Some tags work better than others, results get more random the more tags you add in it seems

• Tags for types of laughs (for V2 add "text" after the chosen types):

"hahaha" "nohoho" "ohoho" "stohohop" "hehehe" "heehee" "heh heh" "huhuhu" "nyehehe" "bahaha" "bwahaha" "keke" "kyahaha" "mwahaha" "gahaha" "aaahahaha" "pffhahaha" "whahaha"

In V2 there's also "ha ha ha text" and "he he he text" which are spaced out versions of their counterparts

• Types of speech bubbles:

"speech bubble" (Replaced by "white speech bubble" in V2) "black speech bubble" "huge speech bubble" "pointy speech bubble" "conjoined speech bubble"

• Types of text:

"outlined text" "white/black/red/orange/green/blue text" "bold text" "stylized text" "thin text" "big text" "slanted text" "pixel text" "underlined text"

V2 adds "small text" and "simple text"

• Locations of text:

V1: "text in background" "hahaha background" "hehehe background"

V2: "text everywhere" "text next to body"

Both: "text behind head" "text above head" "text next to head" "text above body"

Putting "two speech bubbles, dialogue, talking" in the negative prompt seems to help