I like to take nice photos of organisms, but often have trouble deciding which photos are most helpful. Are there any guidelines as to how many photos / how high a resolution is appropriate for iNat observations?
The common sense advice is: try to include only photos that identify novel features, and try not to include higher resolution that image quality (focus / blur / distortion) support. But sometimes photos capture different combinations of features, and it’s difficult (esp. for unfamiliar taxa) to tell how much the fine-resolution is crucial.
Any advice? Would it make sense to add a flag for “redundant” or “over-resolution” photos, to help reduce serve load?
(Finally - if a photo is sourced to Flickr, do Flickr’s servers shoulder the burden, or is this equivalent to uploading from the desktop?)
Number of photos is up to the user. Probably doesn’t make sense to be overly redundant, but you never know when one particular angle might be just what an identifier needed for a positive ID. Personally I try to be pretty selective and only include my sharpest images of whatever features I captured.
iNaturalist resizes all photos uploaded (including from Flickr) to a maximum of 2048 pixels in the long dimension. So there is no need to worry about “over-resolution” photos, except to the extent they might slow down the uploading process while iNat takes the time to resize the big ones. But final storage size is controlled by iNat. And yes, iNat stores its own copies of Flickr photos that get uploaded.
I personally err on the side of adding too many pictures, because you never know what they’ll be used for. Maybe they’re useful for the AI so that it can identify things in bad images, or so that people can identify things that have been run over, or are disfigured in some way, etc. In addition, we don’t know how they’ll be useful in the future. Maybe right now species X is considered to be a very clear cut one with a very specific phenotype, but maybe in the future it’ll evolve to look a bit different (whether through adaptation, drift, or hybridization). Or maybe another variant/mutation/subspecies/form will be discovered and either the AI or people will be able to go back and re-classify some of the observations based on the extra pictures you took. There are lots of unknowns, so I say just go ahead and add as many potentially useful pictures as is reasonable.
I notice some users will post several photos (often from a distance and uncropped) in a record that are all of the exact same angle. That’s probably due to inexperience. Certainly a series of photos taken at different angles and showing different features can be useful, especially if you’re unsure what characteristics might be important for IDing.
Sometimes for insects the depth of field is an issue. Two photos that look identical can actually have very different focal points, and consequently show different features in focus.
There are two ways of looking at it… the more photos the better, because there may be something in one that helps with the ID, but then again too many photos and you make it too much work for the identifier, who then needs to sift through all the photos to make sure you haven’t put up photos of different individuals, or even taxa (happens a lot!)
I tend to put up photos of what I believe will be useful for ID… unusual characters (specific ones if I know them), standard views where possible (dorsal, ventral, eyes… bark, leaves, buds) and habitus. I take lots and don’t put them all up, and if an IDer asks for certain views I can check to see if I have them. Likewise, If I think a certain view might help me ID, I will ask observer if they have it.