You should try adding attention to tokens. If you highlight the words you want to add attention to and press Ctrl + Arrow key up or down, you can change the amount of attention on those tokens and get some interesting effects. Past 1.5 attention things get a little weird, but that might be what you're looking for.
On Automatic1111 you can use the prompt editing feature to do this kind of thing. [from:to:when]. So [hair:leaves:0.3] will swap generating hair to leaves at 30% through the generation. Or you could use Alternating Words[hair|leaves] to swap generation every step.
Posting your generation parameters helps everyone. There's a lot we can learn from each other.
You shouldn't put too much stock in these detection tools. Not only do they not work, they flag non-native English speakers for cheating more than native speakers.
You should try adding attention to tokens. If you highlight the words you want to add attention to and press Ctrl + Arrow key up or down, you can change the amount of attention on those tokens and get some interesting effects. Past 1.5 attention things get a little weird, but that might be what you're looking for.