Why I Still Don't Reach for AI Image Generation
I've tried it a few times, and the results are pretty, but they all kind of feel the same.
And constantly tweaking the prompts feels like it just makes more work, so I end up drawing it myself again lol.
5 answers
True, prompt grunt work has become harder than my actual job lol
I relate to that same-ish feeling. Especially when you only use one model. It gets a bit better if you swap in style LoRAs or mix checkpoints, but then you have to study the setup again... In the end, for hobbyists, just drawing by hand seems faster.
Well, I'm not so sure about that. You're only cherry-picking the ones that come out pretty, and isn't the purpose different from drawing it yourself? It's still unmatched for speed or at the concept-ideation stage, but "not reaching for it" might just be part of getting used to it.
I was the same at first. But once I dropped the habit of writing long, sentence-like prompts and switched to a few keywords plus mostly negative prompts, it became much easier. And instead of trying to get it perfect in one go, I changed my mindset to generating 30 images and using one of them, which drastically reduced my stress.
Got a source? I'd be curious if you've ever seen data showing they're all about the same.