Sharing things with you guys can be a big challenge sometimes in these current times.
I always want to bring you the greatest and most exciting photos to show you the most beautiful bits of nature I can find. But accusations from machines or people can really start to distort what is real.
I’ve shared before some really bad examples of AI owls, and that got a somewhat mixed reaction. Some likes me pointing odd really bad things being presented as real, but some didn’t like seeing them at all.
But today, let’s look at a pair of images I came across that look legitimate, one tagged, likely by AI, as being AI generated, and one that is rather fantastic that people are claiming is AI.
I’m inclined to think both are real, but as many owls as I’ve seen up close, I’m what is call 90+ percent confident, but I don’t know everything there is to know, especially once photos start getting processed, since I know nothing about photography or post processing.
I’ll present both cases, and I’m curious of your thoughts. Many of you are photographers, and while you guys may not look as closely at these pics as I often do, you’re all different sets of eyes and many of you have looked at hundreds, if not thousands, of these photos I’ve shared with you.
From Aniruddha Roy
One posing, one judging, and one pretending not to be involved.
Nature’s cutest surveillance team, reporting for duty.
Spotted Owlet (কোটর ে প্যাঁচা) Canon Eos R10
Facebook has this flagged as AI, which I find rich considering how many obvious AI owl images I see without such a tag.
He states the tag is a mistake, and while I don’t know if I shared his content before, looking through his feed shows a ton of wildlife photos that all look legit and in a style that seems consistent. He has other owl photos I’d say I have no doubts about and are species I’m familiar enough with I feel confident in being able to detect real vs fake photos of them.
On to case #2, the curious case of species that should not get along are getting along.
From time to time, I will see photos of animals of different species being posed together where they should be incompatible with each other. In “posed” images, sometimes these can be captive individuals used to some weird circumstances (the lady with the little owl and the German Shepard are an often seen example) or ones that I feel are more old fashioned edits where images are composited like this:
But what if we have something much more realistic but still fairly improbable, though certainly not impossible?
From Eric Tobias
Well this happened today. Caught me completely off guard and was super excited when I saw it. A Barn Owl and a Great Horned Owl.
And for all the doubters out there, this is absolutely not Al generated.
Normally, a Great Horned Owl would not tolerate anyone, especially another raptor, all up in its business, but here we seem to be seeing it.
Neither one looks bothered by the other, just interested in the photographer. That would seem a very stressful situation for wild owls, another strange predator, and a human all at once. Maybe they didn’t like it, and that’s why we only get a single pic.
He did provide what looks to be the uncropped image,and the owl details are good in the photo, and there are no weird looking patterns in the material of the barn either, to my eyes anyway.
Many in the comments were quick to judge negatively, saying the GHO would kill the Barn Owl, which would many times be true. But nature is weird in a lot of ways. If food is abundant, territoriality can go down significantly.
I didn’t see other owls in his Facebook feed, mostly landscapes, but his standalone website features a bunch of very legit looking owl photos, including some others in barns that may even be the same barn.
So what do you think? I feel both cases are legit. The Spotted owlets look to just be a bit heavily processed, especially in saturation, and the GHO/Barn is just a really uncommon scene, but otherwise I don’t see anything wrong with it.
I originally was not going to share either with you guys, as I’d rather skip something than present something false to you as something true. But I kept thinking about how I really liked the scene of the owlets and that it would suck to capture such a rare image as the GHO/Barnie and have so many people calling you a liar. So I just wanted to see what you guys think.
Even if you have no idea if they’re real or not, do you think I should skip things I’m even the slightest bit skeptical about? I obviously can’t know every animal or all their behaviors, but how cautious do you feel I should be as the one presenting things to you?
If you do think either image is fake, what do you spot that I don’t? I’ve tried some AI detectors before on photos I took myself and it called them AI generated, so I’m not sold on that as a solution.
I think since you pretty much run this community, it’s no big deal to make an honest mistake every now and then. If you’re just a little skeptical of some image, you could just add a tag to the post.
Thank you for the opinion. I’ve always tried to make it clear that I’m just someone very interested in owls, and I wouldn’t call myself an expert. This has lead me to working with owls and other animals in a hands-on way, but every week I volunteer there, I still see things I’d never thought I’d see, so while it’s made me more confident in spotting some things, it’s also opened me up to seeing potential truth in less regular scenarios.
Many questionable things I come across are of the “this looks very post processed, but it’s ordinary enough of a photo I can easily find a better one of the same thing” but these 2 were both very unique. Usually, it is bad image generation or editing that grabs attention, but both of these made me exceptionally sympathetic to the photographers, having their reputations questioned due solely to them being in what would otherwise be extremely lucky opportunities, but it now brings the taint of AI accusation.
As Lemmy has what I feel is an extreme but not so different than my own feeling about AI content, I wanted to see if I could get some consensus about what margin of error was acceptable. I don’t necessarily post things for your likes, but I also don’t want to waste your time. A lot of less popular owls like Great Grays or Barred Owls don’t seem to perform well here, and I still like them and appreciate them on my own, but will share something else here that I think you will like more. If you guys tell me you only want to share things I have no doubts about, then that is what I’ll do. If you don’t want to miss something like these 2 examples where I’m almost fully sure, but skeptical, I want to know that and how you’d like that presented. If you trust me enough to make the decision, I’m good with that. A disclaimer like you suggest is good, but I can see how that may also detract for some other people, as in if I’m not confident, why should you be, so some may be good trusting me and not want me to toss in a question mark on the validity.
End of the day, I want to do right by you all as a group because I am here to consume your limited free time. I learn about owls because I want to, and try to hook you on them as well. I take the time to share my research and learning with you that want to see it, so you don’t have to dig through all the crap like I do to get to the good stuff. I want you to fall in love with owls, and all animals, and hopefully get active in their protection, so I don’t want to do anything to alienate you when possible.
Again, I appreciate your time taken to comment. I always like hearing your opinions on the content.
i think both are legit. the first one is maybe mildly suspect but the visual context and overall visuals check out. the second one is very odd because of the circumstances but just like you i see no signs of AI.
AI detectors are almost always bullshit from what i’ve heard, a.k.a. they just use another ai model to look at the input image and return whatever the model’s verdict is. i think some imagegen models are starting to introduce some like embedded watermarks or something along those lines which could lead to the development of actual reliable detectors, but the false positive and false negative rates are allegedly still high.
it’s honestly getting scary how indistinguishable from reality some AI generated images can get. lately i’ve been having more and more cases where I got duped by an image that looked genuine but then turned out to be generated. they don’t immediately strike you with that “AI aura” like they did previously. i’m honestly losing my grip on what’s real and what’s not anymore.
as for AI images in this comm i’ll repeat what another commenter said. we’re human and we make mistakes, so don’t stress about it. if a seemingly genuine image turns out to be ai-assisted, then i’d add a visible note to the post, and if it turns out to be ai-generated i’d either remove the post or maybe rather leave it up and turn it into an example on what to look out for when verifying an image
they don’t immediately strike you with that “AI aura” like they did previously
I’m unsure if it’s the tech getting better, people getting better at writing prompts, or both, but I agree it is getting harder to distinguish when it’s a subject someone isn’t familiar with. Even some video clips I’ve seen have been better at not getting strange artifacts. Since I’ve examined enough photos and real birds to be familiar with the pattern in plumage, I can still pick out a lot that way, but that it’s coming down to that level of detail is concerning. I feel it would be really disheartening to fall in love with an “animal” and maybe build some real concern over it, to find out later it was fake and there is no such creature and they’d get discouraged from learning about other animals.
I think that is why I take this to what a lot of you see as somewhat of a ridiculously high standard. My goal is to get you guys to learn more about your local wildlife and to get encouraged to become active in its preservation like I have. I don’t want any of you to end up with a negative connection to animals or conservation. It’s important to me for you all to see these as real and individual animals that aren’t just NPCs and they have unique lives and that want to survive and have happy and safe families just like we do. Fake images and stories take away from the animals’ fight for existence and the work of the people that spend hours photographing, studying, and caring for them.
I’m glad I’ve earned enough grace that you guys all seem supportive and forgiving if I mess up, but I still don’t want to test it. A while back everyone decided against AI posts here, so they will probably be deleted, and it sounds so far like most of you will be happy if I say I’m pretty confident in something, but maybe not, which I appreciate. I thought a good bit how to present these photos in a way that didn’t seem like I was advocating slop or coming off as 100% in support of the images, and it seems I did so adequately.
Thanks for your opinions!
I think you’re doing a great job sharing owl content on a regular basis. I don’t personally think about real vs AI unless it’s obviously a bad AI
I think it’s just a bit of the modern day
I personally don’t like generative AI but I’m not sure how we can just avoid it in this world.
I think ethics is the best approach and ai products should be labeled as such. The deception is what really bothers me
I appreciate the support!
I keep a critical eye, as I’m primarily doing my research to teach myself.
I’m somewhat ambivalent about generative images. If people like the style, that’s their prerogative, I think they all kinda look the same and that takes a lot of the fun out of art to me. I do hate the waste of resources to produce them though, and I especially hate when they are almost decent enough to pawn off as real images. If there was a guaranteed way to label or detect them, I’d be more ok with it. That’s what irked me about these images. Overly aggressive humans calling out images without proof, or especially Meta’s AI calling out probable real images as AI both really bothered me to the point I made a huge post about it. I feel both photographers are being mildly punished for doing too good a job, which feels really wrong.
I am in total agreement
the cost of AI is an abomination and people that just use it without care for what it costs makes me very upset
I still get upset when I have to run Deepl to get a translation for a good post in a more difficult language for the built in Meta or Google translations to give me what I feel is the intended meaning. I don’t know if that’s the same kind of AI stuff or not, but being a standalone service, it seems, I dunno, more AI. But it does seem to deliver better results to give us an accurate translation. I’ve subconsciously stopped the mild use of generated images for jokes some months back as there was some pretty big backlash at other users sharing generated comics, so occasional translation is really my only knowing usage at this point.
translation work is indeed an amazing thing
To me, every photo here looks real except for the one with absolutely perfect lighting and a whole bunch of predators standing next to each other.
I can accept two predators next to each other. I deal with that all the time. But when you get six different ones from six different species and they’re not known to have undergone any domestication and you combine that with perfect lighting reality falls apart.
i think that’s just an edited photo.
Yes, the arms full of owls is a composite image. That’s how I make the community banner image with the 4 Owl of the Year finalists. I at least take the extra moment to add drop shadows to the layers so it looks like a little more effort was put into it. If you zoom in, you can see there’s no shadows and the edges are just very clean and neat like if you cut images out of a magazine and pasted them on a photo. It’s nice as an estimation of size, though I think Saw Whet is a little large. I’d like it for that if it was ever labeled as fake for demonstration purposes.
