When we upload observations with multiple images, sometimes it’s not very easy to decide which image to put first!
When I started iNat back in 2019 I didn’t really think of this seriously, but when I started using iNat more I realized how important this is, as only the first image of an observation is seen by people most of the time.
For example, when I have a really beautiful portrait of a millipede and a close-up shot of their reproductive organs (gonopods), I would want to put a portrait as the first image but often end up with the gonopod image, because it’s crucial for the ID and is very rarely photographed by people.
Usually this is resolved by finding two individuals of the same species and putting two observation, one with the portrait as the first image, the other with the gonopods as the first image - but when I can find just one, it’s pretty annoying.
Same struggle with UV photos - I always take photos under normal light as well as UV when I do UV macro photography, but then I have a hard time decide between which one to put first.
If it’s a common species UV comes first, but if it’s a rarely observed species, I sometimes opt for the normal light so people don’t get confused.
For the first image I aim to express the general image of what I see with my eyes when I focus on the subject. I dont want the background to be cluttered but im also not immediately looking through a lens. As far as specific anatomical features I personally would keep those as secondary photos unless I planned on adding the ob. to a specific grouping or project thats relevant. For UV photos it depends on if i can re-visit the subject; if I can, I make an entirely new ob. at night. If not I put it as a secondary photo as its not what i would naturally see with my eyes. Good topic.
I like to use the image that clearly shows the most identifiable features, for the following reasons. When I am selecting a few observations for identification from the mass of thumbnails in my feed, I want to know that the observation photos likely include the features required for accurate identification. If the first photo includes a flower and a leaf that fill the whole frame, I’m interested. If the first photo is a wide landscape shot or a drive-by car window shot, I’ll pass. I try to keep this in mind when taking my own observation photos too. The best photo to put first will vary by subject.
Also, since this feature request - https://forum.inaturalist.org/t/use-computer-vision-on-each-photo-in-an-observation/4210 has not been implemented, it is a lot of effort for an identifier to get CV suggestions for any images other than the first observation photo. I often use the Windows Snip & Sketch tool to clip out useful parts of observation photos to run through the CV demo at https://www.inaturalist.org/computer_vision_demo.
I’m always jumping between these two. I wish iNat will implement something like this: observations showing flowers is labelled as flowering, just like how it’s labelled Needs ID - or observations showing male reproductive organs labelled as so (important for many insects) - so IDers can easily see whether the observation includes what they want to see or not.
It shouldn’t be too hard to create different kinds of labels for different broad taxa.
I find the CV suggestions a useful backstop for age-related forgetfulness. It also seems to be the only way to keep up with the constant unannounced taxon name changes.
Being an amateur at this, I often find suggestions for new-to-me species that may be “expected nearby” that I am not familiar with but probably should be.
Except that what is relevant really depends. Some plants in some places can be ID’d even without flowers. Others cannot be ID’d unless you photograph the stem, or the underside of the leaves, or the fruits. Some male insects require genitalia to be visible. For others, it may be necessary to see the antennae, or the front legs, or the hair on the eyes, or the face, or the sterna, or something else. Or they may be so distinctive that they can be ID’d even from a slightly blurry photo from a couple meters away. This may vary even within a single genus.
In the time it would take for me to annotate an observation to indicate whether the features relevant for ID are visible, I could just as easily ID it.
I’m reluctant to assign (plant) IDs in most cases without looking at all the photos for the observation. I don’t find it overly important which one comes first.
If it’s something that is easy to ID, I often just pick the most aesthetically pleasing photo. If it’s something where the ID isn’t obvious I choose the image that shows the most diagnostic characters.
I pick whichever image shows the whole organism most clearly, unless it’s something really large (like a big tree partially obscured by other vegetation), in which case I’ll pick whichever image shows a relevant section of it most clearly.
I usually start with the whole animal (invertebrate usually) so that an identifier skimming the thumbnail pictures can see if it is an observation they would be interested in. The picture of the genitalia may be crucial for an identification, but it won’t tell a skimmer whether it is a fly, a millipede or a fungus.
Agree here, you are trying to catch the IDers attention, but with one additional consideration - making it easier for an observation to be included in a project. This may be the animals interaction with the environment, another organism, or most frequently for me, roadkill observations.
(I know that the final point comes with some diagreement about what is considered polite on iNat).
I honestly don’t think it is crucial which image is the first one in an observation. People should be looking at all images before making an ID, and close-up detail photos are generally hard to interpret unless there is also a larger photo of the entire organism, so I don’t see one as inherently being better than the other. And for a lot of organisms, it is not feasible to capture all the relevant characteristics in a single photo anyway. My personal preference is to put a whole-body photo first because this feels like it better reflects our natural thought processes – we don’t start to figure out what something is by looking at genitalia or hairs on the underside of a leaf, we start by putting it into a broad category by looking at macro characteristics/gestalt.
I do think it is considerate to make sure that the first image is relatively unambiguous about the subject of the observation and the context of the find. In other words: I avoid having the first image be one where the organism of interest is off at the edge of the frame or where it is easy to overlook because there are other organisms that are more prominent. If there is nonetheless potential for confusion I will mark the photo. My experience is that it is easy to be “primed” by the thumbnail to have particular expectations about the subject of the observation, which can lead to accidentally overlooking the relevant organism and IDing the wrong one, even if the observer includes a note.
For escaped garden plants I use a zoomed out shot that shows the larger environment, even if this means that the photo does not show important ID characteristics, because I find that this reduces the likelihood that people will reflexively mark it as “not wild” without reading my notes.
The computer suggestions are not a factor in deciding what to put first. I mainly upload via the website and if I want to compare suggestions I will simply leave the photos in separate cards and check the suggestions for each one before merging them.
If there are no strategic considerations, I may simply choose a photo that I like. Other times I might be interested in documenting an interaction or a behavior and I will upload the photos in chronological order, even if the first photo is not the most useful one for ID purposes.
There are a few users who upload all their images in a single montage photo. I think a montage can be a useful way to present all the relevant features at a glance, but I often find myself wishing they would additionally upload the individual component images separately, because the image size limit means there is often a significant loss of detail – I can’t enlarge the photo to see the original images at anything close to full size.
Amateur here… I try to choose the best view that depicts the organisms. It’s not aways the photo CV likes, I change the order to deal with that when I need.
I want the regular folks, like me, to see what I saw, the other photos, different parts, views, angles come after. The ones that ID know what they want to see, the rest of us want… enticed?
We want ‘everyone who IDs to look at all the photos, all the info from the observer’. But - reality check - if an identifier is scrolling thru a page to ID - it is the observer’s choice to offer
something blurry, but 3 is better …
a detailed field mark for the knowledgeable taxon specialists
a reasonable image, but wait, there’s more!
In the Identify modal we can see, there are more pictures, there are comments. But the hook to click for more is the chosen first picture.
I tend to put what I think is the most interesting or cool photo first, if I think one is worthy. Ya know, like when a hummingbird photobombs your thistle shot.
Putting content in the first photo may help train computer vision. I want iNaturalist computer vision to recognize Yellow Birch roots, so I put a lot of those first. In case you haven’t seen them, Yellow Birches will send their roots out above the wet forest floor, out in the air when the stump they grew on rots away, down ledges, etc.