Yea, he could have chat GPT remove the filler words, and anything not a noun, or descriptive word. Would avoid SD using up tokens for filler words.

@Feral thats bs, if its a filler word meaning something sd cant recognize its not using up tokens, everything producing a token is something understood, eventho not every token seems to have an impact in every given situation. so theres nothing wrong if he wants to describe a scene in human language in most cases depending on the model as long as he doesnt accidently use words producing tokens that have a negative effect on what hes trying to achieve. also most prompts i see here have dozens of words not recognized by the model used, or words that produce actual tokens but dont have any effect on the scene, people just see what others used and copy that, because of course nobody can know all possible words and their effect from their head and most dont take the time to actually test the exact effect of each word. if this is not a lesson in learning/teaching prompt engineering do whatever you think you have to if it achieves a positive outcome

@Bombalurina I've never seen prompt descriptions like these before. I felt it was super cool regardless of how it interprets it.