My main gripe with the current state of fully or mostly generated AI images is that they fall into the uncanny valley or look generic as dirt. Even the better stuff like Midjourney. Stable diffusion often looks like an attempt at doctoring an image by someone who is bad at photoshop. Gen AI has this theme of trying to look real but failing.
And that distracts quite strongly from the actual content. So in my view, gen AI is not an appropriate tool for the task at hand because the results are simply not good enough.
Copying a cartoon character into an actual photo is quite a different thing, because it is obviously symbolic. Just like a drawing of stick figures would be.
Its not just a problem with images, generated text has similar issues.
Jan. 28, 2025, 9:15 a.m. - Timer
My main gripe with the current state of fully or mostly generated AI images is that they fall into the uncanny valley or look generic as dirt. Even the better stuff like Midjourney. Stable diffusion often looks like an attempt at doctoring an image by someone who is bad at photoshop. Gen AI has this theme of trying to look real but failing. And that distracts quite strongly from the actual content. So in my view, gen AI is not an appropriate tool for the task at hand because the results are simply not good enough. Copying a cartoon character into an actual photo is quite a different thing, because it is obviously symbolic. Just like a drawing of stick figures would be. Its not just a problem with images, generated text has similar issues.