph could be described as a static, two-dimensional, hyper-real artwork made up of tiny dots, which has been optimised to suit human visual perception. Obviously, no photograph is never going to be a substitute for the real thing. But for our purposes, all that really matters is whether the subject is represented in a way that can be reliably identified by multiple independent observers (human or otherwise).
As we all know, there are a great many unmodified photos on iNaturalist that do not come very close to accurately portraying their intended subjects (to put it mildly). In artistic terms, they would be most fairly described as being impressionistic, rather than being remotely realistic. That doesn’t always prevent them from being reliably identified, though. So it seems reasonable to say that, in general, identification itself is as much of an art as it is a science. Identifying subjects from photograps (as opposed to physical specimens) is a specialised skill (or art) than needs lots of practice.
Okay - but why do you want to use these photos on iNaturalist? Why not just upload a group of unmodified photos that show all the relevant features? This is surely what the vast majority of observers on iNaturalist do. Personally, when I’m taking photos of smallish subjects in the field, I usually take about 20-30 from multiple angles and that is usually good enough to capture enough detail for identification purposes. Of course, it would be nicer to see more of the features with greater clarity in each photo - but it’s not actually necessary, right?
The only time I consider using focus-stacking is with very small subjects that can only be identified using microscopic features. Most microscopes have a very narrow field of view, so stacking is truly essential in some cases - for example, when trying to assess whether there’s more of one type of hair and/or puncturation than another across a given area. Again, it would certainly be nicer to achieve greater clarity across the whole frame of view, but for most practical purposes it isn’t really necessary.
Although I use focus-stacking quite rarely, I don’t feel it’s ever worth the bother of advertising that fact. The relevant features would hardly be visible at all if I didn’t use stacking, so the modifications are always much more than mere “enhancements”. In fact, I make a point of ruthlessly rejecting any focus-stacks that don’t significantly aid identification.
This is what I came here to say. The amount of processing that happens to turn the light bouncing off an organism into a digital image file is mind-boggling. The additional clean-ups, focus stacking, cropping, etc. that we do after-the-fact pale in comparison to what the digital camera is already doing to create an image representing what’s in front of the lens. Just changing around camera settings, changing the lighting conditions, or swapping out lenses will cause more significant alterations and warping to the image than most of the “processing” we do to the image later. So I’m a lot more forgiving of “enhancements” than some purist photographers seem to be.
I’d argue that all photography is a form of art- it’s a way we try to express our experience in a way that we can share it with others. These artistic renditions of what we experienced can be used for scientific pursuits, but they are ultimately all attempts at imitation of what we actually saw. So while the ethics of disclosing the methods used to create an image file might be very important in the photographer community, I think it’s less of a concern to those of us using iNat exclusively for scientific purposes.
I bought an OM TG-7 specifically for in-camera stacking. Most of my grass and moss close-up images are stacked. I need to capture features that are 0.1 mm and less: orientation of barbs, details of stellate hairs etc.
I use notes for describing smell, taste, iridescent hues and other details that the camera cannot capture or I just forgot to do so.
This is my disclosure of using stacking, I won’t be adding it to individual observations.
Often I feel I am a technician or detective just gathering facts.
That’s when I stop and take a couple shots just for myself, trying to capture the essence of the subject in a single image.
Distorting proportions on photographs is inevitable and its in no way a specific feature of focus stacking. Is literally impossible to represent all dimensions of a 3D object on a single 2D photograph. The only dimensions that are ever reliably measurable from a photograph are those in the plane perpendicular to the line of sight and even that has possible limitations due to lens distortion, in particular when it’s a wide-angle close-up (it’s negligible for telephoto).