Browse Library

DALL-E Prompts, Tools & Guides

OpenAI's image generator, built into ChatGPT. Works best as one natural-language sentence, not comma-separated tags. Notably better than most at legible in-image text, though still limited to a few words. ChatGPT sometimes rewrites your prompt before generating -- ask it not to if exact wording matters.

FAQ

Should I write DALL-E prompts like Midjourney's tag-and-parameter style?

No -- DALL-E responds better to a single natural-language sentence describing the image, closer to how you'd describe it to a person, rather than a comma-separated list of tags.

Why doesn't my result match my exact prompt wording?

ChatGPT sometimes expands or rewrites a prompt before generating the image. Explicitly asking it to use your exact wording without embellishing it keeps the result closer to what you actually typed.

Is DALL-E good for images that need readable text in them?

Yes, relative to most competitors -- but keep any requested text short. It handles a few words reliably; a full sentence in-image often comes out distorted.

Can DALL-E place multiple objects in a precise spatial arrangement?

It handles this more loosely than a detailed spatial description might suggest -- for compositions where exact positioning matters, expect to regenerate or adjust rather than getting it exactly right on the first attempt.