Investigating Prompt Engineering in Diffusion Models

Abstract

With the spread of the use of Text2Img diffusion models such as DALL-E 2, Imagen, Mid Journey and Stable Diffusion, one challenge that artists face is selecting the right prompts to achieve the desired artistic output. We present techniques for measuring the effect that specific words and phrases in prompts have, and (in the Appendix) present guidance on the selection of prompts to produce desired effects.

Many illustrations in the Appendix!

Link to Paper

• The paper's arXiv page

And a suitable BiBTeX entry for the paper :

@misc{witteveen2022investigating,
      title={Investigating Prompt Engineering in Diffusion Models}, 
      author={Sam Witteveen and Martin Andrews},
      year={2022},
      eprint={2211.15462},
      archivePrefix={arXiv},
      primaryClass={cs.CV}
}