Twice in recent weeks I’ve seen an uncommon bird species and haven’t had my camera with me - a flock of 12 Goosanders (uncommon round me, anyway) and a Red-legged partridge (ditto).
I should make a habit of carrying a camera more often.
Mind you, I also found a Firecrest but even though I had my camera, I didn’t try and photograph it because it was very actively flitting in the foliage of a Holly tree. I decided to enjoy watching it rather than try for a photo which I was most unlikely to get.
1 Like
I try to enjoy what I’m seeing, too, Ken - even with a camera in hand sometimes.
1 Like
Just wondering when people will stop taking pictures altogether.
Seems that Adobe stock, one of the huge libraries of images that publishers use, now take vastly more AI generated images than actual photos. 29million of them per month and photographers who send in real pictures are seeing their rejection rates suddenly become very high because the library already has an image like that (from an AI?).
How long before you just think about what you have just seen and the image appears without all that inconvenience of actually carrying a camera.
Is civilisation going backwards or forwards!
Having AI generated images of species which are rapidly disappearing is small consolation.
1 Like
I guess people may assume that AI generated images are copyright free… but we know AI gets the data to construct them (and AI text too, of course) by trawling the webthing without any regard for copyright. As Ken says, is this progess or regress?
So I will continue to take photos, identify and share the information for the joy of doing so; my own progress is what brings me pleasure.
1 Like
My opinion is that IPR law doesn’t properly account for AI generated material.
One of the principles of copyright law (specifically in America, but I expect that applies elsewhere) is the only something produced a human is copyrightable. (There was a case where copyright was denied on a photograph taken by a monkey.)
One argument is that AI images are therefore not copyrightable because they are produced by a computer. I don’t think that is clear cut. At the one end just sticking a prompt such as majestic lion into an image generator has negligible creative input. But at the other hand people would do hundreds of rounds of inpainting (getting the computer to rerender part of the image) in order to get a result that achieved their intention. Or in a different direction generating 1,000 images of botanical illustration of a daffodil, and picking out the best.
One could just make up a rule that AI generated images are not copyrightable, but otherwise you have to problems if you try to define domain independent principles. On the one hand your rules might also exclude collage or found art or even photography and CGI; on the other hand if your rules include some AI generated images but not others the uncertainty over what is covered is not good.
There are also arguments that AI generated images are under copyright as derivative works of either the software (I don’t think these are taken seriously) or of the training materials. The latter depends on the scope of fair use. The pro-AI argument is that human artists can study and copy another artist’s style; why is it any different if a software process does the same.
I would draw a distinction between training a system on trawled images for research into generative AI, and for use in a commercial image generating service. There’s a stronger argument for fair use for the former than for the latter.
I believe that there are now copyright clean models, trained on public domain and Creative Commons resources, but they’re not the best available models (by far).
Property is a social construct - for example what rights land ownership gives you depends on the jurisdiction. This applies especially to intellectual property. Intellectual property rights were invented to promote the common good - to maximise societal benefit by balancing the returns to creators, intermediaries and consumers. My opinion is that generative art may be a sufficiently different domain that we have to return to this basic principle.
There’s another issue which may be more pressing. AI does not generate accurate representations of real world objects. For example, I looked at a sample of Adobe Stock’s AI images of Hibiscus flowers. Lots of them have the wrong number of petals. Nearly all have the wrong number of stigmas or have anthers and stamens indistinguishable. If it does get these right it still might have less obvious details like venation or indumenta wrong.
1 Like
I mostly take photos for my own interest. That I share some with iSpor users is the Educator in me but photography is often just for me, now that the grandchildren are adults. I may delete 85% of my daily photos and sometimes the whole day’s run.
I do sometimes, much more often lately, take photos specially for iSpot projects.
I have always been more interested in rarities and so go no-where without my camera in a pocket, just in case I find one. I often collect things to photograph at home in controlled lighting.
Long ago I realised that bird photography required extra commitment, though I do have the gear.
Photography is an art form, I get a lot or personal pleasure from it. I do have more gear than I need and almost NEVER use my phone-cam.
Bird photography requires a lot of patience and skill. I just dabble. In my view, you either do bird watching or bird photography. I like to observe and keep records. Photography is an occasional way of illustrating that. I don’t have all the equipment that serious photographers need. And there’s a limit to how much stuff I can carry! (I follow a guy on Youtube who takes all his camera gear around the world - almost more than I could lift, yet alone carry in tropical conditions. He’s a taxi driver and goes under the handle stormcabbirds, if anyone wants to check it out. (See https://www.youtube.com/watch?v=dAjEekCUtQg)
With invertebrates, I am often struggling to make and ID so I often take photos with a view to being able to clinch the ID.