I agree for anatomy. The many possible subtle errors are funny, e.g., an extra division of the abs muscles below the navel. And we humans are excellent at recognizing anatomy errors subconsciously (a result of sexual selection). Something feels “off” with the image. However, as I said in my post, special effects are different. Light effects, mist, aurorae, paint splatter, window rain splashes, etc., are high-entropy and under-constrained. So I don't see errors. Example: I drew the dolphins by hand and led the AI to create a graffiti from it. This image isn't great artistically, sure, but I don't think you can point to a hallucination. Of course, contrary to software engineering, in the art world, using AI is stigmatized. It would be interesting, though, for what it would be used for if the stigma was not there. Most professional art needs: • consistency • compositionality • correctness of things with heavy global constraints (e.g., anatomy) Which AI struggles with. But for rendering effects, those issues do not appear, and it could save a lot of time.
