Diluted Dante wrote:

Some tips for people on Mage. You get 77 'tokens' per use. It ignores anything after that. You can check how many tokens you're using with this: https://beta.openai.com/tokenizer

A helpful rule of thumb is that one token generally corresponds to ~4 characters of text for common English text. This translates to roughly ¾ of a word (so 100 tokens ~= 75 words).

Punctuation uses a token, and is often unnecessary. You can use negative prompts to tell it what you don't want. I generally use the below:

Low quality, ugly, mutated, mutilated, unrealistic, duplicate, multiple limbs, crossed arms, text, cutoff, no face, deformed

Full sentences are not needed, and can sometimes confuse the ai as to what you want. Things are weighted towards the start of your prompt, so if something isn't appearing or is not as prominent as you are going for, try re-ordering your prompt. The AI was trained on square images. I'd generally avoid using cinema or phone. If you want a rectangular image, landscape and tablet will generally work better. You can use the same prompt to get different images, as long as you aren't typing in a specific seed. If you want to tweak an image, you can type in the seed of the first one, then change your prompt. This can keep a consistent style, and result in much smaller changes than simply searching the same prompt again.