somewhat of a faulty lora, appearing concept such as speech bubble, multiple views, holding mask or condom, by default without using negative prompt

it turn out the booru tagbase and its child model wd14 family auto tagger didn't possess the required corresponding capability to properly caption some images coming from feipin zhanshi, so concept like crowd & mascot in the foreground & background, weird objects, speech bubbles, got baked into the lora