For now, resolution is still a bit of a giveaway. High resolution requires AI skills, to not have it introduce phantom elements, low resolution is easy to fake with AI. You can often tell by the obvious fake shadows. This image here is too low resolution, might as well be real.
I’ve seen 4k images people made using stable diffusion XL models. They are apparently trained from higher resolution source images. Giving better resolution and less artifacting.
Its becoming easier and easier to fake things. All it needs is a decent gpu and an open source program.
Been fooled before by seemingly real images with near perfect lighting and high resolution. Its the noise patterns you need to look for.
Almost had me thinking AI. But I could be wrong.
Then again, AI has gotten so good that, with a little effort you can make something this convincing.
For now, resolution is still a bit of a giveaway. High resolution requires AI skills, to not have it introduce phantom elements, low resolution is easy to fake with AI. You can often tell by the obvious fake shadows. This image here is too low resolution, might as well be real.
I’ve seen 4k images people made using stable diffusion XL models. They are apparently trained from higher resolution source images. Giving better resolution and less artifacting.
Its becoming easier and easier to fake things. All it needs is a decent gpu and an open source program.
Been fooled before by seemingly real images with near perfect lighting and high resolution. Its the noise patterns you need to look for.
I don’t think IA is the answer, this picture is at least 10 years old, going around in social networks.