Rainbow Dash, cup is so much less tiresome.
It’s less expressive. It’s why people keep adding convoluted tags like the aforementioned Skirt on Pinkie Pie ONLY
. They wouldn’t if they didn’t care about the distinction. The official answer is to create a gallery instead, but that requires each concerned user to sift through huge amounts of images to find the applicable ones and to put them in their own gallery. And that effort doesn’t even get reused, since in practice people looking for specific images use regular image searches, not gallery searches. With structured tags this effort could at least get reused.
No thanks, sick of AI in everything.
I’m sick of some AI, like “smart” stuff (because usually it’s just as stupid but less predictable) or LLM-based “assistants” that will very eloquently tell you that they can’t help you. Though I honestly believe AI has its uses (such as offline translations) and automatically tagging images could be one of them. It seems way more useful than image generation to me.
Another possible use case is providing accessible interfaces. Some people don’t even know they can do rd, cup
, let alone foo && (bar || baz) && NOT quux
. Empowering them to use complex searches is a good thing, even if other users would write their own queries instead.